SOLAR RESOURCE MAPPING IN ZAMBIA 24 MONTH SITE RESOURCE REPORT August 2018 This report was prepared by Solargis, under contract to The World Bank. It is one of several outputs from the solar resource mapping component of the activity “Renewable Energy Resource Mapping and Geospatial Planning – Zambia” [Project ID: P145271]. This activity is funded and supported by the Energy Sector Management Assistance Program (ESMAP), a multi-donor trust fund administered by The World Bank, under a global initiative on Renewable Energy Resource Mapping. Further details on the initiative can be obtained from the ESMAP website. This document is a final output from the above-mentioned project, and the content is the sole responsibility of the consultant authors. Users are strongly advised to exercise caution when utilizing the information and data contained, as this may include preliminary data and/or findings, and the document has not been subject to full peer review. Final outputs from this project will be marked as such, and any improved or validated solar resource data will be incorporated into the Global Solar Atlas. Copyright © 2018 THE WORLD BANK Washington DC 20433 Telephone: +1-202-473-1000 Internet: www.worldbank.org The World Bank, comprising the International Bank for Reconstruction and Development (IBRD) and the International Development Association (IDA), is the commissioning agent and copyright holder for this publication. However, this work is a product of the consultants listed, and not of World Bank staff. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work and accept no responsibility for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for non-commercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: +1-202-522-2625; e-mail: pubrights@worldbank.org. Furthermore, the ESMAP Program Manager would appreciate receiving a copy of the publication that uses this publication for its source sent in care of the address above, or to esmap@worldbank.org. Annual Solar Resource Report Six solar meteorological stations after completion of ground measuring campaign (24 months) Republic of Zambia Solargis reference 128-07/2017 Date: 15 August 2018 Customer Consultant World Bank Solargis s.r.o. Energy Sector Management Assistance Program Contact: Mr. Marcel Suri Contact: Mr. Oliver Knight Mytna 48, 811 07 Bratislava, Slovakia 1818 H St NW, Washington DC, 20433, USA Phone +421 2 4319 1708 Phone: +1-202-473-3159 E-mail: marcel.suri@solargis.com E-mail: oknight@worldbank.org http://solargis.com http://www.esmap.org/RE_Mapping Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 TABLE OF CONTENTS Table of contents ............................................................................................................................................ 4 Acronyms........................................................................................................................................................ 6 Glossary.......................................................................................................................................................... 7 Executive summary ....................................................................................................................................... 8 1 Introduction ........................................................................................................................................... 9 1.1 Background.................................................................................................................................. 9 1.1 Delivered data sets ....................................................................................................................... 9 1.2 Information included in this report............................................................................................... 10 2 Position of solar meteorological sites ................................................................................................... 11 3 Ground measurements in Zambia .......................................................................................................... 13 3.1 Instruments and measured parameters ....................................................................................... 13 3.2 Station operation and calibration of instruments ......................................................................... 14 3.3 Quality control of measured solar resource data .......................................................................... 17 3.3.1 University of Zambia (UNZA) Lusaka .............................................................................. 18 3.3.2 Mount Makulu (Chilanga) ............................................................................................... 22 3.3.3 Mochipapa (Choma) ...................................................................................................... 26 3.3.4 Longe (Kaoma).............................................................................................................. 30 3.3.5 Misamfu (Kasama) ........................................................................................................ 33 3.3.6 Mutanda........................................................................................................................ 36 3.4 Recommendations on the operation and maintenance................................................................. 39 4 Solar resource model data .................................................................................................................... 40 4.1 Solar model................................................................................................................................ 40 4.2 Site adaptation of the solar model − method ............................................................................... 41 4.3 Results of the model adaptation at six sites ................................................................................ 43 5 Meteorological model data.................................................................................................................... 52 5.1 Meteorological model ................................................................................................................. 52 5.2 Validation of meteorological data ............................................................................................... 52 5.2.1 Air temperature at 2 metres............................................................................................ 53 5.2.2 Relative humidity ........................................................................................................... 56 5.2.3 Wind speed and wind direction at 10 metres ................................................................... 60 5.3 Uncertainty of meteorological model data .................................................................................... 63 6 Solar resource: uncertainty of long-term estimates ............................................................................... 64 6.1 Uncertainty of solar resource yearly estimate .............................................................................. 64 6.2 Uncertainty due to interannual variability of solar radiation........................................................... 65 6.3 Combined uncertainty................................................................................................................. 67 7 Time series and Typical Meteorological Year data ................................................................................. 70 7.1 Delivered data sets ..................................................................................................................... 70 7.2 TMY method .............................................................................................................................. 71 7.3 Results....................................................................................................................................... 71 8 Conclusions .......................................................................................................................................... 76 Annex 1: Site related data statistics ............................................................................................................... 77 Yearly summaries of solar and meteorological parameters ..................................................................... 77 © 2018 Solargis page 4 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Monthly summaries of solar and meteorological parameters .................................................................. 79 Frequency of occurrence of GHI and DNI daily model values for a period 1994 to 2017 ............................ 82 Frequency of occurrence of GHI and DNI 15-minute model values for a period 1994 to 2017 .................... 88 List of figures .............................................................................................................................................. 103 List of tables ............................................................................................................................................... 106 References .................................................................................................................................................. 108 Support information .................................................................................................................................... 110 Background on Solargis ....................................................................................................................... 110 Legal information ................................................................................................................................ 110 © 2018 Solargis page 5 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 ACRONYMS AP Atmospheric Pressure CFSR Climate Forecast System Reanalysis. The meteorological model operated by the US service NOAA (National Oceanic and Atmospheric Administration) CFSv2 Climate Forecast System Version 2 CFSv2 model is the operational extension of the CFSR (NOAA, NCEP) DIF Diffuse Horizontal Irradiation, if integrated solar energy is assumed. Diffuse Horizontal Irradiance, if solar power values are discussed DNI Direct Normal Irradiation, if integrated solar energy is assumed. Direct Normal Irradiance, if solar power values are discussed. GFS Global Forecast System. The meteorological model operated by the US service NOAA (National Oceanic and Atmospheric Administration) GHI Global Horizontal Irradiation, if integrated solar energy is assumed. Global Horizontal Irradiance, if solar power values are discussed. MACC Monitoring Atmospheric Composition and Climate – meteorological model operated by the European service ECMWF (European Centre for Medium-Range Weather Forecasts) Meteosat MFG and MSG Meteosat satellite operated by EUMETSAT organization. MSG: Meteosat Second Generation; MFG: Meteosat First Generation PWAT Precipitable water (water vapour) QC Quality control RH Relative Humidity at 2 metres TEMP Air Temperature at 2 metres WD Wind Direction at 10 metres WS Wind Speed at 10 metres © 2018 Solargis page 6 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 GLOSSARY Aerosols Small solid or liquid particles suspended in air, for example desert sand or soil particles, sea salts, burning biomass, pollen, industrial and traffic pollution. All-sky irradiance The amount of solar radiation reaching the Earth's surface is mainly determined by Earth- Sun geometry (the position of a point on the Earth's surface relative to the Sun which is determined by latitude, the time of year and the time of day) and the atmospheric conditions (the level of cloud cover and the optical transparency of atmosphere). All-sky irradiance is computed with all factors taken into account Bias Represents systematic deviation (over- or underestimation) and it is determined by systematic or seasonal issues in cloud identification algorithms, coarse resolution and regional imperfections of atmospheric data (aerosols, water vapour), terrain, sun position, satellite viewing angle, microclimate effects, high mountains, etc. Clear-sky The clear sky irradiance is calculated similarly to all-sky irradiance but without taking into irradiance account the impact of cloud cover. Long-term Average value of selected parameter (GHI, DNI, etc.) based on multiyear historical time average series. Long-term averages provide a basic overview of solar resource availability and its seasonal variability. P50 value Best estimate or median value represents 50% probability of exceedance. For annual and monthly solar irradiation summaries it is close to average, since multiyear distribution of solar radiation resembles normal distribution. P90 value Conservative estimate, assuming 90% probability of exceedance (with the 90% probability the value should be exceeded). When assuming normal distribution, the P90 value is also a lower boundary of the 80% probability of occurrence. P90 value can be calculated by subtracting uncertainty from the P50 value. In this report, we apply a simplified assumption of normal distribution of yearly values. Root Mean Square Represents spread of deviations given by random discrepancies between measured and Deviation (RMSD) modelled data and is calculated according to this formula: 6 * ∑7 (* +,-./0,1 − +31,4,1 5 = & *89 On the modelling side, this could be low accuracy of cloud estimate (e.g. intermediate clouds), under/over estimation of atmospheric input data, terrain, microclimate and other effects, which are not captured by the model. Part of this discrepancy is natural - as satellite monitors large area (of approx. 3.2 x 4.0 km for MSG satellite), while sensor sees only micro area of approx. 1 sq. centimetre. On the measurement side, the discrepancy may be determined by accuracy/quality and errors of the instrument, pollution of the detector, misalignment, data loggers, insufficient quality control, etc. Solar irradiance Solar power (instantaneous flow of energy) falling on a unit area per unit time [W/m2]. Solar resource or solar radiation is used when considering both irradiance and irradiation. Solar irradiation Amount of solar energy falling on a unit area over a stated time interval [Wh/m2 or kWh/m2]. Uncertainty Is a parameter characterizing the possible dispersion of the values attributed to an of estimate, Uest estimated irradiance/irradiation values. In this report, uncertainty assessment of the solar resource model estimate is based on a detailed understanding of the achievable accuracy of the solar radiation model and its data inputs (satellite, atmospheric and other data), which is confronted by an extensive data validation experience. The second source of uncertainty is ground measurements. Their quality depends on accuracy of instruments, their maintenance and data quality control. Third contribution to the uncertainty is from the site adaptation method where ground-measured and satellite-based data are correlated. © 2018 Solargis page 7 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 EXECUTIVE SUMMARY This report is prepared within Phase 2 of the project Renewable Energy Resource Mapping for the Republic of Zambia. The project objectives is to deliver high quality solar resource mapping and measurement services for renewable energy development implemented by the World Bank in Zambia. This report describes results of 24+ months of the measuring campaign at six solar meteorological stations, installed in Zambia. The report accompanies delivery of site-specific measurements and model data prepared for six sites, where meteorological measurement campaign has been conducted. These solar meteorological stations were installed and operated by GeoSUN Africa (South Africa) with their partner SGS Zambia, and commissioned by the World Bank over the years 2015 to 2017 under the same activity. The data quality control and further processing and integration to the solar models has been conducted by Solargis (Slovakia) under the same project. The 2-year campaign brought a unique set of solar resource measurements for a region in Africa, and climate zone, that have been mapped insufficiently so far. The data helps specialists to better understand solar resource availability and variability, as both are crucial for development of solar power plants and for their efficient integration into existing energy infrastructure. This campaign also contributes to better understanding of performance and uncertainty of solar measuring sensors in tropical conditions. The knowledge based on the analysis of measured and modelled data in this region improves confidence of engineers, designing solar power plants, and investors and banks, providing the financing. One of key benefits of having these type of measurements is that they can be used for improving the solar and meteorological models. We used data from six meteo sites for adaption of Solargis model to the regional climate, and this results in reduced uncertainty of the model outputs, see the summary table below. This way, the improved model is able to generate more accurate solar and meteorological historical data, which reduces uncertainties in technical and financial evaluation of any solar energy project in Zambia. This report describes technical parameters of the measurement campaign, features of the measured data, adaptation of models, summary statistics for the outputs at six meteorological sites, and relevant uncertainties. Summary table: Uncertainty of solar model estimates for original and site-adapted annual long-term values at 80% probability of occurrence Uncertainty of long-term Acronym Uncertainty of the original Uncertainty of the Solargis model after annual values Solargis model site adaptation based on solar measured data After 1st year After 2nd year Global Horizontal Irradiation GHI ±7.5% (up to ±10.0%*) ±4.5% ±4.0% Direct Normal Irradiance DNI ±12.0% (up to ±18.0%*) ±6.0% ±5.5% * in complex microclimate © 2018 Solargis page 8 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 1 INTRODUCTION 1.1 Background This report is prepared within Phase 2 of the project Renewable Energy Resource Mapping for the Republic of Zambia. This part of the project focuses on solar resource mapping and measurement services as part of a technical assistance in the renewable energy development implemented by the World Bank in Zambia. It is being undertaken in close coordination with the Department of Energy (DoE) of Zambia, the World Bank’s primary country counterpart for this project, and Zambia Meteorological Department (ZMD). This project is funded by the Energy Sector Management Assistance Program (ESMAP), administered by the World Bank and supported by bilateral donors. This report summarizes results of 24+ months of the measuring campaign at six solar meteorological stations, installed as part of the World Bank’s ESMAP mission in Zambia. The report describes delivery of site-specific measurements, and site-adaptation model data. Data uncertainty and summary statistics are also described in this report. This report accompanies delivery of site-specific solar resource and meteorological data for six sites, where solar meteorological stations have been in operation. High-quality measurements were used for the adaptation of the Solargis model for regional climate of Zambia. The model was run for the six sites and delivered high accuracy time series and Typical Meteorological Year. The improved model can be used for delivery of similar type of data for any location in Zambia, for the needs of technical and financial evaluation of solar energy projects. The measurements are provided by GeoSUN Africa company (South Africa). The model data for the same sites and related calculations, together with this report are supplied by Solargis company (Slovakia). 