New airGR developments: semi-distribution and data assimilation

airGR (Coron et al., 2017, 2020) is an R package that offers the possibility to use the GR rainfall-runoff models developed in the Hydrology Research Group at INRAE (formerly at Irstea). It allows running seven hydrological models (including GR4J) dedicated to different time steps (hourly to annual) that can be combined to a snow accumulation and melt model (CemaNeige).Thanks to the success of the airGR package, that was downloaded 45,000 times so far among 50 countries in the world and was used in dozen of publications since its release[1], its development team carries on its efforts to offer new features and improve the computer codes. This is how after offering a first add-on, the airGRteaching package, expressly developed for educational purposes, the team now offers tools dedicated to semi-distribution and data assimilation.Using (semi-)distributed models is often necessary to explicitly represent spatial climatic and physiographic heterogeneities and to allow an analysis of their impact on the watershed response. Consequently, in the latest version of the airGR package, we introduced the semi-distribution of GR models, which are traditionally lumped, on a sub-basin basis. This development will also ultimately enable possibilities of implementing on a modular way different transfer functions as well as integrated water resource management (see package airGRiwrm in Abstract EGU21-2190).In addition, a new package, called airGRdatassim, was recently proposed (Piazzi et al., 2021a, b) as an add-on to the airGR package. airGRdatassim enables the user to assimilate discharge observations via both Ensemble Kalman filter (EnKF) and particle filter (PF) schemes. Besides improving the simulations of GR models, this new package extends the potential applications of airGR to forecasting purposes by allowing for a reliable assessment of the initial conditions of streamflow forecasts.&#160;&#160;References:Coron L., Thirel G., Delaigue O., Perrin C., Andr&#233;assian V. (2017). The Suite of Lumped GR Hydrological Models in an R package, Environmental Modelling & Software, 94, 166-171. DOI: 10.1016/j.envsoft.2017.05.002.Coron, L., Delaigue, O., Thirel, G., Perrin, C. and Michel, C. (2020). airGR: Suite of GR Hydrological Models for Precipitation-Runoff Modelling. R package version 1.4.3.65. URL: https://CRAN.R-project.org/package=airGR.Piazzi, G., Delaigue, O. (2021a). airGRdatassim: Suite of Tools to Perform Ensemble-Based Data Assimilation in GR Hydrological Models. R package version 0.0.3.13. URL: https://gitlab.irstea.fr/HYCAR-Hydro/airgrdatassim.Piazzi, G., Thirel, G., Perrin, C., Delaigue, O. (2021b, accepted). Sequential data assimilation for streamflow forecasting: assessing the sensitivity to uncertainties and updated variables of a conceptual hydrological model. Water Resources Research.<div> <div> [1] https://hydrogr.github.io/airGR/page_publications.html </div> </div>

Download Full-text

Predictability of Seasonal Streamflow Forecasting Based on CSM: Case Studies of Top Three Largest Rivers in China

Water ◽

10.3390/w13020162 ◽

2021 ◽

Vol 13 (2) ◽

pp. 162

Author(s):

Lyuliu Liu ◽

Ying Wu ◽

Peiqun Zhang ◽

Jianqing Zhai ◽

Li Zhang ◽

...

Keyword(s):

