Optimization Of Snow Model Algorithm Based On Microwave Remote Sensing And Verification In Xinjiang And Qinghai-Tibet Region

Validation of the snow process model is an important preliminary work for the snow parameter estimation.The snow grain growth is a continuous and accumulative process,which cannot be evaluated without comparing with the observations in snow season scale.In order to understand the snow properties in the Asian Water Tower region(including Xinjiang province?Qinghai province and the Tibetan Plateau)and enhance the use of modeling tools,an extended snow experiment at the foot of the Altay Mountain was designed to validate and improve the coupled physical Snow Thermal Model(SNTHERM)and the Microwave Emission Model of Layered Snowpacks(MEMLS).First,I revise and improve the model.After the model verification,the improved model is used to calculate the snow depth and snow grain size in Xinjiang,Qinghai,and Tibet.After comparing with the snow depth and satellite brightness temperature data of the site,it is found that when the meteorological driving data is more accurate,SNTHERM can better simulate the snow depth,and mainly draws the following conclusions:(1)Add the soil moisture movement process into the SNTHERM model framework and set up Three comparative experiments(original SNTHERM code,clay type;new SNTHERM code,clay type;new SNTHERM code,specific soil type).The first two schemes simulated snow temperature is 2-3K lower than the measured snow temperature;and the third option is to match the simulated snow temperature with the actual measurement.In the influence of the soil parameter setting on the brightness temperature simulation,it is found that the average difference of the soil parameter setting on the snow grain size is 0.1mm.The average difference of the brightness temperature under the vertical polarization of 36.5GHz is within 1K,which means the snow cover properties simulated by SNTHERM are credible regardless of the soil parameter settings.(2)The grain size growth coefficient of SNTHERM(=1.044×10-6m4/kg)is updated.When2)1 is the default value,the simulated₈(6)is lower than the average0.55mm(RMSE=0.74mm),and the ratio of simulated and measured₈(6)is 0.69±0.22;Compared the updated calculation2)1 and the actually measured₈(6)with the actually measured₈(6)of the snow pit.The average deviation is reduced to-0.15mm(RMSE=0.47mm),and the ratio between the simulated and measured values is 0.97±0.33.So the improved simulated snow grain size is more accurate.When the grain size growth factor is used together with the snow temperature and temperature gradient,SNTHERM's grain size growth factor2)1 reduces its dependence on other snow parameters and the corrected2)1 can increase the accuracy of the snow season end date by 3 days.(3)In the verification of the MEMLS model of the site,the measured bright temperature of the snow pit is compared with the simulated bright temperature,and the root mean square error(RMSE)of the bright temperature frequency at 18.7V,36.5V,18.7H and 36.5H is 2.56 respectively K,5.54K,8.07K and 6.11K.The average deviation(MB)is 1.86K,-3.16K,7.51K and-3.51K respectively.(4)The verified coupled model is used to simulate the snow depth,snow grain size and brightness temperature of 102 sites in Xinjiang,Qinghai and Tibet.Comparing the simulated snow depths of 102 meteorological stations with the snow depths of the stations,it is found that the root mean square error(RMSE)in northern Xinjiang is4.03cm,southern Xinjiang is 2.57cm,Qinghai is 2cm,and Tibet is 3.07cm.For all stations,the simulated seasonal mean snow depth root mean square error(RMSE)is1.27cm,and the correlation(R)is 0.98.In order to consider the sub-pixel effect,I introduce snow cover products in the simulation.Comparing the daily simulated brightness temperature data of all stations with the AMSR-E daily brightness temperature data,the brightness temperature greater than 10.65 GHz frequency has a higher consistency between simulated and observed values.Under the vertical polarizations of 18.7,23.8 and 36.5 GHz,the root mean square error(RMSE)is 6.91k,8.2k and 13.34k,respectively.At 10.65 GHz,the correlation between simulated brightness temperature and observed brightness temperature is small,and the root mean square error(RMSE)of vertical polarization and horizontal polarization are 6.46K and 12.93K respectively.The root means square error(RMSE)of the 89GHz simulated brightness temperature is the largest(24.95K).