Global-scale validation of model-based load deformation of the Earth's crust from continental watermass and atmospheric pressure variations using GPS

Temporal mass variations in the continental hydrosphere and in the atmosphere lead to changes in the gravitational potential field that are associated with load-induced deformation of the Earth's crust. Therefore, models that compute continental water storage and atmospheric pressure can be validated by measured load deformation time series. In this study, water mass variations as computed by the Water-GAP Global Hydrology Model (WGHM) and surface pressure as provided by the reanalysis product of the National Centers for Environmental Prediction describe the hydrological and atmospheric pressure loading, respectively. GPS observations from 14 years at 208 stations world-wide were reprocessed to estimate admittance factors for the associated load deformation time series in order to determine how well the model-based deformation fits to real data. We found that such site-specific scaling factors can be identified separately for water mass and air pressure loading. Regarding water storage variation as computed by WGHM, weighted global mean admittances are 0.74 +/- 0.09, 0.66 +/- 0.10, 0.90 +/- 0.06 for the north, east and vertical component, respectively. For the dominant vertical component, there is a rather good fit to the observed displacements, and, averaged over all sites, WGHM is found to slightly overestimate temporal variations of water storage. For Europe and North America, with a dense GPS network, site-specific admittances show a good spatial coherence. Regarding regional over- or underestimation of WGHM water storage variations, they agree well with GRACE gravity field data. Globally averaged admittance estimates of pre-computed atmospheric loading displacements provided by the Goddard Geodetic VLBI Group were determined to be 0.88 +/- 0.04, 0.97 +/- 0.08, 1.13 +/- 0.01 for the north, east and vertical, respectively. Here, a relatively large discrepancy for the dominant vertical component indicates an underestimation of corresponding loading predictions. (C) 2011 Elsevier Ltd. All rights reserved.