1.1 Delivered data sets The site-specific data, provided as part of this delivery, include: • Solar and meteorological ground measurements, after data quality assessment, 25+ months of data (11/2015 – 12/2017) • Time series of satellite-based model data, adapted for regional climate, representing last 24+ years (1994 to 2017) • Typical Meteorological Year data, representing 24 calendar years (1994 to 2017) The data is delivered in formats ready to use in solar energy simulation software. This report provides detailed insight into the methodologies and results. Table 1.1 Characteristics of the delivered data Feature Time coverage Primary time Delivered files step Measurements Nov 2015 to Dec 2017 1 minute Quality controlled measurements: 1- minute (GeoSUN Africa) time resolution Model data – original Jan 1994 to Dec 2017 15 minutes Time series: hourly, monthly and yearly time (Solargis) aggregation Model data – site adapted Jan 1994 to Dec 2017 15 minutes Time series: hourly, monthly and yearly time (Solargis) aggregation Model data – site adapted Jan 1994 to Dec 2017 hourly Typical Meteorological Year for P50 and P90 (Solargis) © 2018 Solargis page 9 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Table 1.2 Parameters in the delivered time series (TS) and TMY data (hourly time step) Parameter Acronym Unit TS TMY P50 TMY P90 2 Global horizontal irradiance GHI W/m X X X 2 Direct normal irradiance DNI W/m X X X 2 Diffuse horizontal irradiance DIF W/m X X X 2 Global tilted irradiance (at optimum angle) GTI W/m X - - Solar azimuth SA ° X X X Solar elevation SE ° X X X Air temperature at 2 metres TEMP °C X X X Wind speed at 10 metres WS m/s X X X Wind direction at 10 metres WD ° X X X Relative humidity RH % X X X Air Pressure AP hPa X X X 2 Precipitable Water PWAT kg/m X X X 1.2 Information included in this report This report presents: • Solar resource and meteorological measurements after 24 months of operation o Review and quality check of the measured data o Calibration procedures and results o List and explanation of the occurred disturbances and failures • Comparison of the measurements with the Solargis model; uncertainty analysis o Comparison of solar and meteo measurements with the model data o Site adaptation of satellite data based on ground measurements and uncertainty estimate o Estimate of model data uncertainty • Data analysis (measured vs. modelled) o Monthly summaries of solar and meteorological parameters captured at the site o Variability of measured solar parameters o Frequency of occurrence of GHI and DNI 1-minute and 15-minute values o Frequency of occurrence of GHI and DNI 1-minute and 15-minute ramps © 2018 Solargis page 10 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 2 POSITION OF SOLAR METEOROLOGICAL SITES In Zambia, six measuring stations were installed within the ESMAP Solar initiative. They have been located within the premises of Zambia Meteorological Department (ZMD), Zambia Agriculture Research Institute (ZARI) and School of Agricultural Sciences at University of Zambia (UNZA) (Figure 2.1, Table 2.1). Figure 2.1: Position of solar meteorological stations in Zambia Map of global horizontal irradiation in the background Table 2.1 Solar meteorological stations installed in Zambia: Overview Altitude Host of Site name Closest town Latitude [º] Longitude [º] [m a.s.l.] measurement station Lusaka UNZA Lusaka -15.39463° 28.33722° 1263 UNZA Mount Makulu Chilanga -15.54830° 28.24817° 1227 ZARI/ZMD Mochipapa Choma -16.83828° 27.07046° 1282 ZARI/ZMD Longe Kaoma -14.83900° 24.93100° 1169 ZARI Misamfu Kasama -10.17165° 31.22558° 1380 ZARI/ZMD Mutanda Mutanda -12.42300° 26.21500° 1316 ZARI/ZMD © 2018 Solargis page 11 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 The position of solar meteorological station is selected to achieve a representative geographical distribution within the territory of Zambia, as well as in proximity to the population centres, where solar energy installations will be mostly deployed. In addition to geographical and socio-economic criteria, the sites fulfil the criteria for the operation and maintenance of the solar measuring stations: • Existence of free horizon, • Availability of GSM networks, • Availability of local work force for maintenance, • Easy to access and high level of security During two years of the measurements, the measured data was analysed and harmonized with the objective to acquire reference solar radiation data for reducing the uncertainty of the model Chapter 6). The quality measurements from one Tier 1 and five Tier 2 meteorological stations is available for this assessment (Chapter 3). Position and detailed information about the measurement sites is available also in the Global Solar Atlas by the World Bank: https://globalsolaratlas.info/. © 2018 Solargis page 12 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 3 GROUND MEASUREMENTS IN ZAMBIA 3.1 Instruments and measured parameters Basic information about measurements sites is in Table 3.1. Solar parameters at stations are measured by high accuracy equipment (CMP 10 for GHI measurements and CHP 1 for DNI measurements) at Tier 1 station, and by CMP 10 and additional medium accuracy equipment (RSR, for GHI, DNI and DIF, Tables 3.2 to 3.4) at all other stations. The measurement campaign in Zambia has been performed by GeoSUN Africa company (South Africa). Table 3.1 Overview information on measurement stations operated in the region ID Site name Closest town Station type Installation date 1 Lusaka UNZA Lusaka Tier 1 7 November 2015 2 Mount Makulu Chilanga Tier 2 13 November 2015 3 Mochipapa Choma Tier 2 5 November 2015 4 Longe Kaoma Tier 2 10 November 2015 5 Misamfu Kasama Tier 2 18 November 2015 6 Mutanda Mutanda Tier 2 21 November 2015 Table 3.2 Solar instruments installed at the solar meteorological stations Site name GHI DIF DNI GHI 2 DIF 2 DNI 2 Lusaka UNZA CMP 10 CMP 10 CHP 1 RSR 2 RSR 2 RSR 2 Mount Makulu CMP 10 - - RSR 2 RSR 2 RSR 2 Mochipapa CMP 10 - - RSR 2 RSR 2 RSR 2 Longe CMP 10 - - RSR 2 RSR 2 RSR 2 Misamfu CMP 10 - - RSR 2 RSR 2 RSR 2 Mutanda CMP 10 - - RSR 2 RSR 2 RSR 2 Table 3.3 Meteorological instruments installed at the solar meteorological stations Site name WS WD TEMP RH AP PWAT Lusaka UNZA 05103 05103 CS215 CS215 PTB110 TE525 Mount Makulu 03002 03002 CS215 CS215 PTB110 TE525 Mochipapa 03002 03002 CS215 CS215 PTB110 TE525 Longe 03002 03002 CS215 CS215 PTB110 TE525 Misamfu 03002 03002 CS215 CS215 PTB110 TE525 Mutanda 03002 03002 CS215 CS215 PTB110 TE525 © 2018 Solargis page 13 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Table 3.4 Technical parameters and accuracy class of the instruments at Tier 1 and Tier 2 stations Parameter Instrument Type Manufacturer Uncertainty GHI Secondary standard pyranometer CMP 10 Kipp & Zonen < ±2 % (daily) DIF Tier 1 station Secondary standard pyranometer CMP 10 Kipp & Zonen < ±2 % (daily) DNI Tier 1 station Pyrheliometer CHP 1 Kipp & Zonen < 1 % (daily) GHI 2 Rotating Shadowband Radiometer with LI200 RSR 2 Irradiance Inc. Indicatively ±5 % DIF 2 Rotating Shadowband Radiometer with LI200 RSR 2 Irradiance Inc. Indicatively ±8 % DNI 2 Rotating Shadowband Radiometer with LI200 RSR 2 Irradiance Inc. Indicatively ±5 % WS Tier 1 station wind speed sensor (at 10 m) 05103 R.M. Young ±0.3 m/s Tier 2 station wind speed sensor (at 3 m) 03002 R.M. Young ±0.5 m/s WD Tier 1 station wind direction sensor (at 10 m) 05103 R.M. Young ±3 ° Tier 2 station wind direction sensor (at 3 m) 03002 R.M. Young ±5 ° TEMP Temperature probe (at 2 m) CS215 Campbell Scientific ±0.9°C RH Relative humidity probe CS215 Campbell Scientific ±4% RH AP Atmospheric pressure sensor PTB110 Vaisala ±1.5 hPa PWAT Tipping-bucket rain gage TE525 Texas Instrument ±1% ± (0.06% of reading + - Data logger CR1000 Campbell Scientific offset) 3.2 Station operation and calibration of instruments In this report, complete set of data from of 2-year measurement campaign is analysed. As the measurement stations have been installed during November 2015, the period for the data analysis starts in November 2015 and ends in November 2017. On some stations, where limit of 95% of high quality data availability was not fulfilled by the end 2-year campaign, the measurements continued until December 2017. Overview of the data availability, time step and measured parameters is shown in Tables 3.5 to 3.8. Table 3.5 Overview information on solar meteorological stations operating in the region Site name Closest town Measurement period Primary time step Lusaka UNZA Lusaka 7 November 2015 – 31 December 2017 1 minute Mount Makulu Chilanga 13 November 2015 – 31 December 2017 1 minute Mochipapa Choma 5 November 2015 – 31 December 2017 1 minute Longe Kaoma 10 November 2015 – 31 December 2017 1 minute Misamfu Kasama 18 November 2015 – 31 December 2017 1 minute Mutanda Mutanda 21 November 2015 – 31 December 2017 1 minute Table 3.6 show data recovery statistics for the whole measurement period for each station. In this statistics, only serious issues (missing data for a longer period, erroneous data - initial installation problem or shading ring problem) are accounted. Short-term operational issues (shading by surrounding objects, morning dew on the instrument, etc.) are not considered. The column Data loss represents amount of missing data or data excluded during quality control process. Percentage share is calculated from daytime values and days represent cumulative amount of missing data (one day may be composed from several shorter missing data periods). The column influenced days represents number of days with fully or partially missing data or days excluded by quality control process. The column Exceeding two years show count of days of measurements exceeding the two years period. All stations fulfilled criteria of 2-years availability of high quality ground measurements. Periods of missing or erroneous data were substituted by additional measurement days, beyond the originally assigned measurement campaign. © 2018 Solargis page 14 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Table 3.6 Data recovery statistics of the measurement campaign Zambia Data loss* Influenced days** Exceeding Acceptance criteria Measurement period Erroneous + Missing data Total Length of individual periods two years *** 95% 15+ days 11/2015-12/2017 [%] (days) Description (days) (days) (days) Lusaka 0 0 - 0 54 OK OK Mount Makulu 0.5 4 initial RSR instrument problem (mornings) 13 1, 1, 1, 10 48 OK OK Mochipapa 0 0 - 0 56 OK OK Longe 2.5 18 missing data 18 1, 8, 9 51 OK OK Misamfu 1.9 14 missing/NAN data, battery failure 14 14 43 OK OK Mutanda 0 0 - 0 40 OK OK Table 3.7 Period of measurements analysed in this report Year, month 2015 2016 2017 Station 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Lusaka Mount Makulu Mochipapa Longe Misamfu Mutanda During measurement campaign, the local staff of Zambia Meteorological Department, Zambia Agriculture Research Institute and School of Agricultural Sciences at University of Zambia was fully trained by GeoSUN and provided instruments inspection, monitoring and cleaning, with frequency of 1 to 5 days. All stations measured a rare phenomenon of dip in solar irradiance on 1 September 2016 - the solar eclipse. Table 3.8 Meteorological stations maintenance and instruments field verification Lusaka UNZA Comments and issues Station type Tier 1 • Occasional distortions of GHI and DNI values occurred for some mornings due to dew. Instruments cleaning interval Average: 2.1 [days] Longest: 12 Verification visits date by 6 Jun 2016 GeoSUN Africa 1 Nov 2016 23 Jun 2017 3 to 10 April 2018 Instruments field verification GHI – reference CMP 10 DIF – reference CMP10 DNI – reference CHP1 GHI 2 – reference Li200 Mount Makulu - Chilanga Comments and issues Station type Tier 2 • The station did not face North in November and December 2015. This affects the DIF and the DNI Instruments cleaning interval Average: 1.4 reading, especially in the mornings. [days] Longest: 8 • Atmospheric pressure data was lost in November 2016 due to sensor verification. Verification visits date by 6 Jun 2016 • The station is located between trees, which causes GeoSUN Africa 1 Nov 2016 minimum shading on the instruments during some 23 Jun 2017 mornings and afternoons. 3 to 10 April 2018 • Dew on the instruments cause inaccurate readings in some mornings. Instruments field verification GHI – reference CMP 10 GHI 2 – reference Li200 © 2018 Solargis page 15 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Mochipapa - Choma Comments and issues Station type Tier 2 • Shading from a tree in late afternoon. • Several distortions of GHI and DNI occurred in some Instruments cleaning interval Average: 2.2 mornings due to dew. [days] Longest: 9 • During February, March and April 2016, malfunction of TEMP and RH sensor occurred (CS215); thus datalogger Verification visits date by 27 Apr 2016 internal temperature sensor readings were used for GeoSUN Africa 2 Nov 2016 correction of RSR measurements (GHI, DNI, DIF). 24 Jun 2017 3 to 10 April 2018 Instruments field verification GHI – reference CMP 10 GHI 2 – reference Li200 Longe – Kaoma Comments and issues Station type Tier 2 • The station is located close to trees, which cause slight shading in the mornings and afternoons. Instruments cleaning interval Average: 4.3 • Several distortions of GHI and DNI values occurred on [days] Longest: 17 some mornings due to dew. • Short data period loss occurred in February and May Verification visits date by 7 Jun 2016 2016 due to lost connection with datalogger. GeoSUN Africa 4 Nov 2016 • Several longer periods without cleaning (9 to 17 days) 25 June 2017 3 to 10 April 2018 • Unknown shadow cast over the CMP 10 instrument on 6 June 2017 causing inaccurate measurements for a couple of minutes. Instruments field verification GHI – reference CMP 10 GHI 2 – reference Li200 Misamfu – Kasama Comments and issues Station type Tier 2 • Shading on the irradiance sensors every morning due to overhead utility power cables close to the station. Instruments cleaning interval Average: 2.1 • Several distortions of GHI and DNI values occurred on [days] Longest: 9 some mornings due to dew. • Incorrect readings from RSR and CMP10 in February Verification visits date by 7 Nov 2016 2016 due to faulty RSR and battery GeoSUN Africa 29 Jun 2017 • Small data loss in February 2016 due to faulty battery 2 March 2017 3 to 10 April 2018 Instruments field verification GHI – reference CMP 10 GHI 2 – reference Li200 Mutanda – Solwezi Comments and issues Station type Tier 2 • Several distortions of GHI and DNI values occurred on some mornings due to dew. Instruments cleaning interval Average: 3.0 [days] Longest: 17 Verification visits date by 8 Jun 2016 GeoSUN Africa 5 Nov 2016 27 Jun 2017 3 to 10 April 2018 Instruments field verification GHI – reference CMP 10 GHI 2 – reference Li200 GeoSUN Africa performed detailed visits and station maintenance after 6, 12, 18 and 24 months of operation. Instruments field verification [26], i.e. comparative measurements of solar radiation parameters and cross check with the reference instruments was performed by GeoSUN Africa after first year of operation (Table 3.8). The objective was to proof that calibration constants remained stable within the instrument specifications. At the end of measurement campaign, the re-calibration of instruments was performed. To perform a verification of the measurements, the spare instruments of the same category were used. The solar sensors (thermopile pyrheliometer and thermopile pyranometer) were side-by-side compared and only clear-sky © 2018 Solargis page 16 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 values were used. In case of RSR, the GHI values of the RSR were compared to the GHI of the reference thermopile pyranometer for a 12-month period, to assess possible drift. Results of field instruments verification are listed in Table 3.9. During the data analysis, it was found that incorrect multipliers were applied on RSR sensors, thus the calibration coefficients at all meteorological stations had been updated and the measurements have been post- processed to correct this issue. The detailed results and discussion is supplied in Sensor verification report delivered in July 2017 [26]. The sensors are found to operate within the expected uncertainty limits. Table 3.9 Results of field instruments verification performed by GeoSUN Africa on April 2018 Measured bias Station Site name GHI DNI DIF GHI 2 type 2 2 2 2 [W/m ] [%] [W/m ] [%] [W/m ] [%] [W/m ] [%] Lusaka UNZA Tier 1 -11.42 -1.10 -6.75 -0.77 -1.83 -1.32 2.05 0.21 Mount Makulu Tier 2 -3.74 -1.38 - - - - -11.79 -4.42 Mochipapa Tier 2 -6.44 -0.72 - - - - -2.57 -0.33 Longe Tier 2 -3.60 -0.35 - - - - -6.98 -0.73 Misamfu Tier 2 -4.59 -0.55 - - - - -15.07 -1.85 Mutanda Tier 2 -17.27 -1.99 - - - - -22.79 -2.74 On the April 2018 GeoSUN Africa swopped the pyranometers with newly-calibrated ones. A field verification was performed on the meteorological instruments. Swap of the pyrheliometer at UNZA remains to be done. The data from all stations is regularly downloaded on a PC at the ZMD headquarters. After the end of Phase 2 (24-months of the measuring campaign) the operation of the stations is still overseen by GeoSUN Africa. All is ready for final hand-over of the equipment to ZMD. 3.3 Quality control of measured solar resource data Prior to correlation with satellite-based solar data, the ground-measured solar radiation was quality-controlled by Solargis. Quality control (QC) was based on methods defined in SERI QC procedures and Younes et al. [1, 2] and the in-house developed tests. The ground measurements were inspected also visually, mainly for identification of shading and other error patterns such as RSR shading ring malfunction. Figures 3.1, 3.2, 3.4, 3.9, 3.12, 3.16 and 3.18 show results of quality control for individual stations. The colours in the quality control pictures indicate the following flags: • Blue: data excluded by visual inspection - mainly shading, shading ring issues and effect of dew • Green: data passed all tests • Grey: un below horizon • White and brown strips: missing data • Red and violet: GHI, DNI and DIF consistency issue or exceedance of physical limits The data records not passing the quality control test were flagged and excluded from further processing. The results show relatively small amount of excluded data readings (Tables 3.11, 3.14, 3.17, 3.20, 3.23 and 3.26), predominantly during first year of operation. The most frequent is the shading from surrounding trees and the GHI values affected by dew. © 2018 Solargis page 17 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 3.3.1 University of Zambia (UNZA) Lusaka Table 3.10 Occurrence of data readings for UNZA Lusaka meteorological station Data availability DNI CHP1, GHI CMP10 GHI, DNI RSR Sun below horizon 555 472 49.3% 555 472 49.3% Sun above horizon 570 244 50.7% 570 244 50.7% Total data readings 1 125 716 100.0% 1 125 716 100.0% Table 3.11 Excluded ground measurements after quality control (Sun above horizon) in UNZA Lusaka Occurrence of data samples (Sun above horizon) Type of test DNI CHP1 GHI CMP10 DNI RSR GHI RSR Physical limits test 232 0.0% 3652 0.6% 0 0.0% 3041 0.5% Consistency test (GHI – DNI – DIF) 4091 0.7% 4091 0.7% 12 0.0% 12 0.0% Visual test (incorrect data) 12726 2.2% 10581 1.9% 11172 2.0% 10663 1.9% Other (non-valid data) 0 0.0% 0 0.0% 17280 3.0% 0 0.0% Total excluded data samples 17 049 3.0% 18 324 3.2% 28 464 5.0% 13 716 2.4% Total samples 570 244 100.0% 570 244 100.0% 570 244 100.0% 570 244 100.0% Main findings: • Occurrence of morning dew on sensors influencing mainly the measurements from thermopile instruments (CMP10 and CHP1) • Short periods of inconsistency between independent GHI, DNI and DIF measurements is present in the data (Figure 3.1 and 3.2). This might be a result of morning dew occurrence and insufficient cleaning. • Seasonal early morning and late afternoon shading from the surrounding objects. • A systematic difference between GHI measurements from the secondary standard pyranometer CMP10 and RSR (Figure 3.3 right). GHI from CMP10 is in average higher by 1.6% than GHI from RSR. The difference was 1.9% in 2016 and 1.2% in 2017. In the noon time, it can exceed 3% to 4%. • A systematic difference between DNI measurements from first class pyrheliometer CHP1 and RSR (Figure 3.3 left). DNI from CHP1 is in average higher by 2.0% than DNI from RSR. The difference was 2.4% in 2016 and 1.5% in 2017. In the noon time, it can exceed 4% to 5%. © 2018 Solargis page 18 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.1 Results of DNI (CHP1) and GHI (CMP10) quality control at UNZA Lusaka. Green – data passing all tests; grey – sun below horizon; red – consistency issue, violet – physical limit, blue excluded by visual inspection. Top: DNI (CHP1); bottom: GHI (CMP10) © 2018 Solargis page 19 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.2 Results of DNI (RSR) and GHI (RSR) quality control at UNZA Lusaka. Green – data passing all tests; grey – sun below horizon; red – consistency issue, violet – physical limit, blue excluded by visual inspection. Top: DNI (RSR); bottom: GHI (RSR) © 2018 Solargis page 20 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.3 Difference of DNI and GHI between two sensors – UNZA Lusaka. Left: DNI from CHP1 and RSR; right: GHI from CMP10 and RSR Table 3.12 Quality control summary – UNZA Lusaka Indicator