关键词: CLDAS, 土壤相对湿度, 自动土壤水分观测站, 农业生产生态类型区, 精度评估, 时空对比

Abstract:

To develop high−precision, high−timeliness and high−resolution relative soil moisture products, based on the daily relative soil moisture product from the China Meteorological Administration Land Data Assimilation System (CLDAS) and hourly relative soil moisture data from 309 automatic soil moisture observation stations in Henan from March to November during 2022−2024, the CLDAS relative soil moisture products for the 0−10cm, 0−20cm and 0−50cm layers in six agro−ecological zones of Henan were evaluated by using correlation coefficient, root mean square error（RMSE）, and bias at the site, regional and regional monthly scales. Regional correction models of three layers were developed by regression−based correction, and the models were validated through a comparative analysis of RMSE and the bias before and after correction. The results showed that the CLDAS simulated values of relative soil moisture in Henan were consistent with the observed values from March to November during 2022−2024. At the site scale, the CLDAS simulated values of relative soil moisture at depths of 0−10cm and 0−20cm were close to the observed values, while the CLDAS simulated values of relative soil moisture at a depth of 0−50cm were smaller than the observed values. At three depths, the correlation coefficient between the CLDAS simulated and observed values was generally greater than 0.50, with the RMSE below 30pp. At the regional scale, the correlation coefficient between the CLDAS simulated and observed values at a depth of 0−10cm ranged from 0.83 to 0.91, with the highest values in southwestern and southern Henan; the RMSE ranged from 5.72pp to 9.00pp, with the smallest value in eastern Henan; and the bias ranged from −6.20pp to 5.54pp, with the smallest absolute value of bias in eastern Henan. The correlation coefficient between the CLDAS simulated and observed values at a depth of 0−20cm ranged from 0.85 to 0.92, with the highest value in eastern Henan; the RMSE ranged from 4.13pp to 9.07pp, with the smallest value in eastern Henan; and the bias ranged from −6.36pp to 4.74pp, with the smallest absolute value of bias in western Henan. The correlation coefficient between the CLDAS simulated and observed values at a depth of 0−50 cm ranged from 0.80 to 0.91, with the highest value in eastern Henan; the RMSE ranged from 4.68pp to 12.51pp, with the smallest value in northern Henan; and the bias ranged from −11.05pp to 1.39pp, with the smallest absolute value of bias in northern Henan. At the regional monthly scale, the overall monthly correlation between the CLDAS simulated and observed values was good, with correlation coefficient exceeding 0.70 in most areas from May to October. The RMSE was generally less than 15%, with the smallest values in western and eastern Henan at three depths. After regional correction, the RMSE between the CLDAS simulated and observed values decreased to 3.47−7.25pp, with a reduction range of 0.04−7.20pp; and the bias decreased to −0.04−0.02pp, with a reduction range of 0.45−11.42pp. Additionally, the correction models for northern, eastern, and central−southern Henan demonstrated good generalization capabilities. Overall, the CLDAS relative soil moisture product has high simulation accuracy in northern, western, and eastern Henan, with higher simulation accuracy from May to October. After regional correction, the simulation accuracy of CLDAS relative soil moisture was further improved, providing important support for agricultural drought and flood monitoring, assessment, and early warning.

Key words: CLDAS, Relative soil moisture, Automatic soil moisture observation station, Agro?ecological zone, Precision assessment, Spatiotemporal comparison