Data assimilation of satellite-based actual evapotranspiration in a distributed hydrological model of a controlled water system

doi:10.1016/j.jag.2016.12.015

International Journal of Applied Earth Observation and Geoinformation

Volume 57, May 2017, Pages 123-135

https://doi.org/10.1016/j.jag.2016.12.015 Get rights and content

Highlights

•
The particle filter data assimilation framework that uses the spatial satellite-based evapotranspiration data has been implemented.
•
It has been shown that use of data assimilation in the highly controlled water system results in the increased accuracy of discharge simulation.
•
Open-source software is utilized for the data assimilation framework.

Abstract

Advances in earth observation (EO) and spatially distributed hydrological modelling provide an opportunity to improve modelling of controlled water systems. In a controlled water system human interference is high, which may lead to incorrect parameterisation in the model calibration phase. This paper analyses whether assimilation of EO actual evapotranspiration (ETa) data can improve discharge simulation with a spatially distributed hydrological model of a controlled water system. The EO ETa estimates are in the form of eight-day ETa composite maps derived from Terra/MODIS images using the ITA-MyWater algorithm. This algorithm is based on the surface energy balance method and is calibrated for this research for a low-lying reclamation area with a heavily controlled water system: the Rijnland area in the Netherlands. Data assimilation (DA) with the particle filter method is applied to assimilate the ETa maps into a spatially distributed hydrological model. The hydrological model and DA framework are applied using the open source software SIMGRO and PCRaster-Python respectively. The analysis is done for a period between July and October 2013 in which a high discharge peak followed a long dry-spell. The assimilation of EO ETa resulted in local differences in modelled ETa compared to simulation without data assimilation, while the area average ETa remained almost the same. The modelled cumulative discharge graphs, with and without DA, showed distinctive differences with the simulation, with DA better matching the measured cumulative discharge. The bias of simulated cumulative discharge to the observed data reduced from 14% to 4% when using DA of EO ETa. These results showed that assimilating EO ETa may not only be effective in the more common applications of soil moisture and crop-growth modelling, but also for improving discharge modelling of controlled water systems.

Introduction

In a controlled water system (CWS), hydrological variables and system states are partly determined by man-made control structures. Usually, in such systems the water flow is controlled by regulating structures aiming at maintaining the water level or discharge to a certain target level for a certain period. A typical example of a CWS is an irrigation system, with weirs, channels and gates to regulate the water flow. Inside the CWS, human influences highly affect the hydrological states. In low-lying land reclamation areas, such as in large parts of the Netherlands, human influence may be stronger still, for example because of the use of pre-pumping determined by forecasted heavy rainfall events, or because of flushing of the canals to maintain good water quality in the system. Modelling a CWS differs from modelling a natural water system since modellers should not only focus on the hydrodynamic and numerical challenges of the modelling system (Clemmens et al., 2005) but also take into account a high degree of freedom with many unknown events triggered by structures or human decisions (often unregistered). At the same time, data availability for CWS is typically higher than that for a natural system (van Andel et al., 2010).

Remote sensing images can provide additional data for distributed hydrological models. A number of satellite missions and platforms are of specific interest for hydrological studies. These are mainly sensor systems that combine thermal infrared and optical data, such as the Thematic Mapper (TM), Enhanced Thematic Mapper Plus (ETM + ), Operational Land Imager (OLI) and Thermal Infrared Scanner (TIRS) on board of the Landsat satellite series, the Advanced Very High Resolution Radiometer (AVHRR) on board NOAA (National Oceanic and Atmospheric Administration) satellite series, the Moderate Resolution Imaging Spectro-radiometer (MODIS) on board of the Terra and Aqua satellites, and the upcoming Sentinels (Sentinel-2 and Sentinel-3). Of equal interest are the active and passive radar systems, such as the RADARSAT, TerraSAR-X, COSMO-SkyMed, and Sentinel-1. Apart from the sensor’s spectral definition, revisit time and spatial resolution are important considerations to determine the suitability of a sensor for hydrological monitoring and modelling.

