A comparison of global reanalyses to satellite observations: quantifying errors in Cloud Fraction and TOA cloud radiative forcings

Abstract:

As the improvements of observing systems, model parameterizations, and advanced assimilation techniques, a few reanalyses data sets become available; however, it is necessary to understand their strengths and biases before exploiting them. In this study, we investigate five contemporary reanalyses (20CR, CFSR, Era-Interim, JRA-25, and MERRA). In detail, we first compared the cloud fraction, TOA radiation fluxes and cloud radiative forcings using CERES MODIS SYN1deg and EBAF products on the global scale. We then analyze the differences and similarities between these parameters in tropical and mid-latitudes under convective and subsidence regimes. It is important to note that CERES MODIS and EBAF results are used as reference to evaluate these simulated parameters, and these data cover the period from 03/2000 to 02/2012. The preliminary results show that most of the reanalyses under estimate cloud fraction on global average, and the SW, LW, and Net CRFs are biased by as much as 3.9 (CFSR), −8.7 (JRA-25), and −5.7 (JRA-25) Wm⁻², respectively. Convective and subsidence regimes are identified as the strong ascending and descending branches of the large general circulation in the tropics and mid-latitudes. This metric helps to identify the physical characteristics in specific areas and the types of clouds that reside there. Clouds are typically over estimated in convective areas while in the subsidence region clouds are under estimated. Sensitivity studies show that these regimes exhibit different regressive behaviors when comparing cloud fraction to CRFs, especially in the LW CRF. The error analysis reveals information that can be useful for modelers so that they may develop better parameterizations for clouds and how they interact with incoming solar and outgoing infrared radiation.