Search

In recent years, there has been an increasing occurrence of compound heat-humidity (HH) events worldwide, leading to excessive morbidity and mortality. Southeast China has been particularly affected by HH events over the past decade. However, the seasonal prediction of HH events in Southeast China remains a considerable challenge. Based on the year-to-year increment (DY) approach and the empirical orthogonal function analysis, a robust prediction model is developed in this study for the midsummer HH events in Southeast China during 1981–2022. Three prediction models are individually constructed for the first three principal components (PCs) of the DY for HH events (hereafter referred to as HH_DY). The correlation coefficients (CCs) between the predicted and observed PCs exceed 0.75 in the cross-validation test for 1981–2022 and 0.80 in independent hindcasts for 2013–2022. The predicted graphical maps of HH_DY are reconstructed by the projecting the fitted three PCs onto the observed first three empirical orthogonal function modes, and the geographical patterns of HH events are obtained by adding the simulated HH_DY to the observed HH events for the preceding year. In the cross-validation test, the spatial features of the predicted HH events are highly accordant with those of the observations in nine extreme compound HH years, with area CC values ranging from 0.89 to 0.95. In the independent hindcasts for the past decade, the area CC values exceed 0.76 in all years and are >0.80 in eight years. Moreover, the predicted high-incidence areas of HH events are highly consistent with the observations in both the cross-validation test and independent hindcasts. The reliable prediction for geographical distributions of HH events in Southeast China is of great significance for human adaptation to future climate warming. • Using interannual increment methods, we developed a prediction model of midsummer compound HH events in Southeast China. • The area CCs range from 0.89 to 0.95 in nine extreme years of compound HH years in cross-validation test. • The predicted high-incidence areas of HH events agree with the observations in cross-validation and independent hindcasts. [ABSTRACT FROM AUTHOR]