Available online at www.sciencedirect.com ScienceDirect Energy Procedia 57 (2014) 1037 – 1043 2013 ISES Solar World Congress A combination of Heliosat-1 and Heliosat-2 methods for deriving solar radiation from satellite images Iđigo Pagolaa,*, Martín Gastóna, Ana Bernardosa, Carlos Fernández-Peruchenaa a National Renewable Energy Centre (CENER), Ciudad de la Innovación 7, Sarriguren 31621, Spain Abstract Solar radiation estimation from geostationary satellite images is accepted by the international scientific community, especially where no previous ground radiometric measurements are available The most accepted methodology is Heliosat In this work, a combination of the existing methods Heliosat-1 and Heliosat-2 has been implemented for this purpose, and the obtained results are presented To analyze the results provided by the implemented methodology a validation against measurements has been made Firstly, solar Global Horizontal Irradiance (GHI) has been estimated from satellite images using data of the Meteosat Second Generation (MSG) satellite corresponding to the period from 2009 to 2011 Secondly, measurements recorded during the same period of time at the Cener station which belongs to the Baseline Surface Radiation Network (BSRN) have been obtained Both estimated and measured data have been integrated into hourly values for the validation process The obtained values for Cener BSRN station are a MBE of 2% and a RMSE of 113 W/m2, which are smaller than the recommended values for hourly GHI data (MBE smaller than 5% and RMSE smaller than 160 W/m2) During the months of summer the errors in terms of W/m2 are bigger than during the months of winter However, since the irradiance is higher during the months of summer, the errors are smaller in terms of % during the months of summer Although the methodology can be applied to locations where no ground measurements are available, it is preferable to analyze locations with available measured data © Authors Published by Elsevier This is Ltd an open access article under the CC BY-NC-ND license ©2014 2013The The Authors Published byLtd Elsevier (http://creativecommons.org/licenses/by-nc-nd/3.0/) Selection and/or peer-review under responsibility of ISES Selection and/or peer-review under responsibility of ISES Keywords: solar resource assessment; satellite images; Heliosat * Corresponding author Tel.: +34 948 25 28 00; fax: +34 948 27 07 74 E-mail address: ipagola@cener.com 1876-6102 © 2014 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/) Selection and/or peer-review under responsibility of ISES doi:10.1016/j.egypro.2014.10.088 1038 Iñigo Pagola et al / Energy Procedia 57 (2014) 1037 – 1043 Introduction Before the planning of any solar energy project, solar resource assessment is an essential first step Solar resource assessment provides the means to accurately determine the availability of solar radiation, to understand its characteristics, and to validate its quality It also provides the resources for developing, deploying, and operating cost-effective solar energy technologies When assessing the solar resource, one of the most important factors is the availability of ground measurements Ground based pyranometers are capable of measuring solar radiation at specific networks but not at large spatial resolutions This fact motivated the use of satellite data as an important alternative to assess the solar resource for large areas Solar radiation estimation from geostationary satellite images is accepted by the international scientific community, especially where no previous ground radiometric measurements are available Methodology 2.1 Deriving solar radiation from satellite images Nowadays, solar radiation derived from geostationary satellites is a commonly used methodology in solar resource assessment studies [1-3] This methodology has been in continuous progress and evolution since the first approaches [4] and up to now [5, 6] Solar radiation derived from satellite images is accepted by the international scientific community as one of the most useful methodologies to be applied where the spatial distribution of solar radiation needs to be estimated The most accepted methodology is Heliosat, with which images acquired by meteorological geostationary satellites such as Meteosat (Europe), GOES (USA) or GMS (Japan) are converted into data and maps of solar radiation received at ground level As it became popular some modifications were proposed creating several versions of the Heliosat methodology In the Heliosat-1 methodology, the digital count values recorded by the radiometer of the satellite are converted into the cloud index normalized parameter The solar radiation incoming to the earth’ surface can be estimated by the cloud index and some empirical/statistical methodologies The relationship between the cloud index and the clearness index is empirically defined, and its parameters are computed by the means of a comparison between the cloud index and measurements made by ground stations in the area under concern All these parameters were well tuned during the construction of the method using ground measurements In Heliosat-2 methodology, the cloud index parameter is maintained, but the inputs to the method are not the numerical counts of the satellite image These counts are calibrated and thus converted into radiances, which allows taking into account the change of sensor, gain and calibration Rigollier and Wald [7] proposed a relationship between the cloud index and the clear sky index, which is the ratio of the GHI to the same quantity but for clear skies This relationship is used in the