The CliMond climate dataset consists of gridded historical climate data and some future climate scenario data at 10' or 30' spatial resolution. The underlying historical data is sourced from the Worldclim and the Climate Research Unit (CRU) (CL1.0 and CL2.0) datasets. These data were reformatted, adjusted and recombined to generate all of the required variables as detailed in Kriticos et al. (2012). The Worldclim dataset draws primarily on data between 1961-1990, though station records from 1950 to 2000 were used occasionally to fill gaps in records. The CRU datasets draw exclusively on data from 1961-1990.

The precipitation data was taken from the Worldclim dataset because the 10' data in the CRU dataset was found to suffer from inconsistencies. Conversely, the CRU dataset had more reliable temperature data, which was most apparent in Greenland. Relative Humidity data for 09:00 and 15:00 hours was calculated using temperature and vapour pressure variables. The CliMond data have been produced for only those locations where a full set of variables are available. In the southern part of the island of Hawai'i, the CRU radiation dataset is missing data in the 10' dataset. We are exploring ways to patch this hole.

A set of climate change projections has been applied to the baseline climatologies using methods described in Kriticos et al. (2012). We used data from two Global Climate Models to create estimates of the future climate for a range of future dates. The currently available GCM data were prepared for the IPCC fourth assessment report (AR4). We selected the CSIRO Mk3.0 model and the Miroc-H models because during 'spin-up', they performed well when compared with a subset of historical observation data, they had a relatively small spatial resolution, and because they included all the necessary variables at the correct temporal resolution. Spin-up is the process where the GCMs are run for some period (often 1900-1999) so that the effects of the starting conditions are washed out by the time they are forced using injections of greenhouse gases. The degree of congruence between the models and historical data was measured using the M-skill score (Watterson 1996). These methods preserve the topographic signal in the baseline climatology. Modellers considering using the CliMond future climate scenario data may wish to read Harris et al. (2014) for a gentle introduction to the subject.