|Title||A critical evaluation of modeled solar irradiance over California for hydrologic and land surface modeling|
|Publication Type||Journal Article|
|Year of Publication||2017|
|Authors||Lapo K.E, Hinkelman L.M, Sumargo E., Hughes M, Lundquist J.D|
|Journal||Journal of Geophysical Research-Atmospheres|
|Type of Article||Article|
|Keywords||atmospheric rivers; central valley; cloud properties; Complex terrain; energy budget; Mountain meteorology; radiation budget; reference evapotranspiration; sierra-nevada; snow water equivalent; solar radiation; surface energy budget; united-states|
Studies of land surface processes in complex terrain often require estimates of meteorological variables, i.e., the incoming solar irradiance (Q(si)), to force land surface models. However, estimates of Q(si) are rarely evaluated within mountainous environments. We evaluated four methods of estimating Q(si): the CERES Synoptic Radiative Fluxes and Clouds (SYN) product, MTCLIM, a regional reanalysis product derived from a long-term Weather Research and Forecast simulation, and Mountain Microclimate Simulation Model (MTCLIM). These products are evaluated over the Central Valley and Sierra Nevada mountains in California, a region with meteorology strongly impacted by complex topography. We used a spatially dense network of Q(si) observations (n=70) to characterize the spatial characteristics of Q(si) uncertainty. Observation sites were grouped into five subregions, and Q(si) estimates were evaluated against observations in each subregion. Large monthly biases (up to 80 Wm(-2)) outside the observational uncertainty were found for all estimates in all subregions examined, typically reaching a maximum in the spring. We found that MTCLIM and SYN generally perform the best across all subregions. Differences between Q(si) estimates were largest over the Sierra Nevada, with seasonal differences exceeding 50 Wm(-2). Disagreements in Q(si) were especially pronounced when averaging over high-elevation basins, with monthly differences up to 80 Wm(-2). Biases in estimated Q(si) predominantly occurred with darker than normal conditions associated with precipitation (a proxy for cloud cover), while the presence of aerosols and water vapor was unable to explain the biases. Users of Q(si) estimates in regions of complex topography, especially those estimating Q(si) to force land surface models, need to be aware of this source of uncertainty.