USING DOWNSCALED GRACE-FO DATA, 2022 TO ASSESS TOTAL WATER STORAGE IN COASTAL LOWLAND AQUIFERS OF LOUISIANA, MISSISSIPPI, AND ALABAMA

The Gravity Recovery and Climate Experiment Follow-On (GRACE-FO) instrument data relies on minor gravitational field variations to map total water storage (TWS) anomaly. The resolution of GRACE-FO is too coarse (3^o x 3^o) to be useful for small-scale variations in TWS. Downscaling the GRACE-FO mascon data is needed to assess the TWS in regional study areas, like the Coastal Lowland Aquifer System of Louisiana, Mississippi, and Alabama. This study applied Random Forest Model (RFM) to downscale the monthly GRACE-FO of 2022 from 3 degree (~333 km) to a 4 km spatial resolution. The parameters used included monthly total precipitation, monthly mean temperature, monthly Normalized Differences of Vegetation Index (NDVI), monthly Evapotranspiration of 2022; Shuttle Radar Topography Mission (SRTM) DEM, Slope (Generated from SRTM DEM), dominant soil type, land cover of 2022, and lithology were used to generate the 4 km spatial resolution monthly TWS anomaly map of 2022. The validating data R² for each model of the downscaled monthly TWS anomaly map ranges from 0.79 to 0.95. Large R² and small RMSE and MAE values suggest a good fit and accurate predictions by RFM. The downscaled monthly TWS anomaly map of the study area shows the western part of the area is more depleted in TWS than the eastern part. These results agree with a 2018 study by the U.S. Geological Survey that report groundwater depletion in the western Louisiana area. Therefore, using the same methods, potential TWS stresses zone may be identified using the downscaled time series GRACE and GRACE-FO data of the study area, which may be helpful for groundwater management activities.

Keywords: GRACE-FO Mascon; downscaling; total water storage anomaly; random forest model