| Soil moisture is a key parameter in hydrology.It is of great significance to estimate evaporation,transpiration,erosion,runoff,infiltration and irrigation water demand.Compared with site observation,satellite remote sensing technology can provide larger soil moisture data in space and time.However,satellite remote sensing can only detect the moisture of a few centimeters on the surface of the soil.Root zone soil moisture is a key parameter at the junction of land and atmosphere.An in-depth understanding of its temporal and spatial dynamics is very important for the study of regional and global climate change and water resources management.At the same time,the estimation of root zone soil moisture with high temporal and spatial resolution can also provide decision support for guiding agricultural management,ensuring food security and maintaining the sustainable utilization of water resources.Supported by the National Natural Science Foundation of China"Research on optical and radar remote sensing collaborative inversion algorithm of soil moisture and surface roughness"(No.41971323)and the Strategic Priority Research Programm of the Chinese Academy of Science(No.XDA28000000),this research is carried out based on Changchun Jingyuetan remote sensing experimental station of Chinese Academy of Sciences.By coupling crop growth model and hydrological model and assimilating multi-source remote sensing data,the estimation and regional expansion of soil moisture in maize root zone were studied.The main research results are as follows:(1)According to the climatic characteristics,crop varieties and field management of the study area,the parameter sensitivity analysis and parameter calibration of the crop growth model and hydrological model in this study were carried out,and the applicability evaluation of the model in this study area was completed.For the key output parameter(leaf area index)of crop growth model,the sensitivity of parameters related to the change of leaf area index in the model is quantified,and localized calibration is carried out in combination with the environmental characteristics of the study area and field management mode.Aiming at the hydraulic parameters in the hydrological model,combined with the soil properties in the study area,the soil water potential is analyzed,and the relevant parameters are calibrated based on the sensitivity.(2)Based on the principle of photogrammetry,the digital surface model of the study area is obtained by UAV equipped with true color camera,and the soil rough surface is characterized by statistical parameters such as root mean square height and correlation length commonly used in microwave remote sensing;By comparing and analyzing the measurement results with the ground needle plate method,the accuracy of soil surface roughness obtained by UAV photogrammetry method is quantitatively evaluated,and the reliability of this method is proved.On this basis,the vegetation index calculated by the spectral reflectance of sentinel-2 satellite,the backscattering coefficient obtained by Sentinel-1 satellite and the soil moisture obtained on the ground are used to construct the water cloud model to realize the inversion of surface soil moisture.In this study,the root mean square error of surface soil moisture inversion results is better than 0.055 cm~3/cm~3.At the same time,based on optical reflectance(Sentinel-2,UAV)and ground observation leaf area index,combined with radiative transfer model and statistical regression analysis,multi-scale inversion of leaf area index is realized.In this study,the deviation and root mean square error of leaf area index estimated by the two methods are 0.80 and 0.40 respectively.(3)Based on the theoretical simulation of crop growth process and the principle of water transport,this study quantified the absorption degree of soil water in the root zone by water extraction from crop roots.At the same time,based on the utilization of soil evaporation and plant transpiration to soil water in the root zone estimated by the hydrological model,the proportion of carbohydrate distribution to various organs in the crop growth model was updated.Based on the above process,the coupling of crop growth model and hydrological model is realized,the soil moisture in maize root zone(0-100 cm)is estimated,and the longitudinal change process of soil moisture is simulated.In addition,in the operation of the coupling model,combined with remote sensing observation and inversion parameters,the data assimilation method is used to analyze the change parameters,and the single parameter and double parameter assimilation analysis are completed.By comprehensively comparing the simulation results of the coupling model,the single parameter assimilation results and the two parameter assimilation results,the root mean square errors of the two parameter assimilation results at the depths of 5 cm,10 cm and 60 cm are 0.060 cm~3/cm~3,0.079cm~3/cm~3 and 0.059 cm~3/cm~3 respectively,which is the highest accuracy among the four results.At the same time,the accuracy meets the application requirements for soil moisture at different depths.(4)Based on the daily reflectance data of MODIS satellite and Sentinel-2 high spatial resolution multispectral data,the temporal advantage of low spatial resolution satellite data and the spatial advantage of high spatial resolution satellite are complementary by using spatial-temporal data fusion algorithm(ESTARFM),so as to realize the fusion estimation of high spatial-temporal resolution remote sensing vegetation index.The results of spatiotemporal data fusion based on vegetation index show that the correlation coefficient between the fused vegetation index and the vegetation index calculated from Sentinel-2 satellite data is higher than 0.87.(5)Based on the high spatio-temporal resolution remote sensing data after the above spatio-temporal data fusion,combined with the coupling model to assimilate the soil moisture data in the root area estimated by multi-source remote sensing data,the random forest algorithm is used for regression analysis,and the random forest regression model is constructed cm by cm in the root depth.Based on the above random forest model and the high spatial-temporal resolution remote sensing data after data fusion,the forest model is run to realize the estimation of soil moisture in the root zone at the regional scale with high spatial resolution,and the horizontal regionalization expansion of soil moisture in the root zone is completed.By analyzing the soil moisture in the root zone with a regional scale of 10 m spatial resolution,it can be found that the soil moisture increases with the increase of longitudinal depth;In terms of time scale,affected by the near surface climate factors and the water lifting effect of crop growth roots,the soil moisture in the root zone has dynamic changes with time.In this study,the data assimilation framework is constructed by combining the coupling model with multi-source remote sensing data,which provides a method support for the high-precision estimation of soil moisture in the root zone.The high-precision estimation of soil moisture in the root zone provides data support for drought and flood disaster monitoring,early warning,precision agriculture and agricultural water resources management and efficient utilization.In addition,the longitudinal centimeter by centimeter resolution of soil water in the root zone is of great significance for the study of crop root water extraction and water extraction process,and also provides quantitative indicators for crop growth monitoring and yield estimation. |
