关键词: 土壤水分, 风云卫星, 中国气象局陆面数据同化系统, 多尺度, 精度验证

Abstract: Soil moisture is an important variable for the surface hydrological cycle. High-precision soil moisture products, which have consistent time and space consistency, play a vital role in studying hydrological-driven climate and vegetation growth. In this paper, several soil moisture products were evaluated, including Fengyun satellite soil moisture products(FY-3B/3C) and Chinese Land Data Assimilation System Version 2.0(CLDAS-V2.0), aiming to clarify their accuracy characteristics of the time and space sequence. This study is expected to provide a reference for improving accuracy and inversion models of soil moisture products. The verification experiment was carried out in the Naqu region of the Qinghai-Tibet Plateau, based on the daily observed soil moisture data of 0-5cm, 0-10cm and 10-40cm at 1°×1°, 0.3°×0.3° and 0.1°×0.1° scale sparse observation networks. In comparison, the Global Land Data Assimilation System Vision 1.0(GLDAS-1) Noah soil moisture products were added to conduct the evaluate as a reference. The results showed that: (1) in terms of time sequence, CLDAS-V2.0 soil moisture products showed good temporal and spatial continuity. The null values of FY series soil moisture products were common existed, especially in freezing periods. FY and CLDAS-V2.0 soil moisture products usually overestimated the in-situ measurements, especially after the precipitation and in the vegetation growth period. (2) During the multi-scale evaluation, compared with the large-scale sparse observation station, the results were more stable and excellent on the small-scale sparse observation network. Correlation coefficients between soil moisture products and in-situ data at small scale sparse observation network were higher than those at large and middle scale sparse observation network, but Bias, RMSE and ubRMSD were lower. For example, correlation coefficient of CLDAS-V2.0 soil moisture product of 0-5cm at three(large, middle and small) scale sparse observation networks were 0.77, 0.78 and 0.79. (3) At multi-scale sparse observation networks, the results of validation were consistent. At 0−5cm, the quality of CLDAS-V2.0 soil moisture products was better than that of FY series soil moisture products on the whole. The daytime retrieved FY series soil moisture products were more accurate than the night observed data. At 0−10cm, GLDAS-1 Noah soil moisture products estimated closely soil moisture in Naqu region. However, R between GLDAS-1 Noah soil moisture products and in-situ data at multi-scale sparse observation networks were less than CLDAS-V2.0. At 10−40cm, CLDAS-V2.0 soil moisture products performed better than GLDAS-1 soil moisture products. The spatial statistical indices(Bias, RMSE and ubRMSD) of CLDAS-V2.0 and GLDAS-1 soil moisture products were closed, but correlation coefficients of CLDAS-V2.0 soil moisture products were much higher than those of GLDAS-1 soil moisture products. It showed that the quality of FY-3B/3C and CLDAS soil moisture products was relatively stable and reliable.

Key words: Soil moisture, Fengyun satellites, CLDAS, Multi-scale, Accuracy evaluation