Flag Description Station description, Installation report available metadata Instrument accuracy 2x Secondary standard pyranometer CMP10 (GHI, DIF) 1x First class pyrheliometer CHP1 (DNI) 1x Rotating Shadowband Radiometer RSR 2 (GHI, DIF, DNI) Instrument calibration Instruments were calibrated and calibration verified after 12 months Data structure Clear Cleaning and maintenance Cleaning log available information Diligent cleaning and maintenance Time reference Correct and clear time reference Quality control complexity RSR data, full QC CMP10 and CHP1 data, with (GHI-DNI-DIF) consistency test, comparison of GHI and DNI from RSR and CHP1 and CMP10 Quality control results Occurrence of morning dew influencing mainly data from CMP10 and CHP1 Small issues with early morning and late afternoon shading Small inconsistency of RSR and CMP10/CHP1 measurements Period More than 25 months Other issues - Legend: Quality flag Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 21 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 3.3.2 Mount Makulu (Chilanga) Table 3.13 Occurrence of data readings for Mount Makulu meteorological station Data availability GHI CMP10 GHI, DNI RSR Sun below horizon 553 295 49.3% 553 295 49.3% Sun above horizon 567 963 50.7% 5679 63 50.7% Total data readings 1 121 258 100.0% 1 121 258 100.0% Table 3.14 Excluded ground measurements after quality control (Sun above horizon) in Mount Makulu Occurrence of data samples (Sun above horizon) Type of test GHI CMP10 DNI RSR GHI RSR Physical limits test 5 063 0.9% 0 0.0% 3 756 0.7% Consistency test (GHI – DNI – DIF) 0 0.0% 0 0.0% 0 0.0% Visual test (incorrect data) 29 406 5.2% 32 748 5.8% 30 148 5.3% Other (non-valid data) 19 0.0% 19 0.0% 19 0.0% Total excluded data samples 34 488 6.1% 32 767 5.8% 33 923 6.0% Total samples 567 963 100.0% 567 963 100.0% 567 963 100.0% Main findings: • Incorrect orientation of RSR instrument towards the North results in incorrect data (Figure 3.5). This issue influenced morning data readings in the initial period of operation (November and December 2015). • Early morning shading (Figure 3.6) from the surrounding objects or trees. • A systematic difference between GHI measurements from the secondary standard pyranometer CMP10 and RSR (Figure 3.7). GHI from CMP10 is in average higher by 2.2% than GHI from RSR. The difference was 2.3% in 2016 and 2.0% in 2017. In the noon time, the difference can exceed 3% to 4%. • Occurrence of morning dew on the sensors influencing mainly GHI from the thermopile instrument (CMP10). © 2018 Solargis page 22 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.4 Results of GHI and DNI quality control in Mount Makulu. Green – data passing all tests; grey – sun below horizon; violet – physical limit issue, blue excluded by visual inspection; brown – missing data. Top: DNI (RSR); middle: GHI (RSR); bottom: GHI (CMP10) © 2018 Solargis page 23 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.5 Effect of RSR alignment issues – drop of DNI in Mount Makulu Green: DNI RSR; red: GHI CMP10; blue: GHI RSR; yellow: DIF RSR; dashed: theoretical clear-sky profile Figure 3.6 Effect of morning shading - Mount Makulu Green: DNI RSR; red: GHI CMP10; blue: GHI RSR; yellow: DIF RSR; dashed: theoretical clear-sky profile Figure 3.7 Systematic difference between GHI from CMP10 and RSR - Mount Makulu © 2018 Solargis page 24 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Table 3.15 Quality control summary – Mount Makulu Indicator Flag Description Station description, Installation report available metadata Instrument accuracy 1x Secondary standard pyranometer CMP10 (GHI) 1x Rotating Shadowband Radiometer RSR 2 (GHI, DIF, DNI) Instrument calibration Instruments were calibrated, and calibration was verified after 12 months Data structure Clear Cleaning and maintenance Cleaning log available information Diligent cleaning and maintenance Time reference Correct and clear time reference Quality control complexity RSR data, full QC comparison of GHI from RSR and CMP10 Quality control results Incorrect orientation of RSR instrument in initial phase of measurements Occurrence of morning dew influencing mainly data from CMP10 Early morning shading Small inconsistency of RSR and CMP10 measurements Period More than 25 months Other issues Legend: Quality flag Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 25 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 3.3.3 Mochipapa (Choma) Table 3.16 Occurrence of data readings for Mochipapa meteorological station Data availability GHI CMP10 GHI, DNI RSR Sun below horizon 558 340 49.3% 5583 40 49.3% Sun above horizon 574 316 50.7% 574 316 50.7% Total data readings 1 132 656 100.0% 1 132 656 100.0% Table 3.17 Excluded ground measurements after quality control (Sun above horizon) in Mochipapa Occurrence of data samples (Sun above horizon) Type of test GHI CMP10 DNI RSR GHI RSR Physical limits test 5 184 0.9% 0 0.0% 3 322 0.6% Consistency test (GHI – DNI – DIF) 0 0.0% 0 0.0% 0 0.0% Visual test (incorrect data) 34 992 6.1% 32 417 5.6% 30 785 5.4% Other (non-valid data) 0 0.0% 0 0.0% 0 0.0% Total excluded data samples 40 176 7.0% 32 417 5.6% 34 107 5.9% Total samples 574 316 100.0% 574 316 100.0% 574 316 100.0% Main findings: • Late afternoon shading from surrounding objects or trees (Figure 3.9) • A systematic difference between GHI measurements from secondary standard pyranometer CMP10 and RSR (Figure 3.10). GHI from CMP10 is in average higher by 1.8% than GHI from RSR. The difference was 2.1% in 2016 and 1.5% in 2017. In the noon time it can exceed 4%. • Occurrence of morning dew on sensors influencing mainly the GHI from thermopile instrument (CMP10) • Slight asymmetry of diurnal profiles may indicate problems with misalignment of instruments (Figure 3.11). © 2018 Solargis page 26 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.8 Results of GHI and DNI quality control − Mochipapa. Green – data passing all tests; grey – sun below horizon; violet – physical limit issue, blue - excluded by visual inspection. Top: DNI (RSR); middle: GHI (RSR); bottom: GHI (CMP10) © 2018 Solargis page 27 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.9 Shading effects on GHI and DNI in Mochipapa Green: DNI RSR; red: GHI CMP10; blue: GHI RSR; yellow: DIF RSR; dashed: theoretical clear-sky profile Figure 3.10 Systematic difference between GHI from CMP10 and RSR − Mochipapa Figure 3.11 Asymmetry of GHI diurnal profiles from CMP10 and RSR − Mochipapa Red: GHI CMP10; blue: GHI RSR; dashed: theoretical clear-sky profile © 2018 Solargis page 28 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Table 3.18 Quality control summary – Mochipapa Indicator Flag Description Station description, Installation report available metadata Instrument accuracy 1x Secondary standard pyranometer CMP10 (GHI) 1x Rotating Shadowband Radiometer RSR 2 (GHI, DIF, DNI) Instrument calibration Instruments were calibrated, and calibration was verified after 12 months Data structure Clear Cleaning and maintenance Cleaning log available information Diligent cleaning and maintenance Time reference Correct and clear time reference Quality control complexity RSR data, full QC comparison of GHI from RSR and CMP10 Quality control results Occurrence of morning dew influencing mainly data from CMP10 Late afternoon shading Small inconsistency of RSR and CMP10 measurements Slight instrument misalignment Period More than 25 months Other issues Legend: Quality marker Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 29 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 3.3.4 Longe (Kaoma) Table 3.19 Occurrence of data readings for Longe meteorological station Data availability GHI CMP10 GHI, DNI RSR Sun below horizon 534 958 49.4% 534 958 49.4% Sun above horizon 548 940 50.6% 548 940 50.6% Total data readings 1 083 898 100.0% 1 083 898 100.0% Table 3.20 Excluded ground measurements after quality control (Sun above horizon) in Longe Occurrence of data samples (Sun above horizon) Type of test GHI CMP10 DNI RSR GHI RSR Physical limits test 6 216 1.1% 0 0.0% 3 080 0.6% Consistency test (GHI – DNI – DIF) 0 0.0% 0 0.0% 0 0.0% Visual test (incorrect data) 47 163 8.6% 45 888 8.4% 42 726 7.8% Other (non-valid data) 26 0.0% 0 0.0% 0 0.0% Total excluded data samples 53 405 9.7% 45 888 8.4% 45 806 8.3% Total samples 548 940 100.0% 548 940 100.0% 548 940 100.0% Main findings: • Several periods with missing data, due to issues with the datalogger (Figure 3.12) • Early morning and late afternoon shading from the surrounding trees (Figure 3.13) for the whole period of measurements. • A systematic difference between GHI measurements from secondary standard pyranometer CMP10 and RSR (Figure 3.14). GHI from CMP10 is in average higher by 1.2% than GHI from RSR. The difference was 1.1% in 2016 and 1.3% in 2017. In the noon time, it can exceed 4%. • Higher occurrence of morning dew on sensors influencing mainly GHI from the thermopile instrument (Figure 3.15) © 2018 Solargis page 30 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.12 Results of GHI and DNI quality control − Longe Green – data passing all tests; grey – sun below horizon; violet – physical limit issue, blue – excluded by visual inspection; Top: DNI (RSR); middle: GHI (RSR); bottom: GHI (CMP10) Figure 3.13 Systematic shading effects on GHI and DNI in Longe. Green: DNI RSR; red: GHI CMP10; blue: GHI RSR; yellow: DIF RSR; dashed: theoretical clear-sky profile © 2018 Solargis page 31 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.14 Systematic difference between GHI from CMP10 and RSR - Longe. Figure 3.15 Morning dew effect on GHI and DNI measurements in Longe. Green: DNI RSR; red: GHI CMP10; blue: GHI RSR; yellow: DIF RSR; dashed: theoretical clear-sky profile Table 3.21 Quality control summary – Longe Indicator Flag Description Station description, Installation report available metadata Instrument accuracy 1x Secondary standard pyranometer CMP10 (GHI) 1x Rotating Shadowband Radiometer RSR 2 (GHI, DIF, DNI) Instrument calibration Instruments were calibrated, and calibration was verified after 12 months Data structure Clear Cleaning and maintenance Cleaning log available information Several periods without cleaning (9 to 17 days) Time reference Correct and clear time reference Quality control complexity RSR data, full QC comparison of GHI from RSR and CMP10 Quality control results Occurrence of morning dew influencing mainly data from CMP10 Early morning and late afternoon shading Small inconsistency of RSR and CMP10 measurements Missing data Period More than 24 months Other issues Legend: Quality flag Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 32 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 3.3.5 Misamfu (Kasama) Table 3.22 Occurrence of data readings for Misamfu meteorological station Data availability GHI CMP10 GHI, DNI RSR Sun below horizon 545 304 49.3% 545 304 49.3% Sun above horizon 560 524 50.7% 560 524 50.7% Total data readings 1 105 828 100.0% 1 105 828 100.0% Table 3.23 Excluded ground measurements after quality control (Sun above horizon) in Misamfu Occurrence of data samples (Sun above horizon) Type of test GHI CMP10 DNI RSR GHI RSR Physical limits test 5 733 1.0% 0 0.0% 3208 0.6% Consistency test (GHI – DNI – DIF) 0 0.0% 0 0.0% 0 0.0% Visual test (incorrect data) 33 459 6.0% 30 335 5.4% 29 276 5.2% Other (non-valid data) 205 0.0% 6376 1.1% 280 0.0% Total excluded data samples 39 397 7.0% 36 711 6.5% 32 764 5.8% Total samples 560 524 100.0% 560 524 100.0% 560 524 100.0% Main findings: • Early morning shading for the whole period of measurements. • Missing data in February 2016, due to faulty RSR and battery (Figure 3.16). • A systematic difference between GHI measurements from the secondary standard pyranometer CMP10 and RSR (Figure 3.17). GHI from CMP10 is in average higher by 2.2% than GHI from RSR. The difference was 2.0% in 2016 and 2.4% in 2017. In the noon time, it can exceed 4%. • Occurrence of morning dew on sensors influencing mainly the GHI from thermopile instrument © 2018 Solargis page 33 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.16 Results of GHI and DNI quality control − Misamfu. Green – data passing all tests; grey – sun below horizon; violet – physical limit issue, blue excluded by visual inspection, brown – missing or non-valid values; Top: DNI (RSR); middle: GHI (RSR); bottom: GHI (CMP10) © 2018 Solargis page 34 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.17 Systematic difference between GHI from CMP10 and RSR - Misamfu. Table 3.24 Quality control summary – Misamfu Indicator Flag Description Station description, Installation report available metadata Instrument accuracy 1x Secondary standard pyranometer CMP10 (GHI) 1x Rotating Shadowband Radiometer RSR 2 (GHI, DIF, DNI) Instrument calibration Instruments were calibrated, and calibration was verified after 12 months Data structure Clear Cleaning and maintenance Cleaning log available information Diligent cleaning and maintenance Time reference Correct and clear time reference Quality control complexity RSR data, full QC comparison of GHI from RSR and CMP10 Quality control results Occurrence of morning dew influencing mainly data from CMP10 Early morning shading Small inconsistency of RSR and CMP10 measurements Missing data Period 25 months Other issues Legend: Quality flag Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 35 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 3.3.6 Mutanda Table 3.25 Occurrence of data readings for Mutanda meteorological station Data availability GHI CMP10 GHI, DNI RSR Sun below horizon 548 670 49.4% 548 670 49.4% Sun above horizon 561 866 50.6% 561 866 50.6% Total data readings 1 110 536 100.0% 1 110 536 100.0% Table 3.26 Excluded ground measurements after quality control (Sun above horizon) in Mutanda Occurrence of data samples (Sun above horizon) Type of test GHI CMP10 DNI RSR GHI RSR Physical limits test 6 541 1.2% 20 0.0% 3 228 0.6% Consistency test (GHI – DNI – DIF) 0 0.0% 0 0.0% 0 0.0% Visual test (incorrect data) 28 855 5.1% 18 301 3.3% 16 696 3.0% Other (non valid data) 0 0.0% 0 0.0% 0 0.0% Total excluded data samples 35 396 6.3% 18 321 3.3% 19 924 3.5% Total samples 561 866 100.0% 561 866 100.0% 561 866 100.0% Main findings: • Late afternoon shading for the whole period of measurements (Figure 3.18). • Negligible systematic difference between GHI measurements from secondary standard pyranometer CMP10 and RSR. GHI from CMP10 is in average higher by 0.6% than GHI from RSR. The difference was 0.3% in 2016 and 0.9% in 2017. In the noon time, it can exceed 3% (Figure 3.19). • High occurrence of morning dew on sensors influencing mainly the GHI from thermopile instrument (Figure 3.20). © 2018 Solargis page 36 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.18 Results of GHI and DNI quality control − Mutanda. Green – data passing all tests; grey – sun below horizon; violet – physical limit issue, blue excluded by visual inspection; Top: DNI (RSR); middle: GHI (RSR); bottom: GHI (CMP10) © 2018 Solargis page 37 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 3.19 Systematic difference between GHI from CMP10 and RSR - Mutanda. Figure 3.20 Morning dew effect on GHI measurements in Mutanda. Green: DNI RSR; red: GHI CMP10; blue: GHI RSR; yellow: DIF RSR; dashed: theoretical clear-sky profile Table 3.27 Quality control summary – Mutanda Indicator Flag Description Station description, Installation report available metadata Instrument accuracy 1x Secondary standard pyranometer CMP10 (GHI) 1x Rotating Shadowband Radiometer RSR 2 (GHI, DIF, DNI) Instrument calibration Instruments were calibrated, and calibration was verified after 12 months Data structure Clear Cleaning and maintenance Cleaning log available information Diligent cleaning and maintenance Time reference Correct and clear time reference Quality control complexity RSR data, full QC comparison of GHI from RSR and CMP10 Quality control results High occurrence of morning dew influencing mainly data from CMP10 Late afternoon shading Period More than 25 months Other issues Legend: Quality flag Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 38 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 3.4 Recommendations on the operation and maintenance of the meteorological stations in Zambia Based on the results of quality control (Tables 3.11, 3.14, 3.17, 3.20, 3.23 and 3.26), we conclude that the solar radiation measurements come from the high accuracy (CMP10, CHP1) and medium accuracy (RSR) measuring equipment that is professionally installed, diligently operated and carefully maintained. Small issues were identified during the data quality control: • Higher occurrence of the measurements by the thermopile instruments that are affected by dew. These data values were flagged and excluded from further processing. • Early morning and late afternoon shading from surrounding objects in several sites. The data were flagged and excluded from further processing. • Several short periods with insufficient cleaning at Longe and Mutanda station. For future works, we recommend: • Maintain and improve the cleaning frequency of solar sensors • Consider installation of ventilation units for the pyranometers to reduce data records affected by morning dew. © 2018 Solargis page 39 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 4 SOLAR RESOURCE MODEL DATA 4.1 Solar model Solar radiation is calculated by Solargis model, which are parameterized by a set of inputs characterizing the cloud transmittance, state of the atmosphere and terrain conditions. A comprehensive overview of the Solargis model is made available in the recent book publication [3]. The methodology is also described in [4, 5]. The related uncertainty and requirements for bankability are discussed in [6, 7]. In Solargis approach, the clear-sky irradiance is calculated by the simplified SOLIS model [8]. This model allows fast calculation of clear-sky irradiance from the set of input parameters. Sun position is deterministic parameter, and it is described by the algorithms with satisfactory accuracy. Stochastic variability of clear-sky atmospheric conditions is determined by changing concentrations of atmospheric constituents, namely aerosols, water vapour and ozone. Global atmospheric data, representing these constituents, are routinely calculated by world atmospheric data centres: • In Solargis, the new generation aerosol data set representing Atmospheric Optical Depth (AOD) is used. The calculation accuracy is strongly determined by quality of aerosols, especially for cloudless conditions. The aerosol data implemented by MACC-II/CAMS and MERRA-2 projects are used [9, 10]. • Water vapour is also highly variable in space and time, but it has lower impact on the values of solar radiation, compared to aerosols. The GFS and CFSR databases (NOAA NCEP) are used in Solargis, and the data represent the daily variability from 1994 to the present time [11, 12, 13]. • Ozone absorbs solar radiation at wavelengths shorter than 0.3 µm, thus having negligible influence on the broadband solar radiation. The clouds are the most influencing factor, modulating clear-sky irradiance. Effect of clouds is calculated from the satellite data in the form of a cloud index (cloud transmittance). The cloud index is derived by relating radiance recorded by the satellite in spectral channels and surface albedo to the cloud optical properties. In Solargis, the modified calculation scheme of Cano has been adopted to retrieve cloud optical properties from the satellite data [14]. To calculate all-sky irradiance in each time step, the clear-sky global horizontal irradiance is coupled with cloud index. Direct Normal Irradiance (DNI) is calculated from Global Horizontal Irradiance (GHI) using modified Dirindex model [15]. Diffuse irradiance for tilted surfaces, which is calculated by Perez model [16]. The calculation procedure includes also terrain disaggregation, the spatial resolution is enhanced with use of the digital terrain model to 250 meters [17]. Solargis model version 2.1 has been used. Table 4.1 summarizes technical parameters of the model inputs and of the primary data outputs. Table 4.1 Input data used in the Solargis and related GHI and DNI outputs for Zambia Inputs into the Solargis model Source Time Original Approx. grid of input data representation time step resolution Cloud index (satellite data) Meteosat MFG 1994 to 2004 30 minutes 2.7 x 3.3 km Meteosat MSG 2005 to date 15 minutes 3.2 x 4.0 km (EUMETSAT) Atmospheric optical depth MACC-II/CAMS* 2003 to date 3 hours 75 km and 125 km (aerosols)* (ECMWF) MERRA-2 (NASA) 1994 to 2002 1 hour 50 km Water vapour CFSR/GFS (NOAA) 1994 to