Modeling high-resolution land water storage with OS LISFLOOD for global Geodesy

Model-derived terrestrial water storage (TWS) and its individual storage compartments soil moisture, groundwater, surface water, and snow are widely used in the geodetic community for, e.g., the evaluation and improvement of satellite gravimetry products, the correction of GNSS-based coordinate time-series, and simulation studies for future satellite gravity missions. We employ the open-source, high-resolution hydrological rainfall-runoff-routing model OS LISFLOOD to generate global daily water storage time series in 1/20° resolution over the time period 2000 – 2023.

The most recent OS LISFLOOD run performed at the GFZ benefits from several model improvements and adjustments to arrive at a highly realistic TWS simulation. These adjustments include an optimized soil depth definition; an improved model initialization; a modified snow routine; and the inclusion of anthropogenic water abstraction used for irrigation and industrial, domestic, energy (cooling), and livestock demands. A particular challenge in hydrological modeling is the representation of surface water variability. While the most recent version of OS LISFLOOD already explicitly simulates the dynamics of 463 lakes and 667 reservoirs, endorheic lakes (i.e. lakes without an outlet like the Caspian Sea, Lake Balkhash, or Lake Chad) have not been so far accounted for. Since 18% of the land surface drains into endorheic lakes, their consideration is a big step towards improved storage estimates. For the verification of the simulated lake levels we utilize time series from satellite altimetry, and even report on first experiments with altimetry data for the calibration of lake parameters in OS LISFLOOD.

With respect to both GRACE-based TWS estimates and GNSS station displacements, TWS from OS LISFLOOD has been shown before to be superior to results from the Land Surface Discharge Model (LSDM), which has been routinely used for many years at GFZ for geodetic applications. In this contribution we further extend the quality assessment of OS LISFLOOD by utilizing additional TWS data sets from alternative hydrological models (e.g., WGHM, GLDAS, W3RA) that provides insights into the specific strengths and weaknesses of those models regarding their ability to represent TWS at a wide range of spatial and temporal scales.