Downscaling precipitation and temperature under climate change over semi-arid regions of southwestern United States of America

Two different space-time models are developed to estimate downscaled precipitation and temperature under the {dollar}rm 2 times COsb2{dollar} scenario of climate change over semi-arid regions of southwestern USA, represented by Arizona and New-Mexico (upper Rio-Grande river basin). Local precipitation and temperature are assumed to be dependent upon two effects: the first one, a global effect, is captured by atmospheric circulation pattern (CP) types and the other, a local effect, is reflected by spatially averaged daily pressure heights of the 500 hPa pressure field (h) within the region. CP classification is performed for the 500 hPa pressure fields of observed data and that obtained from the output of the Max Plank Institute (MPI) general circulation (GCM) model T21 for the {dollar}rm 1times COsb2{dollar} and {dollar}rm 2 times COsb2{dollar} scenarios. The evolution of CP types for different scenarios are modeled by a Markov process. Daily precipitation and temperature conditioned on a CP type are modeled by multivariate autoregressive processes. The daily precipitation probability is linked to h through a parametric regression and daily precipitation amount is modeled by a gamma distribution. The daily temperature is modeled by a two sided normal distribution whose parameters are estimated conditioned on fitted values of h. Models are validated using split sampling.; Simulations are performed to generate a series of daily rainfall and temperature (maximum and minimum) both in Arizona and New Mexico stations. Statistical properties of model outputs and statistical significance tests are carried out for current conditions and under climate change using {dollar}rm 2 times COsb2{dollar} scenarios. The results show that precipitation and temperature are increasing significantly with the increase in CO{dollar}sb2{dollar} content. Increases in temperature are more prominent in spring and fall. However the actual amounts of increase in precipitation and temperature depend both on the season and station location.