Premium
Sahel rainfall in multimodel CMIP5 decadal hindcasts
Author(s) -
Martin E. R.,
Thorncroft C.
Publication year - 2014
Publication title -
geophysical research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.007
H-Index - 273
eISSN - 1944-8007
pISSN - 0094-8276
DOI - 10.1002/2014gl059338
Subject(s) - climatology , coupled model intercomparison project , initialization , environmental science , ensemble average , subtropics , sea surface temperature , forecast skill , climate model , climate change , geology , oceanography , fishery , computer science , biology , programming language
This study assesses the capability of Coupled Model Intercomparison Project 5 (CMIP5) decadal hindcasts to represent Sahel rainfall and a relative sea surface temperature (SST) index (RSI). The RSI measures the relative difference between subtropical North Atlantic SST and tropical SST and is highly correlated with Sahel rainfall. Ten year predictions from 15 models initialized every 5 years (six initialized every year) beginning in 1960 are evaluated. The hindcasts show increased decadal variability compared to the uninitialized historical simulations, with a larger magnitude drought and smaller RSI in the 1970s and 1980s. The multimodel ensemble mean shows skillful predictions that are more accurate than persistence and historical simulations. Models that were able to realistically simulate the correlation between the RSI and Sahel rainfall in uninitialized historical simulations produced more skillful decadal hindcasts of Sahel rainfall as these models were able to successfully translate the SST that was improved through initialization to more skillful Sahel rainfall predictions.