Heliosat-2 method In this version of the methodology, the clear sky models of the 4th European Solar Radiation Atlas (ESRA) are utilized in the calculations In particular, the methodology developed by CENER is a combination of Heliosat-1 and Heliosat-2 methods Broadly speaking, in this methodology the cloud index is derived from the digital count values recorded by the satellite as in the Heliosat-1 methodology However, a relationship between the cloud index and the clearness index is not used In the developed methodology, the same relationship between the cloud index and the clear sky index used in the Helioast-2 method has been implemented In addition, the ESRA clear sky models are used as in the Heliosat-2 methodology, needing as inputs the Linke 1039 Iñigo Pagola et al / Energy Procedia 57 (2014) 1037 – 1043 turbidity factor for an air mass of and the elevation of the site, besides the parameters related to the solar geometry 2.2 Comparison with measurements To analyze the results provided by the implemented methodology, a validation against measurements has been made Firstly, solar GHI has been estimated from satellite images using data of the Meteosat Second Generation satellite corresponding to the period from 2009 to 2011 The frequency of this data is 15 minutes, and data of the high resolution visible channel have been used Secondly, measurements recorded during the same period of time at the Cener station which belongs to the Baseline Surface Radiation Network (BSRN) have been obtained Both estimated and measured data have been integrated into hourly values for the validation process BSRN is a project of the World Climate Research Program (WCRP) and the Global Energy and Water Experiment (GEWEX) This project aims at detecting changes in the Earth's radiation field at the Earth's surface that may be related to climate change The data are of primary importance for the validation and evaluation of satellite and model estimates of radiative quantities The BSRN stations are located in contrasting climatic zones, where solar and atmospheric radiation is measured with instruments of the highest available accuracy and with high time resolution (1 to minutes) The selected station for the validation is CNR BSRN station operated by Cener in Sarriguren (Spain) In Table 1, the coordinates and the principal characteristics of the selected BSRN station for the validation are presented Table BSRN station selected for the validation Label Location Country Latitude (º) Longitude (º) Elevation (m) Start Date CNR Sarriguren Spain 42.816 -1.601 471 01/07/2009 For the validation, the data measured at this BSRN station during the period of time corresponding to the satellite images (2009-2011) were obtained The MSG satellite images have a temporal resolution of 15 minutes, and the measured data are recorded at the CNR BSRN station every minute In order to make the comparison possible, both estimated and measured data have been integrated into hourly values, which is the most typical temporal resolution of the solar data bases All the nocturnal data have been filtered and not compute in the comparison Some statistical parameters have been calculated to compare the results provided by the methodology for the different locations These parameters are: x MAE: Mean Absolute Error MAE n ¦ pi mi n i1 (1) Being pi the estimated values, mi the measured values and n the number of compared values 1040 Iñigo Pagola et al / Energy Procedia 57 (2014) 1037 – 1043 x MAE (%): Mean Absolute Error (%) MAE (%) 100 MAE m (2) Being m the mean of the measured values x MBE: Mean Bias Error MBE n ¦ pi mi n i1 (3) x MBE (%): Mean Bias Error (%) MBE (%) 100 MBE m (4) x RMSE: Root Mean Square Error RMSE n pi mi ¦ n i1 (5) x RMSE (%): Roor Mean Square Error (%) RMSE (%) 100 RMSE m (6) Results In this section, the results obtained by applying the developed methodology for deriving the solar radiation from satellite images to the selected location are presented The methodology has been put into practice for the Meteosat Second Generation satellite images corresponding to the period from 2009 to 2011 The obtained results of GHI derived from satellite have been compared to the measured GHI at CNR BSRN station operated by Cener in Sarriguren (Spain) Some statistical parameters of the comparison are presented in the following tables for each location Table Results for Cener BSRN station Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total MAE (W/m2) 39 62 79 101 86 99 79 73 72 61 53 46 72 MAE (%) 26 30 28 26 23 26 18 18 20 22 33 35 23 1041 Iñigo Pagola et al / Energy Procedia 57 (2014) 1037 – 1043 MBE (W/m2) -12 -10 -6 27 39 23 12 -11 -20 MBE (%) -8 -5 -2 -7 -15 RMSE (W/m ) 60 98 122 156 135 148 113 108 109 97 81 74 113 RMSE (%) 40 48 43 41 35 38 25 26 30 36 50 56 36 The results for the total data of the analyzed period 2009-2011 presented in Table are shown graphically in Fig 120 100 80 60 40 20 MAE (W/m2) MAE (%) MBE (W/m2) MBE (%) RMSE (W/m2) RMSE (%) Fig Parameters of error obtained for the CNR BSRN station As it can be seen in Table 2, during the months of summer the errors in terms of W/m2 are bigger than during the months of winter However, since the irradiance is higher during the months of summer, the errors are smaller in terms of % during the months of summer For this location, the MBE and RMSE in GHI related to all the period of data are smaller than the recommended by The United Nations Environmental Programme (UNEP) in [8] The recommended values for hourly GHI data are a MBE smaller than 5% and a RMSE smaller than 160 W/m2, and the obtained values for CNR BSRN station