Earth observation (EO) instruments measure the hydrological variables or parameters indirectly, so an interpretative model has to be used to estimate their actual values. The variables important for hydrological modelling that can be retrieved by EO imagery, are, e.g. land use/land cover, natural vegetation cover, leaf area index, biomass, ground surface elevation, and surface energy balance parameters (Schultz and Engman, 2000). However, the accuracy of estimated variables may vary, depending on the sensor, region, atmospheric conditions and local surface conditions, so the accuracy of EO products differs in space and time.

Data assimilation (DA) is expected to help in combining the strengths of the EO information and the hydrological model to better characterize the modelled state (Dorigo et al., 2007). EO provides the hydrological model with a measured estimation of a certain hydrological variable at the moments when the EO data is available (Errico, 1999, Errico et al., 2000), and the DA process modifies the model states to result in a hydrological state that is closer to the state as estimated by EO.

Use of DA of EO data for hydrological modelling has been demonstrated in a number of studies. For example, the soil moisture estimation from EO data has been assimilated in hydrological models by, amongst others, Hoeben and Troch (2000), Reichle et al., 2002, Reichle et al., 2001, Lee et al. (2011), Han et al. (2012), and Lievens et al. (2015). Assimilation of soil moisture combined with leaf area index (LAI) has been shown to improve crop-growth model results (Pauwels et al., 2007). In addition, the combination of soil moisture with stream flow DA has been carried out, e.g. by Aubert et al. (2003) and Wanders et al. (2014). Snow cover area has been successfully assimilated to improve stream flow simulation, e.g. by Clark et al. (2006), Nagler et al. (2008), and De Lannoy et al. (2012). It should be noted however, that microwave remote sensing gives an estimate of soil moisture in the top few centimetres, while thermal infrared images combined with surface energy balance algorithms may provide information extending to the root zone (Alexandridis et al., 2016).

Studies assimilating EO-based actual evapotranspiration (ETa) into distributed hydrological models are more limited, and usually not always with the aim of improving discharge simulation. For example, Olioso et al. (2005), Vazifedoust et al. (2009), and Irmak and Kamble (2009) performed DA of EO ETa for improving crop-growth models. Schuurmans et al. (2003) demonstrated that the assimilation of 1 km resolution NOAA-AVHRR latent heat flux data could improve the ETa simulation of a distributed hydrological model. Qin et al. (2008) assimilated ETa into a distributed hydrological model and analysed the improvement in ETa simulation. Xie and Zhang (2010) tried to improve discharge simulation using ETa assimilation, however, the ETa assimilation was only supplementary to the assimilation of measured discharge.

So far no studies have been done on DA of ETa in controlled water systems. The aim of this paper is to study the effectiveness of assimilation of EO ETa in a distributed hydrological model for improving discharge simulation of a controlled water system.

Section snippets

Rijnland water system

The Rijnland area is located in the western part of the Netherlands, the location is presented in Fig. 1a. 70% Of the 1200 km² area is a low-lying region where the ground surface is below the mean sea level (0.00 m +NAP, Normaal Amsterdams Peil), as shown in Fig. 1a where the white to blue colours indicate levels below mean sea level. Rijnland is divided into more than 200 polders and sub-polders (Fig. 1b): controlled catchment sub-areas where dikes prevent water from outside to flood the area

Results and discussion

The SIMGRO hydrological model is validated against the measured total discharge of the four main pumping stations mentioned in section 2.1, for the year 2011. However, the human influence is high in the operation of the pumps. The human influence may lead to un-natural patterns in the observed discharge time series. For instance, there can be changes in the target water level in the canals or need of flushing, hence additional water needs to be supplied to the system. Furthermore, there might

Conclusions

This research has implemented data assimilation of EO ETa in a distributed hydrological model to analyse the effect on simulated discharge from controlled water systems. A particle filter algorithm with residual resampling was used. The results for the controlled water system of Rijnland in the Netherlands showed an increased accuracy of simulated discharge. This indicates that assimilating EO ETa may not only have value for the more common applications in soil moisture and crop-growth

