880 Evaluation of Nine Soil Moisture Reanalysis/Analysis Products over China Using over 2300 in Situ Observations

Wednesday, 9 January 2019
Hall 4 (Phoenix Convention Center - West and North Buildings)
Yong Chen, Nanjing Univ., Nanjing, China; and H. Yuan

Soil moisture (SM) is a critical variable in land surface modeling. Due to the limited SM observations, model produced SM reanalysis/analysis products often serve as alternatives. The uncertainty of these products is affected by various factors, such as the forcing data, the model structure and the quality of assimilated observations in the land surface models. This study aims to evaluate nine of the SM reanalysis/analysis products over China, including SM data generated by NOAA’s Global Forecast System (GFS), NOAA’s National Centers for Environmental Prediction (NCEP)/Department of Energy (DOE) reanalysis 2 (R2), NASA’s Global Land Data Assimilation System (GLDAS) Noah, CLM, VIC and Mosaic products, ECMWF’s ERA-Interim and ERA 5, and China Meteorological Administration Land Data Assimilation System (CLDAS). According to observational network and geographical locations, the study area is divided into eight research regions, namely northeast China (I), north China (II), Jianghuai (III), southeast China (IV), the east part (V) and the west part (VII) of northwest China, southwest China (VI), and Tibet (VIII). These SM products are statistically evaluated in terms of spatial distribution and temporal (daily, monthly, and interannual) variability at the surface layer (0-10 cm) and the middle layer (40-50 cm) during 2010 to 2017 using over 2300 in situ automatic observations.

The results show that the nine products are capable of capturing characteristics of the spatial variations of SM in most regions, while all the products poorly represent the time series of SM in region VII. Overall, CLDAS shows the best agreement with the in situ observations over most parts of China at different time scales, benefiting from higher model resolution, more integrated ground station observations and better atmospheric background forcing. GLDAS CLM has relatively better seasonal change and smaller mean bias than others, and it is even better than CLDAS in region VIII. In contrast, GLDAS Mosaic shows the worst overall performance. ERA Interim has lower correlation and smaller mean bias than ERA5 for most regions. In terms of daily time series, these products have better skill at the surface layer than the middle layer, and perform better in the eastern China than the western regions. GFS, ERA 5, GLDAS VIC, GLDAS Noah and NCEP R2 fail to reproduce the seasonal variability and interannual variability, with either unrealistic extreme SM in the winter/spring or show little variation. The evaluation provides a general guidance to choose the relatively high quality SM products over different regions and seasons in China.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner