Evaluation of Tropical cyclone genesis frequency derived from Six Reanalysis Datasets

The tropical cyclone genesis frequency (TCGF) provides essential insights to understand TC activities and mitigate their impacts on human communities. Cyclogenesis is significantly governed by large-scale environment, but this relationship is weak in general circulation models. Generated by advanced operational forecast models with observational constraints, the atmospheric reanalysis represents the best estimate of the large-scale environments, bridging the gap between TCGF simulations and observations. Here, to identify whether the large-scale environments factors are sufficient to the successful reproduction of TCGF over different time scales, we evaluate the performance of reanalysis datasets by using two genesis potential indices. The six sets of reanalysis data include ERA5, JRA-55, CRA-40, CFSR, NCEP-2, and MERRA-2. We show evidence that, based solely on large-scale environmental factors, all reanalysis datasets generally reproduce the climatological distribution, the seasonal cycle, and the interannual variability in the basins expect for the South Indian Ocean (SI), but show limited abilities to reproduce the observed linear trend of TCGF. The bad performance over SI is partly attributed to the processes related to TC seed vortices. Meanwhile, there are large inter-reanalysis discrepancies in capturing the linear trend of TCGF over the western North Pacific, which are attributed to the uncertainties in 500 hPa vertical velocity and the thermodynamic factors. The temporally discontinuous observing system affects the accuracy of reanalysis data in depicting the linear trends of TCGF. Our findings highlight the importance of considering the temporal heterogeneity in reanalysis datasets when investigating the linear trend of TCGF, with particular attention to the consistency and stability of assimilated observations over time.