Evaluation of mixed layer depth in Korean waters obtained by reanalysis data sets: spatial distribution and interannual variability

The mixed layer depth (MLD) plays a vital role in regulating climate by controlling the exchange of momentum, heat, and moisture between the ocean and atmosphere. Improving the simulation of MLD is therefore crucial for reliable climate predictions and projections. However, studies on the interannual variability of MLD using global reanalysis data are insufficient. In this study, we examined the interannual variability of winter (February) MLD in Korean waters over a 25-year period (1994–2018) using two reanalysis data sets, the HYbrid Coordinate Ocean Model (HYCOM) and CMEMS Global Ocean Reanalysis and Simulation (GLORYS), which have been widely used in climate change studies in Korean waters. The reanalysis MLD data were compared with observational estimates from the Korea Oceanographic Data Center (KODC) and NIFS Serial Oceanographic observations (NSO) for February, the month with the deepest MLD. The spatial distribution is relatively well simulated, but the long-term trend is poorly reproduced. Notably, the models underestimate the long-term mean MLD by approximately 25% in regions influenced by the Ulleung Eddy and the Yellow Sea Warm Current. The underestimated bias in the Ulleung Eddy can be attributed to the insufficient resolution of the reanalysis data sets in capturing the fine-scale structure of the Ulleung Eddy, while the bias in the Yellow Sea Warm Current region is possibly due to lack of tidal mixing in the reanalysis. Furthermore, while the observed MLD shows a deepening trend over most Korean waters during the study period, the models show negligible changes or even a shallowing trend, except in the East/Japan Sea showing a underestimated deepening trend. The models also tend to underestimate the magnitude of interannual variability of the MLD. Empirical Orthogonal Function (EOF) analysis reveals that MLD interannual variability is influenced primarily by variabilities of 10 m wind and 2 m air temperature (~18%), and secondarily by Tsushima Warm Current transport (TWC; ~11%). The TWC transport is closely related to the path of the East Korea Warm Current, suggesting that changes in the current's interannual variability could influence the MLD. Additionally, in other regions, TWC transport is influenced by the Kuroshio current transport, which determines the volume of transport entering Korean waters, thus explaining its association with MLD variability. This finding highlights the importance of oceanic processes in interannual variability in the winter MLD in Korean waters.