Characteristics of top-of-atmosphere radiation budget over the Tibetan Plateau and its bias sources in climate models

Abstract

Observations indicate that the Tibetan Plateau (TP) has an annual-mean ~9.3 W m−2 positive radiation budget (RT) at the top of the atmosphere, the largest at the same continental latitudes. This unique radiative heating is critical to the TP's thermal forcing and hydrological cycle. Here we use satellite and reanalyzed data to investigate the characteristics of RT over the TP in the observation and 28 CMIP6 models. The positive observational RT is mainly caused by low surface temperature associated with high elevation. Most models underestimate annual mean RT over the whole TP, with a multimodel average of 2.0 W m−2. This RT bias results mainly from weaker absorbed shortwave radiation (ASR) that exhibits substantial seasonal and regional differences. Serious RT, ASR, and surface albedo biases appear over the western TP. For instance, their wintertime multimodel-mean biases of −44.9 W m−2, −53.3 W m−2, and 0.23 are larger than the eastern TP's counterparts (−38.6 W m−2, −40.6 W m−2, and 0.15). Simulated RT also shows a large intermodel spread. In the models with worse RT's performance, weaker ASR is primarily attributed to higher surface albedo that coincides well with lower surface temperature. In contrast, the models that reproduce well the RT have less surface albedo bias but somewhat overestimate surface temperature. Weaker simulated cloud radiative cooling reduces reflected shortwave radiation that alleviates the RT bias, especially over the springtime eastern TP. This study highlights the importance of land surface states and clouds in modeling the TP's climate. Significance statementObservations indicate that the Tibetan Plateau (TP) has an annual-mean ~ 9.3 W m−2 positive radiation budget (RT) at the top of the atmosphere, the largest among land regions in the same latitudes. This study aims to investigate the characteristics of RT over the TP in the observation and CMIP6 models. Most models underestimate annual mean RT over the whole TP, with a multimodel average of 2.0 W m−2. This bias results mainly from weaker absorbed shortwave radiation, exhibiting substantial seasonal and regional differences. CMIP6 simulations show a large multimodel spread, especially over the wintertime and springtime western TP. This study highlights the importance of land surface states (e.g., surface temperature and albedo) and clouds in modeling the TP's climate.