Water Resource Management ◽

Climate Models ◽

Initial Conditions ◽

Global Climate Models ◽

Climate System ◽

Hydrological Models ◽

Streamflow Forecasting ◽

Relative Operating Characteristic ◽

Climate System Model ◽

Seasonal Streamflow

Accurate seasonal streamflow forecasting is important in reservoir operation, watershed planning, and water resource management, and streamflow forecasting is often based on hydrological models driven by coupled global climate models (CGCMs). To understand streamflow forecasting predictability, this study considered the three largest rivers in China and explored deterministic and probabilistic skill metrics on the monthly scale according to ensemble streamflow hindcasts from the hydrological model Hydrologiska Byråns Vattenbalansavdelning (HBV) driven by multiple climate forcings from the climate system model by the Beijing Climate Center (BCC_CSM1.1m). The effects of initial conditions (ICs) and meteorological forcings (MFs) on skill were investigated using the conventional ensemble streamflow prediction (ESP) and reverse-ESP (revESP). The results revealed the following: (1) Skill declines as lead time increases, and forecasting is generally the most skillful for lead month 1; (2) skill is higher for dry rivers than wet rivers, and higher for dry target months than wet months for the Yellow and Yangtze Rivers, suggesting greater skill in potential drought forecasting than flood forecasting; (3) the relative operating characteristic (ROC) area is greater for abnormal terciles than the near-normal tercile for all three rivers, greater for the above-normal tercile than the below-normal tercile for the Yellow and Yangtze Rivers, but slightly greater for the below-normal tercile than the above-normal tercile for the Xijiang River; and (4) the influence of ICs outweighs that of MFs in dry months, and the period of influence varies from 1 to 3 months; however, the influence of MFs is dominant in wet target months. These findings will help improve the understanding of both the seasonal streamflow forecasting predictability based on coupled climate system/hydrological models and of streamflow forecasting for variable rivers and seasons.

Download Full-text

airGRiwrm: an extension of the airGR R-package for handling Integrated Water Resources Management modeling

10.5194/egusphere-egu21-2190 ◽

2021 ◽

Author(s):

David Dorchies ◽

Olivier Delaigue ◽

Guillaume Thirel

Keyword(s):

Water Resources ◽

River Basin ◽

Water Resource Management ◽

R Package ◽

Hydrological Models ◽

Environmental Modelling ◽

Seine River ◽

Local Flow ◽

Distributed Models ◽

Seine River Basin

IWRM modeling aims at representing interactions between humans and their environment (Badham et al. 2019), which can involve hydrological, surface-hydraulic, and groundwater models. Semi-distributed models implementing a simplified hydraulic propagation between sub-catchments are often used as IWRM model (Ficchi et al. 2014, Dorchies et al. 2016) because of the good trade-off they offer between simplification and result relevancy. The R-package airGR (Coron et al., 2017, 2020) is widely used in the R language hydrology community and its recent development with semi-distributive (see Abstract EGU21-1371) capabilities allows to use it for IWRM modeling. The R-package airGRiwrm has been developed for multiple purposes linked to IWRM. First, it proposes a simplified network description for building semi-distributed models containing several sub-basins with diverse connections, which greatly simplifies the calibration and modeling steps. Then, it allows to easily integrate predefined flows (feedforward control) into the model, namely local flow injections or withdrawals. Finally, it integrates controllers that apply user-defined decision algorithms given model outputs during simulation (feedback control). The controllers allows for example to apply withdrawal restriction in case of drought, or to simulate a reservoir behaviour with complex management rules.In this presentation, we will introduce the airGRiwrm possibilities and we will demonstrate its use on the case of the Seine River basin in France.&#160;&#160;References:Badham, J., et al., 2019. Effective modeling for Integrated Water Resource Management: A guide to contextual practices by phases and steps and future opportunities. Environmental Modelling & Software 116, 40&#8211;56. https://doi.org/10.1016/j.envsoft.2019.02.013Coron, L., Delaigue, O., Thirel, G., Perrin, C., Michel, C., 2020. airGR: Suite of GR Hydrological Models for Precipitation-Runoff Modelling. R package version 1.4.3.65. https://doi.org/10.15454/EX11NACoron, L., Thirel, G., Delaigue, O., Perrin, C., Andr&#233;assian, V., 2017. The suite of lumped GR hydrological models in an R package. Environmental Modelling & Software 94, 166&#8211;171. https://doi.org/10.1016/j.envsoft.2017.05.002Dorchies, D., Thirel, G., Perrin, C., Bader, J.-C., Thepot, R., Rizzoli, J.-L., Jost, C., Demerliac, S., 2016. Climate change impacts on water resources and reservoir management in the Seine river basin (France). La Houille Blanche 32&#8211;37. https://doi.org/10.1051/lhb/2016047 Ficchi, A., Raso, L., Malaterre, P.-O., Dorchies, D., Jay-Allemand, M., 2014. Short Term Reservoirs Operation On The Seine River: Performance Analysis Of Tree-Based Model Predictive Control. Presented at the International Conference on Hydroinformatics, New York.

