Downscaling regional climate model outputs for the Caribbean using a weather generator

Abstract

Locally relevant scenarios of daily weather variables that represent the best knowledge of the present climate and projections of future climate change are needed by planners and managers to inform management and adaptation to climate change decisions. Information of this kind for the future is only readily available for a few developed country regions of the world. For many less-developed regions, it is often difficult to find series of observed daily weather data to assist in planning decisions. This study applies a previously developed single-site weather generator (WG) to the Caribbean, using examples from Belize in the west to Barbados in the east. The purpose of this development is to provide users in the region with generated sequences of possible future daily weather that they can use in a number of impact sectors. The WG is first calibrated for a number of sites across the region and the goodness of fit of the WG against the daily station observations assessed. Particular attention is focussed on the ability of the precipitation component of the WG to generate realistic extreme values for the calibration or control period. The WG is then modified using change factors (CFs) derived from regional climate model projections (control and future) to simulate future 30-year scenarios centred on the 2020s, 2050s and 2080s. Changes between the control period and the three futures are illustrated not just by changes in average temperatures and precipitation amounts but also by a number of well-used measures of extremes (very warm days/nights, the heaviest 5-day precipitation total in a month, counts of the number of precipitation events above specific thresholds and the number of consecutive dry days).