Research On The Distributed Model Simulation Of The Spatial/Temporal Hydrological Characteristics In A Cold Alpine Basin

Cold alpine basins are widely distributed in areas with high latitude and altitude around the world.In our country,cold alpine basins are mainly distributed at arid area in West,represented by the Qinghai-Tibet Plateau.The cold alpine basins has unique water/heat balance processes with impacts of special underlying conditions and climatic characteristics.The different frozen ground types:permafrost and seasonally frozen ground(SFG)have heterogeneous hydrothermal processes with different spatial/temporal distribution.This cause the different impacts on basin's hydrology processes of this two frozen ground types.The seasonal precipitation/temperature changes also has strong influence on the water cycle characteristics in cold alpine basin.At present,the hydrological models rarely consider the hydrothermal processes of frozen ground and the seasonal changes of precipitation and temperature conditions.The accurate hydrological processes simulation of cold alpine basins is still facing enormous challenges.In this study,the Babao River basin(BRB)in the Qilian Mountain of northwestern China,a typical cold alpine basin was chosen as the study area.The Distributed Hydrology-Soil-Vegetation Model(DHSVM)was selected as the research tool to investigate the hydrology characteristics of BRB.A new hydrological model calibration method with stratifying frozen ground types and seasons was proposed to improve the model's simulation accuracy in BRB.A set of frozen soil process simulation scheme for the BRB was established and integrated into the DHSVM model to improve the physical process of the model.The purpose is to establish an effective hydrological process simulation system for cold alpine basin for exploring the effects of different frozen ground types and climatic characteristics on the hydrothermal processes,and to provide theoretical basis and technical Support for the research of hydrological characteristics and water resources management in cold alpine basins.The research contents and results of this paper are as follows:(1)A spatial/temporal resolution sensitive analysis was applied for DHSVM in BRB to determine the best suitable resolution and time step.The 300 m spatial resolution and 3 h temporal resolution were determined as the suitable resolution.The sensitive parameters of DHSVM in BRB was analyzed by eFAST method,and six sensitive parameters were determined:Field Capacity,Monthly LAI,Min Resistance,Exponential Decrease,Bubbling Pressure and Albedo.The PSO method was applied to calibrate the sensitive parameters.The observed streamflow for 2009 was set as the calibrating target and the streamflow for 2005 to 2008 were set to validate the calibrating accuracy.The Nash-Sutcliffe efficiency coefficient(NSE)of calibrating period and validating period are 0.62 and 0.48,respectively.In general,DHSVM can reflect the trend characteristics of runoff in BRB effectively.The simulation accuracy is higher in winter than summer.However,DHSVM has two main shortcoming in BRB:(1)DHSVM overestimated the evapotranspiration in summer.This caused the poor simulation accuracy in summer,resulted the calibrated albedo of vegetation were much larger than its normal range;(2)the model lacked the freezing/thawing processes of frozen ground and overestimated the melting speed of snowpack in winter.This caused the lower simulation streamflow in April and the and abnormal runoff peaks in winter.(2)A new parameters calibration method by stratifying frozen ground types and seasons was proposed to enhance the simulation accuracy of DHSVM in BRB.The sensitivity of parameters were analyzed under different frozen ground/seasons,4 frozen ground sensitive parameters:Lateral Conductivity,Field Capacity,Bubbling Pressure,Exponential Decrease,and three seasonal sensitive parameters:Albedo,LAI,Min Resistance were determined.The frozen ground sensitive parameters and seasonal sensitive parameters were calibrated under four sets of scenarios:(1)calibrating with no stratifying as baseline;(2)calibrating with stratifying frozen ground types;(3)calibrating with stratifying seasons;(4)calibrating with stratifying frozen ground types and seasons together.The calibrating period was set from 10/1/2008 to 9/30/2009.The NSE of four scenarios were 0.58,0.63,0.65 and 0.69,respectively.The calibrated parameters of(1)and(4)were applied for a long time validation from 2005 to 2008 and the NSE in validation periods were 0.48 and 0.60,respectively.The simulation accuracy of DHSVM was significantly improved both in calibration period and validation periods.Calibrating by stratifying frozen ground types could improve the simulation of base flow,and had a significant accuracy improvement in winter,when the base flow plays a major role.With calibrating by stratifying seasons,the evapotranspiration different between dry and wet season was effectively distinguished.This enhanced the streamflow simulation accuracy in precipitation periods.Calibrating parameters by stratifying frozen ground types and seasons could obviously improve the simulation accuracy of DHSVM in BRB.However,there is still the problem that the frozen ground processes in April cannot be effectively expressed and Albedo were overestimated in summer.Further improvements to the model processes are needed to enhance the performance of the DHSVM in the alpine mountain basin.(3)Based on the relationship between potential evapotranspiration,observed precipitation and actual evapotranspiration in the Qilian Mountains,a conversion method between the calculated potential evapotranspiration and the true evapotranspiration of the BRB was established for the DHSVM model to improve the evapotranspiration module.The improved model effectively solves the problem of overestimating vegetation albedo parameters under the premise of ensuring the simulation accuracy.According to the observation data of snowpack albedo in the BRB,the parameters of snowpack albedo decay curve were recalibrated.The max snowpack albedo was increased to 0.92 and the decay rate was slowed down.The improved snowpack albedo could accurately estimate the melting period and speed of snow in winter and early spring,eliminate the abnormal runoff peaks in winter,and improve the runoff simulation accuracy DHSVM in melting period of snow and frozen ground in April.This paper also established a frozen ground freezing/thawing simulating algorithm for DHSVM to compensate for the lack of model processes.The simulation accuracy of the model had a significant raising during the melting periods after integrating the frozen ground module.Through the improvement and complementation of the model's physical processes,the NSE has increased from 0.69 to 0.71 in calibrating period,from 0.60 to 0.66 in validation period,respectively.The simulation performance of model in the cold alpine basins has been significantly improved to further strong the applicability of model in this typical basins.