Regional four‐dimensional hydrological mass variations from GRACE, atmospheric flux convergence, and river gauge data

Abstract

The spatiotemporal behavior of the Earth's gravity field is traditionally modeled as a series expansion in terms of spherical harmonics with a number of time‐dependent coefficients. In this contribution we study regional effects, namely, hydrological mass variations observed by the Gravity Recovery And Climate Experiment (GRACE) satellite mission since 2002. As compared to the global spherical harmonics, scaling and wavelet functions are quasi‐compactly supported, i.e., highly localizing with respect to space and time. Consequently, they establish a system of base functions well suited for regional modeling. We determine a spatiotemporal regional gravity model from GRACE using the spherical wavelet technique for the spatial part and a one‐dimensional B spline expansion for the temporal variations; that is, we end up with a four‐dimensional multiresolution representation (MRR) of the geopotential over selected regional areas and time intervals. For an exemplary area consisting of three large river basins in South America we compute mass variations expressed in terms of equivalent water heights. We demonstrate that a B spline expansion for modeling the temporal behavior is more appropriate than the classical Fourier series approach. In order to corroborate the results, our storage estimates are balanced with the net effect of river discharge from gauges (outflow) and the difference of precipitation and evaporation (inflow) from the convergence of vertically integrated water vapor fluxes. Furthermore, our results are compared with monthly mass grids from GRACE spherical harmonic solution data products. We conclude that the MRR means a robust tool to maximize the observational accuracy for GRACE, it considers the spatiotemporal resolution of the observations due to resolution‐dependent highest levels, and it is applicable elsewhere globally to model hydrological or other phenomena of mass change.