Following the Clausius-Clapeyron (CC) relationship, the atmospheric moisture holding capacity should increase with warming temperatures at a rate of approximately 7%/℃. This scaling has been regarded as an important starting point for projecting Extreme precipitation (P_e). However, a large and growing body of evidence reports divergent sensitivities of P_e to near-surface temperatures (T), varying from super CC-scaling rates (i.e., >7%/℃) to decreases with rising temperatures. The reason for this divergence is that CC scaling only tells part of the story: P_e is also a function of local vertical motion (atmospheric dynamics) and available atmospheric moisture (thermodynamics). Variations in large-scale atmospheric circulation and local weather patterns can also contribute to deviations from CC scaling. For instance, enhanced vertical velocity associated with deep convection in the tropics or extratropical cyclones may intensify P_e and lead to super CC rates. Thermodynamic factors such as a less steep moist-adiabatic lapse rate with warming can also decrease the vertically integrated saturation specific humidity and thus weaken precipitation sensitivity.

Different physical processes are involved in precipitation generation in diverse geographical areas. Three EPS (i.e., P_e-T scaling relationship) regimes have been widely reported in the literature. Monotonically increasing EPS is usually found in high latitudes, while the tropics are dominated by monotonically decreasing scaling. Over most regions of the globe, both observational records and model simulations exhibit a “hook-like” structure, in which precipitation intensity generally increases with warming but decreases beyond a peak-point temperature (T_pp). More recently, it has been found that the EPS is not stationary but can shift along wetter and warmer directions in future climates. However, the physical mechanisms underpinning the contribution of atmospheric thermodynamics and dynamics to EPS, as well as the shifting pattern of thermodynamic versus dynamic effects under climate warming, remain unexamined.

Here, we decompose the dynamic and thermodynamic components of EPS to explore the underlying physical mechanisms behind EPS and its shifting patterns under climate change. First, we employ a physical diagnostic scaling approach to detect three EPS regimes using ERA5 reanalysis data (20 pressure levels) and climate simulations from the latest Coupled Model Inter-comparison Project phase 6 (CMIP6, all available runs with vertical velocity and temperature profiles at 19 pressure levels) during both reference (1985-2014) and future (2071-2100) climates. Then, the three EPS regimes are attributed to one dynamic component (represented by changes in vertical velocity, ω) and three thermodynamic components, i.e., the pressure (pPR), temperature (pT) and lapse rate (pLR) components (Figure 1). Moreover, we project future shifts in the three regimes under climate change (i.e., we diagnose EPS anomalies by comparing future EPS relative to historical EPS) and disentangle the drivers of future EPS anomalies within the CMIP6 experiments. This study represents, to our knowledge, the detailed understanding of the physical mechanisms responsible for changing EPS regimes in a warming world.

Counter to our intuition, we find that the thermodynamic components do not always contribute to precipitation intensification. Although the thermodynamic temperature (pT) term, or CC scaling, strongly enhances EPS, especially in mid-to-high latitudes, the lapse rate term (pLR) and the pressure component (pPR) can weaken EPS. Specifically, the pLR term not only correlates with saturation specific humidity, but is also affected by atmospheric stability and convective. However, these processes are difficult to capture with current convection parameterizations in GCMs, which may result in an underestimation of this term. The mechanisms behind extreme precipitation scaling are quite complex in some regions. In tropical regions where EPS is governed by the dynamic term, extreme precipitation is typically associated with storms and cyclones. Other synoptic patterns, including moisture transport from low level jets and upper-level atmospheric rivers, also play a role in modulating EPS. In mid-latitude land regions such as over the Southeast and Midwestern US, Southeast China, and Southern Australia, deep convection dominates extreme precipitation, as indicated by very large convective available potential energy (CAPE) and convective inhibition (CIN) anomalies. This convection is accompanied by high total column water vapor and strong moisture convergence during extreme precipitation. Interestingly, these regions all firmly exhibit a hook-like scaling, using both the binning scaling and quantile regression approaches, at both hourly and daily temporal scales.

In conclusion, this work provides the global quantitative assessment of how thermodynamic and dynamic factors regulate P_e shifts in a warming climate. We systematically disentangle the physical mechanisms, in terms of updraft velocity and moist-adiabatic saturation specific humidity, and assess how they may govern future EPS anomalies. We find that the internal thermodynamic components do not always contribute positively to precipitation intensification. Moreover, large EPS anomalies are projected in the future climate, with strongly positive anomalies emerging over oceans and negative anomalies over land areas. These large EPS anomalies are predominantly driven by atmospheric dynamics over ~80.9% of the globe, particularly in tropical and subtropical regions, whereas the thermodynamic effects are much more stable in a continued warming world. Under the impacts of atmospheric dynamics and thermodynamics, extreme precipitation events are projected to continue increasing across the globe, both in terms of their mean and variability, intensifying future runoff extremes. Our findings suggest an urgent need to increase societal resilience to this changing environment, as precipitation and runoff extremes are likely to intensify in a warming climate, causing major challenges for existing infrastructure and human society.