Tropical cyclone characteristics in the MERRA-2 reanalysis and AMIP simulations

This study evaluates the tropical cyclone (TC) activity in two high-resolution data sets-MERRA-2 Reanalysis (Modern-Era Retrospective Analysis for Research and Applications, Version 2) and MERRA-2 AMIP (Atmospheric Model Intercomparison Project). These data sets use the same atmospheric model, the Goddard Earth Observing System Model, Version 5 (GEOS-5) during the same period. However, while MERRA-2 AMIP is a free-running atmospheric simulation forced only with sea surface temperature (SST), MERRA-2 Reanalysis uses an advanced data assimilation system to include a large variety of data sets. Thus, we analyze (1) the sensitivity of TC activity to the model forcing, (2) how well the TCs in both data sets replicate observed TC characteristics, (3) the sensitivity of these results to tracking schemes and thresholds. Standard diagnostics such as the number of tropical cyclones and their intensity distribution are very similar in the AMIP model and the reanalysis. TCs in both data sets are weaker than observed, as is typical for the spatial resolution of these global models. Overall, the use of data assimilation in the MERRA-2 Reanalysis does not lead to a significantly better TC climatology than in AMIP. Furthermore, comparison of the MERRA-2 Reanalysis to two other reanalysis data sets shows that MERRA-2 generates fewer, but more intense TCs, than those reanalysis products.