% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@PHDTHESIS{Zhang:850282,
author = {Zhang, Hongjuan},
title = {{I}mproved characterization of root zone soil moisture by
assimilating groundwater level and surface soil moisture
data in an integrated terrestrial system model},
volume = {427},
school = {RWTH Aachen},
type = {Dissertation},
address = {Jülich},
publisher = {Forschungszentrum Jülich GmbH Zentralbibliothek, Verlag},
reportid = {FZJ-2018-04324},
isbn = {978-3-95806-335-8},
series = {Schriften des Forschungszentrums Jülich Reihe Energie $\&$
Umwelt / Energy $\&$ Environment},
pages = {x, 125 S.},
year = {2018},
note = {RWTH Aachen, Diss., 2018},
abstract = {Soil moisture is an important variable for the cycling of
water and energy at the catchment/regional/global scale.
Soil moisture content is usually simulated by land surface
models, monitored by ground-based sensors, or observed by
remote sensing techniques. However, land surface models
often have high uncertainties due to simplified
parameterizations and uncertainties from input forcing data
and hydraulic parameters. For example, in most land surface
models, the interaction between groundwater and root zone
soil moisture is neglected. The availabilities of in situ
monitoring networks are limited because of high costs.
Remote sensing can provide soil moisture at the global scale
but is limited to the top 5 cm and resolution is coarse.
Data assimilation can take advantage of these three
different sources of information by assimilating the
observations (e.g. from the ground sensors or remote sensing
data) into the land surface models to improve soil moisture
predictions in both space and time. Furthermore, soil
hydraulic parameters in land surface models can also be
estimated jointly with soil moisture by data assimilation to
further improve soil moisture characterization. In order to
make land surface models more robust, integrated land
surface-subsurface models have been developed which consider
the effect of groundwater on root zone soil moisture in a
fully two-way coupled fashion. In this work, we firstly
compared four data assimilation methods in terms of joint
estimation of soil moisture and soil hydraulic parameters
for two land surface models. The four assimilation methods
included Ensemble Kalman Filter (EnKF) with state
augmentation (EnKF-AUG) or dual estimation (EnKF-DUAL), the
residual resampling Particle Filter (RRPF) and the
MCMC-based parameter resampling method (PMCMC). The two land
surface models used were the Variable Infiltration Capacity
Model (VIC) and the Community Land Model (CLM version 4.5).
Real world data (soil properties, soil moisture measurements
at 5, 20 and 50 cm depth, climate forcing data) from the
Rollesbroich site located in the western Germany were used.
We evaluated the usefulness and applicability of the four
different data assimilation methods for joint parameter and
state estimation of the VIC and the CLM using a 5-month
calibration (assimilation) period of the soil moisture
measurements. The performance of the “calibrated” VIC
and CLM were investigated using water moisture measurements
of a 5-month evaluation period. Results from the first study
showed that all of the four assimilation methods were able
to improve the model predictions of soil moisture after soil
hydraulic parameters (for VIC) or sand/ clay/ organic matter
fraction (for CLM) were jointly estimated with soil
moisture. Overall, EnKF (EnKF-AUG and EnKF-DUAL) performed
better than PF (RRPF and PMCMC). The differences between the
soil moisture simulations of VIC and CLM were much larger
than the discrepancies among the four data assimilation
methods. CLM performed better than VIC in the soil moisture
simulations at 50 cm depth. The large systematic
underestimation of water storage at 50cm depth in VIC is
most probably related to the fact that groundwater is not
well represented in VIC. [...]},
cin = {IBG-3},
cid = {I:(DE-Juel1)IBG-3-20101118},
pnm = {255 - Terrestrial Systems: From Observation to Prediction
(POF3-255)},
pid = {G:(DE-HGF)POF3-255},
typ = {PUB:(DE-HGF)3 / PUB:(DE-HGF)11},
urn = {urn:nbn:de:0001-2018080300},
url = {https://juser.fz-juelich.de/record/850282},
}