Soil moisture data assimilation based on NDVI optimization

Soil moisture could be better estimated through assimilation at various observations into ecosystem models to use all sources of information well. However, different assimilation results could be got from different observations. And the results might have big differences. Three different remote sensed surface soil moisture derived from MODIS red、near infrared and shortwave infrared bands was assimilated to initialize the soil moisture in BEPS(boreal ecosystem production simulator) in May to July in 2008, taking Guguan city in Ningxia as a case. Three different drought indexes PDI (perpendicular drought index)、SPSI (shortwave infrared perpendicular water stress index)and MPDI (modify perpendicular drought index)to derive the surface soil moisture were chose based on the time-series NDVI (normalized difference vegetation index) threshold. An Ensemble Kalman Filter was used to perform the data assimilation. The in-situ sites’ observations were used to verify the assimilation results which were got from three different remote sensed results. It was demonstrated that the method of remote sensed soil moisture assimilation could help to improve the results in BEPS model. And the assimilation using accurate remote sensed result as observation in different time series could help to improve the soil moisture results.