Download Full-text

Performance of ensemble streamflow forecasts under varied hydrometeorological conditions

10.5194/hess-2016-584 ◽

2016 ◽

Cited By ~ 2

Author(s):

Harm-Jan F. Benninga ◽

Martijn J. Booij ◽

Renata J. Romanowicz ◽

Tom H. M. Rientjes

Keyword(s):

Hydrological Model ◽

Initial Conditions ◽

Snow Accumulation ◽

Lead Times ◽

Model Parameters ◽

Streamflow Forecasting ◽

Weather Forecasts ◽

Streamflow Forecasts ◽

Forecasting System ◽

Precipitation And Temperature

Abstract. The paper presents a methodology to give insight in the performance of ensemble streamflow forecasting systems. We developed an ensemble forecasting system for the Biała Tarnowska, a mountainous river catchment in southern Poland, and analysed the performance for lead times from 1 day to 10 days for low, medium and high streamflow and related runoff generating processes. Precipitation and temperature forecasts from the European Centre for Medium-Range Weather Forecasts serve as input to a deterministic lumped hydrological (HBV) model. Due to inconsistent bias, the best streamflow forecasts were obtained without pre- and post-processing of the meteorological and streamflow forecasts. Best forecast skill, relative to alternative forecasts based on historical measurements of precipitation and temperature, is shown for high streamflow and for snow accumulation low streamflow events. Forecasts of medium streamflow events and low streamflow events generated by precipitation deficit show less skill. To improve the performance of the forecasting system for high streamflow events, in particular the meteorological forecasts require improvement. For low streamflow forecasts, the hydrological model should be improved. The study recommends improving the reliability of the ensemble streamflow forecasts by including the uncertainties in hydrological model parameters and initial conditions, and by improving the dispersion of the meteorological input forecasts.

Download Full-text

Performance of ensemble streamflow forecasts under varied hydrometeorological conditions

Hydrology and Earth System Sciences ◽

10.5194/hess-21-5273-2017 ◽

2017 ◽

Vol 21 (10) ◽

pp. 5273-5291 ◽

Cited By ~ 6

Author(s):

Harm-Jan F. Benninga ◽

Martijn J. Booij ◽

Renata J. Romanowicz ◽

Tom H. M. Rientjes

Keyword(s):

Hydrological Model ◽

Initial Conditions ◽

Snow Accumulation ◽

Lead Times ◽

Model Parameters ◽

Streamflow Forecasting ◽

Weather Forecasts ◽

Streamflow Forecasts ◽

Forecasting System ◽

Hydrometeorological Conditions

Abstract. The paper presents a methodology that gives insight into the performance of ensemble streamflow-forecasting systems. We have developed an ensemble forecasting system for the Biała Tarnowska, a mountainous river catchment in southern Poland, and analysed the performance for lead times ranging from 1 to 10 days for low, medium and high streamflow and different hydrometeorological conditions. Precipitation and temperature forecasts from the European Centre for Medium-Range Weather Forecasts served as inputs to a deterministic lumped hydrological (HBV) model. Due to a non-homogeneous bias in time, pre- and post-processing of the meteorological and streamflow forecasts are not effective. The best forecast skill, relative to alternative forecasts based on meteorological climatology, is shown for high streamflow and snow accumulation low-streamflow events. Forecasts of medium-streamflow events and low-streamflow events under precipitation deficit conditions show less skill. To improve performance of the forecasting system for high-streamflow events, the meteorological forecasts are most important. Besides, it is recommended that the hydrological model be calibrated specifically on low-streamflow conditions and high-streamflow conditions. Further, it is recommended that the dispersion (reliability) of the ensemble streamflow forecasts is enlarged by including the uncertainties in the hydrological model parameters and the initial conditions, and by enlarging the dispersion of the meteorological input forecasts.