date 1 hour 35 and 55 km Elevation and horizon SRTM-3 (SRTM) - - 250 m Solargis primary data (GHI, DNI) - 1994 to date 15 minutes 250 m * Aerosol data for 2003-2012 come from the reanalysis database; the data representing years 2013-present are derived from near- real time operational model © 2018 Solargis page 40 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 4.1: Sites with solar meteorological stations used for site adaptation of Solargis model in Zambia 4.2 Site adaptation of the solar model − method This chapter describes accuracy improvement of the delivered model time series for six sites. This improvement has been achieved by site adaptation of the model, based on the use of local measurements. The fundamental difference between a satellite observation and a ground measurement is that signal received by the satellite radiometer integrates a large area, while a ground station represents a pinpoint measurement. This results in a mismatch when comparing instantaneous values from these two observation instruments, mainly during intermittent cloudy weather and changing aerosol load. Nearly half of the hourly Root Mean Square Deviation (RMSD) for GHI and DNI can be attributed to this mismatch (value at sub-pixel scale), which is also known as the “nugget effect” [18]. The satellite pixel is not capable describing the inter-pixel variability in complex regions, where within one pixel diverse natural conditions mix-up (e.g. fog in narrow valleys or along the coast). In addition, the coarse spatial resolution of atmospheric databases such as aerosols or water vapour is not capable to describe local patterns of the state of atmosphere. These features can be seen in the satellite GHI and DNI data by increased bias due to imperfect description of aerosol load and satellite GHI mainly due to inaccurate identification of highly variable clouds. Satellite data have inherent inaccuracies, which have certain degree of geographical and time variability. Especially DNI is strongly sensitive to variability of cloud information, aerosols, water vapour, and terrain shading. The relation between uncertainty of global and direct irradiance is nonlinear. Often, a negligible error in global irradiance may have high counterpart in the direct irradiance component. This relation may be present also in opposite order. The solar energy projects require representative and accurate GHI and DNI time series. The satellite-derived databases are used to describe long-term solar resource for a specific site. However, their problem when compared to the high-quality ground measurements is a slightly higher bias and partial disagreement of frequency distribution functions, which may limit their potential to record the occurrence of extreme situations (e.g. very low atmospheric turbidity resulting in a high DNI and GHI). A solution is to correlate satellite-derived data with ground measurements to understand the source of discrepancy and subsequently to improve the accuracy of the resulting time series. The Solargis satellite-derived data are correlated with ground measurement data with two objectives: • Improvement of the overall bias (removal of systematic deviations) © 2018 Solargis page 41 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 • Improvement of the fit of the frequency distribution of values. Limited spatial and temporal resolution of the input data and simplified nature of the models results in the occurrence of systematic and random deviations of the model outputs when compared to the ground observations. The deviations in the satellite-computed data, which have systematic nature, can be reduced by site adaptation or regional adaptation methods. The terminology related to the procedure improving accuracy of the satellite data is not harmonized, and various terms are used: • Correlation of ground measurements and satellite-based data; • Calibration of the satellite model (its inputs and parameters); • Site adaptation or regional adaptation of satellite-based data. The term site adaptation is more general and best explains the concept of adapting the satellite-based model (by correlation, calibration, fitting and recalculation) to the ground measured data. Site adaptation aims to adapt the characteristics of the satellite-based time series to the site-specific conditions described by local measurements. Three conditions are important for successful adaptation of the satellite-based model: 1. High quality DNI and GHI ground measurements for at least 12 months must be available; optimally data for 2 or 3 years should be used; 2. High quality satellite data must be used, with consistent quality over the whole period of data; 3. There must be identified a systematic difference between both data sources. Systematic difference can be stable over the year or it can slightly change seasonally for certain meteorological conditions (e.g. typical cloud formation during a day, seasonal air pollution). The data analysis should distinguish systematic differences from those arising at occasional events, such as extreme sand storms or forest fires. The episodically-occurring differences may mislead the results of adaptation, especially if short period of ground measurements is only available. If one of the three above-mentioned conditions is not fulfilled, site adaptation (regional adaptation) will not provide the expected results. On the contrary, such an attempt may provide worse results. For the quantitative assessment of the accuracy enhancement procedures, the following metrics is used: • Metrics based on the comparison of all pairs of the hourly daytime data values: Mean Bias, and Root Mean Square Deviation (RMSD), histogram, in an absolute and relative form (divided by the daytime mean DNI values); • Metrics based on the difference of the cumulative distribution functions: KSI (Kolmogorov-Smirnov test Integral) [19] The normalized KSI is defined as an integral of absolute differences of two cumulative distribution functions D normalized by the integral of critical value acritical: , where critical value depends on the number of the data pairs N. As the KSI value is dependent on the size of the sample, the KSI measure may be used only for the relative comparison of fit of cumulative distribution of irradiance values. For the accuracy enhancement of solar resource parameters in this study, a combination of two methods was used. First, systematic deviations due to influence of aerosols were partially removed. Afterwards, to improve the distribution of values, the fitting of cumulative frequency distribution curves of ground measurements and satellite data was used. The site-adaptation procedure first identifies the sources of discrepancies by comparing the ground-measured data with Solargis model data, for the period of the overlap between both data sets approx. 24-25 months). Based on this analysis, correction coefficients to improve the fit between the measured and the model Solargis data are developed. In the second step, these coefficients are used for the adaptation of the full (24 years) time series. © 2018 Solargis page 42 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 The satellite data is available in 15-minute time step and the ground measurements in 1-minute time step. To partially remove the conceptual difference of point and satellite pixel measurements, prior to site adaptation, all the measures are calculated using aggregated data in the hourly time step. The adaptation was based on measured DNI data from CHP1 instrument at UNZA Lusaka and RSR instrument for other stations. Measured GHI data from the secondary standard CMP10 instrument is used at all stations. GHI measured by the RSR was not used because of higher uncertainty of the outputs (Chapter 3.3). More about the Solargis site adaptation is in [21], more general description can be found in [22]. 4.3 Results of the model adaptation at six sites The original Solargis data show a regional pattern of overestimation, compared to the ground measurements, for both GHI and DNI. The biggest difference between ground measurements and satellite data is found at Mutanda station (GHI). In Mutanda, the mismatch between the measured and modelled GHI is 9.5% (Table 4.2). Such discrepancy is beyond usual uncertainty interval known for Solargis GHI data in this region (6% to 8% for GHI). The detailed inspection of the measurements and the satellite data indicates two possible sources (or their combination) of this difference: • Performance of satellite models is in general lower for conditions with high occurrence of scattered and fast-changing clouds. • Ground measurements for Mutanda site, but partly also for other sites, indicate some issues, mostly related to the occurrence of dew influence on measurements and the local shading. The high frequency variability (small scattered clouds) makes it difficult to distinguish the shading from surrounding objects from drop of solar irradiance due to clouds. The model adaptation allowed removing a large part of the mismatch between the satellite-based data and the ground measurements. Tables 4.2 to 4.5 summarize the site-adaptation results for all solar meteorological stations in Zambia. Table 4.2 Direct Normal Irradiance: bias and KSI before and after model site-adaptation Meteo station Original DNI data DNI after site adaptation Bias Bias KSI Bias Bias KSI [kWh/m2] [%] [-] [kWh/m2] [%] [-] UNZA Lusaka 44 10.5 213 0 0.0 79 Mount Makulu 42 9.9 200 0 0.0 74 Mochipapa 41 9.0 197 0 0.0 87 Longe 32 6.9 156 0 0.0 68 Misamfu 44 10.1 206 0 0.1 83 Mutanda 43 10.5 202 0 0.0 84 Mean 41 9.5 196 0 0.0 79 Standard deviation 5 1.4 0 0.0 © 2018 Solargis page 43 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Table 4.3 Global Horizontal Irradiance: bias and KSI before and after model site-adaptation Meteo station Original GHI data GHI after site adaptation Bias Bias KSI Bias Bias KSI [kWh/m2] [%] [-] [kWh/m2] [%] [-] UNZA Lusaka 32 6.8 156 0 0.0 24 Mount Makulu 30 6.4 148 0 0.1 26 Mochipapa 26 5.4 127 0 0.0 21 Longe 33 6.6 156 0 0.0 24 Misamfu 32 6.4 154 0 0.0 22 Mutanda 46 9.5 218 0 0.0 35 Mean 33 6.9 160 0 0.0 25 Standard deviation 7 1.4 0 0.0 Table 4.4 Direct Normal Irradiance: RMSD before and after model site-adaptation Meteo station RMSD of original DNI data RMSD of DNI after site adaptation Hourly Daily Monthly Hourly Daily Monthly [%] [%] [%] [%] [%] [%] UNZA Lusaka 32.3 18.2 14.6 30.0 13.7 5.8 Mount Makulu 34.8 19.5 14.9 33.0 15.9 7.0 Mochipapa 30.4 17.1 12.6 28.9 13.4 4.0 Longe 30.9 18.3 13.7 29.5 14.7 5.6 Misamfu 35.3 19.2 14.2 33.4 15.5 7.0 Mutanda 36.0 20.9 16.5 33.5 16.2 7.0 Mean 33.3 18.9 14.4 31.4 14.9 6.1 Table 4.5 Global Horizontal Irradiance: RMSD before and after model site-adaptation Meteo station RMSD of original GHI data RMSD of GHI after site adaptation Hourly Daily Monthly Hourly Daily Monthly [%] [%] [%] [%] [%] [%] UNZA Lusaka 19.1 10.4 8.6 17.4 7.6 4.0 Mount Makulu 21.2 11.0 9.0 19.7 8.6 4.9 Mochipapa 18.4 9.1 7.1 17.2 7.1 3.5 Longe 18.4 10.3 8.7 16.8 7.7 4.4 Misamfu 19.8 9.9 7.9 18.3 7.4 4.0 Mutanda 21.8 12.6 11.2 18.9 7.9 4.5 Mean 19.8 10.6 8.8 18.1 7.7 4.2 As a result, at the level of individual meteorological sites in Zambia, the mean bias of the site-adapted values was reduced to zero. The values of RMSD and KSI accuracy parameters are also reduced, both for GHI and DNI. The effect of the site adaptation is presented in a detail for all sites (Figures 4.2 to 4.7). © 2018 Solargis page 44 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 UNZA Lusaka: Original DNI UNZA Lusaka: DNI after adaptation UNZA Lusaka: Original GHI UNZA Lusaka: GHI after adaptation Figure 4.2: Correction of DNI and GHI hourly values for UNZA Lusaka. Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. © 2018 Solargis page 45 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Mount Makulu: Original DNI Mount Makulu: DNI after adaptation Mount Makulu: Original GHI Mount Makulu: GHI after adaptation Figure 4.3: Correction of DNI and GHI hourly values for Mount Makulu Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. © 2018 Solargis page 46 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Mochipapa: Original DNI Mochipapa: DNI after adaptation Mochipapa: Original GHI Mochipapa: GHI after adaptation Figure 4.4: Correction of DNI and GHI hourly values for Mochipapa. Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. © 2018 Solargis page 47 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Longe: Original DNI Longe: DNI after adaptation Longe: Original GHI Longe: GHI after adaptation Figure 4.5: Correction of DNI and GHI hourly values for Longe. Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. © 2018 Solargis page 48 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Misamfu: Original DNI Misamfu: DNI after adaptation Misamfu: Original GHI Misamfu: GHI after adaptation Figure 4.6: Correction of DNI and GHI hourly values for Misamfu. Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. © 2018 Solargis page 49 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Mutanda: Original DNI Mutanda: DNI after adaptation Mutanda: Original GHI Mutanda: GHI after adaptation Figure 4.7: Correction of DNI and GHI hourly values for Mutanda. Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. The change of model GHI and DNI after site adaptation is presented on an example of UNZA Lusaka (Figure 4.8). Both the adapted GHI and DNI values are lower than the original values. The other sites show similar pattern (Table 4.6). The site-adapted model values better represent the geographical variability of DNI and GHI solar resource, and they also improve the distribution and match of hourly values. The measurements show that the Solargis model overestimates GHI and DNI in the region, and the results of site adaptation significantly improve the model performance. The results increase the confidence about the reliability of the measured and modelled solar resource data for Zambia. © 2018 Solargis page 50 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 4.8: Comparison of Solargis original and site-adapted data for UNZA Lusaka site. Left: DNI; Right: GHI; Data represent years 1994 to 2017. Table 4.6 Comparison of long term average of yearly summaries of original and site-adapted values Meteo station DNI annual values GHI annual values Original Site-adapted Difference Original Site-adapted Difference 2 2 2 2 [kWh/m ] [kWh/m ] [%] [kWh/m ] [kWh/m ] [%] Lusaka UNZA 2030 1846 -9.0 2131 1996 -6.3 Mount Makulu 2033 1861 -8.5 2128 2001 -6.0 Mochipapa 2134 1969 -7.7 2150 2041 -5.1 Longe 2127 1995 -6.2 2212 2076 -6.2 Misamfu 1944 1763 -9.3 2190 2059 -6.0 Mutanda 1930 1755 -9.1 2165 1978 -8.7 © 2018 Solargis page 51 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 5 METEOROLOGICAL MODEL DATA 5.1 Meteorological model For the territory of Zambia, the last 24 years of the Solargis model-based meteorological data is derived from the meteorological models CFSR and CFSv2, with original characteristics specified in Table 5.1. Table 5.1 Source of Solargis meteorological data: models CFSR and CFSv2 and their characteristics. Climate Forecast System Reanalysis Climate Forecast System (CFSR) (CFSv2) Period 1994 to 2010 2011 to the present time Original spatial resolution 30 x 35 km 19 x 22 km Original time resolution 1 hour 1 hour Table 5.2 shows meteorological parameters available in Solargis, their specifications and it also indicates, which of them have been delivered within this study. The original spatial resolution of the models is enhanced to 1 km for air temperature and relative humidity by spatial disaggregation and use of the Digital Elevation Model SRTM- 3. The spatial resolution (spatial representation) of other parameters is unchanged. Table 5.2 Solargis meteorological parameters delivered within this project Time Spatial Data Data Meteorological parameter Acronym Unit resolution representation delivered validated Air temperature at 2 metres TEMP °C 60 minute 1 km Yes Yes (dry bulb temperature) Relative humidity at 2 metres RH % 60 minute Original model Yes Yes 2 Wind speed at 10 metres WS m/s 60 minute Original model Yes Yes Wind direction at 10 metres WD ° 60 minute Original model Yes Yes Atmospheric pressure AP hPa 60 minute 1 km Yes Yes Precipitable water PWAT 60 minute Original model Yes No Important note: meteorological parameters are derived from the numerical weather model outputs and these models have lower spatial and temporal resolution. Thus, they do not represent the same accuracy as the solar resource data. Especially wind speed data has higher uncertainty, and it provides only overview information for solar energy projects. Thus, the local microclimate at the meteorological stations may deviate from the values derived from the Solargis meteorological database. 5.2 Validation of meteorological data The validation procedure was carried out to compare the modelled data with ground-measurements from the 6 meteorological stations installed within the ESMAP project: Lusaka, Mount Makulu, Mochipapa, Longe, Misamfu and Mutanda. In general, the data from the meteorological model outputs represent larger area, and it is not capable to represent accurately the local microclimate. © 2018 Solargis page 52 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 5.2.1 Air temperature at 2 metres Air temperature is derived from the CSFR and CSFv2 meteorological models and recalculated at the spatial resolution of 1 km (Table 5.3 and Figures 5.1 to 5.6). Considering spatial and time interpolation, the deviation of the modelled values to the ground observations for hourly values can occasionally reach several degrees of Celsius. Figures 5.1 to 5.6 show graphical representation of the model values accuracy at the meteorological stations. In general, the model matches the ground measurements quite well. The main issue identified is underestimation or overestimation of night-time temperature by the model. Day-time temperature is represented with higher accuracy compared to night-time. Table 5.3 Air temperature at 2 m: accuracy indicators of the model outputs [ºC]. CFSv2 model Meteorological station Bias Bias Bias Bias Bias RMSD RMSD RMSD mean min max nigh-time day-time hourly daily monthly Lusaka UNZA -1.6 -1.7 -0.9 -1.9 -1.3 2.5 1.8 1.6 Mount Makulu -1.7 -1.6 -1.3 -1.8 -1.7 2.7 2.0 1.8 Mochipapa -1.1 -0.6 -0.9 -1.0 -1.3 2.2 1.5 1.2 Longe 0.2 1.3 -0.4 1.0 -0.7 2.5 1.4 0.9 Misamfu -1.7 -0.8 -2.0 -1.0 -2.3 2.7 2.0 1.8 Mutanda 0.8 2.8 -1.2 2.2 -0.6 3.4 2.2 1.9 Figure 5.1: Scatterplots of air temperature at 2 m at Lusaka UNZA meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements © 2018 Solargis page 53 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.2: Scatterplots of air temperature at 2 m at Mount Makulu meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements Figure 5.3: Scatterplots of air temperature at 2 m at Mochipapa meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements © 2018 Solargis page 54 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.4: Scatterplots of air temperature at 2 m at Longe meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements Figure 5.5: Scatterplots of air temperature at 2 m at Misamfu meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements © 2018 Solargis page 55 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.6: Scatterplots of air temperature at 2 m at Mutanda meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements 5.2.2 Relative humidity Relative humidity is calculated from the specific humidity, atmospheric pressure and the air temperature. The comparison of the model values with ground measurements at all 6 meteorological stations is shown in Table 5.4 and Figures 5.7 to 5.12. In general, the model matches the ground measurements quite well, representing both daily and yearly profiles. Table 5.4 Relative humidity: accuracy indicators of the model outputs [%]. CFSv2 model Meteorological station Bias Bias Bias Bias Bias RMSD RMSD RMSD mean min max nigh-time day-time hourly daily monthly Lusaka -1 0 -2 -1 -1 10 6 2 Mount Makulu 0 0 -1 -1 0 10 6 2 Mochipapa -3 1 -6 -5 -1 11 7 4 Longe -9 -3 -12 -14 -4 16 12 11 Misamfu 1 3 -2 -2 4 11 6 4 Mutanda -9 1 -15 -16 -2 19 14 13 © 2018 Solargis page 56 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.7: Scatterplots of relative humidity at 2 m at Lusaka UNZA meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. Figure 5.8: Scatterplots of relative humidity at 2 m at Mount Makulu meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. © 2018 Solargis page 57 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.9: Scatterplots of relative humidity at 2 m at Mochipapa meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. Figure 5.10: Scatterplots of relative humidity at 2 m at Longe meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. © 2018 Solargis page 58 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.11: Scatterplots of relative humidity at 2 m at Misamfu meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. Figure 5.12: Scatterplots of relative humidity at 2 m at Mutanda meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. © 2018 Solargis page 59 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 5.2.3 Wind speed and wind direction at 10 metres Wind speed and direction values delivered within Solargis data represent the height at 10 meters and they are calculated from the CFSR and CFSv2 models, from 10 m wind u- and v- components. The spatial resolution is kept the same, as in the original data. Wind measurements take place at the height of 10 metres (Lusaka) and 3 metres (other stations), while the model data represent values at height of 10 m above ground. Comparison of the modelled wind speed with ground measurements is shown in Table 5.5 and Figures 5.13 to 5.18. The model values underestimate the wind conditions measured at the meteorological stations. Similar to relative humidity, the data representation for wind speed and wind direction strongly depends on the local conditions; therefore the model values are only indicative and better characterize a larger region rather than the local microclimate. The important source of systematic difference is different height of the installed wind sensor at Tier 2 stations (3 metres above ground), compared to the model assumptions (10 metres above ground). Table 5.5 Wind speed: accuracy indicators of the model outputs [m/s]. CFSR and CFSv2 models Meteorological station Bias Bias Bias Bias Bias RMSD RMSD RMSD mean min max nigh-time day-time hourly daily monthly Lusaka UNZA -0.1 0.2 -0.2 0.1 -0.2 1.0 0.5 0.3 Mount Makulu 1.8 1.4 2.2 1.7 1.8 2.1 1.9 1.9 Mochipapa 1.4 1.3 0.9 1.8 1.0 1.9 1.6 1.5 Longe 1.7 1.5 1.3 2.2 1.3 2.0 1.9 1.8 Misamfu 1.1 1.1 0.7 1.4 0.8 1.4 1.2 1.2 Mutanda 1.7 1.4 1.2 1.9 1.4 1.9 1.8 1.7 Figure 5.13: Scatterplots of wind speed at Lusaka UNZA meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations and model data at 10 m) © 2018 Solargis page 60 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.14: Scatterplots of wind speed at Mount Makulu meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations at 3 m height and model data at 10 m) Figure 5.15: Scatterplots of wind speed at Mochipapa meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations at 3 m height and model data at 10 m) © 2018 Solargis page 61 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.16: Scatterplots of wind speed at Longe meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations at 3 m height and model data at 10 m) Figure 5.17: Scatterplots of wind speed at Misamfu meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations at 3 m height and model data at 10 m) © 2018 Solargis page 62 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.18: Scatterplots of wind speed at Mutanda meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations at 3 m height and model data at 10 m) 5.3 Uncertainty of meteorological model data The meteorological parameters are derived from two very similar numerical meteorological models covering periods from 1994 to 2010 (CFSR model) and 2011 to 2017 (CFSv2). Considering the comparison results, the uncertainty of the estimate for the main meteorological parameters is summarised in Table 5.6. The uncertainty is expressed for 80% occurrence. It was found that the modelled air temperature fits reasonably well the measured data though (logically) due to the spatial resolution there are some issues like underestimation of night-time temperature. Similar to air temperature, the model relative humidity fits well the measured data representing both daily and yearly amplitude. Wind speed data, obtained from the meteorological model, represents an area of larger region, in comparison to the point measurements collected at the meteorological sites. The model values represent wind conditions at 10 metres height above ground, while the measurements represent 3 metres height. Atmospheric pressure from the models fits well the measured data with a very small bias for all meteorological stations not exceeding 3 hPa. Table 5.6 Expected uncertainty of modelled meteorological parameters at the project sites. Unit Annual Monthly Hourly Air temperature at 2 m °C ±2.0 ±2.0 ±3.5 Relative humidity at 2 m % ±10 ±10 ±15 Average wind speed at 10 m m/s ±1.5 ±1.5 ±2.0 Atmospheric pressure hPa ±3 ±3 ±3 © 2018 Solargis page 63 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 6 SOLAR RESOURCE: UNCERTAINTY OF LONG-TERM ESTIMATES 6.1 Uncertainty of solar resource yearly estimate The uncertainty of site-adapted satellite-based GHI and DNI is determined by the uncertainty of the model and of the ground measurements [7], more specifically it depends on: 1. Parameterization and adaptation of numerical models integrated in Solargis for the given data inputs and their ability to generate accurate results for various geographical and time-variable conditions: • Data inputs into Solargis model (accuracy of satellite data, aerosols and water vapour and Digital Terrain Model). • Clear-sky model and its capability to properly characterize various states of the atmosphere • Simulation accuracy of the satellite model and cloud transmittance algorithms, being able to properly distinguish different types of desert surface, clouds, fog, but also snow and ice. • Diffuse and direct decomposition models • Site adaptation methods. 2. Uncertainty of the ground-measurements, which is determined by: • Accuracy of the instruments • Maintenance practices, including sensor cleaning, calibration • Data post-processing and quality control procedures. Solargis model uncertainty is compared to the high-quality data measured by the meteorological instruments. Representativeness of such data comparison (satellite and ground-measured) is determined by the precision of the measuring instruments, the maintenance and operational practices, and by quality control of the measured data – in other words, by the measurement accuracy achieved at each meteorological station. Accuracy statistics, such as bias and RMSD (Chapter 4.3) characterize accuracy of the Solargis model in the given validation points, relative to the ground measurements. The validation statistics is affected by local geography and by quality and reliability of the ground-measured data. Therefore, the validation statistics only indicates performance of the model in the region. From the user’s perspective, the information about the model uncertainty has probabilistic nature. It generalizes the validation accuracy and it has to be considered at different confidence levels. The expert estimate of the calculation uncertainty in this report assumes 80% probability of occurrence of values. The solar model uncertainty is discussed in Chapters 4 and 6.1. The main findings are summarized in Table 6.1. The site-adaptation procedure reduced uncertainty of estimate of all parameters. Chapter 6.3 evaluates combined uncertainty, in which also interannual variability is included (Chapter 6.2). The physical reduction of the model uncertainty was significant already after site adaptation using ground measurements after first year, and it was further reduced after second year of measurement campaign (Table 6.1). In addition, the site adaptation increases confidence in the model data values. Table 6.1 Uncertainty of the model estimates for original and site-adapted annual long-term values (Considers 80% probability of occurrence) Uncertainty of long-term Acronym Uncertainty of the original Uncertainty of the Solargis model after annual values Solargis model site adaptation After 1st year After 2nd year Global Horizontal Irradiation GHI ±7.5% (up to ±10.0%*) ±4.5% ±4.0% Direct Normal Irradiance DNI ±12.0% (up to ±18.0%*) ±6.0% ±5.5% * in complex microclimate © 2018 Solargis page 64 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 6.2 Uncertainty due to interannual variability of solar radiation Weather changes in cycles and has also stochastic nature. Therefore, annual solar radiation in each year can deviate from the long-term average in the range of few percent. The estimation of the interannual variability below shows the magnitude of this change. The uncertainty of GHI and DNI prediction is highest if only one single year is considered, but when averaged for a longer period, weather oscillations even out and approximate to the long- term average. In this report, the interannual variability is calculated from the unbiased standard deviation stdev of GHI and DNI over 24 years, considering, in the long-term, the normal distribution of the annual sums for n years, where xi is any particular year and ̅ is longterm yearly average. Due to the limited number of years of available data, for the calculation we apply simplified assumption of normal distribution of yearly values: 9 = B7C9 ∑7 E89(E − ̅ ) 6 Tables 6.2 and 6.3 show GHI and DNI values that are to be exceeded at P90 for a consecutive number of years. The variability (varn) for a number of years (n) is calculated from the unbiased standard deviation (stdev): .I1,J 7 = √7 The uncertainty, which characterises 80% probability of occurrence (Uvar), is calculated from the variability (varn), multiplying it with 1.28155: = 1.28155 The lower boundary (negative value) of uncertainty represents 90% probability of exceedance, and it is used for calculating the P90 value. Table 6.2 Annual GHI that should be exceeded with 90% probability in the period of 1 to 10 (25) years Lusaka UNZA│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 3.6 2.5 2.1 1.8 1.6 1.5 1.3 1.3 1.2 1.1 0.7 Uncertainty P90 [±%] 4.6 3.2 2.6 2.3 2.0 1.9 1.7 1.6 1.5 1.4 0.9 Minimum GHI P90 1905 1932 1943 1950 1955 1959 1961 1964 1966 1967 1978 Mount Makulu│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 3.3 2.3 1.9 1.7 1.5 1.4 1.3 1.2 1.1 1.0 0.7 Uncertainty P90 [±%] 4.2 3.0 2.5 2.1 1.9 1.7 1.6 1.5 1.4 1.3 0.8 Minimum GHI P90 1916 1941 1952 1959 1963 1967 1969 1971 1973 1974 1984 Mochipapa │Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 3.0 2.1 1.7 1.5 1.3 1.2 1.1 1.1 1.0 0.9 0.6 Uncertainty P90 [±%] 3.8 2.7 2.2 1.9 1.7 1.6 1.5 1.4 1.3 1.2 0.8 Minimum GHI P90 1963 1986 1996 2002 2006 2009 2011 2013 2015 2016 2025 Longe │Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 2.7 1.9 1.6 1.3 1.2 1.1 1.0 1.0 0.9 0.9 0.5 Uncertainty P90 [±%] 3.4 2.4 2.0 1.7 1.5 1.4 1.3 1.2 1.1 1.1 0.7 Minimum GHI P90 2004 2025 2034 2040 2044 2046 2049 2050 2052 2053 2061 Misamfu│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 2.8 2.0 1.6 1.4 1.3 1.2 1.1 1.0 0.9 0.9 0.6 Uncertainty P90 [±%] 3.6 2.6 2.1 1.8 1.6 1.5 1.4 1.3 1.2 1.1 0.7 Minimum GHI P90 1984 2006 2016 2021 2025 2028 2031 2032 2034 2035 2044 Mutanda│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 2.7 1.9 1.5 1.3 1.2 1.1 1.0 0.9 0.9 0.8 0.5 Uncertainty P90 [±%] 3.4 2.4 2.0 1.7 1.5 1.4 1.3 1.2 1.1 1.1 0.7 Minimum GHI P90 1910 1930 1939 1944 1947 1950 1952 1954 1955 1956 1964 © 2018 Solargis page 65 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Table 6.3 Annual DNI that should be exceeded with 90% probability in the period of 1 to 10 (25) years. Lusaka UNZA│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 8.7 6.2 5.0 4.4 3.9 3.6 3.3 3.1 2.9 2.8 1.7 Uncertainty P90 [±%] 11.2 7.9 6.5 5.6 5.0 4.6 4.2 4.0 3.7 3.5 2.2 Minimum DNI P90 1639 1700 1727 1743 1754 1762 1768 1773 1777 1781 1805 Mount Makulu│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 8.0 5.6 4.6 4.0 3.6 3.3 3.0 2.8 2.7 2.5 1.6 Uncertainty P90 [±%] 10.2 7.2 5.9 5.1 4.6 4.2 3.9 3.6 3.4 3.2 2.0 Minimum DNI P90 1670 1726 1751 1765 1775 1783 1789 1793 1797 1800 1822 Mochipapa│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 7.6 5.4 4.4 3.8 3.4 3.1 2.9 2.7 2.5 2.4 1.5 Uncertainty P90 [±%] 9.8 6.9 5.7 4.9 4.4 4.0 3.7 3.5 3.3 3.1 2.0 Minimum DNI P90 1777 1833 1858 1873 1883 1891 1897 1901 1905 1908 1931 Longe│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 6.5 4.6 3.7 3.2 2.9 2.6 2.4 2.3 2.2 2.0 1.3 Uncertainty P90 [±%] 8.3 5.9 4.8 4.1 3.7 3.4 3.1 2.9 2.8 2.6 1.7 Minimum DNI P90 1830 1878 1899 1912 1921 1927 1932 1936 1940 1943 1962 Misamfu│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 6.5 4.6 3.7 3.2 2.9 2.6 2.4 2.3 2.2 2.0 1.3 Uncertainty P90 [±%] 8.3 5.8 4.8 4.1 3.7 3.4 3.1 2.9 2.8 2.6 1.7 Minimum DNI P90 1617 1660 1679 1690 1698 1703 1708 1711 1714 1717 1734 Mutanda│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 6.3 4.5 3.6 3.2 2.8 2.6 2.4 2.2 2.1 2.0 1.3 Uncertainty P90 [±%] 8.1 5.7 4.7 4.0 3.6 3.3 3.1 2.9 2.7 2.6 1.6 Minimum DNI P90 1613 1655 1673 1684 1692 1697 1701 1705 1708 1710 1727 We can interpret the above Table 6.2 and 6.3 on the example of Lusaka UNZA site: i. GHI interannual variability at P90 of 4.6% has to be considered for any single year at Lusaka UNZA site. In other words, assuming that the long-term average is 1996 kWh/m2, it is expected (with 90% probability) that annual GHI exceeds, at any single year, the value of 1905 kWh/m2. ii. Within a period of three consecutive years, it is expected at P90 that annual average of GHI exceeds value of 1943 kWh/m2; iii. For a period of 25 years, it is expected at 90% probability that due to interannual variability the estimate of the long-term annual DNI average will deviate within the range of ±2.2% in Lusaka UNZA. Thus, assuming that the estimate of the long-term average is 1846 kWh/m2, it can be expected at P90 that due to variability of weather, it should be at least 1805 kWh/m2. It is to be underlined that prediction of the future irradiation is based on the analysis of the recent historical data (period 1994 to 2017). Future weather changes may include man-induced or natural events such as volcano eruptions, which may have impact on this prediction. Based on the existing scientific knowledge [23, 24], an effect of extreme volcano eruptions, with an emission of large amount of stratospheric aerosols, can be estimated on the example of Pinatubo event in 1991 (the second largest volcano eruption in 20th century). It can be expected that in such a case, the annual DNI in the affected year can decrease by approx. 16% or more, compared to the long-term average, still influencing another two consecutive years. In the same way, the volcano eruption of the comparable size may reduce long-term average estimate of DNI by about 4%. The decrease of GHI is much lower; the annual value in the particular year of eruption could be reduced by about 2% compared to the long-term average. © 2018 Solargis page 66 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 6.3 Combined uncertainty In this Chapter, the combined uncertainty of the annual GHI and DNI values is quantified. Taking into account uncertainties of both types of data (satellite and ground measured), the combined effect of two components of the uncertainty of the site-adapted GHI and DNI values has to be considered. 1. Uncertainty of the estimate (Uest) of the annual solar resource values, which is ±4.0% for GHI and ±5.5% for DNI (Chapter 6.1); 2. Interannual variability (Uvar) in any particular year, due to changing weather. In six Zambian sites, it varies from ±3.4% to ±4.6% for GHI and from ±8.1% to ±11.2% for DNI. The uncertainty due to weather variability decreases over the time with square root of the number of years (Chapter 6.2). The two above-mentioned uncertainties combine in Uc (see Glossary), which represents a conservative expectation of the minimum GHI and DNI assuming various number of years N (Tables 6.4 and 6.5). Considering a simplified assumption of normal distribution of the annual values, probability of exceedance can be calculated at different confidence levels. GHI and DNI minimum annual values expected for combined uncertainty in any single year are shown on Figure 6.2 and 6.3. Table 6.4 Combined probability of exceedance of annual GHI for uncertainty of the estimate ±4.0%. Nr. of Uncertainty Interanual Combined Expected minimum │Lusaka UNZA 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 4.0 4.6 6.1 2216 2151 2117 2060 1996 1932 1875 1840 1776 5 4.0 2.0 4.5 2159 2111 2086 2043 1996 1949 1906 1881 1833 10 4.0 1.4 4.3 2150 2105 2081 2041 1996 1951 1911 1887 1842 25 4.0 0.9 4.1 2145 2101 2078 2039 1996 1953 1914 1891 1847 Nr. of Uncertainty Interanual Combined Expected minimum │Mount Makulu 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 4.0 4.2 5.8 2213 2151 2118 2063 2001 1940 1885 1852 1789 5 4.0 1.9 4.4 2162 2115 2090 2048 2001 1955 1913 1888 1841 10 4.0 1.3 4.2 2155 2110 2086 2046 2001 1957 1917 1893 1848 25 4.0 0.8 4.1 2150 2106 2083 2044 2001 1958 1920 1896 1853 Nr. of Uncertainty Interanual Combined Expected minimum │Mochipapa 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 4.0 3.8 5.5 2246 2186 2154 2101 2041 1981 1928 1896 1836 5 4.0 1.7 4.4 2202 2155 2130 2088 2041 1994 1952 1927 1880 10 4.0 1.2 4.2 2196 2151 2126 2086 2041 1996 1956 1932 1886 25 4.0 0.8 4.1 2192 2148 2124 2085 2041 1997 1958 1934 1890 Nr. of Uncertainty Interanual Combined Expected minimum│Longe 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 4.0 3.4 5.3 2275 2216 2185 2133 2076 2018 1966 1935 1877 5 4.0 1.5 4.3 2237 2190 2165 2123 2076 2029 1987 1961 1914 10 4.0 1.1 4.1 2232 2186 2162 2121 2076 2030 1990 1965 1919 25 4.0 0.7 4.1 2229 2184 2160 2120 2076 2031 1991 1968 1923 Nr. of Uncertainty Interanual Combined Expected minimum│Misamfu 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 4.0 3.6 5.4 2260 2201 2170 2117 2059 2000 1948 1916 1857 5 4.0 1.6 4.3 2220 2173 2148 2106 2059 2012 1970 1945 1897 10 4.0 1.1 4.2 2214 2169 2144 2104 2059 2014 1973 1949 1903 25 4.0 0.7 4.1 2211 2166 2142 2103 2059 2015 1975 1951 1907 © 2018 Solargis page 67 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Nr. of Uncertainty Interanual Combined Expected minimum│ Mutanda 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 4.0 3.4 5.3 2167 2112 2082 2033 1978 1923 1874 1844 1789 5 4.0 1.5 4.3 2132 2087 2063 2022 1978 1933 1893 1869 1824 10 4.0 1.1 4.1 2127 2083 2060 2021 1978 1935 1896 1873 1829 25 4.0 0.7 4.1 2123 2081 2058 2020 1978 1936 1898 1875 1832 Table 6.5 Combined probability of exceedance of annual DNI for uncertainty of the estimate ±5.5%. Nr. of Uncertainty Interanual Combined Expected minimum │Lusaka UNZA 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 5.5 11.2 12.5 2263 2141 2076 1967 1846 1725 1616 1551 1428 5 5.5 5.0 7.4 2095 2022 1983 1918 1846 1774 1709 1670 1597 10 5.5 3.5 6.5 2065 2001 1967 1909 1846 1782 1725 1691 1627 25 5.5 2.2 5.9 2045 1987 1955 1904 1846 1788 1736 1705 1647 Nr. of Uncertainty Interanual Combined Expected minimum │Mount Makulu 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 5.5 10.2 11.6 2253 2138 2077 1974 1861 1747 1644 1583 1468 5 5.5 4.6 7.2 2102 2032 1994 1931 1861 1791 1727 1690 1619 10 5.5 3.2 6.4 2076 2013 1979 1923 1861 1798 1742 1708 1645 25 5.5 2.0 5.9 2059 2001 1970 1918 1861 1803 1751 1720 1662 Nr. of Uncertainty Interanual Combined Expected minimum │Mochipapa 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 5.5 9.8 11.2 2371 2253 2191 2086 1969 1853 1748 1686 1568 5 5.5 4.4 7.0 2221 2147 2108 2042 1969 1897 1831 1792 1718 10 5.5 3.1 6.3 2195 2129 2094 2035 1969 1904 1845 1810 1744 25 5.5 2.0 5.8 2178 2117 2084 2030 1969 1909 1854 1822 1761 Nr. of Uncertainty Interanual Combined Expected minimum │Longe 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 5.5 8.3 10.0 2355 2250 2193 2099 1995 1890 1796 1740 1635 5 5.5 3.7 6.6 2235 2165 2127 2065 1995 1925 1863 1825 1755 10 5.5 2.6 6.1 2216 2151 2116 2059 1995 1931 1873 1839 1774 25 5.5 1.7 5.7 2203 2142 2110 2055 1995 1935 1880 1848 1787 Nr. of Uncertainty Interanual Combined Expected minimum│Misamfu 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 5.5 8.3 9.9 2080 1987 1938 1855 1763 1671 1588 1538 1445 5 5.5 3.7 6.6 1975 1913 1880 1824 1763 1701 1646 1613 1551 10 5.5 2.6 6.1 1958 1901 1870 1819 1763 1706 1655 1625 1568 25 5.5 1.7 5.7 1947 1893 1864 1816 1763 1709 1662 1633 1579 Nr. of Uncertainty Interanual Combined Expected minimum│Mutanda 2 years of estimate variability uncertainty [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 5.5 8.1 9.8 2066 1975 1927 1845 1755 1665 1583 1535 1444 5 5.5 3.6 6.6 1965 1903 1871 1816 1755 1694 1640 1607 1545 10 5.5 2.6 6.1 1948 1892 1861 1811 1755 1699 1649 1618 1562 25 5.5 1.6 5.7 1938 1884 1856 1808 1755 1702 1654 1626 1572 © 2018 Solargis page 68 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 This analysis is based on the data representing a history of year 1994 to 2017, and on the expert extrapolation of the related weather variability. This report may not reflect possible man-induced climate change or occurrence of extreme events such as large volcano eruptions in the future (see the last paragraph in Chapter 6.2). Graphical visualisation of Tables 6.4 and 6.5 on the example of Lusaka UNZA site is shown in Figures 6.1 and 6.2, where the expected probabilities of exceedance (different Pxx scenarios) are drawn on the cumulative distribution curve showing yearly GHI and DNI values. 