are a MBE of 2% and a RMSE of 113 W/m2 The same calculations have made using an implementation of the Heliosat-1 methodology in order to compare the new methodology with The results obtained for CNR BSRN station using both methodologies are presented in Table Table Results obtained for Cener BSRN station using Heliosat-1 and the new methodology Heliosat-1 New methodology MAE (W/m2) 80 72 MAE (%) 25 23 MBE (W/m2) -7 1042 Iñigo Pagola et al / Energy Procedia 57 (2014) 1037 – 1043 MBE (%) -2 RMSE (W/m ) 124 113 RMSE (%) 39 36 As it can be seen in Table 3, the errors obtained for CNR BSRN station with the new methodology are smaller than the obtained with the implementation of the Heliosat-1 methodology At present, a complete version of the Heliosat-2 methodology is being implemented in order to a result comparison, and the developed methodology is going to be applied to other locations with BSRN stations Although the methodology can be applied to locations where no ground measurements are available, it is preferable to analyze locations with available measured data Therefore, tuning and fitting of the methodology for each location could be carried out, obtaining a specific relationship between the cloud index and the clear sky index Conclusions A combination of Heliosat-1 and Heliosat-2 methods for deriving solar radiation from satellite images has been implemented In this methodology developed by CENER, the ESRA clear sky models are used, needing as inputs the Linke turbidity factor for an air mass of and the elevation of the site, besides the parameters related to the solar geometry A validation of the methodology against measurements has been made Solar radiation has been estimated from satellite images of the MSG satellite for the period from 2009 to 2011 It has been compared with measurements that correspond to the same period of time obtained at CNR BSRN station During the months of summer the errors in terms of W/m2 are bigger than during the months of winter However, since the irradiance is higher during the months of summer, the errors are smaller in terms of % during the months of summer The obtained values for CNR BSRN station are a MBE of 2% and a RMSE of 113 W/m2, which are smaller than the recommended values for hourly GHI data (MBE smaller than 5% and RMSE smaller than 160 W/m2) Although the methodology can be applied to locations where no ground measurements are available, it is preferable to analyze locations with available measured data Therefore, tuning and fitting of the methodology for each location could be carried out, obtaining a specific relationship between the cloud index and the clear sky index In this paper, the importance of using established resource assessment methods to lower project risk and to improve project and site characterization has been shown Acknowledgements The authors would like to thank the BSRN for the measured data at the selected station of Cener (CNR) in Spain References [1] Zelenka A., Perez R., Seals R., Renné D Effective accuracy of satellite-derived hourly irradiances Theoretical and Applied Climatology, 62; 1999, pp 199-207 Iñigo Pagola et al / Energy Procedia 57 (2014) 1037 – 1043 [2]Vignola F., Harlan P., Perez R., Kmiecik M Analysis of satellite derived beam and global solar radiation data, Solar Energy, Volume 81, Issue 6, June 2007, Pages 768-772, ISSN 0038-092X, 10.1016/j.solener.2006.10.003 [3] Hoyer-Klick, C., et al Getting Solar Energy to Work: Resource Assessment by Remote Sensing as a Base for Investment Decisions 2009 "www.earthzine.org/2009/05/05/getting-solar-energy-work-resource-assessment-remote-sensing-baseinvestmentdecisions/ [4] Cano D., Monget J.M., Albuisson M., Guillard H., Regas N., Wald L A method for the determination of the global solar radiation from meteorological satellite data, Solar Energy, Volume 37, Issue 1, 1986, Pages 31-39, ISSN 0038-092X, 10.1016/0038092X(86)90104-0 [5] Perez R., Ineichen P., Moore K., Kmiecik M., Chain C., George R., Vignola F A new operational model for satellite-derived irradiances: description and validation, Solar Energy, Volume 73, Issue 5, November 2002, Pages 307-317, ISSN 0038-092X, 10.1016/S0038-092X(02)00122-6 [6] Polo J., Martín L., Cony M Revision of ground albedo estimation in Heliosat scheme for deriving solar radiation from SEVIRI HRV channel of Meteosat satellite, Solar Energy, Volume 86, Issue 1, January 2012, Pages 275-282, ISSN 0038-092X, 10.1016/j.solener.2011.09.030 [7] Rigollier C., Wald L Using Meteosat images to map the solar radiation: improvements of the Heliosat method In Proceedings of the 9th Conference on Satellite Meteorology and Oceanography Published byEumetsat, Darmstadt, Germany, 1998, EUM P 22, pp 432-433 [8] Hoyer-Klick C., McIntosh J., Moner-Girona M., Renné D., Perez R., Puig D., Developing of a benchmarking tool for solar energy resource datasets, a guide for non-expert users to determine the most appropriate use of solar energy resource information International Solar Energy Society July 2010 1043 ... 72 61 53 46 72 MAE (%) 26 30 28 26 23 26 18 18 20 22 33 35 23 10 41 Iñigo Pagola et al / Energy Procedia 57 (2 014 ) 10 37 – 10 43 MBE (W/m2) - 12 -10 -6 27 39 23 12 -11 -20 MBE (%) -8 -5 -2 -7 -15 ... GHI data are a MBE smaller than 5% and a RMSE smaller than 16 0 W/m2, and the obtained values for CNR BSRN station are a MBE of 2% and a RMSE of 11 3 W/m2 The same calculations have made using an... are available Methodology 2 .1 Deriving solar radiation from satellite images Nowadays, solar radiation derived from geostationary satellites is a commonly used methodology in solar resource assessment