Download Full-text

Latest developments of the airGR rainfall-runoff modelling R-package: inclusion of an interception store in the hourly model

10.5194/egusphere-egu2020-15275 ◽

2020 ◽

Author(s):

Guillaume Thirel ◽

Olivier Delaigue ◽

Andrea Ficchi

Keyword(s):

R Package ◽

Snow Accumulation ◽

Recent Model ◽

Hydrological Models ◽

Rainfall Runoff ◽

Environmental Modelling ◽

Model Improvement ◽

Improved Model ◽

The Impact ◽

Daily Time

airGR (Coron et al., 2017, 2019) is an R package that offers the possibility to use the GR rainfall-runoff models developed in the Hydrology Research Group at INRAE (formerly at Irstea), including the daily GR4J model as well as hourly, monthly and annual models. Recent model developments are regularly introduced in airGR.Recently, an hourly model including an interception store was implemented in airGR. The additional interception store, developed by Ficchi et al. (2019), aims at better representing the impact of vegetation on evaporation fluxes. This improved model showed a better consistency of model fluxes across time and enhanced performance.In addition, the possibility to run the hourly GR models together with the CemaNeige snow accumulation and melt module was added to airGR.&#160;References:Coron L., Thirel G., Delaigue O., Perrin C., Andr&#233;assian V. (2017). The Suite of Lumped GR Hydrological Models in an R package, Environmental Modelling & Software, 94, 166-171. DOI: 10.1016/j.envsoft.2017.05.002.Coron, L., Delaigue, O., Thirel, G., Perrin, C. and Michel, C. (2019). airGR: Suite of GR Hydrological Models for Precipitation-Runoff Modelling. R &#160;&#160;package version 1.4.3.30. URL: https://CRAN.R-project.org/package=airGR.Ficch&#236;, A., Perrin, C., and Andr&#233;assian, V., 2019. Hydrological modelling at multiple sub-daily time steps: model improvement via flux-matching, Journal of Hydrology, 575, 1308-1327, https://doi.org/10.1016/j.jhydrol.2019.05.084.

Download Full-text

Hybridizing Bayesian and variational data assimilation for robust high-resolution hydrologic forecasting

10.5194/hess-2017-431 ◽

2017 ◽

Author(s):

Felipe Hernández ◽

Xu Liang

Keyword(s):

High Resolution ◽

Data Assimilation ◽

Predictive Accuracy ◽

Initial Conditions ◽

Time Estimation ◽

Variational Data Assimilation ◽

Streamflow Forecasting ◽

Infiltration Capacity ◽

Computational Overhead ◽

Capacity Model

Abstract. The success of real-time estimation and forecasting applications based on geophysical models has been possible thanks to the two main frameworks for the determination of the models’ initial conditions: Bayesian data assimilation and variational data assimilation. However, while there have been efforts to unify these two paradigms, existing attempts struggle to fully leverage the advantages of both in order to face the challenges posed by modern high-resolution models – mainly related to model indeterminacy and steep computational requirements. In this article we introduce a hybrid algorithm called OPTIMISTS (Optimized PareTo Inverse Modeling through Integrated STochastic Search) which is targeted at non-linear high-resolution problems and that brings together ideas from particle filters, 4-dimensional variational methods, evolutionary Pareto optimization, and kernel density estimation in a unique way. Streamflow forecasting experiments were conducted to test which specific parameterizations of OPTIMISTS led to higher predictive accuracy. The experiments analysed two watersheds, one with a low resolution using the VIC (Variable Infiltration Capacity) model and one with a high-resolution using the DHSVM (Distributed Hydrology Soil Vegetation Model). By selecting kernel-based non-parametric sampling, non-sequential evaluation of candidate particles, and through the multi-objective minimization of departures from the streamflow observations and from the background states, OPTIMISTS was shown to outperform a particle filter and a 4D variational method. Moreover, the experiments demonstrated that OPTIMISTS scales well in high-resolution cases without imposing a significant computational overhead and that it was successful in mitigating the harmful effects of overfitting. With these combined advantages, the algorithm shows the potential to increase the accuracy and efficiency of operational prediction systems for the improved management of natural resources.

