Climate Change Scenarios and the Impact on Runoff

Abstract

(1) The performance of General Circulation Models (GCMs) in simulating near-surface air temperature and precipitation is assessed in the Tarim River Basin (TRB). The criteria included skills on reproducing the observed statistics of two variables, such as long-term mean and standard deviation, seasonal variation, temporal and spatial distributions, and probability density functions (PDFs). The results show that some GCMs performed relatively better than others, usually simulating temperature better than precipitation. (2) Two statistical downscaling models, the Non-Homogeneous hidden Markov Model (NHMM) and the Statistical Down-Scaling Model (SDSM), which have been widely applied in the world and were proved skillful in term of downscaling purpose, were evaluated based on observed daily precipitation in the Tarim River Basin (TRB). The evaluated merits included relative errors of residual functions for annual mean and several percentile values, correlation analysis of annual cycle and spatial distributions, and two skill scores based on probability density functions (PDF). Results indicated that NHMM had a better model performance on daily precipitation than SDSM in the study area, and the former was a more stable model than the later. The main reason is that NHMM could capture precipitation spatial distribution pattern, while SDSM is a single-site model, which downscales each individual station independently. SDSM also lost its skill on modeling extreme values of precipitation amount. (3) The NHMM and SDSM were applied to generate future scenarios of both mean and extremes in the Tarim River Basin, which were based on nine combined scenarios including three GCMs (CSIRO30, ECHAM5, GFDL21) predictor sets and three special report on emission scenarios (SRES A1B, SRES A2, SRES B1). Results showed that trends magnitude projected by statistical downscaling models were the greatest under SRES A2 scenario, and was the smallest under B1 scenario, with A1B scenario in–between. In general, the trends magnitude was greater in the period of 2081–2100 than that in the period of 2046–2065. (4) Driven by these climate change scenarios, a distributed macro-scale hydrological model (Variable Infiltration Capacity model) was applied to assess the impact of climate change on hydrological processes in the Headwater Catchment (HC) of the TRB. Results showed that it tends to be warmer and drier for the HC under the combined climate change scenarios. It is different from air temperature that the magnitude of changes for extreme values of precipitation is obviously greater than that for mean values. It tends to show a decreasing trend for runoff in the HC, driven by the combined climate change scenarios. But it showed an increasing trend for winter runoff. It showed an inconsistent intra-annual distribution for the changes of precipitation and runoff in the HC, which might be explained by the increasing snowmelt runoff resulted from the increasing air temperature. It was concluded that uncertainties from different GCMs outputs are more significant than emission scenarios in the assessment on the impact of climate change.