Evaluation Of Atmospheric Temperature Trend In CMIP5’s Historical Experiments For Climate Models Using Satellite-derived MSU Records

New static microwave sounding unit (MSU) weighting functions are obtained from using fifth phase Coupled Model Inter-comparison Project (CMIP5) historical multimodel simulations as inputs into the fast Radiative Transfer Model for TOVS (RTTOV v10). For the same CMEP5model simulations, it is demonstrated that the computed MSU channel4brightness temperature (T4) trends in the lower stratosphere and MSU channel2brightness temperature (T2) trends in the middle troposphere over globe using the proposed weighting function are equivalent to those calculated by RTTOV, but T4(T2) trends show more cooling (less warming) than those computed using the traditional UAH (University of Alabama at Huntsville) or RSS (Remote Sensing Systems in Santa Rosa, California) static weighting functions. The new static weighting function not only reduces the computational cost, but also reveals reasons why trends using a radiative transfer model are different from those using a traditional static weighting function. This study also shows that CMIP5model simulated T4(T2) trends using the traditional UAH or RSS static weighting functions show less cooling (more warming) than satellite observations over the globe and the tropics. Although not completely removed, this difference can be reduced using the proposed weighting function to some extent, especially over the tropics. This work aims to explore the reasons for the trend differences and to see to what extent they are related to the inaccurate weighting functions. This would also help distinguish other sources for trend errors and thus better understand the climate change.The next part of work is to evaluate trends difference between equvelant MSU channels brightness temperature simulated by16CMIP5climate models and satellite observations from three study groups. The results show that in the period of1979-2005, warming trends of CMIP5models simulations are over-estimated in the tropospheric layer, and cooling trends of CMIP5models simulations are under-estimated in the stratospheric layer related to trends of satellite observations.