2022 Volume 100 Issue 5 Pages 807-824
Twentieth-century atmospheric reanalysis datasets are important for understanding climate in the early era of the century. This paper first compares two sets of the 20th-century atmospheric reanalyses, the NOAA-CIRES-DOE 20th-Century Reanalysis Version 3 (20CRv3) and the European Centre for Medium-Range Weather Forecasts (ECMWF) 20th-Century Reanalysis (ERA20C), as far as the summer low-level cross-equatorial flows (CEFs) over the Asian-Australian monsoon region are concerned. The results show evident regional differences in the intensity of individual branches of CEFs between the two reanalyses despite an overall agreement in the climatological seasonal mean and variability. At the interannual timescale, significant differences are observed prior to 1925 and in the 1940s. During the two periods, there are often opposite variations in Somali CEF in the two datasets, along with obvious different amplitudes in the Bay of Bengal (BOB) and Australian CEFs. At the interdecadal timescale, the two datasets have different periodicities in Somali CEF and have a greater fluctuation of BOB CEF after 1925 in ERA20C than 20CRv3, as well as an opposite decadal variation in the Australian CEF prior to 1940 and in the 1960s. As for the long-term trend, both the Somali and BOB CEFs exhibit intensification in both the datasets, but the intensification amplitude is bigger in 20CRv3 than in ERA20C for Somali CEF; the Australian CEF exhibits a weakening trend in both the datasets but is less evident in 20CRv3. To figure out which of the two datasets is relatively more reliable, the observed cross-equatorial meridional gradient of sea-level pressure index and the Indian summer monsoon rainfall index, which both have longer instrumental records, are used as benchmarks to validate the CEFs in view of their close connections. The results indicate that ERA20C is more reliable and thus more suitable for investigating decadal climate variability of the 20th century across the hemispheres.
