Impact of atmospheric moisture in a rainfall downscaling framework for catchment-scale climate change impact assessment

摘要

Three formulations of a non-parametric stochastic downscaling framework for simulation of daily rainfall at multiple point locations for catchment-scale climate change impact assessment under enhanced greenhouse conditions are evaluated. In the first case, rainfall is downscaled using a system of variables that combine atmospheric circulation indicators (geo-potential height, pressure and derived variables) with time-lagged wetness indicators reflecting a continuous rainfall state. In the second and third cases, the above system of variables is augmented using indicators of atmospheric moisture found relevant in defining the downscaling relationship. The downscaling framework is specified using 43 years of daily rainfall observations at 30 locations near Sydney, Australia, with reanalysis data being used to represent observed atmospheric variables. A single ensemble member of the CSIRO Mark 3 general circulation model (GCM) (corresponding to the SRES A2 emission scenario) is used for comparison of downscaled rainfall for current and future (year 2070) conditions. While all the downscaling formulations show an overall similarity in the downscaled rainfall for the current climate, variations for some rainfall attributes are observed for year 2070 conditions. A variable convergence score (VCS) is employed to assess the relative skill of atmospheric predictors used in the different formulations. The VCS indicates use of temperature depression (difference of dew point and atmospheric temperatures) as an indicator of atmospheric moisture to obtain more realistic estimates of rainfall for year 2070. In general, a decrease in winter and spring and an increase in summer in the frequency of wet days with a decrease in the winter rainfall amount in the year 2070 is projected. Also, inland stations are found to be more sensitive to the expected future climate changes in comparison to the coastal ones.