Publications

Li, YT; Gao, YH; Chen, GX; Wang, GY; Zhang, M (2024). Decomposition and reduction of WRF-modeled wintertime cold biases over the Tibetan Plateau. CLIMATE DYNAMICS.

Abstract
Land surface temperature (LST) is a critical thermal variable of the ground surface. However, accurate LST simulation is still challenging over the Tibetan Plateau (TP), having a large cold bias in many global and regional climate models (especially in the winter season). In this study, the LST in winter simulated by WRF was compared to three global land data assimilation datasets (GLDAS), and two reanalysis datasets (ERA-Interim and ERA5). All of these datasets were evaluated against satellite observation. Three GLDAS datasets generally outperform the WRF simulation and reanalysis datasets with smaller cold biases and were taken as the references in the attribution analysis. By decomposing the LST biases using the decomposed temperature metric (DTM), we investigated the contributions of relevant factors to the cold surface temperature biases and the underlying mechanisms. Result shows that the too-less incoming longwave radiation (LW) contributes the most to cold biases, and too-bright surface albedo effect ranks the second. Comparison with MODIS demonstrates an underestimation in the simulated cloud fractions (CF), causing the large contribution of LW simulation to the cold biases. Using a developed neural network-based scale-adaptive CF parameterization, the cold bias over the mainland of TP is greatly reduced. In addition, the improvement of the snow cover fraction (SCF) parameterization leads to the surface albedo decreasing and sensible heat flux increasing, the cold bias can also be reduced by half. Reduction of simulated cold bias possesses great significance and implications in water resources responses over high mountain to global warming.

DOI:
10.1007/s00382-024-07126-0

ISSN:
1432-0894