Arctic cloud properties and associated radiative effects in the three newer reanalysis datasets (ERA5, MERRA-2, JRA-55): Discrepancies and possible causes

Arctic cloud properties and associated radiative effects were investigated by comparing three newly available reanalysis datasets (i.e., ERA5, MERRA-2, and JRA-55) with CloudSat-CALIPSO (CC) observations. Cloud fractions (CFs) in different datasets exhibited different spatial and temporal variations: ERA5 and MERRA-2 exhibited large cloudiness over sea ice in winter, while JRA-55 exhibited relatively larger cloudiness over the open ocean in summer. This difference was related primarily to excessive low-level clouds over sea ice in winter in ERA5 and MERRA-2, especially near the surface (i.e., 1000 hPa). All datasets used in this study exhibited similar spatial distribution of ice water path and liquid water path (LWP), but the reanalyses underestimated both ice and liquid compared to CloudSat retrievals. Despite their similar spatial patterns, relative proportions were considerably different among the reanalyses, which can be attributed to the different phase-partitioning parameterizations. Longwave (LW) cloud radiative effects (CREs) exhibited more consistent variations among the datasets compared to CFs, with large discrepancies being evident in winter. In addition, MERRA-2 exhibited significantly larger LW CRE at low LWPs (i.e., LWP < 30 g m(-2)) owing to the large CF near the surface. Shortwave (SW) CREs were also affected by the CF and cloud water path (CWP), thereby exhibiting lower values in JRA-55 owing to low CF and CWP. The SW CRE change rate with increasing LWP was largest in MERRA-2, followed by JRA-55 and ERA5; these differences are associated with different dependence of the reduced downward SW radiation by clouds on LWP among the different datasets.