Heat waves are among the deadliest climate hazards. Yet the relative importance of the physical processes causing their near-surface temperature anomalies (𝑇′)—advection of air from climatologically warmer regions, adiabatic warming in subsiding air and diabatic heating—is still a matter of debate. Here we quantify the importance of these processes by evaluating the 𝑇′ budget along air-parcel backward trajectories. We first show that the extreme near-surface 𝑇′ during the June 2021 heat wave in western North America was produced primarily by diabatic heating and, to a smaller extent, by adiabatic warming. Systematically decomposing 𝑇′ during the hottest days of each year (TX1day events) in 1979–2020 globally, we find strong geographical variations with a dominance of advection over mid-latitude oceans, adiabatic warming near mountain ranges and diabatic heating over tropical and subtropical land masses. In many regions, however, TX1day events arise from a combination of these processes. In the global mean, TX1day anomalies form along trajectories over roughly 60 h and 1,000 km, although with large regional variability. This study thus reveals inherently non-local and regionally distinct formation pathways of hot extremes, quantifies the crucial factors determining their magnitude and enables new quantitative ways of climate model evaluation regarding hot extremes.

热浪是最致命的气候危害之一。然而，导致其近地表温度异常 ( 𝑇 ′) 的物理过程的相对重要性——来自气候较温暖地区的空气平流、下沉空气中的绝热变暖和绝热加热——仍然是一个有争议的问题。在这里，我们通过评估沿空运包裹后向轨迹的𝑇 ′ 预算来量化这些过程的重要性。我们首先表明，2021 年 6 月北美西部热浪期间的极端近地表𝑇 '主要是由绝热加热产生的，在较小程度上是由绝热变暖产生的。系统分解𝑇' 在 1979 年至 2020 年全球每年最热的日子（TX1day 事件）期间，我们发现强烈的地理变化以中纬度海洋上的平流为主，山脉附近的绝热变暖以及热带和亚热带陆地上的绝热加热。然而，在许多地区，TX1day 事件是由这些过程的组合引起的。在全球平均值中，TX1day 异常沿着大约 60 小时和 1,000 公里的轨迹形成，尽管具有很大的区域变异性。因此，这项研究揭示了极端高温的内在非局部和区域不同的形成途径，量化了决定其强度的关键因素，并为极端高温气候模型评估提供了新的定量方法。