We are currently working on an climate grids and a software package for South America, similar to those that we have developed for Africa, North America and Europe.

Historical data will comprise monthly climate estimates from 1901-present, in addition to seasonal, annual, decadal and 30-year climate normal averages. Future projections will be based on the latest Coupled Model Intercomparison Project Phase 6 (CMIP6) archives, from which we develop custom regional ensemble projections for the 2020s, 2050s and 2080s.

Climate grids will include 4 monthly (Tmin, Tmax, Tave, Prec), 4 seasonal, and 25 annual variables, including many economically or biologically relevant variables such as growing and chilling degree days, heating and cooling degree days, Hargrave's moisture deficit and reference evaporation, beginning and end of the frost-free period, etc. In total, this database will comprise more than 20,000 climate grids for South America.

To develop the climate grids, we employ a deep neural network that learns climatic patterns from geographic and topographic predictor variables, such as elevation, topographic indices, distances to lakes and ocean, and MERRA wind data to modify distance and topographic variables as described by Namiiro et al (2025). This allows for accurate representation of local climate patterns in complex terrain, for example precipitation due to orographic lift on the windward side of mountain ranges, and rain shadows on leeward facing slopes.

Please contract andreas.hamann@ualberta.ca to receive updates, help with testing, or contribute data once we have a release candidate of the database and software.