References (56)

D. Aubert et al.
Sequential assimilation of soil moisture and streamflow data in a conceptual rainfall–runoff model
J. Hydrol.
(2003)
W.G.M. Bastiaanssen et al.
A remote sensing surface energy balance algorithm for land (SEBAL). 1. Formulation
J. Hydrol.
(1998)
M.P. Clark et al.
Assimilation of snow covered area information into hydrologic and land-surface models
Adv. Water Resour.
(2006)
C. Dechant et al.
Radiance data assimilation for operational snow and streamflow forecasting
Adv. Water Resour.
(2011)
W.A. Dorigo et al.
A review on reflective remote sensing and data assimilation techniques for enhanced agroecosystem modeling
Int. J. Appl. Earth Obs. Geoinf.
(2007)
E. Han et al.
Implementation of surface soil moisture data assimilation with watershed scale distributed hydrological model
J. Hydrol.
(2012)
D. Karssenberg et al.
A software framework for construction of process-based stochastic spatio-temporal models and data assimilation
Environ. Modell. Softw.
(2010)
H. Lee et al.
Assimilation of streamflow and in situ soil moisture data into operational distributed hydrologic models: effects of uncertainties in the data and initial model soil moisture states
Adv. Water Resour.
(2011)
H. Lievens et al.
SMOS soil moisture assimilation for improved hydrologic simulation in the Murray Darling Basin, Australia
Remote Sens. Environ.
(2015)
Q. Mu et al.
Improvements to a MODIS global terrestrial evapotranspiration algorithm
Remote Sens. Environ.
(2011)

T. Nagler et al.

Assimilation of meteorological and remote sensing data for snowmelt runoff forecasting

Remote Sens. Environ.

(2008)

J.M. Schuurmans et al.

Assimilation of remotely sensed latent heat flux in a distributed hydrological model

Adv. Water Resour.

(2003)

X. Xie et al.

Data assimilation for distributed hydrological catchment modeling via ensemble Kalman filter

Adv. Water Resour.

(2010)

J. Yu

On leverage in a stochastic volatility model

J. Econometr.

(2005)

P.E.V. van Walsum et al.

Integration of models using shared state variables: implementation in the regional hydrologic modelling system SIMGRO

J. Hydrol.

(2011)

AHN

Actueel Hoogtebestand Nederland [WWW Document]

(2011)

T.K. Alexandridis et al.

Integrated methodology for estimating water use in mediterranean agricultural areas

Remote Sens.

(2009)

T. Alexandridis et al.

Spatial and temporal distribution of soil moisture at the catchment scale using remotely-sensed energy fluxes

Water

(2016)

R.G. Allen et al.

Crop Evapotranspiration-Guidelines for Computing Crop Water Requirements-FAO Irrigation and Drainage Paper 56

(1998)

CPTEC/INPE

Center for Weather Forecasting and Climate Research (Centro De Previsão De Tempo E Estudos Climáticos) CPTEC/INPE [WWW Document]

(2014)

I. Cherif et al.

Improving remotely sensed actual evapotranspiration estimation with raster meteorological data

Int. J. Remote Sens.

(2015)

A.J. Clemmens et al.

Simulation of automatic canal control systems

J. Irrig. Drain. Eng.

(2005)

G.J.M. De Lannoy et al.

Multiscale assimilation of Advanced Microwave Scanning Radiometer–EOS snow water equivalent and Moderate Resolution Imaging Spectroradiometer snow cover fraction observations in northern Colorado

Water Resour. Res.

(2012)

European Soil Database (distribution Version v2.0)

(2003)

R.M. Errico et al.

2000. NOAA-NASA-DoD workshop on satellite data assimilation

Bull. Am. Meteorol. Soc.

(2000)

R.M. Errico

Workshop on assimilation of satellite data

Bull. Am. Meteorol. Soc.

(1999)

R.A. Feddes et al.

