MULTIVARIATE MULTISITE STATISTICAL DOWNSCALING OF ATMOSPHERE-OCEAN GENERAL CIRCULATION MODEL OUTPUTS OVER THE CANADIAN PRAIRIE PROVINCES

Abstract

Atmosphere-Ocean General Circulation Models (AOGCMs) are the primary tool for modelling global climate change in the future. However, their coarse spatial resolution does not permit direct application for local scale impact studies. Therefore, either dynamical or statistical downscaling techniques are used for translating AOGCM outputs to local scale climatic variables. The main goal of this study was to improve our understanding of the historical and future climate change at local-scale in the Canadian Prairie Provinces (CPPs) of Alberta, Saskatchewan and Manitoba, comprising 47 diverse watersheds. Given the vast nature of the study area and paucity of recorded data, a novel approach for identifying homogeneous regions for regionalization of precipitation characteristics for the CPPs was proposed. This approach incorporated information about predictors ― large-scale atmospheric covariates from the National Center for Environmental Prediction (NCEP) Reanalysis-I, teleconnection indices and geographical site attributes that impact spatial patterns of precipitation in order to delineate homogeneous precipitation regions using a combination of multivariate approaches. This resulted in the delimitation of five homogeneous climatic regions which were validated independently for homogeneity using statistics computed from observations recorded at 120 stations across the CPPs. For multisite multivariate statistical downscaling, an approach based on the Generalized Linear Model (GLM) framework was developed to downscale daily observations of precipitation and minimum and maximum temperatures from 120 sites located across the CPPs. First, the aforementioned predictors and observed daily precipitation and temperature records were used to calibrate GLMs for the 1971–2000 period. Then the calibrated GLMs were used to generate daily sequences of precipitation and temperatures for the 1962–2005 historical (conditioned on NCEP predictors), and future period (2006–2100) using outputs from six CMIP5 (Coupled Model Intercomparison Project Phase-5) AOGCMs corresponding to Representative Concentration Pathway (RCP): RCP2.6, RCP4.5, and RCP8.5 scenarios. The results indicated that the fitted GLMs were able to capture spatiotemporal characteristics of observed climatic fields. According to the downscaled future climate, mean precipitation is projected to increase in summer and decrease in winter while minimum temperature is expected to warm faster than the maximum temperature. Climate extremes are projected to intensify with increased radiative forcing.