Technical workshop on assimilating GRACE/-FO data into land surface and hydrological models

Format: Online workshop one or two 3-4h sessions
suitable in US (morning) and Europe (afternoon) time zones

Date: January 18 and 19, 2024

Organizers: Jürgen Kusche (University of Bonn), Gabrielle de Lannoy (KU Leuven), Matt Rodell (NASA GSFC)

 Several groups worldwide are now assimilating GRACE and GRACE-FO data, typically in form of monthly TWSA maps, into hydrological or land surface models. However, these data sets differ very much from more conventional data sets used in assimilation systems, and open questions remain with respect to data products (spherical harmonics vs. mascon grids, uncertainty assumptions), pre-processing (e.g. filtering, rescaling, dealing with mass signals that cannot be represented through model variables such as glaciers), prognostic model requirements (e.g. presence of groundwater), assimilation approaches (e.g. temporal disaggregation, multi-data assimilation, ensemble Kalman or particle filtering or smoothing), reconciling spatial and temporal mismatches between observation and model (model space vs. observation space data assimilation, model time steps vs. monthly observations, fine scale model features (e.g., mountains and surface water) vs. spatially smoothed observations), and tuning (e.g. ensemble size and generation), interpretation and validation of assimilation results, model comparisons and bias assessment, and the potential of future e.g. quantum instrumentation / multi-pair space-gravimetric data within assimilative frameworks. Meanwhile also some model-free downscaling techniques have been proposed, as well as the use of machine learning for e.g. bridging data gaps.

So far neither a „standard approach“ for assimilating GRACE/-FO data exists nor has a standard validation procedure been defined. This hampers the comparison across different groups, as well as the development of new applications (e.g. for coupled ESMs).

The purpose of the workshop would be to bring groups working on GRACE/-FO assimilation together, exchange information on e.g. data preprocessing and DA system tuning, identify problems and seek a common understanding, and see how/whether e.g. a „benchmark“ data set can be established.