Exploring a method to estimate real-sky global solar radiation in mountainous areas at high resolutions with an open-source geospatial solar radiation model and GOES-13 geostationary meteorological satellite data.

Abstract

<p>With their complex topography and orographic effects, the amount of solar radiation reaching a surface (insolation) can vary over short distances and time frames in mountainous areas, affecting the spatio-temporal variability of hydrological and ecological processes and contributing to the biodiversity of mountain ecosystems.  The combined effects of variable topography and meteorological conditions on insolation can complicate assessments of how land cover changes affect insolation in mountainous regions as measurement data is often limited to few locations. To incorporate the spatio-temporal variability of sky conditions as well as the spatial variability of terrain in estimates of solar radiation across a montane headwater basin over two summers, we extended an open-source geospatial model of surface solar radiation, r.sun.hourly, to permit the spatially and temporally explicit parameterization of atmospheric conditions at user-specified spatial and temporal resolutions with temporal raster datasets.  Sensitivity analyses indicated that of the three atmospheric parameters in the model, the coefficient of real-sky direct beam solar radiation (coeff_bh) (an index of cloudiness to clear sky conditions) had the greatest influence on insolation estimates for our study area, located in the southern Appalachian Mountains of the southeastern USA, and we developed a workflow for estimating coeff_bh from publicly available geostationary meteorological satellite images (GOES-13, 1 km spatial and 15-minute temporal resolutions).  The extended r.sun.hourly model was parameterized with a bare-earth digital elevation model (1/9 arc-second DEM obtained from the USGS National Map 3-D Elevation Program), the estimated coeff_bh temporal raster datasets (downscaled from 1 km to the DEM resolution), and monthly mean Linke Turbidity values to estimate global solar radiation across the basin at a 15-minute resolution over two summers.  Estimates of instantaneous (15-minute interval) and cumulative total (for 12-hour period bracketing solar noon) global solar radiation were evaluated with pyranometer (Eppley 8-48) measurements of global solar radiation collected by the U.S. Department of Agriculture Forest Service Coweeta Hydrological Laboratory for a total of 144 days (dates with recorded precipitation during daylight hours or incomplete imagery datasets were excluded from analyses).   Despite the low spatial resolution of the satellite images from which the real-sky direct beam radiation coefficient was estimated and the proximity of the ground measurement location to the edges of a GOES-13 cell (< 90 m north and < 230 m east), both instantaneous and daily total estimates of global solar radiation corresponded well with measurements (R<sup>2</sup> = 0.81, p-value < 0.001, n = 7056 and R<sup>2</sup> = 0.89, p-value < 0.001, n = 144). Estimate errors tended to be lower on cloudier days (MAPE = 6.8%, n = 61) than less cloudy days (MAPE = 13.6%, n = 83).  Offsets in the timing and magnitude of peaks and troughs in insolation over time (with passing clouds) between measurements and estimates were also greater during partly cloudy conditions.  Although these findings are for one location, they suggest the potential of the extended r.sun.hourly model to provide high-resolution estimates of solar radiation, over extensive areas and timeframes, in mountainous regions with remotely sensed elevation and meteorological data.</p>