Interdecadal Variability and Trends in Global Precipitation during the Satellite Era: CMIP6 Simulations vs Satellite Estimates

Abstract:

Interdecadal variability and trends in precipitation and intensity characteristics during the satellite era (1979-present) are explored by means of the satellite-based monthly precipitation analysis from the Global Precipitation Climatology Project (GPCP) in comparison to the outputs of CMIP6 historical and AMIP runs. The analysis includes the comparisons between GPCP and models and also between CMIP and AMIP, possibly distinguishing the effects from surface and radiative forcings and the impact of the relatively short observational record. The ensemble means from various models and from the same models with relatively large ensemble members are applied. In addition to mean precipitation, precipitation intensity especially for wet and dry extremes on the high and low ends, respectively, is quantified using monthly percentiles over large domains and within various latitude bands in the tropics and extra-tropics. Monthly wet extremes are related to wet/deluge conditions under which daily and sub-daily extremes could often, though not necessarily, occur, while monthly dry extremes are directly associated with droughts.

Models (CMIP and AMIP) can reproduce similar climatological characteristics of precipitation as the observation, including both mean precipitation amount and intensity. For example, mean precipitation increases can be seen during the period specifically over tropical ocean. AMIP can generally reproduce similar spatial structures of precipitation trends as GPCP, however discrepancies can readily be seen between GPCP and CMIP. For precipitation intensity, especially in the tropics, GPCP shows a positive trend for the upper one third of the percentiles (Pct≥70^th) and a much weaker positive trend for the lowest percentiles (Pct≤10^th), while negative trends appear for the middle one-half percentiles (20^th-65^th). AMIP agrees with GPCP in terms of the sign of the trends for high and intermediate percentiles. CMIP historical results also agree in the sign of the trends, but the trends are weaker. Comparisons with other types of simulations including the GHG-only, aerosol-only, and nature-only suggest that the observed trends in precipitation amount and intensity during the GPCP period are dominated by a combination of the effects of the Pacific Decadal Oscillation and GHG-related warming.