Assessing the applicability of GLDAS monthly precipitation data in China

Global Land Data Assimilation System (GLDAS) is an important data source for global change and water cycle research. GLDAS-1 and GLDAS-2 monthly precipitation data during 1979—2012 are compared with Chinese ground-based observations for evaluating their capabilities to detect spatial and temporal changes, and evaluating data quality in terms of correlation coefficient, mean bias error, relative absolute error and root mean square error. The results show that: The quality of GLDAS-1 data sets has a problem of discontinuity, especially serious anomalies in 1996 and poor quality in 2000; the qualities of both GLDAS-1 data and GLDAS-2 data decline during the period from 1979 to 2012 according to their fitness with the observed precipitation data; GLDAS data sets show better quality in eastern China wet areas than in western China arid areas; in terms of correlation and error measures, GLDAS-1 precipitation data are slightly better than GLDAS-2 data (mostly in the period before 1995), but GLDAS-2 data are significantly superior to GLDAS-1 data in the temporal consistency, seasonal stability and their capability to describe temporal variations.