The Global High-Resolution Soil-Water Balance dataset provides hydrological raster data (ESRI Grid format) describing actual evapo-transpiration and soil water deficit with resolution of 30 arc seconds (~920 m at equator). This Soil-Water Balance utilizes the WorldClim and Global-PET database as primary input. The method used to derive these datasets is described in the dataset documentation. The results highlight specifically the climatic influence on hydrological dimensions regulating vegetation suitability. This dataset is based on modeling and analyses by Antonio Trabucco (Forest Ecology and Management Research Group, K.U. Leuven), with the support of the International Water Management Institute (IWMI)and the International Centre for Integrated Mountain Development (ICIMOD), and are provided online by the CGIAR-CSI Consortium for Spatial Information with the support of the International Center for Tropical Agriculture (CIAT).

CIAT has been spatially-downscaling GCM data also up to 30 arc-second (a.k.a., 1 km) and posting their collection online at http://gisweb.ciat.cgiar.org/GCMPage.

The datasets, being developed by Julian Ramirez This e-mail address is being protected from spambots. You need JavaScript enabled to view it  and Andy Jarvis This e-mail address is being protected from spambots. You need JavaScript enabled to view it , are part of the International Centre for Tropical Agriculture (CIAT) climate change downscaled data, developed in the Decision and Policy Analysis (DAPA) program. The data have been originally downloaded from the IPCC data portal and re-processed using an spline interpolation algorithm of the anomalies and the current distribution of climates from the WorldClim database developed by Hijmans et al. (2005).

All GCM's presented here come from the fourth and third IPCC Assessment Reports, but in further updates of the webpage, only models from the 4AR will be kept.

From the website:

We assume that the geographies of changes in climates don't vary too much at regional scales and that the relationships between the different variables will remain basically the same in the future. The surfaces provided here are thus generated using an empirical downscaling approach instead of re-modeling the climate patterns using an RCM (Regional Climate Model).

The downscaling process we follow is mainly the following: (1) calculation of anomalies (if they're not provided directly by IPCC) by simply subtracting each variable's future values with the baseline (both provided by IPCC), (2) interpolation of anomalies to a 30 arc-seconds resolution (approx. 1km) and (3) addition of the interpolated anomalies to the current distribution of climates in WorldClim, for temperature we make an absolute sum, but for precipitation (as there are differences between the GCM baseline and our WorldClim baseline), we use the
relative difference. 

All the datasets are available to direct-download from the site:
http://gisweb.ciat.cgiar.org/GCMPage/download.html

In January 2010, the University of East Anglia officially released the CRU-TS 3.0 Climate Database (See the official data release at http://badc.nerc.ac.uk/data/cru). This new version of database covers from 1901 to 2006*, globally at 0.5 degree spatial resolution on land areas. To help facilitate the use of this database in the CSI community using GIS software, we have converted the raw data into the ESRI ASCII raster format.

Spatially-downscaled climate projection datasets by Peter Jones (Waen Associates/CIAT), Philip Thornton (ILRI), and Jens Heinke (PIK) are now available to download at http://futureclim.info. The datasets currently include 5 arc-minute (a.k.a., 10 km) resolution, global coverage of three core variables (i.e., monthly rainfall, tmin, and tmax) and two derived variables (solar radiation and rainy days) for biophysical modeling applications, for four GCM's (CNRM-CM3, CSIRO-Mk3.0, ECHam5, and MIROC 3.2), three emission scenarios (A1B, A2, B1) for two time slices (2030 and 2050).