Error assessment of grid‐based terrain shading algorithms for solar radiation modeling over complex terrain. Issue 1 (21st October 2019)

• Record Type: Journal Article
• Authors: Zhang, Shuhua; Li, Xingong; She, Jiangfeng
• Is Part Of: Transactions in GIS. Volume 24:Issue 1(2020)
• Page Start: 230
• Page End: 252
• Publication Date: 2019-10-21
• Subjects: Geographic information systems -- Periodicals; 910.285
• DOI: 10.1111/tgis.12594
• Languages: English
• ISSNs: 1361-1682

Abstract

Abstract: In mountainous regions, solar radiation exhibits a strong spatial heterogeneity due to terrain shading effects. Terrain shading algorithms based on digital elevation models can be categorized into two types: area‐based and point‐specific. In this article, we evaluated two shading algorithms using designed mathematic surfaces. Theoretical shading effects over four Gauss synthetic surfaces were calculated and used to evaluate the terrain shading algorithms. We evaluated the area‐based terrain shading algorithm, Hillshade tool of ArcGIS, and the point‐specific shading algorithm from Solar Analyst (SA) in ArcGIS. Both algorithms showed shading overestimation, and Hillshade showed more accuracy with a mean absolute error (MAE) of 1.20%, as compared to the MAE of 1.66% of SA. The MAE of Hillshade increases exponentially as the spatial extent of the study area increases because the solar position for all locations on the surface is the same in Hillshade. Consequently, we suggest that the surface should be divided into more tiles in Hillshade when the discrepancy in the latitude of the whole surface is greater than 4°. Skyshed, which represents the horizon angle distribution in SA, is error‐prone over more complex terrain because horizon angle interpolation is problematic for such areas. We also propose a new terrain shading algorithm, with solar positions calculated using local latitude for each cell and the horizon angle calculated for every specific time interval, but without projections. The new model performs better than Hillshade and SA with an MAE of 0.55%.