Snow Hydrological Simulation In Alpine Areas Using Remote Sensing And GIS Technologies

The snow hydrological model is an important method to study the dynamics of snowmelt, snow hydrothermal process simulation has become one of the focuses of global climate change research. China is the most abundant country in cryosphere in middle-low latitudes area and extremely sensitive to climate change. As the vital fresh water resource, snow plays an important role in prataculture, animal husbandry and other industries. Therefore, establishing the snow hydrological model in alpine areas has great significance for snow hydrological simulation and forecasting.In this study, Manas river basin in Xinjiang and Babao river basin in Qinghai were selected as the study area, a sub-grid snow distribution model was established with multi-source remote sensing data. A snow hydrological model was established by coupling many kinds of hydrological models’ advantages. Particle swarm optimization algorithm was adopted to define the model parameters. In the Babao river basin, the role of remote sensing data for simulation result of the snow hydrological model was analyzed. In the Manas river basin, the snow hydrological model was verified by the observed data. The results show that:1) Using multi-source remote sensing data, the sub-grid snow distribution model was established, and the parameter K value was defined. When the K value was0.0028, the simulation had the highest accuracy. The overall accuracy was70.36%, Kappa coefficient was0.5262, RMSE of slope was3.8854°, RMSE of sine aspect was0.3877, and error of aspect was22.8115°, the parameter K value was determined as0.0028.2) The sub-grid snow distribution model had a high simulating precision, which could be used in sub-grid snow distribution studies. The sub-grid snow distribution model was verified by the eighteen higher resolution remote sensing images (i.e., TM. CHRIS, and EO-1). The mean overall accuracy was70.96%, the mean Kappa coefficient was0.5346, the mean RMSE of slope was3.8794°, the mean RMSE of sine aspect was0.3862, and the mean error of aspect was22.7183°.3) Using remote sensing snow area data could improve the SWE (snow water equivalent) simulation accuracy. In the Babao river basin, the sixteen model parameters were determined by the particle swarm optimization algorithm. Snow hydrological simulation accuracies were compared before and after using the remote sensing snow area data. The Yakou station SWE Kappa coefficient was0.5916without remote sensing snow area data, and was0.6128with remote sensing snow area data. The determination coefficient R2of SWE observed in eleven sites was0.1040without remote sensing snow area data, and0.4819with remote sensing snow area data.4) Using remote sensing snow area data could improve the SCF (snow cover fraction) simulation accuracy. In the Babao river basin, the overall SCF accuracy were compared before and after using the remote sensing snow area data from2001to2008. The determination coefficient R2value in snowmelt season was generally higher than that of the whole year, the SCF simulation results in snowmelt season superior to it in snow season. The average value of R2in the whole years was0.4429without remote sensing snow area data, and was0.5467with remote sensing snow area data. The average value of R2in snowmelt season was0.5790without remote sensing snow area data, and was0.6181with remote sensing snow area data.5) Using remote sensing snow area data could improve the runoff simulation accuracy, which played an important role in improving the simulation result of the snow hydrological model. In the Babao river basin, the simulation accuracies of runoff were compared before and after using the remote sensing snow area data from2001to2008. The average values of Nash coefficient was0.3869without remote sensing snow area data, and was0.6111with remote sensing snow area data. The average values of Dv (volume difference) was12.08%without remote sensing snow area data, and was9.77%with remote sensing snow area data.6) In the Manas river basin, the snow hydrological model had high simulating precisions for SWE and runoff, which could be used in snow hydrological simulation studies. The Nash coefficient of SWE in Hankazi station in2012was0.8334. The average value of Nash coefficient of runoff from2008to2012was0.8113, and the average value of Dv was10.99%.