Download Full-text

Assimilation of soil moisture data for improving streamflow prediction: Is there a role for the hydrological model structure?

10.5194/egusphere-egu2020-19109 ◽

2020 ◽

Author(s):

Aruna Kumar Nayak ◽

Basudev Biswal ◽

Kulamulla Parambath Sudheer

Keyword(s):

Soil Moisture ◽

Data Assimilation ◽

Ensemble Kalman Filter ◽

Hydrological Model ◽

Assimilation Efficiency ◽

Hydrological Models ◽

Streamflow Forecasting ◽

Streamflow Prediction ◽

The Past ◽

Soil Moisture Data

Soil moisture data assimilation has found increased applicability in hydrology due to easily available remotely sensed soil moisture data. Numerous studies in the past have explored the possibility of assimilating soil moisture information for improving streamflow forecasting. The general understanding is that if better soil moisture data can provide better streamflow forecast. However, to our knowledge no study has so far focused on understanding if the hydrological model itself has a role in assimilation of soil moisture data. In this regard, here we use three different conceptual hydrological models for soil moisture assimilation: (1) Dynamic Budyko (DB), (2) GR4J, and (3) PDM model. Assimilation of GLDAS observed soil moisture is carried out for four MOPEX basins using Ensemble Kalman Filter. DB model&#8217;s performance improved after soil moisture data assimilation for all the study basins. However, deterioration in performance was observed for GR4J and PDM for all the basins after the assimilation exercise. The performance of the assimilated models is evaluated in terms of Assimilation Efficiency (AE), which was found to be varying from 17.11 to 22.56%, from -20.98 to -41.29%, and from -8.4 to -38.23%, respectively, for DB, GR4J, and PDM. Overall, our results highlight the importance of the hydrological models structure in a soil moisture data assimilation exercise.

Download Full-text

Hybridizing Bayesian and variational data assimilation for high-resolution hydrologic forecasting

Hydrology and Earth System Sciences ◽

10.5194/hess-22-5759-2018 ◽

2018 ◽

Vol 22 (11) ◽

pp. 5759-5779 ◽

Cited By ~ 5

Author(s):

Felipe Hernández ◽

Xu Liang

Keyword(s):

High Resolution ◽

Data Assimilation ◽

Predictive Accuracy ◽

Initial Conditions ◽

Time Estimation ◽

Variational Data Assimilation ◽

Streamflow Forecasting ◽

Infiltration Capacity ◽

Probabilistic Forecasts ◽

Non Linear

Abstract. The success of real-time estimation and forecasting applications based on geophysical models has been possible thanks to the two main existing frameworks for the determination of the models' initial conditions: Bayesian data assimilation and variational data assimilation. However, while there have been efforts to unify these two paradigms, existing attempts struggle to fully leverage the advantages of both in order to face the challenges posed by modern high-resolution models – mainly related to model indeterminacy and steep computational requirements. In this article we introduce a hybrid algorithm called OPTIMISTS (Optimized PareTo Inverse Modeling through Integrated STochastic Search) which is targeted at non-linear high-resolution problems and that brings together ideas from particle filters (PFs), four-dimensional variational methods (4D-Var), evolutionary Pareto optimization, and kernel density estimation in a unique way. Streamflow forecasting experiments were conducted to test which specific configurations of OPTIMISTS led to higher predictive accuracy. The experiments were conducted on two watersheds: the Blue River (low resolution) using the VIC (Variable Infiltration Capacity) model and the Indiantown Run (high resolution) using the DHSVM (Distributed Hydrology Soil Vegetation Model). By selecting kernel-based non-parametric sampling, non-sequential evaluation of candidate particles, and through the multi-objective minimization of departures from the streamflow observations and from the background states, OPTIMISTS was shown to efficiently produce probabilistic forecasts with comparable accuracy to that obtained from using a particle filter. Moreover, the experiments demonstrated that OPTIMISTS scales well in high-resolution cases without imposing a significant computational overhead. With the combined advantages of allowing for fast, non-Gaussian, non-linear, high-resolution prediction, the algorithm shows the potential to increase the efficiency of operational prediction systems.

