Detection‐attribution of global warming at the regional scale: How to deal with precipitation variability?

Abstract

Given the strong natural variability of precipitation, detection of anthropogenic climate change has been mainly based on surface thermal indicators. It is here argued that precipitation variability, whatever its natural or anthropogenic origin, is likely to affect surface temperature and, thereby, to hamper the detection‐attribution of climate change at the regional scale. This issue is illustrated over Sudan and Sahel using the outputs of eleven coupled general circulation models having participated in the latest IPCC simulations. Like in situ observations, models show a strong relationship between detrended seasonal anomalies of surface air temperature and precipitation during the summer monsoon season. Linear regressions are then used to remove the precipitation influence on observed and simulated linear trends in surface air temperature. The results indicate that this strategy is efficient to reduce the spread in the simulated surface warming and make it more consistent with the instrumental record over recent decades.