100 P99: 1776 P95: 1840 GHI Value at Pxx P50 P90: 1875 P75 P90 90 P95 P99 80 P75: 1932 70 60 P50: 1996 Pxx 50 40 30 20 10 0 1650 1750 1850 1950 2050 2150 2250 2350 GHI [kWh/m2] Figure 6.1: Expected Pxx values for GHI at Lusaka UNZA site 100 P99: 1428 P95: 1551 DNI Value at Pxx P50 P90: 1616 P75 P90 90 P95 P99 80 P75: 1725 70 60 P50: 1846 Pxx 50 40 30 20 10 0 1250 1350 1450 1550 1650 1750 1850 1950 2050 2150 2250 2350 2450 DNI [kWh/m2] Figure 6.2: Expected Pxx values for DNI at Lusaka UNZA site © 2018 Solargis page 69 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 7 TIME SERIES AND TYPICAL METEOROLOGICAL YEAR DATA 7.1 Delivered data sets This report is accompanied by data sets delivered individually for position of each of six solar meteorological stations in Zambia. The data include (Tables 7.1 and 7.2): • Solar and meteorological measurements, after second level quality assessment (first level was delivered by GeoSUN Africa) representing minimum of 24 months of the measuring campaign; • Time series, representing last 24+ years; • Typical Meteorological Year data, representing last 24 years. The data is delivered in formats ready to use in energy simulation software. This report provides detailed insight of the methodologies and results. Table 7.1 Delivered data characteristics Feature Time coverage Primary time step Delivered files Ground measurements Nov 2015 to Dec 2017 1 minute Quality controlled measurements (GeoSUN Africa) 1- minute time step Model data Jan 1994 to Dec 2017 15 minutes Time series – hourly original model Time series – monthly (Solargis) Time series – yearly Model data Jan 1994 to Dec 2017 15 minutes Time series – hourly site adapted model Time series – monthly (Solargis) Time series – yearly Model data Jan 1994 to Dec 2017 hourly Typical Meteorological Year P50 site adapted model Typical Meteorological Year P90 (Solargis) Table 7.2 Parameters in the delivered site-adapted time series and TMY data (hourly time step) Parameter Acronym Unit Time series TMY P50 TMY P90 2 Global horizontal irradiance GHI W/m X X X 2 Direct normal irradiance DNI W/m X X X 2 Diffuse horizontal irradiance DIF W/m X X X 2 Global tilted irradiance (at optimum angle) GTI W/m X - - Solar azimuth SA ° X X X Solar elevation SE ° X X X Air temperature at 2 metres TEMP °C X X X Wind speed at 10 metres WS m/s X X X Wind direction at 10 metres WD ° X X X Relative humidity RH % X X X Air Pressure AP hPa X X X Precipitable Water PWAT kg/m2 X X X © 2018 Solargis page 70 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 7.2 TMY method The Typical Meteorological Year (TMY) data sets are delivered, together with Solargis time series data and this report. TMY contains hourly data derived from the time series covering complete 24 years (1994 to 2017). The data history of 24 years is compressed into one year (Figure 7.1 to 7.4) following two criteria: • Minimum difference between statistical characteristics (annual average, monthly averages) of TMY and long-term time series. This criterion is given about 80% weighting. • Maximum similarity of monthly Cumulative Distribution Functions (CDF) of TMY and full-time series, so that occurrence of typical hourly values is well represented for each month. This criterion is given about 20% weighting. TMY P50 data set is constructed on the monthly basis. For each month, the long-term average monthly value and cumulative distribution for each parameter is calculated: Direct Normal Irradiance (DNI), Global Horizontal Irradiance (GHI), Diffuse Horizontal Irradiance (DIF) and Air Temperature (TEMP). Following the monthly data for each individual year from the set of 24 years are compared to the long-term parameters. The monthly data from the year, which resembles the long-term parameters more closely, is selected. The procedure is repeated for all 12 months, and the TMY is constructed by concatenating the selected months into one artificial (but representative) year. The method for calculation P90 data set is based on the TMY P50 method. It has been modified in a way of how a candidate month is selected. The search for set of twelve candidates is repeated in iteration until a condition of minimization of difference between annual P90 value and annual average of new TMY is reached (instead of minimization of differences in monthly means and CDFs, as applied in P50 case). Once the selection converges to minimum difference, the TMY is created by concatenation of selected months. The P90 annual values are calculated for each confidence limit − from the combined uncertainty of estimate and inter-annual variability, which can occur in any year (Chapter 6.3). To derive TMY that fits specific needs of the selected energy application the different weights are given to individual parameters – thus highlighting important properties. In solar energy applications, the higher importance is given to GHI and DNI. In assembling TMY P50, the values of DNI, GHI, DIF and TEMP are only considered, where the weights are set as follows: 0.9 is given to DNI, 0.3 to GHI, 0.02 to diffuse horizontal irradiance, and 0.07 to air temperature (divided by the total of 1.29). To derive solar resource parameters with the hourly time step, the original satellite data with time resolution of 15-minutes are aggregated by time integration. The meteorological parameters are available in the original 1-hourly time step. The TMY datasets were constructed from solar radiation and meteorological data (Chapters 4 and 5). Time zone was adjusted to Central Africa Time CAT (UTC +02:00). More about the Solargis TMY method in [25]. 7.3 Results Two data sets are derived from the Solargis historical time series for the six sites: P50 and P90. In graphs and tables below we show the values for the example of Lusaka UNZA meteorological site. Important note: Due to the inherent features of the underlying methods, monthly values in TMY data set in Tables 7.3 to 7.6 do not fit to the values generated from full time series (Figures 7.1 to 7.4). © 2018 Solargis page 71 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Table 7.3 Monthly and yearly long-term GHI averages as calculated from time series and from TMY representing P50, and P90 cases at Lusaka UNZA site Global Horizontal 1 2 3 4 5 6 7 8 9 10 11 12 Year Irradiation [kWh/m2] Time series (24 years) 161 144 159 156 158 140 150 175 193 206 183 172 1996 TMY for P50 case 163 145 155 157 156 139 150 176 193 207 182 173 1996 TMY for P90 case 130 103 145 153 157 139 149 172 189 203 181 154 1875 Table 7.4 Monthly and yearly long-term DNI averages as calculated from time series and from TMY representing P50, and P90 cases at Lusaka UNZA site Direct Normal 1 2 3 4 5 6 7 8 9 10 11 12 Year Irradiation [kWh/m2] Time series (24 years) 94 89 115 158 204 188 191 194 181 177 141 113 1846 TMY for P50 case 92 89 112 162 202 188 191 192 183 179 144 111 1846 TMY for P90 case 56 32 78 155 202 186 180 182 176 162 128 81 1616 Table 7.5 Monthly and yearly long-term DIF averages as calculated from time series and from TMY representing P50, and P90 cases at Lusaka UNZA site Diffused Horizontal 1 2 3 4 5 6 7 8 9 10 11 12 Year Irradiation [kWh/m2] Time series (24 years) 88 75 75 52 35 32 36 47 60 70 74 84 729 TMY for P50 case 90 74 74 52 34 32 36 48 60 70 71 86 728 TMY for P90 case 85 76 83 51 32 31 40 49 59 75 79 90 751 Table 7.6 Monthly and yearly long-term TEMP averages as calculated from time series and from TMY representing P50, and P90 cases at Lusaka UNZA site Air temperature [°C] 1 2 3 4 5 6 7 8 9 10 11 12 Year Time series (24 years) 19.9 19.5 19.1 17.9 15.9 14.0 13.6 16.4 20.2 22.6 22.8 20.8 18.6 TMY for P50 case 19.3 19.3 20.2 17.0 15.5 13.3 15.0 16.6 20.2 22.3 23.6 21.7 18.7 TMY for P90 case 18.3 18.3 18.1 16.2 15.9 13.4 12.8 15.8 19.5 23.1 22.6 20.6 17.9 As an example of interpretation of the tables above, the TMY P50 and P90 data can be described as follows: 1. P50 TMY data set represents, for each month, the average climate conditions and the most representative cumulative distribution function, therefore extreme situations (e.g. extremely cloudy weather) are not represented in this dataset. The long-term annual summary of GHI and DNI are considered as the most critical parameters to consider, and in this data set P50 GHI value is 1996 kWh/m2 and DNI value is 1846 kWh/m2. 2. P90 TMY data set represents for each month the climate conditions, which after summarization GHI and DNI for the whole year results in the value equal or close to P90 derived by the analysis of uncertainty of the estimate and of the interannual variability for any single year (Chapter 6.3). Thus, TMY for P90 represents generally a conservative estimate, i.e. a year with the long-term value of GHI of 1875 kWh/m2 and DNI of 1616 kWh/m2. © 2018 Solargis page 72 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 250 200 GHI [kWh/m 2 ] 150 100 50 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Time series TMY P50 TMY P90 Figure 7.1: GHI monthly values derived from time series and TMY P50 and P90 at Lusaka UNZA site. 250 200 DNI [kWh/m 2 ] 150 100 50 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Time series TMY P50 TMY P90 Figure 7.2: DNI monthly values derived from time series and TMY P50 and P90 at Lusaka UNZA site. 250 200 DIF [kWh/m 2 ] 150 100 50 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Time series TMY P50 TMY P90 Figure 7.3: DIF monthly values derived from time series and TMY P50 and P90 at Lusaka UNZA site. © 2018 Solargis page 73 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 30.0 25.0 TEMP [°C] 20.0 15.0 10.0 5.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Time series TMY P50 TMY P90 Figure 7.4: TEMP monthly values derived from time series and TMY P50 and P90 at Lusaka UNZA site It is important to note that the data reduction in the TMY data set is not possible without loss of information contained in the original multiyear time series. Therefore, time series data are considered as the most accurate reference suitable for the statistical analysis of solar resource and meteorological parameters of the site. Figure 7.5: Seasonal profile of GHI, DNI and DIF for Typical Meteorological Year P50 Lusaka UNZA site: X-axis – day of the year; Y-axis – solar irradiance W/m2 © 2018 Solargis page 74 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 7.6: Snapshot of Typical Meteorological Year for P50 for Lusaka UNZA site © 2018 Solargis page 75 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 8 CONCLUSIONS This report accompanies delivery of measured solar resource and meteorological data for six sites where solar meteorological stations have been installed during November 2015 and operated for a period of 25 months until December 2017. The measured data is a result of systematic work on (i) setting up a network of solar meteorological stations with high-standard equipment and on (ii) implementation of rigorous practices in operation and maintenance of solar equipment. Well-linked to this infrastructure is satellite-based solar radiation model Solargis, which has proven quality and reliability of time series and derived site-specific data products. The measured data is used in site-adaptation of the Solargis model. As a result, reliable Solargis historical time series and TMY data is computed for six specific sites, and provided in formats ready to use in standard photovoltaic energy simulation software. The site adapted Solargis data prepared for six sites is also part of the delivery. Role of solar measuring stations in maintaining sustainable solar data infrastructure Receiving data from a number of high quality measuring stations enables an improved understanding of the geographical and temporal variability of solar resource in Zambia and a wider region. Even though the model adaptation reduced its uncertainty, it is important to maintain the operation of the solar meteorological stations, with special focus on the following cases: • For new sites, relevant to any larger solar power project, it is important to set-up and operate a solar meteorological station to reduce uncertainty to an achievable minimum of the site-specific long-term estimates. • For existing six sites in Zambia, the meteorological stations together with satellite data make it possible to maintain high quality and bankability of solar resource and meteorological data for sustainable performance assessment of solar power plants in the region. • Keeping solar measuring stations is of strategic importance to maintain quality of satellite models and of solar power forecasts in future. Reduced uncertainty of Solargis model The uncertainty of the Solargis model for site-specific estimates has been reduced from the original range of ±12% to ±18% for DNI yearly values to approximately ±5.5% for accuracy-enhanced values. For site-specific yearly GHI estimates, the uncertainty reduction is seen from the original range of ±7.5% to ±10% to approximately ±4.0% for the accuracy-enhanced values. Besides reducing systematic deviation (bias), the model adaptation for sites also results in the improvement of other data quality indicators such as reducing random deviation (measurable by Root Mean Square Deviation) and by improving the probability distribution of hourly values (measurable by Kolmogorov-Smirnoff Index). Higher- quality DNI and GHI data have substantial benefits in energy simulation, which can in turn be used for more reliable financial predictions. © 2018 Solargis page 76 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 ANNEX 1: SITE RELATED DATA STATISTICS Yearly summaries of solar and meteorological parameters Statistics for site-adapted Solargis model yearly values representing 24 years (1994 to 2017). 6.5 2374 Average yearly sum of Global Horizontal Irradiation [kWh/m2] Average daily sum of Global Horizontal Irradiation [kWh/m2] 6.0 2192 5.5 2009 5.0 1826 4.5 1644 4.0 1461 3.5 1278 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Lusaka 4.6% Mount Makulu 4.2% Mochipapa 3.8% Longe 3.4% Misamfu 3.6% Mutanda 3.4% Figure I: Interannual variability of site-adapted yearly GHI [kWh/m2]. Annual average (avg, solid line) and standard deviation (value behind the names of sites). 6.5 2374 Average yearly sum of Direct Normal Irradiation [kWh/m2] Average daily sum of Direct Normal Irradiation [kWh/m2] 6.0 2192 5.5 2009 5.0 1826 4.5 1644 4.0 1461 3.5 1278 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Lusaka 11.2% Mount Makulu 10.2% Mochipapa 9.8% Longe 8.3% Misamfu 8.3% Mutanda 8.1% Figure II: Interannual variability of site-adapted yearly DNI [kWh/m2]. Annual average (avg, solid line) and standard deviation (value behind the names of sites). © 2018 Solargis page 77 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 25.0 22.5 Monthly air temperature [°C] 20.0 17.5 15.0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Lusaka UNZA Mount Makulu Mochipapa Longe Misamfu Mutanda Figure III: Interannual variability of yearly TEMP [°C]. Annual average (avg, solid line). © 2018 Solargis page 78 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Monthly summaries of solar and meteorological parameters The graphs compare monthly site-adapted time series from Solargis model compared to long-term averages for a historical period 1994 to 2017. 9.0 9.0 Lusaka UNZA Mount Makulu 7.5 7.5 Daily sums of GHI [kWh/m2] Daily sums of GHI [kWh/m2] 6.0 6.0 4.5 4.5 3.0 3.0 1.5 1.5 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year 9.0 9.0 Mochipapa Longe 7.5 7.5 Daily sums of GHI [kWh/m2] Daily sums of GHI [kWh/m2] 6.0 6.0 4.5 4.5 3.0 3.0 1.5 1.5 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year 9.0 9.0 Misamfu Mutanda 7.5 7.5 Daily sums of GHI [kWh/m2] Daily sums of GHI [kWh/m2] 6.0 6.0 4.5 4.5 3.0 3.0 1.5 1.5 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year Figure IV: GHI monthly averages [kWh/m2]. Monthly average shown as solid line; min/max monthly values as boundary lines; last 12 months in red. © 2018 Solargis page 79 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 9.0 9.0 Lusaka UNZA Mount Makalu 7.5 7.5 6.0 Daily sums of DNI [kWh/m2] 6.0 Daily sums of DNI [kWh/m2] 4.5 4.5 3.0 3.0 1.5 1.5 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year 9.0 9.0 Mochipapa Longe 7.5 7.5 6.0 Daily sums of DNI [kWh/m2] 6.0 Daily sums of DNI [kWh/m2] 4.5 4.5 3.0 3.0 1.5 1.5 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year 9.0 9.0 Misamfu Mutanda 7.5 7.5 6.0 Daily sums of DNI [kWh/m2] 6.0 Daily sums of DNI [kWh/m2] 4.5 4.5 3.0 3.0 1.5 1.5 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year Figure V: DNI monthly averages [kWh/m2]. Monthly average shown as solid line; min/max monthly values as boundary lines; last 12 months shown in red. © 2018 Solargis page 80 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 40.0 40.0 Lusaka UNZA Mount Makulu Monthly air temperature [°C] Monthly air temperature [°C] 30.0 30.0 20.0 20.0 10.0 10.0 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year 40.0 40.0 Mochipapa Longe Monthly air temperature [°C] Monthly air temperature [°C] 30.0 30.0 20.0 20.0 10.0 10.0 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year 40.0 40.0 Misamfu Mutanda Monthly air temperature [°C] Monthly air temperature [°C] 30.0 30.0 20.0 20.0 10.0 10.0 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year Figure VI: TEMP monthly averages [°C]. Monthly average shown as solid line; min/max monthly values as boundary lines; last 12 months shown in red. © 2018 Solargis page 81 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Frequency of occurrence of GHI and DNI daily model values for a period 1994 to 2017 The histograms below show occurrence statistics of daily values derived from the satellite-based site-adapted time series − GHI and DNI parameters. The time covered in the graphs below is 24 complete calendar years (1994 to 2017). The occurrence is calculated separately for each month. 50 12 12 50 12 50 January February March Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 12 50 12 50 April May June Percentage of days 10 10 40 10 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 50 12 12 50 12 50 July August September Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 50 12 50 12 12 50 October November December Percentage of days 10 10 10 40 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 10 10 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure VII: Histograms of daily summaries of Global Horizontal Irradiation in Lusaka UNZA. 12 50 50 12 12 50 January February March Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 50 12 50 12 12 50 April May June Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 50 12 12 50 July August September Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 50 12 12 50 50 12 October November December Percentage of days 10 10 10 40 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 10 10 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure VIII: Histograms of daily summaries of Global Horizontal Irradiation in Mount Makulu. © 2018 Solargis page 82 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 12 50 12 50 50 12 January February March Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 50 12 12 50 April May June Percentage of days 10 10 40 10 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 50 12 12 50 July August September Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 50 12 12 50 12 50 October November December Percentage of days 10 10 10 40 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 10 10 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure IX: Histograms of daily summaries of Global Horizontal Irradiation in Mochipapa. 