High resolution future scenario climate data for North America

Eight monthly climate scenario data sets covering North America at ~10 km resolution were constructed using General Circulation Model (GCM) output from four models (Canadian CGCM2, UK HadCM3, Australian CSIRO Mk2 and European Union ECHAM4). Each data set included a projection for the period 1991-2100, forced by one of the IPCC SRES A2 and B2 emissions scenarios. GCM variables were monthly precipitation and solar radiation, mean vapour pressure and wind speed, and mean daily minimum and maximum temperatures. Data were first normalized to differences (temperature), or ratios (other variables), of the means for the simulated period 1961-90. These monthly “delta” terms were then interpolated as functions of latitude and longitude using thin-plate smoothing splines as implemented in ANUSPLIN. The resulting spline surfaces were used to create monthly grids at 0.08333º latitude/longitude resolution. These were combined with interpolated maps of observed 30-year climate normals, to produce physically consistent gridded scenarios of future climate. Each combines spatial variability observed in present-day climate with the spatio-temporal variability simulated by one GCM. Averaged across North America, results (reported here for temperature and precipitation only) reveal quite consistent long-term trends, particularly in the GCMs’ differential responses to the A2 and B2 emission scenarios. The scenarios differ more in their projections of interannual variability, both compared to historical data and in their responses to greenhouse gas forcing. Maps of selected data at 0.5º resolution reveal rich spatial structure dominated by elevation effects on observed climate, but vary in details due mainly to differences among the GCMs. The data sets will be useful for assessing impacts of transient changes in climate on ecosystems and other studies where long-term projections of future climate are required.