Simulation of Field Water Use and Crop Yields Simulation Monographs

(1978)

D. Fox et al.

Monte carlo localization: efficient position estimation for mobile robots

AAAI/IAAI

(1999)

Cited by (31)

Enhancing the capabilities of the Chao Phraya forecasting system through the integration of pre-processed numerical weather forecasts
2024, Journal of Hydrology: Regional Studies
This study focuses on the Chao Phraya (CPY) Basin, Thailand.
This study aims to improve the skill of the CPY flood forecasting system, developed by Hydro-Informatics Institute (HII) and DHI A/S since 2012. It introduces two pre-processing workflows that are applied to the raw numerical weather prediction (NWP) provided by Weather Research and Forecasting model (WRF): quantile mapping bias correction (QM) and random forest regression (RF). Rainfall forecasts, updated with two pre-processing methods, WRF-QM and WRF-RF, were evaluated against daily rainfall measurements from HII’s stations in each subcatchment. Six hydrological re-forecasting experiments were conducted using a hydrological model to compare runoff forecasts with and without preprocessing method as well as with in-situ rainfall, climatology and persistence benchmark. We assessed rainfall and runoff predictions during training (2016–2019) and testing periods (2020–2021).
Utilizing pre-processing methods in rainfall prediction enhances the accuracy for raw rainfall and runoff predictions. The WRF-QM and WRF-RF methods improved rainfall prediction by 12% and 18% in RMSE’s terms during testing period, respectively. Overall performance results indicate runoff forecasting with WRF-QM and WRF-RF pre-processing reduces RMSE by 34% and 40%, respectively, compared to Raw WRF. Wilcoxon signed-test confirmed significant improvement with pre-processing methods. Our study demonstrates the potential of pre-processed NWP to enhance the skill of hydrologic forecasting systems. Pre-processing methods boost flood forecasting reliability, addressing challenges caused by more frequent and severe hydrologic extremes.
Quest to find compromised spatial and temporal resolutions for integrating remote sensing data with an agro-ecosystem model for grasslands
2024, International Journal of Applied Earth Observation and Geoinformation
This paper addressed one of the main challenges in assimilating remote sensing derived variables into process-based crop model simulations, which is the inconsistent spatial and temporal resolution between information obtained from remote sensing and the outputs of process-based agroecosystem model. We proposed an applied method to reduce the number of required simulations by identifying (i) optimal points in time where additional information from remote sensing has the largest positive influence on the model performance and (ii) options to cluster 10 m grid cells to larger cells without compromising their information content. The MONICA (Model for Nitrogen and Carbon) model was applied to simulate above and below ground biomass in two grassland sites located in southern and eastern parts of Germany. The model was calibrated using LAI values obtained from Sentinel-2 and the sensitivity of output variables to two key root parameters (Root Form factor and specific root length) was evaluated. Our results showed that one or two satellite images covering the critical time periods right after cutting events significantly improved the predictions of grass yields produced by a mechanistic agroecosystem models (by up to 30 %). A larger number of images at other grass growth stages would not further improve the predictive power of the model. We also found that the sensitivity to these critical time periods was independent of model parameters. The mixed-resolution scheme (between 10 and 50 m) achieved better results compared with the high-resolution standalone state-updating method, yet it reduced computational costs by more than 50 %. In conclusion, we proposed a methodology to reduce the number of required simulations for data assimilation by aggregating data from fine to coarse resolutions. Our method was promising for applying data assimilation over large areas and benefiting more from satellite information for real-time prediction of agricultural productivity.
GEE can prominently reduce uncertainties from input data and parameters of the remote sensing-driven distributed hydrological model
2023, Science of the Total Environment
The coupling of multisource remote sensing data and the lack of measured runoff introduce input data and model parameters uncertainties to the remote sensing-driven distributed hydrological model (RS-DHM). The PB satellite remote sensing datasets of the Google Earth Engine (GEE) are widely used in RS-DHM and remote sensing runoff inversion research, but whether GEE can reduce the two abovementioned uncertainties is still unknown. To answer this question, twelve remote sensing data sources provided by GEE were used in this study to drive a typical RS-DHM called the remote sensing-driven distributed time-variant gain model (RS-DTVGM) and the remote sensing runoff inversion technology called remote sensing hydrological station (RSHS), and the contribution of GEE to the improving hydrological model uncertainties was quantitatively analyzed from 2001 to 2020. The results showed that (1) the GEE-based improved data preparation not only effectively reduced the uncertainty in the input data with better spatial-temporal continuity and a 6.20 % reduction in the total area occupied by invalid grids, but also enhanced the operational efficiency by reducing the image number, memory size and data processing time of the satellite remote sensing data by 83.63 %, 99.53 %, and 98.73 %, respectively; (2) the GEE-based RSHS technology provided sufficient data support for parameter adjustment and accuracy validation of the RS-DTVGM, which effectively reduced the uncertainty in the model parameters and increased the Nash efficiency coefficient (NSE) in the calibration and validation period from 0.67 to 0.87 and 0.75, respectively; and (3) the calibrated RS-DTVGM was more reliable and robust, and its runoff and evapotranspiration were consistent with the actual statistical data. In the future, GEE and RSHS technology should be widely adopted to drive the RS-DHM to more quickly and easily provide reliable hydrological processes simulation results for integrated water resource management, therefore achieving win–win results in terms of efficiency and accuracy.
