Revisiting the spatiotemporal characteristics of past and future global warming

It is still an open question, which processes lead to the spatiotemporal specifications of observed near-surface temperature changes over recent decades. Here, we contribute to this debate by investigating a large number of theory-based atmospheric fields referring to the radiation and energy budget and to atmospheric dynamics that may serve as predictors for local temperature changes. The predictors are linked to temperature trends from reanalysis and climate model data, using a sophisticated spatial and temporal statistical model. Temperature changes since the mid-20th century exhibit distinct regional and seasonal differences. After 1990, the near-surface warming rate is more enhanced over landmasses rather than oceans and roughly increases with latitude in both hemispheres. While none of the considered predictors solitarily accounts for the spatial heterogeneity of recent temperature trends, their linear combination largely reproduces the observed cooling pattern during the mid-20th century and the enhanced warming pattern after 1990. This excludes high-altitude areas, sea ice margins and upwelling regions where local feedbacks and nonlinear processes prevail. The leading predictors pertain to radiative processes, especially downward longwave radiation, and changes in sensible heat fluxes. In the low latitudes, dynamical processes such as temperature advection and energy flux divergence also play a role. Until the end of the 21st century, the warming rate and its ocean-land contrast steadily increase. The underlying mechanisms are the same as the ones already established in present-day climate, but near-surface temperature follows more straightly the imposed greenhouse gas scenario. Climate models have different skills in reproducing the observed trend pattern but exhibit more or less the same mechanisms of temperature control.