Simulating Temporal Variations Of Snow Albedo Targeting Satellite Remote Sensing Observations

Snow surface spectral reflectance is very important in the Earth’s climate system.The accumulation and development of satellite remote sensing observations and numerical models in the past decades have brought a new round of opportunities for the study of snow cover radiative transfer.However,there are still some problems in the correspondence and fusion of remote sensing observation and numerical simulation of snow radiative transfer.Traditional land surface models with parameterized schemes can simulate broadband snow surface albedo but cannot accurately simulate snow surface spectral reflectance with continuous and fine spectral wavebands,which constitute the major observations of the current satellite sensor.Consequently,there is an obvious gap between land surface model simulations and remote sensing observations.The combination of snow reflection information observed by satellite remote sensing and land surface model is usually to integrate remote sensing data into the land surface model by direct insertion or data assimilation,however,few research methods can directly simulate and predict the snow reflection information in the corresponding spectral range observed by satellite remote sensing based on the land surface model.Given the above scientific problems,this dissertation takes the Upstream Heihe River(UHR)basin as the research area.This dissertation has carried out in-depth research from four aspects,the improvement of snow albedo model and development of a long-time series snow albedo reconstruction model at a single point scale in the alpine mountainous area,the integration of cold region snow radiative transfer model targeting satellite remote sensing observations,the evaluation of integrated cold region snow radiative transfer model and preparation of spatio-temporal seamless snow albedo products,and modeling snow albedo with high spectral resolution in integrated cold region snow radiative transfer model targeting satellite remote sensing observations,respectively,and obtained the following main results.(1)A snow albedo reconstruction scheme with long time series in single point scale is proposed.Combined with the cosine correction method of solar zenith angle and solar azimuth angle,the retrieval method of snow albedo based on the ART(Asymptotic Radiative Transfer)model is improved.Combined with Jordan 91 snow grain size evolution model,the forward simulation method of snow albedo based on the TARTES(Two-stre Am Radiative Transf Er in Snow model)model is improved.The spatial representation of snow observation sites is evaluated by using the spherical model of semi-variogram.In this dissertation,a simple scale conversion method is proposed,which takes the high-resolution remote sensing data as the intermediate scale data,and uses the site data to calibrate the high-resolution remote sensing data,and then to verify the authenticity of low-resolution remote sensing data.Based on the improved snow albedo retrieval model and the improved snow albedo forward radiative transfer model driven by observed snow attribute data from snow observation stations,multi-source satellite remote sensing observation data,and meteorological data,and then a set of snow albedo simulation scheme with long time series in single point scale is established.Compared with the "real snow albedo" measured at the site,the snow albedo data obtained by the new simulation scheme is 11% underestimated,and the simulation result of the new scheme improves the effective accuracy by 6% compared with NASA’s(National Aeronautics and Space Administration)MOD10A1 SAD snow albedo product.(2)A integrated Cold region Snow Radiative Transfer model(CSRT)coupled with snow radiative transfer module,snow hydrological module and cold region land surface processes module is proposed.To solve the problem of snow reflection information simulation and prediction targeting satellite remote sensing observation,the snow mass balance process based on the GBEHM(Geomorphology-Based Eco Hydrological Model)model and the snow grain size evolution process based on the FZ06(Flanner Zender)model is coupled,and the snow energy balance process based on the GBEHM model and snow radiative transfer process based on the SNICAR(Snow,Ice,and Aerosol Radiative)model is coupled.Besides,the transfer and absorption process of solar radiation energy in the snow layer is tracked by coupling the snow hydrology module with the snow radiative transfer module and snow grain size evolution module,which greatly improves the simulation accuracy of snow radiative transfer.The CSRT model can simulate not only the basic snow radiation parameters but also the snow process parameters.The verification results of snow depth observation data and snow binary data based on MODIS show that the RMSE(Root Mean Squared Error)of both the snow depth simulated by the CSRT model and the observed snow depth is 14.5 mm,the RMSE of both the snow cover days simulated by the CSRT model and the remotely sensed snow cover days is 5 days,and the snow depth and snow cover days simulated by the CSRT model have high accuracy.(3)A set of spatiotemporal seamless broadband snow albedo products in the Upstream Heihe River are developed.The broadband snow albedo simulated by the CSRT model has good accuracy.The accuracy evaluation results of the CSRT model based on the "real snow albedo" observed by the station show that the broadband snow albedo simulated by the CSRT model has only 3% and 4% deviations from the measured snow albedo at Dadongshu Yakou and Jingyangling observation stations,and the simulation deviations of the CSRT model are much lower than that of the traditional parameterization method by 10% and 14%.Based on the CSRT model,we prepared a set of spatiotemporal seamless broadband snow albedo products with a spatial resolution of 1km,the temporal resolution of 1 hour in the Upstream Heihe River from2000 to 2015.The annual variation tendency of the newly prepared broadband snow albedo indicates that the average broadband snow albedo in the Upstream Heihe River shows a trend of fluctuating downward from 2000 to 2015,with a decrease of 0.03.(4)A modeling scheme of snow albedo with high spectral resolution targeting satellite remote sensing observations is proposed.In this dissertation,by introducing the incident solar flux spectral distribution function,the CSRT model realizes the simulation of snow reflectance and snow albedo covering the whole shortwave band with the high spectral resolution and then constructs a set of simulation and prediction scheme of snow reflection information with high spectral resolution targeting multi-source satellite remote sensing observations.The results indicate that the CSRT model can accurately simulate snow surface reflectance information over a large spatial scale and continuous time series.The CSRT model extends the range of snow spectral reflectance simulation to the whole shortwave band and can predict snow spectral reflectance changes in the solar spectrum region based on atmospheric forcing data,snow optical characteristics data,and land surface characteristics data.The kappa coefficients(K)of both the narrowband snow albedo targeting Moderate Resolution Imaging Spectroradiometer(MODIS)data simulated by the CSRT model and the retrieved snow albedo based on MODIS reflectance data are 0.5,and both exhibit good spatial consistency.Meanwhile,our proposed narrowband snow albedo simulation scheme targeting satellite remote sensing observations is consistent with remote sensing satellite observations in time series and can predict narrowband snow albedo even during periods of missing remote sensing observations.This CSRT model is a significant improvement over traditional land surface models for the direct spectral observations of satellite remote sensing.The proposed CSRT model scheme could contribute to the effective combination of snow surface reflectance information from multisource remote sensing observations with snow hydrological process and land surface process,and greatly promote the application of satellite remote sensing observation data in land surface modeling.In addition,the spatiotemporal seamless broadband snow albedo products prepared based on the CSRT model will effectively improve the accuracy of land surface process simulation and regional climate change simulation in alpine mountainous areas.