Estimating evapotranspiration using earth observation data: A comparison between hydrological and energy balance modelling approaches
2022, Journal of Hydrology
Citation Excerpt :
Earth observation data can be used in hydrological models at three different levels; (i) direct use of EO data as inputs to hydrological simulation models, (ii) using EO data for the calibration of parameters within the model, and (iii) assimilation of EO data into hydrological models in order to improve model simulations (Xu et al., 2014). There have been ample studies focusing on the use of calibration and data assimilation procedures in the estimation of satellite ETa (Kamble and Irmak, 2008; Vazifedoust et al., 2009; Zhang et al., 2009; Hartanto et al., 2017; Jiang et al., 2020; Bennour et al., 2022). However, there is a general lack of studies focusing on using ETa as a direct input into models.
Hydrological models are often applied to acquire estimates of evapotranspiration (ET_a) using quality inputs of reference evapotranspiration (ET₀). However, their application may be limited by the need for good quality long-term data inputs. In data scarce regions, the use of satellite-earth observation data as inputs to these models may provide a potential solution to overcome data acquisition challenges associated with conventional data collection approaches. In this study, the Agricultural Catchments Research Unit (ACRU) hydrological model was applied to estimate the ET_a using both in-situ and satellite-derived estimates of ET₀ as inputs to the model. The simulated ET_a estimates were then evaluated through comparisons against field-based measurements of ET_a acquired from an Eddy Covariance system (EC_ET), as well as from the application of the Surface Energy Balance System (SEBS) model. The results of these investigations demonstrated that the meteorologically-based ACRU ET_a simulations marginally outperformed the satellite-based ACRU ET_a simulations. While these results highlight the influence which the quality of ET₀ inputs may have on the accuracy of the ACRU derived ET_a estimates, the concurrent application and evaluation of SEBS demonstrated that alternate approaches to hydrological modelling, may be better suited for ET_a estimation in certain instances.
Earth Observation and Hydraulic Data Assimilation for Improved Flood Inundation Forecasting
2021, Earth Observation for Flood Applications: Progress and Perspectives
Accurately simulating floodplain inundation is absolutely vital to minimize damage to life and property. The giant strides made in advanced computing, now allow running increasingly complex models at reasonable resolutions over large areas, with the promise of further improvement in the near future. However, the uncertainty contributed by input, boundary, and forcing data, often leads to highly erroneous predictions. As spatially distributed Earth Observations of flood extent and water level become increasingly available, they pave the way for further constraining, and hence, improving the accuracy of hydraulic flood forecasting models. Effectively using these datasets requires an in-depth understanding of the impacts that the resolution, accuracy, location, timing, and frequency of acquisition of observations may have on model-data integration efforts. This chapter presents a review of the current capabilities in the field of flood data assimilation. The challenges and opportunities of using Earth Observation data for operational flood inundation forecasting are also discussed.
Altimetry-derived surface water data assimilation over the Nile Basin
2020, Science of the Total Environment
Citation Excerpt :
EnKF is a relatively cost-effective filter that does not require to record historical observations or minimizing a chosen a complex cost function contrary to other techniques such as variational methods (Talagrand and Courtier, 1987; Courtier et al., 1994; Hoteit et al., 2005). The filter has been widely used in hydrological studies to integrate satellite data products such as those from satellite soil moisture (e.g., Reichle et al., 2002; Crow and Wood, 2003; Zheng et al., 2018), evapotranspiration (e.g., Guerschman et al., 2009; Hartanto et al., 2017; van Dijk et al., 2018), and the Gravity Recovery And Climate Experiment (GRACE) mission (see also Zaitchik et al., 2008; Khaki and Awange, 2019b; Shokri et al., 2019). In what follows, brief information about the Nile Basin characteristics is first presented (Section 2), then, data and method are described in Sections 3 and 4, respectively.
Global hydrological models facilitate studying of water resources and their variations over time. The accuracies of these models are enhanced when combined with ever-increasing satellite remotely sensed data. Traditionally, these combinations are done via data assimilation approach, which permits the use of improved hydrological outputs to study regions with limited in-situ measurements such as the Nile Basin. This study aims at using the state-of-art satellite radar altimetry data to enhance a land-based hydrological model for studying water storage changes over the Nile Basin. Altimetry-derived surface water storage, for the first time, is assimilated into the model using the ensemble Kalman filter (EnKF) for the period of 2003 to 2016. Multiple datasets from ground measurements, as well as space observations, are used to evaluate the performance of the assimilated satellite altimetry data. Results indicate that the assimilation successfully improves model outputs, especially the surface water component. The process increases the correlation between surface water storage changes and water level variations from satellite radar altimetry by 0.44 and reduces the surface water discharge root-mean-square errors (RMSE) by approximately 33%.