12 50 50 12 12 50 January February March Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 50 12 12 50 April May June Percentage of days 10 10 40 10 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 50 12 12 50 12 50 July August September Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 50 12 50 12 12 50 October November December Percentage of days 10 10 10 40 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 10 10 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure X: Histograms of daily summaries of Global Horizontal Irradiation in Longe. © 2018 Solargis page 83 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 12 50 12 50 50 12 January February March Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 50 12 12 50 April May June Percentage of days 10 10 40 10 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 50 12 12 50 July August September Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 12 50 12 50 October November December Percentage of days 10 10 10 40 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 10 10 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure XI: Histograms of daily summaries of Global Horizontal Irradiation in Misamfu. 12 50 12 50 50 12 January February March Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 12 50 50 12 April May June Percentage of days 10 10 40 10 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 50 12 50 50 12 July August September Percentage of days 10 40 10 40 10 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 2 10 2 10 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 50 12 12 50 50 12 October November December Percentage of days 10 10 10 40 40 40 8 8 8 30 30 30 6 6 6 20 20 20 4 4 4 10 10 10 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure XII: Histograms of daily summaries of Global Horizontal Irradiation in Mutanda. Figures VII to XII show histograms of daily GHI summaries for each month as calculated from Solargis time series representing the years 1994 to 2017. The distribution of daily values is not symmetric: median is drawn by the vertical line, and percentiles P10, P25, and P75, and P90 are displayed with dark grey and light grey colour bands, respectively. The percentiles P10 and P90 show 80% occurrence of daily values within each month and percentiles P25 and P75 show 50% occurrence. © 2018 Solargis page 84 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 15 15 15 January February March Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 April May June Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 July August September Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 October November December Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XIII: Histograms of daily summaries of Direct Normal Irradiation in Lusaka UNZA. 15 15 15 January February March Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 April May June Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 July August September Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 October November December Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XIV: Histograms of daily summaries of Direct Normal Irradiation in Mount Makulu. © 2018 Solargis page 85 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 15 15 15 January February March Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 April May June Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 July August September Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 October November December Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XV: Histograms of daily summaries of Direct Normal Irradiation in Mochipapa. 15 15 15 January February March Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 April May June Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 July August September Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 October November December Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XVI: Histograms of daily summaries of Direct Normal Irradiation in Longe. © 2018 Solargis page 86 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 15 15 15 January February March Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 April May June Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 July August September Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 October November December Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XVII: Histograms of daily summaries of Direct Normal Irradiation in Misamfu. 15 15 15 January February March Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 April May June Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 July August September Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15 15 15 October November December Percentage of days 10 10 10 5 5 5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XVIII: Histograms of daily summaries of Direct Normal Irradiation in Mutanda. Figures XIII to XVIII show histograms of daily DNI summaries for each month as calculated from Solargis time series representing the years 1994 to 2017. The distribution of daily values is not symmetric: median is drawn by the vertical line, and percentiles P10, P25, and P75, and P90 are displayed with dark grey and light grey colour bands, respectively. The percentiles P10 and P90 show 80% occurrence of daily values within each month and percentiles P25 and P75 show 50% occurrence. © 2018 Solargis page 87 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Frequency of occurrence of GHI and DNI 15-minute model values for a period 1994 to 2017 The histograms below show occurrence statistics of 15-minute values derived from the satellite-based site- adapted time series − GHI and DNI parameters. The time covered in the graphs below is 24 complete calendar years (1994 to 2017). The occurrence is calculated separately for each month. Figure XIX: Histograms and cumulative distribution function of 15-minute GHI in Lusaka UNZA Figure XX: Histograms and cumulative distribution function of 15-minute GHI in Mount Makulu © 2018 Solargis page 88 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XXI: Histograms and cumulative distribution function of 15-minute GHI in Mochipapa Figure XXII: Histograms and cumulative distribution function of 15-minute GHI in Longe © 2018 Solargis page 89 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XXIII: Histograms and cumulative distribution function of 15-minute GHI in Misamfu Figure XXIV: Histograms and cumulative distribution function of 15-minute GHI in Mutanda Figures XIX to XXIV show monthly histograms (bars) and cumulative distribution (line) of 15-minute GHI values, calculated from Solargis time series. The values represent the occurrence of GHI values within 50 W/m2 bins, ranging from 0 to 1100 W/m2. © 2018 Solargis page 90 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XXV: Histograms and cumulative distribution function of 15-minute DNI in Lusaka UNZA Figure XXVI: Histograms and cumulative distribution function of 15-minute DNI in Mount Makulu © 2018 Solargis page 91 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XXVII: Histograms and cumulative distribution function of 15-minute DNI in Mochipapa Figure XXVIII: Histograms and cumulative distribution function of 15-minute DNI in Longe © 2018 Solargis page 92 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XXIX: Histograms and cumulative distribution function of 15-minute DNI in Misamfu Figure XXX: Histograms and cumulative distribution function of 15-minute DNI in Mutanda Figures XXV to XXX show monthly histograms (bars) and cumulative distribution (line) of 15-minute DNI values, calculated from Solargis time series. The values represent the occurrence of DNI values within 50 W/m2 bins, ranging from 0 to 1100 W/m2. © 2018 Solargis page 93 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Frequency of occurrence of GHI and DNI measured and model values representing two years of ground measurements. Figures XXXI to XLII show histograms comparing the measured values with the model GHI and DNI data. The period covered in these histograms is 24+ months (two full years of data, i.e. from 1 Dec 2015 to 31 Dec 2017): • 1-minute measured vs. 15-min satellite-based model values • 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values • Daily measured (aggregated from 1-min) vs. daily satellite-based model values Aggregation process deals with the missing values in the ground measurement in three steps: 1. Only those 1-minute measured data values that passed through quality control (Chapter 3.3) is taken into account (satellite time series does not have gaps.); 2. Aggregation of 1-minute measured data values into 15-minute slots (equivalent to satellite time slots) is applied if more than 15 valid data-points is available, otherwise the 15-minute data slot is ignored in further statistical comparison; 3. Daily aggregation of measured data represents the same 15-minute time slots in a day (passing through the two steps above), as those in the satellite-based data. Incorrect data slots found in the measurements are excluded in both the measured and model data. Figure XXXI: Measured vs. satellite-based GHI values in Lusaka UNZA 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXXII: Measured vs. satellite-based GHI values in Mount Makulu 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values © 2018 Solargis page 94 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XXXIII: Measured vs. satellite-based GHI values in Mochipapa 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXXIV: Measured vs. satellite-based GHI values in Longe 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXXV: Measured vs. satellite-based GHI values in Misamfu 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXXVI: Measured vs. satellite-based GHI values in Mutanda 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values © 2018 Solargis page 95 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XXXVII: Measured vs. satellite-based DNI values in Lusaka UNZA 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXXVIII: Measured vs. satellite-based DNI values in Mount Makulu 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXXIX: Measured vs. satellite-based DNI values in Mochipapa 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XL: Measured vs. satellite-based DNI values in Longe 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values © 2018 Solargis page 96 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XLI: Measured vs. satellite-based DNI values in Misamfu 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XLII: Measured vs. satellite-based DNI values in Mutanda 1-minute measured vs. 15-min satellite-based values. 15-minute measured (aggregated from 1-min) vs. 15-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values © 2018 Solargis page 97 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Frequency of occurrence of GHI and DNI ramps Figures XLIII to LIV show histograms of instantaneous changes (ramps) calculated from the measurements and compared to the instantaneous changes calculated for the model data. Figures show both negative (-) and positive (+) changes. Two versions for GHI and DNI are shown: • Ramps calculated from 1-minute measured values compared to ramps calculated from 15-minute satellite-based data (figure on the left) • Ramps calculated from 15-minute aggregated valid measurement compared to ramps calculated from 15-minute satellite-based data (figure on the right). Occurrence of gaps in the measurements is managed in the same way as described about in this Chapter: 1. For measurements, only those 1-minute data values (measurements) that passed through quality control (Chapter 3.3) is taken into account (satellite time series does not have gaps.); 2. For measurements, the aggregation (averaging) of 1-minute measured data values into 15-minute slots (equivalent to satellite time slots) is applied if more than 15 valid data-points is available, otherwise the 15-minute data slot is ignored in further statistical comparison; Figure XLIII: 1-minute and 15-minute GHI ramps (measured and satellite data) at Lusaka UNZA. Figure XLIV: 1-minute and 15-minute GHI ramps (measured and satellite data) at Mount Makulu © 2018 Solargis page 98 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XLV: 1-minute and 15-minute GHI ramps (measured and satellite data) at Mochipapa Figure XLVI: 1-minute and 15-minute GHI ramps (measured and satellite data) at Longe Figure XLVII: 1-minute and 15-minute GHI ramps (measured and satellite data) at Misamfu © 2018 Solargis page 99 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XLVIII: 1-minute and 15-minute GHI ramps (measured and satellite data) at Mutanda Figure XLIX: 1-minute and 15-minute DNI ramps (measured and satellite data) at Lusaka UNZA Figure L: 1-minute and 15-minute DNI ramps (measured and satellite data) at Mount Makulu © 2018 Solargis page 100 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure LI: 1-minute and 15-minute DNI ramps (measured and satellite data) at Mochipapa Figure LII: 1-minute and 15-minute DNI ramps (measured and satellite data) at Longe Figure LIII: 1-minute and 15-minute DNI ramps (measured and satellite data) at Misamfu © 2018 Solargis page 101 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure LIV: 1-minute and 15-minute DNI ramps (measured and satellite data) at Mutanda © 2018 Solargis page 102 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 LIST OF FIGURES Figure 2.1: Position of solar meteorological stations in Zambia .......................................................................11 Figure 3.1 Results of DNI (CHP1) and GHI (CMP10) quality control at UNZA Lusaka. .....................................19 Figure 3.2 Results of DNI (RSR) and GHI (RSR) quality control at UNZA Lusaka. ...........................................20 Figure 3.3 Difference of DNI and GHI between two sensors – UNZA Lusaka. .................................................21 Figure 3.4 Results of GHI and DNI quality control in Mount Makulu.................................................................23 Figure 3.5 Effect of RSR alignment issues – drop of DNI in Mount Makulu ......................................................24 Figure 3.6 Effect of morning shading - Mount Makulu .....................................................................................24 Figure 3.7 Systematic difference between GHI from CMP10 and RSR - Mount Makulu ...................................24 Figure 3.8 Results of GHI and DNI quality control − Mochipapa. .....................................................................27 Figure 3.9 Shading effects on GHI and DNI in Mochipapa ..............................................................................28 Figure 3.10 Systematic difference between GHI from CMP10 and RSR − Mochipapa .....................................28 Figure 3.11 Asymmetry of GHI diurnal profiles from CMP10 and RSR − Mochipapa ........................................28 Figure 3.12 Results of GHI and DNI quality control − Longe ...........................................................................31 Figure 3.13 Systematic shading effects on GHI and DNI in Longe...................................................................31 Figure 3.14 Systematic difference between GHI from CMP10 and RSR - Longe. ............................................32 Figure 3.15 Morning dew effect on GHI and DNI measurements in Longe. ......................................................32 Figure 3.16 Results of GHI and DNI quality control − Misamfu. .......................................................................34 Figure 3.17 Systematic difference between GHI from CMP10 and RSR - Misamfu. .........................................35 Figure 3.18 Results of GHI and DNI quality control − Mutanda........................................................................37 Figure 3.19 Systematic difference between GHI from CMP10 and RSR - Mutanda. .........................................38 Figure 3.20 Morning dew effect on GHI measurements in Mutanda. ...............................................................38 Figure 4.1: Sites with solar meteorological stations used for site adaptation of Solargis model in Zambia.........41 Figure 4.2: Correction of DNI and GHI hourly values for UNZA Lusaka. ..........................................................45 Figure 4.3: Correction of DNI and GHI hourly values for Mount Makulu ...........................................................46 Figure 4.4: Correction of DNI and GHI hourly values for Mochipapa. ...............................................................47 Figure 4.5: Correction of DNI and GHI hourly values for Longe. ......................................................................48 Figure 4.6: Correction of DNI and GHI hourly values for Misamfu. ..................................................................49 Figure 4.7: Correction of DNI and GHI hourly values for Mutanda. ..................................................................50 Figure 4.8: Comparison of Solargis original and site-adapted data for UNZA Lusaka site. ...............................51 Figure 5.1: Scatterplots of air temperature at 2 m at Lusaka UNZA meteorological station. ..............................53 Figure 5.2: Scatterplots of air temperature at 2 m at Mount Makulu meteorological station. ..............................54 Figure 5.3: Scatterplots of air temperature at 2 m at Mochipapa meteorological station. ..................................54 Figure 5.4: Scatterplots of air temperature at 2 m at Longe meteorological station...........................................55 Figure 5.5: Scatterplots of air temperature at 2 m at Misamfu meteorological station. ......................................55 Figure 5.6: Scatterplots of air temperature at 2 m at Mutanda meteorological station. ......................................56 Figure 5.7: Scatterplots of relative humidity at 2 m at Lusaka UNZA meteorological station. ............................57 Figure 5.8: Scatterplots of relative humidity at 2 m at Mount Makulu meteorological station. ............................57 Figure 5.9: Scatterplots of relative humidity at 2 m at Mochipapa meteorological station. .................................58 Figure 5.10: Scatterplots of relative humidity at 2 m at Longe meteorological station. ......................................58 Figure 5.11: Scatterplots of relative humidity at 2 m at Misamfu meteorological station. ...................................59 © 2018 Solargis page 103 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure 5.12: Scatterplots of relative humidity at 2 m at Mutanda meteorological station. ..................................59 Figure 5.13: Scatterplots of wind speed at Lusaka UNZA meteorological station. ............................................60 Figure 5.14: Scatterplots of wind speed at Mount Makulu meteorological station. ............................................61 Figure 5.15: Scatterplots of wind speed at Mochipapa meteorological station. .................................................61 Figure 5.16: Scatterplots of wind speed at Longe meteorological station. ........................................................62 Figure 5.17: Scatterplots of wind speed at Misamfu meteorological station......................................................62 Figure 5.18: Scatterplots of wind speed at Mutanda meteorological station. ....................................................63 Figure 6.1: Expected Pxx values for GHI at Lusaka UNZA site .......................................................................69 Figure 6.2: Expected Pxx values for DNI at Lusaka UNZA site........................................................................69 Figure 7.1: GHI monthly values derived from time series and TMY P50 and P90.............................................73 Figure 7.2: DNI monthly values derived from time series and TMY P50 and P90 .............................................73 Figure 7.3: DIF monthly values derived from time series and TMY P50 and P90 .............................................73 Figure 7.4: TEMP monthly values derived from time series and TMY P50 and P90 .........................................74 Figure 7.5: Seasonal profile of GHI, DNI and DIF for Typical Meteorological Year P50 ....................................74 Figure 7.6: Snapshot of Typical Meteorological Year for P50 for Lusaka UNZA site.........................................75 Figure I: Interannual variability of site-adapted yearly GHI [kWh/m2]................................................................77 Figure II: Interannual variability of site-adapted yearly DNI [kWh/m2]. ..............................................................77 Figure III: Interannual variability of yearly TEMP [°C]. .....................................................................................78 Figure IV: GHI monthly averages [kWh/m2]. ...................................................................................................79 Figure V: DNI monthly averages [kWh/m2]. ....................................................................................................80 Figure VI: TEMP monthly averages [°C]. ........................................................................................................81 Figure VII: Histograms of daily summaries of Global Horizontal Irradiation in Lusaka UNZA. ...........................82 Figure VIII: Histograms of daily summaries of Global Horizontal Irradiation in Mount Makulu. ..........................82 Figure IX: Histograms of daily summaries of Global Horizontal Irradiation in Mochipapa. .................................83 Figure X: Histograms of daily summaries of Global Horizontal Irradiation in Longe. .........................................83 Figure XI: Histograms of daily summaries of Global Horizontal Irradiation in Misamfu. .....................................84 Figure XII: Histograms of daily summaries of Global Horizontal Irradiation in Mutanda. ...................................84 Figure XIII: Histograms of daily summaries of Direct Normal Irradiation in Lusaka UNZA. ................................85 Figure XIV: Histograms of daily summaries of Direct Normal Irradiation in Mount Makulu. ...............................85 Figure XV: Histograms of daily summaries of Direct Normal Irradiation in Mochipapa. .....................................86 Figure XVI: Histograms of daily summaries of Direct Normal Irradiation in Longe. ...........................................86 Figure XVII: Histograms of daily summaries of Direct Normal Irradiation in Misamfu. .......................................87 Figure XVIII: Histograms of daily summaries of Direct Normal Irradiation in Mutanda.......................................87 Figure XIX: Histograms and cumulative distribution function of 15-minute GHI in Lusaka UNZA ......................88 Figure XX: Histograms and cumulative distribution function of 15-minute GHI in Mount Makulu .......................88 Figure XXI: Histograms and cumulative distribution function of 15-minute GHI in Mochipapa ...........................89 Figure XXII: Histograms and cumulative distribution function of 15-minute GHI in Longe .................................89 Figure XXIII: Histograms and cumulative distribution function of 15-minute GHI in Misamfu .............................90 Figure XXIV: Histograms and cumulative distribution function of 15-minute GHI in Mutanda ............................90 Figure XXV: Histograms and cumulative distribution function of 15-minute DNI in Lusaka UNZA .....................91 Figure XXVI: Histograms and cumulative distribution function of 15-minute DNI in Mount Makulu ....................91 Figure XXVII: Histograms and cumulative distribution function of 15-minute DNI in Mochipapa ........................92 Figure XXVIII: Histograms and cumulative distribution function of 15-minute DNI in Longe ..............................92 © 2018 Solargis page 104 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Figure XXIX: Histograms and cumulative distribution function of 15-minute DNI in Misamfu.............................93 Figure XXX: Histograms and cumulative distribution function of 15-minute DNI in Mutanda .............................93 Figure XXXI: Measured vs. satellite-based GHI values in Lusaka UNZA .........................................................94 Figure XXXII: Measured vs. satellite-based GHI values in Mount Makulu ........................................................94 Figure XXXIII: Measured vs. satellite-based GHI values in Mochipapa ............................................................95 Figure XXXIV: Measured vs. satellite-based GHI values in Longe...................................................................95 Figure XXXV: Measured vs. satellite-based GHI values in Misamfu ................................................................95 Figure XXXVI: Measured vs. satellite-based GHI values in Mutanda ...............................................................95 Figure XXXVII: Measured vs. satellite-based DNI values in Lusaka UNZA ......................................................96 Figure XXXVIII: Measured vs. satellite-based DNI values in Mount Makulu .....................................................96 Figure XXXIX: Measured vs. satellite-based DNI values in Mochipapa............................................................96 Figure XL: Measured vs. satellite-based DNI values in Longe.........................................................................96 Figure XLI: Measured vs. satellite-based DNI values in Misamfu ....................................................................97 Figure XLII: Measured vs. satellite-based DNI values in Mutanda ...................................................................97 Figure XLIII: 1-minute and 15-minute GHI ramps (measured and satellite data) at Lusaka UNZA. ...................98 Figure XLIV: 1-minute and 15-minute GHI ramps (measured and satellite data) at Mount Makulu....................98 Figure XLV: 1-minute and 15-minute GHI ramps (measured and satellite data) at Mochipapa .........................99 Figure XLVI: 1-minute and 15-minute GHI ramps (measured and satellite data) at Longe................................99 Figure XLVII: 1-minute and 15-minute GHI ramps (measured and satellite data) at Misamfu ...........................99 Figure XLVIII: 1-minute and 15-minute GHI ramps (measured and satellite data) at Mutanda ........................ 100 Figure XLIX: 1-minute and 15-minute DNI ramps (measured and satellite data) at Lusaka UNZA .................. 100 Figure L: 1-minute and 15-minute DNI ramps (measured and satellite data) at Mount Makulu ....................... 100 Figure LI: 1-minute and 15-minute DNI ramps (measured and satellite data) at Mochipapa ........................... 101 Figure LII: 1-minute and 15-minute DNI ramps (measured and satellite data) at Longe ................................. 101 Figure LIII: 1-minute and 15-minute DNI ramps (measured and satellite data) at Misamfu ............................. 101 Figure LIV: 1-minute and 15-minute DNI ramps (measured and satellite data) at Mutanda ............................ 102 © 2018 Solargis page 105 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 LIST OF TABLES Table 1.1 Characteristics of the delivered data .......................................................................................... 9 Table 1.2 Parameters in the delivered time series (TS) and TMY data (hourly time step) ............................10 Table 2.1 Solar meteorological stations installed in Zambia: Overview ......................................................11 Table 3.1 Overview information on measurement stations operated in the region ......................................13 Table 3.2 Solar instruments installed at the solar meteorological stations .................................................13 Table 3.3 Meteorological instruments installed at the solar meteorological stations ..................................13 Table 3.4 Technical parameters and accuracy class of the instruments at Tier 1 and Tier 2 stations ..........14 Table 3.5 Overview information on solar meteorological stations operating in the region ...........................14 Table 3.6 Data recovery statistics of the measurement campaign .............................................................15 Table 3.7 Period of measurements analysed in this report ........................................................................15 Table 3.8 Meteorological stations maintenance and instruments field verification .....................................15 Table 3.9 Results of field instruments verification performed by GeoSUN Africa on April 2018 ...................17 Table 3.10 Occurrence of data readings for UNZA Lusaka meteorological station .......................................18 Table 3.11 Excluded ground measurements after quality control (Sun above horizon) in UNZA Lusaka ........18 Table 3.12 Quality control summary – UNZA Lusaka ..................................................................................21 Table 3.13 Occurrence of data readings for Mount Makulu meteorological station ......................................22 Table 3.14 Excluded ground measurements after quality control (Sun above horizon) in Mount Makulu .......22 Table 3.15 Quality control summary – Mount Makulu .................................................................................25 Table 3.16 Occurrence of data readings for Mochipapa meteorological station ...........................................26 Table 3.17 Excluded ground measurements after quality control (Sun above horizon) in Mochipapa ............26 Table 3.18 Quality control summary – Mochipapa ......................................................................................29 Table 3.19 Occurrence of data readings for Longe meteorological station...................................................30 Table 3.20 Excluded ground measurements after quality control (Sun above horizon) in Longe....................30 Table 3.21 Quality control summary – Longe .............................................................................................32 Table 3.22 Occurrence of data readings for Misamfu meteorological station...............................................33 Table 3.23 Excluded ground measurements after quality control (Sun above horizon) in Misamfu ...............33 Table 3.24 Quality control summary – Misamfu .........................................................................................35 Table 3.25 Occurrence of data readings for Mutanda meteorological station ..............................................36 Table 3.26 Excluded ground measurements after quality control (Sun above horizon) in Mutanda ...............36 Table 3.27 Quality control summary – Mutanda .........................................................................................38 Table 4.1 Input data used in the Solargis and related GHI and DNI outputs for Zambia ...............................40 Table 4.2 Direct Normal Irradiance: bias and KSI before and after model site-adaptation ...........................43 Table 4.3 Global Horizontal Irradiance: bias and KSI before and after model site-adaptation ......................44 Table 4.4 Direct Normal Irradiance: RMSD before and after model site-adaptation.....................................44 Table 4.5 Global Horizontal Irradiance: RMSD before and after model site-adaptation................................44 Table 4.6 Comparison of long term average of yearly summaries of original and site-adapted values .........51 Table 5.1 Source of Solargis meteorological data: models CFSR and CFSv2 and their characteristics.........52 Table 5.2 Solargis meteorological parameters delivered within this project ...............................................52 Table 5.3 Air temperature at 2 m: accuracy indicators of the model outputs [ºC]. .......................................53 Table 5.4 Relative humidity: accuracy indicators of the model outputs [%]. ................................................56 © 2018 Solargis page 106 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 Table 5.5 Wind speed: accuracy indicators of the model outputs [m/s]......................................................60 Table 5.6 Expected uncertainty of modelled meteorological parameters at the project sites.......................63 Table 6.1 Uncertainty of the model estimates for original and site-adapted annual long-term values ..........64 Table 6.2 Annual GHI that should be exceeded with 90% probability in the period of 1 to 10 (25) years .......65 Table 6.3 Annual DNI that should be exceeded with 90% probability in the period of 1 to 10 (25) years. ......66 Table 6.4 Combined probability of exceedance of annual GHI for uncertainty of the estimate ±4.0%...........67 Table 6.5 Combined probability of exceedance of annual DNI for uncertainty of the estimate ±5.5%...........68 Table 7.1 Delivered data characteristics ...................................................................................................70 Table 7.2 Parameters in the delivered site-adapted time series and TMY data (hourly time step)................70 Table 7.3 Monthly and yearly long-term GHI averages as calculated from time series and from TMY..........72 Table 7.4 Monthly and yearly long-term DNI averages as calculated from time series and from TMY ..........72 Table 7.5 Monthly and yearly long-term DIF averages as calculated from time series and from TMY ..........72 Table 7.6 Monthly and yearly long-term TEMP averages as calculated from time series and from TMY ......72 © 2018 Solargis page 107 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 REFERENCES [1] NREL, 1993. User’s Manual for SERI_QC Software-Assessing the Quality of Solar Radiation Data. NREL/TP- 463-5608. Golden, CO: National Renewable Energy Laboratory. [2] Younes S., Claywell R. and Muneer T, 2005. Quality control of solar radiation data: Present status and proposed new approaches. Solar Energy 30, 1533-1549. [3] Perez R., Cebecauer T., Šúri M., 2013. Semi-Empirical Satellite Models. In Kleissl J. (ed.) Solar Energy Forecasting and Resource Assessment. Academic press. http://www.sciencedirect.com/science/article/pii/B9780123971777000024 [4] Cebecauer T., Šúri M., Perez R., High performance MSG satellite model for operational solar energy applications. ASES National Solar Conference, Phoenix, USA, 2010. http://solargis.com/assets/publication/2010/Cebecauer-Suri-Perez--ASES2010--High-performance-MSG- satellite-model-for-operational-solar-energy-applications.pdf [5] Šúri M., Cebecauer T., Perez P., 2010. Quality Procedures of Solargis for Provision Site-Specific Solar Resource Information. Conference SolarPACES 2010, September 2010, Perpignan, France. http://solargis.com/assets/publication/2010//Suri-Cebecaue--Perez-SolarPACES2010--Quality-procedures- of-solargis-for-provision-site-specific-solar-resource-information.pdf [6] Cebecauer T., Suri M., Gueymard C., Uncertainty sources in satellite-derived Direct Normal Irradiance: How can prediction accuracy be improved globally? Proceedings of the SolarPACES Conference, Granada, Spain, 20-23 Sept 2011. [7] http://solargis.com/assets/publication/2011/Cebecauer-Suri-Gueymard--SolarPACES2011--Uncertainty- sources-in-satellite-derived-direct-normal-irradiance--How-can-prediction-accuracy-be-improved-globally.pdf [8] Suri M., Cebecauer T., 2014. Satellite-based solar resource data: Model validation statistics versus user’s uncertainty. ASES SOLAR 2014 Conference, San Francisco, 7-9 July 2014. http://solargis.com/assets/publication/2014//Suri-Cebecauer--ASES-Solar2014--Satellite-Based-Solar- Resource-Data--Model-Validation-Statistics-Versus-User-Uncertainty.pdf [9] Ineichen P., A broadband simplified version of the Solis clear sky model, 2008. Solar Energy, 82, 8, 758-762. [10] Benedictow A. et al. 2012. Validation report of the MACC reanalysis of global atmospheric composition: Period 2003-2010, MACC-II Deliverable D83.1. [11] Molod, A., Takacs, L., Suarez, M., and Bacmeister, J., 2015: Development of the GEOS-5 atmospheric general circulation model: evolution from MERRA to MERRA2, Geosci. Model Dev., 8, 1339-1356, doi:10.5194/gmd-8- 1339-2015 [12] GFS model. http://www.nco.ncep.noaa.gov/pmb/products/gfs/ [13] CFSR model. https://climatedataguide.ucar.edu/climate-data/climate-forecast-system-reanalysis-cfsr/ [14] CFSv2 model http://www.cpc.ncep.noaa.gov/products/CFSv2/CFSv2seasonal.shtml [15] Cano D., Monget J.M., Albuisson M., Guillard H., Regas N., Wald L., 1986. A method for the determination of the global solar radiation from meteorological satellite data. Solar Energy, 37, 1, 31–39. [16] Perez R., Ineichen P., Maxwell E., Seals R. and Zelenka A., 1992. Dynamic global-to-direct irradiance conversion models. ASHRAE Transactions-Research Series, pp. 354-369. [17] Perez, R., Seals R., Ineichen P., Stewart R., Menicucci D., 1987. A new simplified version of the Perez diffuse irradiance model for tilted surfaces. Solar Energy, 39, 221-232. [18] Ruiz-Arias J.A., Cebecauer T., Tovar-Pescador J., Šúri M., 2010. Spatial disaggregation of satellite-derived irradiance using a high-resolution digital elevation model. Solar Energy, 84,1644-57. http://www.sciencedirect.com/science/article/pii/S0038092X10002136 [19] Zelenka A., Perez R., Seals R., Renne D., 1997. Effective Accuracy of Satellite-Derived Hourly Irradiances. Theoretical Appl. Climatology, 62, 199-207. [20] Espinar B., Ramírez L., Drews A., Beyer H. G., Zarzalejo L. F., Polo J., Martín L., Analysis of different comparison parameters applied to solar radiation data from satellite and German radiometric stations. Solar Energy, 83, 1, 118-125, 2009. © 2018 Solargis page 108 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 [21] Cebecauer T., Suri M., 2015. Site-adaptation of satellite-based DNI and GHI time series: overview and Solargis approach. SolarPACES conference, September 2015, Cape Town, North Africa. http://solargis.com/assets/publication/2015/Cebecauer-Suri--2015-SolarPaces--Site-adaptation-of-satellite- based-DNI-and-GHI-time-series--Overview-and-SolarGIS-approach.pdf [22] Polo J., Wilbert S., Ruiz-Arias J.A., Meyer R., Gueymard C., Šúri M., Martín L., Mieslinger T., Blanc P., Grant I., Bolan J., Ineichen P., Remund J., Escobar R., Troccoli A., Sengupta M., Nielsen K.P., Renné D., Geuder N., Cebecauer T., 2016. Preliminary survey on site-adaptation techniques for satellite-derived and reanalysis solar radiation datasets. Solar Energy, 132, 25–37. [23] Lohmann S., Schillings C., Mayer B., Meyer R., 2006. Long-term variability of solar direct and global radiation derived from ISCCP data and comparison with reanalysis data, Solar Energy, 80, 11, 1390-1401. [24] Gueymard C., Solar resource assessment for CSP and CPV. Leonardo Energy webinar, 2010. http://www.leonardo-energy.org/webfm_send/4601 [25] Cebecauer T., Šúri M., 2015. Typical Meteorological Year Data: Solargis Approach. Energy Procedia, Volume 69, 1958-1969. http://dx.doi.org/10.1016/j.egypro.2015.03.195 [26] Sensor Verification Report Project: ESMAP Zambia. GeoSUN Africa, August 2016, reference No. 2012-026- 1216-08. © 2018 Solargis page 109 of 111 Annual Solar Resource Report for Solar Meteorological Stations in Zambia At the occasion of completion of ground measuring campaign Solargis reference No. 128-07/2018 SUPPORT INFORMATION Background on Solargis Solargis is a technology company supplying solar resource and meteorological data, software applications and consultancy services to solar energy industry. Our services are used globally in identification of optimum sites for development of solar power plants, for technical and financial evaluation and optimisation of solar energy production. We develop and operate own technology based on a new-generation high-resolution global meteorological database and software applications integrated within the Solargis® online information system. Accurate and standardised energy simulation data reduce the weather-related risks and costs in planning, performance evaluation, forecasting and management of distributed solar power systems. Solargis is a technology company offering solar and meteorological data, software and consultancy services to solar energy. We support industry in the site qualification, planning, financing and operation of solar energy systems for more than 18 years. We develop and operate a new generation high-resolution global database and applications integrated within Solargis® information system. Accurate, standardised and validated data help to reduce the weather-related risks and costs in system planning, performance assessment, forecasting and management of distributed solar power. Legal information Considering the nature of climate fluctuations, interannual and long-term changes, as well as the uncertainty of measurements and calculations, company Solargis cannot take guarantee of the accuracy of estimates. Company Solargis has done maximum possible for the assessment of climate conditions based on the best available data, software and knowledge. Solargis® is the registered trademark of company Solargis. Other brand names and trademarks that may appear in this study are the ownership of their respective owners. © 2018 Solargis, all rights reserved Solargis is ISO 9001:2015 certified company for quality management. Authors: Marcel Suri Tomas Cebecauer Branislav Schnierer Artur Skoczek Daniel Chrkavy Nada Suriova Maps: Juraj Betak Veronika Madlenakova Project manager: Nada Suriova Approved by: Marcel Suri © 2018 Solargis page 110 of 111