Attributing land water storage trends from satellite gravimetry to long-term wetting and drying conditions with global climate models

Global and interactively coupled climate models are important tools for projecting future climate conditions. Even though the quality and reliability of such models has increased during the most recent years, large model uncertainties still exist for various climate elements, so that it is crucial to continuously evaluate them against independent observations. Changes in the distribution and availability of terrestrial water storage (TWS), which can be measured by the satellite gravimetry missions GRACE and GRACE-FO, represent an important part of the climate system in general, and the terrestrial water cycle in particular. However, the use of satellite gravity data for the evaluation of interactively coupled climate models has only very recently become feasible. Challenges mainly arise from large model differences with respect to land water storage-related variables, from conceptual discrepancies between modeled and observed TWS, and from the still rather short time series of satellite data.

This presentation will highlight the latest results achieved from our ongoing research on climate model evaluation based on the analysis of an ensemble of models taking part in the Coupled Model Intercomparison Project Phase 6 (CMIP6). We will focus on long-term wetting and drying conditions in TWS, by deriving several hot spot regions of common trends in GRACE/-FO observations and regions of large model consensus. However, as the observational record currently only covers about 20 years, observed trends may still be obscured by natural climate variability. Therefore, to further attribute the wetting or drying in the identified hot spot regions to either interannual/decadal variability or anthropogenic climate change, we investigate the influence of dedicated climate modes (such as ENSO, PDO, AMO etc.) on TWS variability and trends. Furthermore, we perform a numerical model investigation with 250 years of CMIP6 TWS data to quantify the degree to which trends computed over differently long time intervals can be expected to represent long-term trends, and to discriminate regions of rather robust trends from regions of large fluctuations in the trend caused by decadal climate variability.