Robust global detection of forced changes in mean and extreme precipitation despite observational disagreement on the magnitude of change

Abstract. Detection and attribution (D&A) of forced precipitation change are challenging due to internal variability, limited spatial, temporal coverage observational records model uncertainty. These factors result in a low signal-to-noise ratio potential regional even global trends. Here, we use statistical method – ridge regression create physically interpretable fingerprints for the detection changes mean extreme with high ratio. The constructed using Coupled Model Intercomparison Project phase 6 (CMIP6) multi-model output masked match three gridded datasets GHCNDEX, HadEX3, GPCC then applied these assess degree detectable real-world climate period 1951–2020. We show that signature is detected all metrics precipitation. Forced still from spatial patterns if trend removed data. This shows beyond robust increases confidence method's power as well models' ability capture relevant processes contribute large-scale change. also find, however, detectability depends on dataset used. Not only differences but uncertainty disagreement, exemplified by times emergence variability ranging 1998 2004 among datasets. Furthermore, different choices over which computed levels agreement between observations projections. sensitivities may explain apparent contradictions recent studies whether models under- or overestimate observed increase Lastly, found rely primarily signal extratropical Northern Hemisphere, at least partly potentially presence more Hemisphere general.