Research on Retrieving Statistical Characteristics of Global Subsurface Marine Heatwaves

This paper aims to retrieve global subsurface marine heatwaves (MHW) on a long-term scale, in order to characterize their trends over an extended period. Through preliminary data analysis, this paper found that there was a significant spatial correlation between the statistical characteristics of marine heatwaves in the subsurface layers and the monthly average seawater temperature. Guided by this statistical principle, the multiple-replicate geographically weighted regression (MRGWR) model was used as the retrieval model. The bandwidth was selected based on the characteristic that ocean circulation can be approximately considered constant within a 2° range. Furthermore, two indicator systems, namely univariate indicators and multivariate indicators, were constructed. The retrieval performance of the MRGWR model was compared with that of the generalized linear model (GLM), which was previously used for retrieving marine heatwaves, confirming the significant advantage of using multivariate indicators fitting to the MRGWR model in the retrieval of subsurface marine heatwaves. Using the selected optimal retrieval model and indicators, the statistical characteristics of global subsurface marine heatwaves from 1940 to 2021 were retrieved to characterize their long-term trends, which showed a clear correspondence with the rising trend of seawater temperature. This paper provides strong evidence for the impact of global warming caused by human activities on subsurface marine heatwaves.