Task 2: GRACE Downscaling

Learn more from about this task, which develops algorithms for spatial and temporal downscaling of GRACE-FO data, in the webinar below:

GRACE Webinar

In the first part of the webinar, we will discuss the predictors used for downscaling GRACE data: precipitation, evapotranspiration, soil moisture, and river runoff. We will explain the sources of these datasets, how to download them from various platforms, and how to perform initial processing.

The next part will focus on the details of GRACE data — from explaining the nature of the signals recorded by the mission's satellites, through the sources from which GRACE-based terrestrial water storage (TWS) data can be obtained, to their final processing. In particular, we will highlight two key aspects of GRACE data processing: (1) filling the nearly one-year data gap resulting from the interruption in observations between the end of the GRACE mission and the launch of its successor, GRACE-FO, and (2) data downscaling to enhance the spatial resolution of TWS, which is critical for regional-scale analyses. Using the example of the transboundary Bug River basin, we will walk through the step-by-step process of predicting missing GRACE TWS data using the ARIMA method, followed by an assessment of the uncertainty associated with this approach. Next, we will demonstrate how to perform downscaling of GRACE data using high-resolution predictors and the Random Forests method, and then how to quantify the quality of the downscaled TWS data.

The final part will focus on deriving groundwater storage (GWS) from satellite data. We will explain how to extract the GWS signal from downscaled GRACE TWS data using outputs from the GLDAS model and how to adapt this approach to local hydrogeological conditions by appropriately selecting the accumulation period of GLDAS-derived TWS based on groundwater table depth. Additionally, we will discuss how to estimate GWS from in-situ measurements of groundwater table levels and specific yield data, and how to use these estimates to calibrate and validate satellite-derived GWS.

By the end of the webinar, attendees will gain both theoretical and practical insights into combining satellite, model, and in-situ data to predict GWS, which is particularly crucial for transboundary areas and regions with limited groundwater monitoring networks.