Bias-aware data assimilation in integrated hydrological modelling

Marc-Etienne Ridler; Donghua Zhang; Henrik Madsen; Jacob Kidmose; Jens Christian Refsgaard; Karsten Høgh Jensen

doi:10.2166/nh.2017.117

Bias-aware data assimilation in integrated hydrological modelling

Marc-Etienne Ridler, Donghua Zhang, Henrik Madsen, Jacob Kidmose, Jens Christian Refsgaard, Karsten Høgh Jensen

Geology

4 Citations (Scopus)

Abstract

One of the major challenges in hydrological data assimilation applications is the presence of bias in both models and observations. The present study uses the ensemble transform Kalman filtering (ETKF) method and an observational bias estimation technique to estimate groundwater hydraulic heads. The study was carried out in a relatively complex, groundwater dominated, catchment in Denmark using the MIKE SHE model code. The method is implemented and evaluated using synthetic data and subsequently tested against real observations. The results from the synthetic experiments show that the bias-aware filter outperforms the standard filter, with improved state estimate and correct bias estimate. The assimilation using real observations further demonstrates the robustness of bias-aware ETKF, and the potential improvements using integrated hydrological modelling. Furthermore, the experiments with assimilating over different depths show that the state estimates depend on correlation across layers.

Original language	English
Journal	Hydrology Research
Volume	49
Issue number	4
Pages (from-to)	989-1004
Number of pages	16
ISSN	0029-1277
DOIs	https://doi.org/10.2166/nh.2017.117
Publication status	Published - Aug 2018

Access to Document

10.2166/nh.2017.117

nh0490989.pdfFinal published version, 939 KB

https://iwaponline.com/hr/article-pdf/49/4/989/537715/nh0490989.pdf

Cite this

@article{485823db173e401bbba1797549ddf6d7,

title = "Bias-aware data assimilation in integrated hydrological modelling",

abstract = "One of the major challenges in hydrological data assimilation applications is the presence of bias in both models and observations. The present study uses the ensemble transform Kalman filtering (ETKF) method and an observational bias estimation technique to estimate groundwater hydraulic heads. The study was carried out in a relatively complex, groundwater dominated, catchment in Denmark using the MIKE SHE model code. The method is implemented and evaluated using synthetic data and subsequently tested against real observations. The results from the synthetic experiments show that the bias-aware filter outperforms the standard filter, with improved state estimate and correct bias estimate. The assimilation using real observations further demonstrates the robustness of bias-aware ETKF, and the potential improvements using integrated hydrological modelling. Furthermore, the experiments with assimilating over different depths show that the state estimates depend on correlation across layers.",

author = "Marc-Etienne Ridler and Donghua Zhang and Henrik Madsen and Jacob Kidmose and Refsgaard, {Jens Christian} and Jensen, {Karsten H{\o}gh}",

year = "2018",

month = aug,

doi = "10.2166/nh.2017.117",

language = "English",

volume = "49",

pages = "989--1004",

journal = "Hydrology Research",

issn = "1998-9563",

publisher = "I W A Publishing",

number = "4",

}

TY - JOUR

T1 - Bias-aware data assimilation in integrated hydrological modelling

AU - Ridler, Marc-Etienne

AU - Zhang, Donghua

AU - Madsen, Henrik

AU - Kidmose, Jacob

AU - Refsgaard, Jens Christian

AU - Jensen, Karsten Høgh

PY - 2018/8

Y1 - 2018/8

N2 - One of the major challenges in hydrological data assimilation applications is the presence of bias in both models and observations. The present study uses the ensemble transform Kalman filtering (ETKF) method and an observational bias estimation technique to estimate groundwater hydraulic heads. The study was carried out in a relatively complex, groundwater dominated, catchment in Denmark using the MIKE SHE model code. The method is implemented and evaluated using synthetic data and subsequently tested against real observations. The results from the synthetic experiments show that the bias-aware filter outperforms the standard filter, with improved state estimate and correct bias estimate. The assimilation using real observations further demonstrates the robustness of bias-aware ETKF, and the potential improvements using integrated hydrological modelling. Furthermore, the experiments with assimilating over different depths show that the state estimates depend on correlation across layers.

AB - One of the major challenges in hydrological data assimilation applications is the presence of bias in both models and observations. The present study uses the ensemble transform Kalman filtering (ETKF) method and an observational bias estimation technique to estimate groundwater hydraulic heads. The study was carried out in a relatively complex, groundwater dominated, catchment in Denmark using the MIKE SHE model code. The method is implemented and evaluated using synthetic data and subsequently tested against real observations. The results from the synthetic experiments show that the bias-aware filter outperforms the standard filter, with improved state estimate and correct bias estimate. The assimilation using real observations further demonstrates the robustness of bias-aware ETKF, and the potential improvements using integrated hydrological modelling. Furthermore, the experiments with assimilating over different depths show that the state estimates depend on correlation across layers.

U2 - 10.2166/nh.2017.117

DO - 10.2166/nh.2017.117

M3 - Journal article

SN - 1998-9563

VL - 49

SP - 989

EP - 1004

JO - Hydrology Research

JF - Hydrology Research

IS - 4

ER -

Bias-aware data assimilation in integrated hydrological modelling

Abstract

Access to Document

Fingerprint

Cite this