Dynamic calibration in hydrologic and hydraulic modelling: exploring the potential of data assimilation for estimation of models' parameters
Конференцијски прилог (Објављена верзија)
Метаподаци
Приказ свих података о документуАпстракт
Hydrological and hydraulic models used for forecasting and providing reliable inputs are essential for effective water management. Throughout their application, models might produce results of unsatisfying accuracy due to many uncertainty sources. Considerable uncertainty stems from model parameters values, which are usually obtained through calibration, i.e., iterative adjustment of the parameter values to achieve the best possible fit between simulated and observed variables. Model calibration yields time-invariant parameter estimates, which can lead to poor-quality simulation outputs that cannot serve as decision support. Specifically, parameter values can be expected to vary due to secondary- or seasonal processes that are not explicitly accounted for in the model, or due to anthropogenic activity (e.g., land-use change). Therefore, models used for operational forecasting should be run with up-to-date parameter values. This necessitates frequent model recalibration, which can be qu...ite impractical due to high time- and computational requirements of the calibration procedure. Therefore, developing fast(er) calibration algorithms could be a viable alternative. This paper explores the potential of control theory-based, tailor-made, data assimilation algorithm intended for continuous update of the parameters of hydrological and hydraulic models. The algorithm enables the parameter values to be regularly updated at each computational time step based on the dynamically assessed goodness-of-fit (GOF) performance indicator. This approach enables one-pass calibration procedure. Using this algorithm instead of traditional, iterative calibration procedure where models’ GOF is assessed at the end of the simulation, can improve efficiency and effectiveness of models’ calibration. The proposed one-pass calibration approach will be tested on two synthetic test cases, one example of a hydrological model and one example of a 1D hydraulic model.
Кључне речи:
water resources management / forecasting / time-variable model parameters / one-pass calibration / control theory-based data assimilationИзвор:
SimHydro 2023: New modelling paradigms for water issues?, Chatou, 8-10. November 2023, 2023Финансирање / пројекти:
- Министарство науке, технолошког развоја и иновација Републике Србије, институционално финансирање - 200092 (Универзитет у Београду, Грађевински факултет) (RS-MESTD-inst-2020-200092)
Колекције
Институција/група
GraFarTY - CONF AU - Milašinović, Miloš AU - Todorović, Andrijana AU - Zindović, Budo PY - 2023 UR - https://grafar.grf.bg.ac.rs/handle/123456789/3318 AB - Hydrological and hydraulic models used for forecasting and providing reliable inputs are essential for effective water management. Throughout their application, models might produce results of unsatisfying accuracy due to many uncertainty sources. Considerable uncertainty stems from model parameters values, which are usually obtained through calibration, i.e., iterative adjustment of the parameter values to achieve the best possible fit between simulated and observed variables. Model calibration yields time-invariant parameter estimates, which can lead to poor-quality simulation outputs that cannot serve as decision support. Specifically, parameter values can be expected to vary due to secondary- or seasonal processes that are not explicitly accounted for in the model, or due to anthropogenic activity (e.g., land-use change). Therefore, models used for operational forecasting should be run with up-to-date parameter values. This necessitates frequent model recalibration, which can be quite impractical due to high time- and computational requirements of the calibration procedure. Therefore, developing fast(er) calibration algorithms could be a viable alternative. This paper explores the potential of control theory-based, tailor-made, data assimilation algorithm intended for continuous update of the parameters of hydrological and hydraulic models. The algorithm enables the parameter values to be regularly updated at each computational time step based on the dynamically assessed goodness-of-fit (GOF) performance indicator. This approach enables one-pass calibration procedure. Using this algorithm instead of traditional, iterative calibration procedure where models’ GOF is assessed at the end of the simulation, can improve efficiency and effectiveness of models’ calibration. The proposed one-pass calibration approach will be tested on two synthetic test cases, one example of a hydrological model and one example of a 1D hydraulic model. C3 - SimHydro 2023: New modelling paradigms for water issues?, Chatou, 8-10. November 2023 T1 - Dynamic calibration in hydrologic and hydraulic modelling: exploring the potential of data assimilation for estimation of models' parameters UR - https://hdl.handle.net/21.15107/rcub_grafar_3318 ER -
@conference{ author = "Milašinović, Miloš and Todorović, Andrijana and Zindović, Budo", year = "2023", abstract = "Hydrological and hydraulic models used for forecasting and providing reliable inputs are essential for effective water management. Throughout their application, models might produce results of unsatisfying accuracy due to many uncertainty sources. Considerable uncertainty stems from model parameters values, which are usually obtained through calibration, i.e., iterative adjustment of the parameter values to achieve the best possible fit between simulated and observed variables. Model calibration yields time-invariant parameter estimates, which can lead to poor-quality simulation outputs that cannot serve as decision support. Specifically, parameter values can be expected to vary due to secondary- or seasonal processes that are not explicitly accounted for in the model, or due to anthropogenic activity (e.g., land-use change). Therefore, models used for operational forecasting should be run with up-to-date parameter values. This necessitates frequent model recalibration, which can be quite impractical due to high time- and computational requirements of the calibration procedure. Therefore, developing fast(er) calibration algorithms could be a viable alternative. This paper explores the potential of control theory-based, tailor-made, data assimilation algorithm intended for continuous update of the parameters of hydrological and hydraulic models. The algorithm enables the parameter values to be regularly updated at each computational time step based on the dynamically assessed goodness-of-fit (GOF) performance indicator. This approach enables one-pass calibration procedure. Using this algorithm instead of traditional, iterative calibration procedure where models’ GOF is assessed at the end of the simulation, can improve efficiency and effectiveness of models’ calibration. The proposed one-pass calibration approach will be tested on two synthetic test cases, one example of a hydrological model and one example of a 1D hydraulic model.", journal = "SimHydro 2023: New modelling paradigms for water issues?, Chatou, 8-10. November 2023", title = "Dynamic calibration in hydrologic and hydraulic modelling: exploring the potential of data assimilation for estimation of models' parameters", url = "https://hdl.handle.net/21.15107/rcub_grafar_3318" }
Milašinović, M., Todorović, A.,& Zindović, B.. (2023). Dynamic calibration in hydrologic and hydraulic modelling: exploring the potential of data assimilation for estimation of models' parameters. in SimHydro 2023: New modelling paradigms for water issues?, Chatou, 8-10. November 2023. https://hdl.handle.net/21.15107/rcub_grafar_3318
Milašinović M, Todorović A, Zindović B. Dynamic calibration in hydrologic and hydraulic modelling: exploring the potential of data assimilation for estimation of models' parameters. in SimHydro 2023: New modelling paradigms for water issues?, Chatou, 8-10. November 2023. 2023;. https://hdl.handle.net/21.15107/rcub_grafar_3318 .
Milašinović, Miloš, Todorović, Andrijana, Zindović, Budo, "Dynamic calibration in hydrologic and hydraulic modelling: exploring the potential of data assimilation for estimation of models' parameters" in SimHydro 2023: New modelling paradigms for water issues?, Chatou, 8-10. November 2023 (2023), https://hdl.handle.net/21.15107/rcub_grafar_3318 .