Surface albedo spatial variability in North America: Gridded data vs. local measurements. (October 2021)

• Record Type: Journal Article
• Authors: Gueymard, Christian A.; Lara-Fanego, Vicente; Sengupta, Manajit; Habte, Aron
• Is Part Of: Solar energy. Volume 227(2021)
• Page Start: 655
• Page End: 673
• Publication Date: 2021-10
• Subjects: Albedo -- Reflectance -- MODIS -- NSRDB -- North America -- Snow; Solar energy -- Periodicals; Solar engines -- Periodicals; 621.47
• DOI: 10.1016/j.solener.2021.05.012
• Languages: English
• ISSNs: 0038-092X

Abstract

Highlights: Surface albedo may be necessary at high spatial resolution for various applications. Satellite-derived albedo products from MODIS and NRSDB are evaluated. Albedo measurements are used to validate predictions at 36 radiometric stations. Spatial resolutions are 500 m, 1 km and 4 km, depending on product. Terrain heterogeneity and snow are found to impact the accuracy of gridded data. Abstract: Considering the current booming interest for the large-scale deployment of bifacial photovoltaic modules, the solar industry now requires accurate estimates of broadband surface albedo at high spatial resolution. In this context, the present study evaluates the adequacy and performance over North America of two Moderate Resolution Imaging Spectroradiometer (MODIS) white-sky albedo products (at 500-m and 1-km resolution) and the National Solar Radiation Database (NSRDB) product at 4-km resolution. Two variations of the 500-m MODIS product are also considered: black-sky albedo and all-sky albedo. Albedo observations from 36 radiometric stations during 2011–2015 are analyzed while considering the apparent homogeneity of the surface characteristics over the 4x4 km NSRDB pixel in which they are located. Even at sites where the albedo around the station has been found "homogeneous" in the literature, marked differences are found between the daily observations and the gridded estimates at any spatial resolution. Differences in seasonal behavior and between the three different albedo types also impact the accuracy of the albedo estimates, with overtones caused by local specificities and inhomogeneities. All this precludes the desirable evaluation of the local albedo at a specific site of relatively small size compared to its corresponding 4x4-km pixel if only the mean albedo over that pixel is known. Significant discrepancies are also found at snow-impacted sites, most importantly in the case of the NSRDB albedo estimates, which are typically much too high.