Uncertainties of the South China Sea summer monsoon and its relationship with sea surface temperature from different reanalysis datasets

The South China Sea summer monsoon (SCSSM) onset and withdrawal have critical impacts on the climate of East Asia. However, using different reanalysis data may lead to the uncertainty of the SCSSM. This study uses reanalysis datasets including JRA-55, ERA5, NCEP-NCAR NCEP-1 and NCEP-DOE NCEP-2 to calculate onset and withdrawal dates and index of SCSSM from 1991 to 2020, analyses the uncertainty and figures out the possible reasons. Results reveal that the four reanalysis datasets have obvious uncertainty in calculating the withdrawal date of SCSSM, with the deviation being up to 3.4 pentads. Further analysis shows that large differences in thermal conditions and lead to the uncertainties. From the perspective of precipitation, Greater uncertainty exists in the relationship between monsoon retreat and September-October precipitation, NCEP1 and NCEP2 datasets show weak correlation in southern China and Indonesia. Moreover, the uncertainty can be also found in the correlations of SCSSM with global sea surface temperature (SST) anomalies. JRA-55 and ERA5 show a strong correlation of with SST in several areas, and exhibit a distinct Pacific decadal oscillation (PDO) mode with monsoon onset, while NCEP-1 and NCEP-2 have a weaker correlation with SST anomalies than JRA-55 or ERA5. Comparing multiple factors comprehensively, this paper concludes that JRA-55 may be more representative in the monsoon study, followed by ERA data. In short, depending on the specific research questions, researchers can choose appropriate data sources to study the SCSSM based on the different aspects and the uncertainty also deserve our special attention when selecting the reanalysis datasets. Using different reanalysis datasets to study the South China Sea summer monsoon may introduces uncertainties. From the perspective of dynamical, thermodynamical elements, and precipitation related to the monsoon, this paper suggests that JRA55 data were more representative in the monsoon study, followed by ERA5. Researchers should select appropriate data based on their own work.image