Quantification Of Water Footprint Of Winter Wheat Production Based On Remote Sensing And Crop Model Assimilation

Water resource shortage is one of the main factors restricting food security in China.In order to achieve scientific water management and improve water use efficiency,it is necessary to evaluate agricultural water use efficiency.Crop production water footprint is effective in accurately assessing the type and amount of agricultural water use.However,current quantitative studies on the water footprint of regional crop production are limited by large spatial scale and poor spatial heterogeneity.This paper takes winter wheat in Xianyang city of Guanzhong Plain as the research object,took leaf area index(LAI)and soil moisture(SM)obtained by remote sensing as state variables,combined the World Food Study(WOFOST)model and Ensemble Kalman filter(En KF)assimilation algorithm,and proposed a method of water footprint quantification of winter wheat production based on crop model-remote sensing information data assimilation.The water footprint of winter wheat production was quantified at the regional grid scale,and its spatial distribution characteristics were analyzed.In order to provide beneficial reference for regional water-saving agriculture planning.The main results are as follows:(1)The applicability of the WOFOST model in the study area was evaluated by parameter calibration,and the comparison with the observed data showed that the WOFOST model could better reflect the growth and development of winter wheat in the whole growth period.Twelve key parameters with strong sensitivity to model output variables were selected by sensitivity analysis method,and the key parameters of the WOFOST model were calibrated using the measured data of Jingyang Station,Yangling Station and Changwu sites.The relative errors(RE)between measured and simulated yields at Jingyang Station,Yangling Station and Changwu Station during 2016-2017 were 4.4%,8.84% and 8.01%,respectively.The results showed that the calibrated WOFOST model had a good simulation effect on the growth and development of winter wheat in the study area.(2)A regional winter wheat yield estimation model based on bivariate(LAI and SM)joint assimilation was constructed,and the simulation accuracy of this model was higher than that of univariate assimilation(only assimilation of LAI or only assimilation of SM).At a single point scale,the difference between the yield of LAI and SM combined assimilation and the measured yield decreased from 486.62kg/ha(assimilated LAI alone)and627.88kg/ha(assimilated SM alone)to 441.98kg/ha.On the regional grid scale,R2=0.61,RMSE=128.78kg/ha between the winter wheat yield after LAI and SM combined assimilation and the statistical yield,and the spatial distribution of the winter wheat yield after combined assimilation was consistent with that of the statistical yield.The relative error RE of official statistical yield and the mean yield after LAI and SM combined assimilation in 12 counties in the study area was all within 10%.The results showed that LAI and SM combined assimilation could effectively improve the accuracy of winter wheat yield estimation in the study area.(3)A quantitative framework of winter wheat production water footprint based on data assimilation was constructed to quantify the blue water,green water and total water footprint of winter wheat at the regional grid scale,revealing the spatial heterogeneity of winter wheat production water footprint in the study area.The production water footprint of winter wheat in the study area was 1.1m3/kg,in which the blue water footprint of winter wheat production was 0.57m3/kg,and the green water footprint of winter wheat production was 0.53m3/kg.In the production water footprint of winter wheat,blue water footprint and green water footprint accounted for 51.85% and 48.15%,respectively.Blue water footprint of winter wheat production was higher in the east than in the west.The spatial heterogeneity of blue water footprint of winter wheat production in the northern and southeastern counties of the study area was more obvious than that in other regions.The low value of green water footprint in winter wheat production coincided with the high value of yield.The extreme value of green water footprint in the eastern counties of the study area was different greatly,and there was significant spatial heterogeneity.The spatial distribution of winter wheat production water footprint in the study area was similar to that of winter wheat blue water production footprint.There was significant spatial heterogeneity in the water footprint of winter wheat production in the eastern and southeastern counties of the study area.In this study,data assimilation algorithm was coupled to crop models.On the basis of existing studies that only used LAI as the assimilation variable,SM as the assimilation variable was also taken into account to analyze the yield estimation accuracy of univariate and bivariate assimilation models.On the basis of assimilation estimation,a quantitative method of water footprint for winter wheat production based on remote sensing and crop model assimilation was proposed,combined with the water footprint theory.The water footprint of winter wheat production was quantified at regional raster scale,and the applicability of data assimilation algorithm to quantify the water footprint of crop production was evaluated,which improved the low resolution of water footprint quantization in the existing studies.The spatial heterogeneity of water footprint of winter wheat production was evaluated at regional raster scale,which fully reflected the spatial heterogeneity of water footprint of winter wheat production.This method can provide reliable guidance and useful reference for agricultural water use efficiency evaluation and water resources management.