Bayesian evaluation of meteorological datasets for modeling snowmelt runoff in Tizinafu watershed in Western China

摘要

The Tizinafu watershed is characterized by extremely scarce precipitation and low temperature, and snow and glacier melting which dominate the main water resource of this area. Thus, assessing the snowmelt water resource has great significance to local residents and ecological systems. Based on snowmelt runoff model (SRM) and Markov chain Monte Carlo (MCMC) simulation, four combinations of meteorological data consisting of two daily temperature data from the Moderate Resolution Imaging Spectroradiometer (MODIS) and the variable infiltration capacity (VIC) model and two daily precipitation data from Tropical Rainfall Measuring Mission (TRMM) and VIC, respectively, are evaluated for this ungauged basin in Bayesian uncertainty analysis. By comparing the performances of the four meteorological data combinations based on SRM (SRMMODIS + TRMM, SRMMODIS + VIC, SRMVIC + TRMM and SRMVIC + VIC, the former and latter subscripts denote temperature and precipitation data, respectively), results show that the four SRMs are both capable of reproducing the runoff process of the Tizinafu watershed on daily runoff simulation, and SRMVIC + VIC has the best performance. In addition, the impact of climate change on the water resources of the Tizinafu watershed is also evaluated in Bayesian uncertainty analysis, and four climate change scenarios under the condition of Representative Concentration Pathway 8.5 of CMIP5 projection are considered. The results demonstrate that the mean annual runoff prediction of the 2090s increased by 17.2%, 20.2%, 24.1%, and 16.2% compared to that of base year for the four scenarios, respectively. In addition, with the increase of the cryosphere area, snow cover area will have an increasing impact on the mean annual runoff of the study area in the 2090s. The runoff components of snowpack melt and new snow are both sensitive to the cryosphere area, while the runoff component of rainfall is not sensitive to the cryosphere area.