Schumacher, Maike: Methods for assimilating remotelysensed water storage changes into hydrological models. - Bonn, 2016. - , . In: Schriftenreihe / Institut für Geodäsie und Geoinformation, 52.
Online-Ausgabe in bonndoc:
author = {{Maike Schumacher}},
title = {Methods for assimilating remotelysensed water storage changes into hydrological models},
school = {},
year = 2016,
series = {Schriftenreihe / Institut für Geodäsie und Geoinformation},
volume = 52,
note = {Understanding physical processes within the water cycle is a challenging issue that requires merging information from various disciplines. The Gravity Recovery And Climate Experiment (GRACE) mission provides a unique opportunity to measure time-variable gravity fields, which can be converted to global total water storage anomalies (TWSA). These observations represent a vertical integral of all individual water compartments, which is difficult to observe by in-situ or other remote-sensing techniques. Knowledge about interactions between hydrological fluxes and terrestrial water storage compartments is reflected in large-scale hydrological models that nowadays increase in complexity to simulate all relevant physical processes within the global water cycle. Hydrological models are driven by climate forcing fields and their parameters are usually calibrated against river discharge to ensure a realistic water balance on river basin scale. However, errors in climate forcing fields, model parameters, and model structure limit the reliability of hydrological models. Therefore, it is necessary to improve model simulations by introducing measurements, which is known as data assimilation or data-model fusion.
In this thesis, a novel calibration and data assimilation (C/DA) framework is developed to merge remotely-sensed large scale TWSA with hydrological models. To implement this framework, the WaterGAP Global Hydrology Model (WGHM) is chosen, which is a sophisticated 0.5°x0.5° conceptual model that simulates daily water changes in surface and sub-surface water compartments (including groundwater), and considers water consumption. In particular, a flexible approach is introduced to assimilate GRACE TWSA as (sub-)basin or gridded averages into WGHM, while for the first time, implementing the observation error correlations in the C/DA system. A sensitivity analysis is performed to identify significant parameters in the largest river basins world-wide. It is also investigated whether GRACE TWSA can be used to calibrate model parameters. To reduce sampling errors and to improve the computational efficiency, the classical ensemble Kalman filter (EnKF) technique is extended to a square root analysis (SQRA) scheme, and the singular evolutive interpolated Kalman (SEIK) filter. The relationships between these algorithms are addressed. A simple model and WGHM are used to describe the mathematical details of the data assimilation techniques.
The observation error model, spatial resolution of observations, choice of filtering algorithm, and model ensemble size are assessed within a realistic synthetic experiment designed for the Mississippi River Basin, USA. Real GRACE products are also integrated into WGHM over this region. Investigations indicate that introducing GRACE TWSA constrains the water balance equation and corrects for insufficiently known climate forcing, in particular precipitation. Individual water states and fluxes are also adjusted but more improvements are expected by assimilating further in-situ and remotely-sensed observations. The processing choices represent important impacts on the final results. The C/DA framework is transferred to the Murray-Darling River Basin, Australia, to improve the simulation of hydrological changes under a long-term drought condition. GRACE C/DA introduces a negative trend to WGHM simulated TWSA. A validation with in-situ groundwater measurements indicates that the trend is correctly associated with the groundwater compartment. Thus, the C/DA helps to identify deficits in model simulations and improves the understanding of hydrological processes. The promising results provide a first step towards more complex C/DA applications on global scale and in conjunction with further terrestrial water storage observations.},

url = {}

The following license files are associated with this item: