پديد آورندگان :
احمدي باصري، نوشين دانشگاه بوعلي سينا - دانشكدة كشاورزي، همدان، ايران , سبزي پرور، علي اكبر دانشگاه بوعلي سينا - دانشكدة كشاورزي، همدان، ايران , خدامرادپور، مهرانه دانشگاه بوعلي سينا - دانشكدة كشاورزي، همدان، ايران , گررو راسكادو، خوان لويئس دانشگاه گرانادا - مؤسسة تحقيقات بين دانشگاهي سيستم زمين در آندلس(IISTA-CEAMA)، اسپانيا , آربولداس، لوكاس آلادوس دانشگاه گرانادا - مؤسسة تحقيقات بين دانشگاهي سيستم زمين در آندلس(IISTA-CEAMA)، اسپانيا
چكيده لاتين :
Surface solar radiation (SSR) is a fundamental key variable in climatological and meteorological studies. The most accurate way to obtain SSR at the surface is to measure solar irradiances using radiometers such as pyranometer. However, ground measurements are limited due to the high cost of the calibration and the regular maintenance of the equipment. To eliminate this problem, many methods have been developed to estimate solar radiation received by the Earth's surface: including, empirical models, radiative transfer models (RTM), semi-empirical models, and models based on satellite products. One of the most popular and easily accessed methods for estimating SSR is satellite-based methods, having advantages of high temporal and spatial resolution, low cost and free access can be a good alternative source for areas without stations and areas with heterogeneous distribution of stations. This work is aimed to evaluate and validate SSR derived from three satellite-based products (CERES, CLARA and SARAH) against ground measurements over Iran daily.
Materials and methods:
Ground measured data: In this study, measured SSR was extracted from 24 radiometric stations of Iran from 2012 to 2015. The SSR ground measurements are more susceptible to error than other meteorological parameters. Therefore, quality control was applied in order to remove likely errors and outliers from the measured data. In this study, Moradi's proposed method was used for data quality control (Moradi, 2009). The studied areas were classified based on Digital Elevation Model (DEM) and UNESCO climate classification approach into three zones of arid, semi-arid and coastal climates.
Satellite-based data: The satellite-based SSR outputs including CERES, CLARA, and SARAH were extracted for the period of 2012 to 2015. The satellite scientific instrument of CERES (Cloud and the Earth's Radiant Energy System) is a component of NASA’s Earth Observation System (EOS) onboard the Terra and Aqua platforms. The CERES datasets provide global coverage available from 2000 to present with a spatial resolution of 1 degree and temporal resolution of daily and monthly (Carmona et al. 2017). The Climate Monitoring Satellite Application Facility (CM SAF) provides high-quality records for climate applications from satellites. The CM SAF CLARA products are based on observations of the AVHRR instruments onboard the NOAA and METOP polar-orbit satellites. The CLARA datasets prepare global coverage with a spatial resolution of 0.25 degrees on a regular latitude-longitude grid and daily and monthly temporal resolution available from 1982 to 2015. The CM SAF SARAH products are based on observations from the MVIRI and SEVIRI (Spinning Enhanced Visible and InfraRed Imager) instruments onboard the Meteosat First- and Second-Generation satellites. The SARAH datasets supply regional coverage (Europe, Africa, the Atlantic and parts of South America) with a spatial resolution of 0.05 degrees on a regular latitude-longitude grid and instantaneous, daily and monthly temporal resolution available from 1983 to 2017 (Riihelä et al. 2015).
Methods: Since the satellite-based products studied are generally distributed in NetCDF format, the nearest neighborhood interpolation method was used to match these data with measured data. Then, satellite-based datasets were generated for station points using CDO software. In this study, satellite-based datasets were compared against measured SSR datasets by four validation metrics. The metrics used are determination coefficient (R2), the mean bias deviation (MBD), relative mean absolute deviation (RMABD) and root mean squared error (RMSE).
Results and discussion:
Evaluation of the performance of CERES, CLARA, and SARAH products in estimating daily SSR in Iran showed that despite the proper performance of all three satellite-based products in this study, SARAH with R2= 0.93 and MBD= -0.1 W.m-2 has the highest agreement with measured SSR compared to other two products. This result is consistent with the study by Journée and Bertrand (2010), Urraca et al. (2017), Alexandri et al. (2017) and Wang et al. (2018). Also, evaluation of the monthly and seasonally variations of daily SSR of three satellite products against the measured daily SSR showed that the studied satellite products are more capable of estimating SSR under clear sky conditions (warm seasons) than cloudy conditions (cold seasons).
Spatial variations of daily SSR showed that the satellite products in the arid and semi-arid climate regions had the best performance, respectively, compared to the coastal regions. In this aspect, SARAH provided the best performance in all three study areas. According to the results, the highest agreement between ground measured SSR and SARAH was observed in the dry climate (R2= 0.94) and the lowest agreement between ground measured SSR and CERES was observed in the coastal region (R2= 0.83). Also, the largest overestimation occurred by CERES in the coastal region with MBD of 21.3 W.m-2 and the smallest underestimation by SARAH in the arid climate region with MBD of -0.1 W.m-2. Also, the smallest RMSE obtained in the arid climate region by SARAH with 20 W.m-2 and largest in the coastal region by CERES with 37.3 W.m-2. These findings are consistent with the results obtained by Thomas et al. (2016), Urraca et al. (2017) and Urraca et al. (2018). Their results showed that in the coastal regions with high humidity, the errors of satellite- based SSR estimates are very high.
Seasonal variations of RMABD showed that the maximum and minimum RMABD in arid and semi-arid climate regions occurred in winter and spring, respectively. These results are consistent with the results shown by Sanchez-Lorenzo et al. (2013) and Wang et al. (2018) that indicate the lowest ability of studied satellite products in cloudy conditions (winter) than in clear sky conditions (summer) to estimate SSR. In coastal regions, the minimum and maximum RMABD were found in spring and summer, respectively. As shown in this study by Thomas et al. (2016) and Urraca et al. (2017), humidity and water vapor in the atmosphere are some of the causes of high error in satellite products. Since the coastal areas studied in this study (Bandar Abbas and Gorgan stations) have warm and humid summers, therefore, the maximum RMABD occurred in these regions in summer and the minimum RMABD in spring with lower humidity and cloudiness than other seasons.
Conclusion:
The results showed that SARAH with the highest spatial resolution compared to CLARA and CERES had the best performance in generating daily SSR in Iran. Also, the high error in cold seasons indicates the high impact of cloudiness in reducing the accuracy of studied satellite products in Iran. Investigation of spatial variations of daily measured SSR and studied satellite products also showed that the satellite products have the highest performance in arid and semi-arid climate regions but the lowest for the coastal regions.
Most regions of Iran located in arid and semi-arid climate region and the growing season of these zones are in accordance with the warm season. Therefore, due to the acceptable performance of satellite products to estimate SSR in arid and semi-arid climate regions and in warm seasons, accordingly, SSR outputs of satellite products can be used in agricultural studies. Also, given the free availability and high spatial and temporal resolution of the satellite products under investigation, the SSR output of these products can be a good alternative for areas where there is no access to the ground based measured SSR datasets.
