Estimating sub-monthly TWS using MODIS and GRACE satellite observations, a case study over Tonlé Sap floodplain

Abstract:

Knowledge of Terrestrial Water Storage (TWS) can provide valuable information that can be used to improve our understanding of the hydrological cycle and the impact of extreme climate events. Global TWS observations are currently only available from the Gravity Recovery And Climate Experiment satellite mission (GRACE) at monthly time scales. In this study, we present a new approach to derive the sub-monthly TWS variation over a regularly inundated area by using MODIS reflectance data in addition to GRACE solutions. In the “training” phase, monthly TWS are computed from filtered GRACE solutions. A signal restoration method is applied to correct for signal leakage caused by filtering. In parallel, a time-series of mean monthly inundated area estimates is computed based on the Normalized Difference Water Index (NDWI), derived from MODIS data. The training phase completes by finding an empirical relationship between the inundated area and the GRACE-based TWS variations, using a regression analysis. Then, the estimated parameters can be used to convert inundated area estimates into TWS variations without a further need in GRACE data. This approach has 3 major advantages over the usage of GRACE data alone. First, it can be used to cross-validate GRACE and MODIS reflectance data in order to identify and eliminate unreliable estimates. Second, it can provide sub-monthly (e.g., 8-day) TWS variations without loss of spatial resolution. Lastly, it can be used to fill gaps in TWS estimates based on GRACE data and to extend the time-series of TWS estimates beyond the time interval when GRACE data are available. The methodology is demonstrated using the Tonlé Sap floodplain located in Central Cambodia as a test case. The analysis shows an excellent agreement between the 8-day NDWI-based TWS estimates averaged over monthly intervals and the GRACE-based monthly TWS variations. The approach developed would have similar application in other areas that experience regular large-scale inundation.