Shah et al. 2000
, , , , and , 2000: AGCM hindcasts with SST and other forcings: Responses from global to agricultural scales. J. Geophys. Res., 105, 20025-20053, doi:10.1029/2000JD900019.
Multiple realizations of the 1969-1998 time period have been simulated by the GISS AGCM to explore its responsiveness to accumulated forcings, particularly over sensitive agricultural regions. A microwave radiative transfer postprocessor has produced the AGCM's lower tropospheric, tropospheric and lower stratospheric brightness temperature (Tb) time series for correlations with Microwave Sounding Unit (MSU) time series. AGCM regional surface air temperature and precipitation were also correlated with GISTEMP temperature data and with rain gauge data. Seven realizations by the AGCM were forced solely by observed sea surface temperatures. Subsequent runs hindcast January 1969 through April 1998 with an accumulation of forcings: observed sea surface temperatures (SSTs), greenhouse gases, stratospheric volcanic aerosols, stratospheric and tropospheric ozone and tropospheric sulfate and black carbon aerosols. Lower stratospheric Tb correlations between the AGCM and MSU for 1979-1998 reached as high as 0.93 globally given SST, greenhouse gases, volcanic aerosol and stratospheric ozone forcings. Mid-tropospheric Tb correlations reached as high as 0.66 globally and 0.84 across the equatorial, 20°S-20°N band. Oceanic lower tropospheric Tb correlations were less high at 0.59 globally and 0.79 across the equatorial band. Of the sensitive agricultural areas considered, Nordeste in northeastern Brazil was simulated best with mid-tropospheric Tb correlations up to 0.80. The two other agricultural regions, in Africa and in the northern mid-latitudes, suffered from higher levels of non-SST-induced variability. Zimbabwe had a maximum mid-tropospheric correlation of 0.54 while the U.S. Cornbelt reached only 0.25. Hindcast surface temperatures and precipitation were also correlated with observations, up to 0.46 and 0.63 respectively for Nordeste. Correlations between AGCM and observed time series improved with addition of certain atmospheric forcings in zonal bands but not in agricultural regions encompassing only six AGCM gridcells.