Download Full-text

Snow Data Assimilation via an Ensemble Kalman Filter

Journal of Hydrometeorology ◽

10.1175/jhm505.1 ◽

2006 ◽

Vol 7 (3) ◽

pp. 478-493 ◽

Cited By ~ 157

Author(s):

Andrew G. Slater ◽

Martyn P. Clark

Keyword(s):

Kalman Filter ◽

Data Assimilation ◽

Snow Water Equivalent ◽

Initial Conditions ◽

Temporal Correlation ◽

Real Data ◽

Hydrologic Model ◽

Streamflow Forecasting ◽

Meteorological Forcing ◽

Ultimate Objective

Abstract A snow data assimilation study was undertaken in which real data were used to update a conceptual model, SNOW-17. The aim of this study is to improve the model’s estimate of snow water equivalent (SWE) by merging the uncertainties associated with meteorological forcing data and SWE observations within the model. This is done with a view to aiding the estimation of snowpack initial conditions for the ultimate objective of streamflow forecasting via a distributed hydrologic model. To provide a test of this methodology, the authors performed experiments at 53 stations in Colorado. In each case the situation of an unobserved location is mimicked, using the data at any given station only for validation; essentially, these are withholding experiments. Both ensembles of model forcing data and assimilated data were derived via interpolation and stochastic modeling of data from surrounding sources. Through a process of cross validation the error for the ensemble of model forcing data and assimilated observations is explicitly estimated. An ensemble square root Kalman filter is applied to perform assimilation on a 5-day cycle. Improvements in the resulting SWE are most evident during the early accumulation season and late melt period. However, the large temporal correlation inherent in a snowpack results in a less than optimal assimilation and the increased skill is marginal. Once this temporal persistence is removed from both model and assimilated observations during the update cycle, a result is produced that is, within the limits of available information, consistently superior to either the model or interpolated observations.

Download Full-text

Effect of joint assimilation of GRACE and discharge observations on simulated water storages and fluxes

10.5194/egusphere-egu2020-13235 ◽

2020 ◽

Author(s):

Kerstin Schulze ◽

Jürgen Kusche ◽

Olga Engels ◽

Petra Döll ◽

Somayeh Shadkam ◽

...

Keyword(s):

Data Assimilation ◽

Mississippi River ◽

Water Resource Management ◽

Water Cycle ◽

Time Frame ◽

Model Parameters ◽

Hydrological Models ◽

Realistic Representation ◽

Single Data ◽

Gravity Recovery

Several applications, from water resource management to the prediction of extreme events, require a realistic representation of the global water cycle. Global hydrological models simulate continental water fluxes and individual storages. However, they poorly reproduce observations of discharge and total water storage anomalies (TWSA). To improve the realism of the simulations, TWSA derived from the Gravity Recovery and Climate Experiment (GRACE) mission are usually assimilated into hydrological models. However, while assimilating GRACE-TWSA yields more realistic TWSA simulations, it is not clear how it affects the simulation of individual storages and fluxes. Therefore, assimilating discharge, in-situ or derived from satellite-altimetry, has been suggested to improve simulated discharge which is especially important for ungauged parts of basins.In this study, we jointly assimilate GRACE-TWSA and discharge observations and, for the first time, simultaneously calibrate the model parameters in order to improve the simulation skills of the model beyond the observational time frame. For this, we couple the WaterGAP 2.2d model with the Parallel Data Assimilation Framework and apply an Ensemble Kalman Filter for the Mississippi River Basin from 2003 to 2016. Furthermore, we compare our results to single-data assimilation and validate them against discharge observations that were not used for calibration/assimilation. Additionally, we analyze the effect of the calibrated parameters on the model&#8217;s realism.

Download Full-text