View all citing articles on Scopus

View full text

Data assimilation of satellite-based actual evapotranspiration in a distributed hydrological model of a controlled water system

Highlights

Abstract

Introduction

Section snippets

Rijnland water system

Results and discussion

Conclusions

J. Hydrol.

J. Hydrol.

Adv. Water Resour.

Adv. Water Resour.

Int. J. Appl. Earth Obs. Geoinf.

J. Hydrol.

Environ. Modell. Softw.

Adv. Water Resour.

Remote Sens. Environ.

Remote Sens. Environ.

Remote Sens. Environ.

Adv. Water Resour.

Adv. Water Resour.

J. Econometr.

J. Hydrol.

Actueel Hoogtebestand Nederland [WWW Document]

Integrated methodology for estimating water use in mediterranean agricultural areas

Remote Sens.

Spatial and temporal distribution of soil moisture at the catchment scale using remotely-sensed energy fluxes

Water

Crop Evapotranspiration-Guidelines for Computing Crop Water Requirements-FAO Irrigation and Drainage Paper 56

Center for Weather Forecasting and Climate Research (Centro De Previsão De Tempo E Estudos Climáticos) CPTEC/INPE [WWW Document]

Improving remotely sensed actual evapotranspiration estimation with raster meteorological data

Int. J. Remote Sens.

Simulation of automatic canal control systems

J. Irrig. Drain. Eng.

Multiscale assimilation of Advanced Microwave Scanning Radiometer–EOS snow water equivalent and Moderate Resolution Imaging Spectroradiometer snow cover fraction observations in northern Colorado

Water Resour. Res.

European Soil Database (distribution Version v2.0)

2000. NOAA-NASA-DoD workshop on satellite data assimilation

Bull. Am. Meteorol. Soc.

Workshop on assimilation of satellite data

Bull. Am. Meteorol. Soc.

Simulation of Field Water Use and Crop Yields Simulation Monographs

Monte carlo localization: efficient position estimation for mobile robots

AAAI/IAAI