Water Vapor Feedbacks in the ECMWF Reanalyses and Hadley Centre Climate Model

Abstract

Many studies have been made of the water vapor feedback, in both satellite data and climate model simulations. Most infer the magnitude of the feedback from the variability present in geographical distributions of the key variables, or from their seasonal variations, often using data only over the oceans. It is argued that a more direct measure of the feedback should come from the interannual variability of global mean quantities, because this timescale and space scale is more appropriate for such a global phenomenon. To investigate this suggestion, the feedback derived from the simulations of clear-sky longwave fluxes (CLERA), which used data from the 15-yr reanalysis project of the European Centre for Medium-Range Weather Forecasts, is compared with simulations by the latest version of the Hadley Centre climate model. Results are taken from an integration of the atmosphere-only version of the climate model with prescribed sea surface temperatures, as well as from a control and a global warming simulation by the coupled ocean?atmosphere version. There is broad consistency between the results from CLERA and the climate model as to the strength of the feedback, although there is considerable scatter in the CLERA results. The signal of changes in the well-mixed greenhouse gases is weak in CLERA but is dominant in the global warming simulation and has to be removed in order to diagnose the water vapor feedback. This result has implications for the exploitation of long time series of satellite and other data to study this and other feedbacks.