A Diagnostic Framework for Hot Temperature Trends over Midlatitude Land

Observations over the past four decades show that the annual hottest temperatures (TXx) over Northern Hemisphere midlatitude land have warmed at nearly twice the rate of globalmean surface temperature (GMST), whereas global climate models span a broad range of TXx-to-GMST sensitivities from approximately 1 to 4.5.To interpret model spread in terms of physical processes, we propose a diagnostic framework based on the near-convective neutrality assumption for hot days in Zhang and Boos (2023).This framework decomposes the sensitivity of TXx to GMST into two factors: the sensitivity of TXx to mid-tropospheric temperature 𝑇 500 , and the sensitivity of 𝑇 500 to GMST.For the first factor, ERA5 indicates stronger near-surface warming per unit midtropospheric warming than most CMIP6 members, which is consistent with its smaller increases in near-surface specific humidity on the hottest days than models.The second factor shows substantial inter-experiment spread, to which aerosol forcing makes an important contribution.The framework also extends to summer means and helps interpret regional behavior, including amplified TXx warming over western Europe and muted warming over the U.S. Midwest.SIGNIFICANCE STATEMENT: This study develops a physically grounded diagnostic framework for hot extremes in near-surface air temperatures.It decomposes trends in the annual hottest temperatures into contributions from free-tropospheric warming and near-surface humidity changes, corresponding respectively to a top-down stability control and a near-surface moisture compensation.The framework can be applied at regional or even grid-point level and, unlike existing ones, does not require rerunning climate model simulations, making it readily applicable to a variety of datasets including observations, reanalysis, and model outputs.This then greatly facilitates identifying the physical causes of model biases in simulating the hot extremes of midlatitude temperatures.