| Soil moisture is a key variable in crop growth,sland-atmosphere water and heat energy and global water cycle.Accurate monitoring and prediction of SM plays a crucial role in crop growth and yield estimation,monitoring and prediction of flood and drought,study on surface hydrology,hydrological and land surface process research,and investigating the global water cycle.Microwave remote sensing has advantages over optical remote sensing and is very sensitive to soil moisture change.The L band in microwave band is considered to be the best band for monitoring soil moisture.Although the SMOS(Soil Moisture Ocean Sanility)and SMAP(Soil Moisture Active Passive)programs have been able to provide global soil moisture products,the uncertainty of these soil moisture products(or inversion algorithms)has not been accurately assessed in the northeast China because the spatial resolution of passive microwave remote sensing soil moisture products is tens of kilometers scale and the heterogeneity of soil moisture exists within the pixel.Therefore,this paper has carried out the research of authenticity test and algorithm improvement for passive microwave pixel scale soil moisture products,and evaluates the uncertainty of L band passive microwave soil moisture products by building and running the soil moisture observation network of microwave pixel scale(36km x 36km).The main results are as follows:(1)Construction of soil moisture observation network in farmland based on pixel scaleNortheast China is the largest commodity grain production base.The plain area is vast,and the main crops are wheat,rice and corn.Accurate monitoring of soil moisture in Northeast farmland plays an important role in grain yield and agricultural development.In order to validate the uncertainty of the passive microwave remote sensing soil moisture products in the Northeast farmland area,firstly,a set of accurate ground soil moisture observation data needs to be obtained.In this paper,firstly,the principle of homogeneity of vegetation type,small elevation difference and homogeneous soil texture was taken into consideration,and the northern farmland area(main crop is corn)of Changchun city,Jilin province was selected as a verifying field for passive microwave remote sensing of soil moisture products;secondly,the spatial heterogeneity of soil moisture in passive microwave pixel was expressed by soil apparent thermal inertia(ATI),and the soil moisture observation nodes were selected by the rule grid method;then the soil moisture sensor of the observation node was calibrated to reduce the uncertainty of the point scale measurement.Through the above work,a set of standards for the ground observation network suitable for the authenticity test of soil moisture products of passive microwave pixel scale was put forward.(2)Accuracy analysis of soil moisture products of passive microwave remote sensing pixel scaleThrough the successful operation of the passive microwave pixel scale soil moisture observation network,the SMAP and SMOS L3 passive microwave remote sensing soil moisture products were verified by using the time series in-situ data from soil moisture observation network from 2016 to 2017.Compared with the in-situ soil moisture,the SMOS L3(ascending and descending overpassing)and the 36km SMAP_L3_P passive microwave remote sensing soil moisture products generally underestimated the ground data in 2016,but there would be an overestimation with the occurrence of rainfall events;the root mean square error(RMSE)of SMOS(ascending and descending overpassing)soil moisture products was 0.125 cm~3/cm~3 and 0.127 cm~3/cm~3,respectively,while the unbiased root mean square error(ubRMSE)was 0.123 cm~3/cm~3 and 0.090 cm~3/cm~3,respectively,indicating that the descending overpassing soil moisture products were affected by random errors;the ubRMSE of SMAP_L3_P passive microwave soil moisture data is slightly lower and is 0.078cm~3/cm~3.The ubRMSE of the these two soil moisture products was greater than 0.07cm~3/cm~3.Compared with the in-situ soil moisture,similar to the validation results of soil moisture product in 2016,the SMAP_L3_P(ascending and descending overpassing)passive soil moisture products in 2017 existed the dry bias.The RMSE of them were0.074 cm~3/cm~3 and 0.062 cm~3/cm~3,respectively,indicating the higher accuracy for SMAP_L3_P descending soil moisture products.But both SMAP_L3_P ascending and descending overpassing passive soil moisture products were higher than 0.06 cm~3/cm~3,without meeting the expected performance of the SMAP project.the RMSE for the majority of the 9-km SMAP_L3_P_E passive soil moisture products within the selected 36-km pixel were more than 0.07 cm~3/cm~3 for descending overpassing and 0.08 cm~3/cm~3 for ascending overpassing,which meant that there was a good consistency between SMAP_L3_P_E descending passive soil moisture products and in-situ data.The SMAP_L3_P_E passive soil moisture products also existed dry bias and their RMSE were higher than those from the corresponding 36-km SMAP_L3_P passive soil moisture products.The evaluation results of passive soil moisture products for two years were not up to the expected accuracy of SMAP and SMOS,and there was a phenomenon of dry bias.It was necessary to analyze the factors of soil moisture inversion.(3)Research on parameter optimization method of SMAP soil moisture retrieval algorithmThrough analysis,the SMAP soil temperature(Ts)is an important parameter that makes the SMAP_L3_P passive soil moisture product underestimated and the precision difference between ascending and descending overpassing SMAP_L3_P passive soil moisture products.In the stage of bare soil/low vegetation,the Bias of SMAP ascending and descending overpassing Ts is-5.46K and-1.33K,respectively,and the Bias of between in-situ Ts and in-situ vegetation temperature(Tc)at ascending and descending overpassing in the stage of high vegetation is 2.11K and-0.21K,respectively,and the uncertainty of SMAP ascending overpassing Ts is greater than that of descending overpassing Ts,and the SMAP ascending overpassing Ts is underestimated.The uncertainty of SMAP descending overpassing Ts adheres to the expected goal of the SMAP project(2K),thus SMAP Ts for ascending overpassing were corrected in this study.After correcting SMAP ascending overpassing Ts,the RMSE of SMAP_L3_P(ascending overpassing)for bare soil/low vegetation stage and high vegetation stage reduced from 0.063 cm~3/cm~3 to 0.041 cm~3/cm~3 and from0.083 cm~3/cm~3 to 0.078 cm~3/cm~3,respectively.The accuracy of the bare soil/low vegetation stage is obviously improved and close to the requirements of the SMAP mission,and the accuracy of the high vegetation stage is not improved obviously.It may need to further study the other factors that affect the soil moisture inversion,such as the vegetation water content. |
PDF Full Text Request