Research On Spring Snowmelt Runoff In The Middle Temperate Zone

Due to its unique climatic characteristics,there is a large amount of seasonal snow cover in the middle temperate zone of China,which leads to completely different runoff process in the spring snowmelt period from other climate zones.As an important part of the middle temperate zone,the northeast of China is the main production base of agriculture,forestry and animal husbandry.On the one hand,snow as an important freshwater resource,can provide a main support for human life,agricultural irrigation and shipping,especially in the spring snowmelt period when rainfall is scarce.On the other hand,if the snow accumulated for a long time in winter meets the continuous rapid rise of the temperature in spring,the rapid melting of a large amount of snow in a short period of time will cause spring flood and delay the optimal sowing time for agriculture,thus causing considerable economic losses in northeast China.Therefore,accurate snowmelt runoff simulation and prediction is of great significance for spring flood&drought control and the efficient development and utilization of water resources in middle temperate zone.This paper taking the Baishan reservoir basin in the source area of the Second Songhua River as an example to explore the general rules of the snowmelt runoff in middle temperate zone.This study collected the measured runoff data from 1971 to2016 of this study area,the meteorological data from 1987 to 2016 of 3 stations,and remote sensing data such as the DEM elevation map,the land use map,he HWSD soil map,the MODIS snow-covered spectral remote sensing products and microwave snow water equivalent products,they are used to analyze the division of the snowmelt runoff period,the key influence factors of the snowmelt runoff and the time-lag.The empirical equation of snowmelt runoff based on genetic algorithm was constructed to reveal the influencing factors and action mechanism of snowmelt runoff.Through comparative analysis with SRM snowmelt runoff model,the empirical equation of snowmelt runoff constructed in this paper was analyzed in precision and evaluated in practicability.The main research contents are as follows:?1?Spring snowmelt runoff period.In this paper based on the ECK digital filtering method of the sliding minimum value combined with the 46 years daily runoff data in Baishan station to separate base flow,through analysis the base flow ratio and relative runoff process evolution trends,identify the spring snowmelt runoff period start time and end time,the start time and end time node of snowmelt runoff were statistically analyzed by box diagram,and it was determined that under the condition of multi-year average?50%?,the snowmelt runoff period was from March 23 to May 4,which lasted a total of 43 days.?2?The key influence factors that affect runoff in snowmelt runoff period and the time-lag are identified.In this paper,a global sensitivity analysis method based on the multi-factor coupling based on the BP neural network method is used to identify the time-lag of 6 main meteorological factors?total daily radiation exposure,daily average wind speed,daily precipitation,daily minimum temperature,daily average temperature and daily maximum temperature?in the snowmelt runoff period of Baishan basin,the daily total radiation exposure quantity SR,the daily average wind speed W and the precipitation P didn't exist time-lag,the daily average temperature MT and the daily maximum temperature HT exist 2 days of time-lag,and the daily minimum temperature LT exist delay of 1 day,identify the key factors under the influence of the effect of the time-lag of each meteorological factor,the snowmelt runoff in Baishan basin is the most sensitive to the precipitation P_T0,followed by the total radiation exposure SR_T0,the average wind speed W_T0 and the average temperature MT_T2.?3?Based on the genetic algorithm optimize the empirical equation of the snowmelt runoff.Focus on the snowmelt runoff period in the study area,based on the key influence factors that affect the runoff in snowmelt runoff period and the time-lag,the early accumulation of snow?the sum of precipitation from last year November 1 to this year February 28/29?,the precipitation P_T0,the total radiation exposure value SR_T0,the daily wind speed W_T0 and the daily average temperature MT_T2 was taken as five simulation factors,the daily runoff in Baishan basin was taken as the simulation object,by judging runoff coefficient greater than or less than1,the differential snowmelt runoff empirical equation was constructed respectively;Empirical equation was optimized by using genetic algorithm to optimize parameters,the simulation period?from 1987 to 2010?,the determination coefficient R²,relative error Re and model efficiency coefficient Ens were 72.1%,25.1%and72.0%,respectively.In the validation period?from 2011 to 2016?,the determination coefficient R²,relative error Re and model efficiency coefficient Ens were respectively 62.3%,27.2%and 61.0%.The simulation precision can meet the requirements of production practice and can be used as a practical method to estimate the snowmelt runoff in spring.?4?The simulation of the SRM snowmelt runoff model.The study is divided into three elevation bands to be respectively simulated,and the snow cover area extracted by MOD10A1 daily snow product in snowmelt runoff period of each elevation zone is taken as preliminary snow condition,,and the precipitation and temperature data of the weather station is used as an input data to calibrate the model parameters.During the simulation period?from 2000 to 2010?,the determination coefficient R²,relative error Re and model efficiency coefficient Ens were 67.4%,33.3%and 54.5%,respectively;During the validation period?from 2011 to 2016?,the determination coefficient R²,relative error Re and model efficiency coefficient Ens were respectively 69.8%,30.5%and 57.6%.Through comparison and analysis,the simulation accuracy of the empirical equation proposed in this paper is better than SRM snowmelt runoff model.