Areal downscaling of global climate models: an approach that avoids data remodeling

Abstract

Since the conceptualization and evolution of the global climate models (GCMs), the necessity to resolve the data generated to areal representations below the very large grid boxes of those models has also been apparent. Different approaches, such as statistical downscaling and regional climate modeling, have been developed over the past few years to address this issue. We present an approach to downscaling GCM output data to finer areal resolutions without having to remodel the GCM data. Our algorithm uses a combination of interpolation equations and the histori- cal relationships of observation points within the geographical area of study to resolve the down- scaled predicted values. A comparative analysis of the algorithm's results with those of a regional cli- mate model which used the same GCM data as boundary conditions indicates that the algorithm's results may be more faithful to both the local climatology and the GCM output. Furthermore, in experiments conducted in Barbados for the case study, a mean of 78% of the values generated by this method were within 30% of the observed data, and a mean of 60% of the generated values fell within 20% of the observed values.