Variation Characteristics And Constraint Projection Of Humid-heat Extreme Events In Southern China Based On Homogenized Observation

The health risks associated with extreme high temperature are widely recognized and have attracted significant attention.The impact of heatwaves,particularly those characterized by high relative humidity,presents a more severe risk to human health.A precise understanding of the features,causes and future risks of such events is a scientific prerequisite for effective intervention and mitigation efforts.While previous studies have investigated this subject extensively,they often neglect the heterogeneity problem in relative humidity data,which leads to biased estimation of the changes in extreme humid-heat events.Studies have shown that this heterogeneity mainly originates from changes in the observation system of Chinese station data,and the resulting biases can further mislead assessments of health risks and decision-making processes.Therefore,accurately assessing the heterogeneity of relative humidity is crucial for evaluating the health impacts of humid-heat events and developing adaptive measures.In this paper,we perform detailed comparisons among more than ten original observational datasets,reanalysis datasets and homogenized datasets to reveal the serious heterogeneity issues in common datasets such as Had ISD(H)and ERA5.By utilizing the Heat Index(HI),Wet-Bulb Temperature(Tw),and "simplified" Wet-Bulb Globe Temperature(s WBGT)as measures of humid-heat extremes,we quantify the estimation error of extreme humid-heat events in southern China caused by the heterogeneity of relative humidity.The paper further conducts constrained projections on future humid-heat extreme temperatures in China based on homogenized observational data and 22 models from the sixth phase of the Coupled Model Intercomparison Project(CMIP6)under two different Shared Socioeconomic Pathways(SSP1-2.6 and SSP2-4.5).We apply advanced constraint methods,such as model selection,emergent constraints,and model weighting,to conduct predictive analyses.The main conclusions obtained are as follows:(1)Through the comparison of more than 10 sets of raw observations,reanalysis datasets,and homogenized corrected datasets,we have discovered that the issue of heterogeneity in relative humidity data is particularly prominent in the humid-heat regions of southern China during 1979～2013.This has resulted in unrealistic strong drying trends in relative humidity,with an exaggeration of 2 to 3 times in the drying magnitude.(2)At the local scale,the heterogeneity of the data has resulted in underestimations of the frequency and intensity of extreme heat and humidity events by over 1.2days/decade and 0.07%/decade in the past,respectively,due to the biased estimation of the dry trend in the relative humidity trend of 1%/decade.At the regional average scale,underestimated trends of 1.9 days/decade and 0.5%/decade have been observed in the frequency and intensity of extreme events in southern China,respectively,due to the heterogeneity of the humidity observations.(3)During extreme humid-heat events,homogenized observations show relative humidity levels approximately 10% lower than non-homogenized observations.(4)Different humid-heat stress indices exhibit varying sensitivity to relative humidity deviations.Among them,the Heat Index(HI)is particularly sensitive to relative humidity deviations as it involves an exponential combination of temperature and relative humidity.When there are substantial deviations in relative humidity,the underestimation of HI events is more significant compared to other types of events.(5)Through constrained projection,we found that under a moderate warming scenario,utilizing homogenized observational constraints on models can effectively reduce the uncertainty range of future changes in the frequency of humid-heat extreme events,with a reduction exceeding 12%.However,the estimated model mean is dependent on the chosen constraint method,humid-heat index,and the uncertainty of observational data.Under a lower warming scenario,there are limitations in improving the uncertainty range of intensity and frequency changes of humid-heat extreme events(the lower limit may be below 0).Our study also indicates that using non-homogeneous observational data as constraint references significantly underestimates the magnitude of the increase in the frequency of future humid-heat extreme events.