Data assimilation (DA) methods continue to evolve in the design of streamflow forecasting procedures. Critical components for efficient DA include accurate description of states, improved model parameterizations, and estimation of the measurement error. Information about these components are usually assumed or rarely incorporated into streamflow forecasting procedures. Knowledge of these components could be gained through the generation of a Pareto-optimal set – a set of competitive members that are not dominated by other members when compared using evaluation objectives. This study integrates Pareto-optimality into the ensemble Kalman filter (EnKF) and the particle filter (PF). Comparisons are made between three methods: evolutionary data assimilation (EDA) and methods based on the integration of Pareto-optimality into the EnKF (ParetoEnKF) and into the PF (ParetoPF). The methods are applied to assimilate daily streamflow into the Sacramento Soil Moisture Accounting model in the Spencer Creek watershed in Canada. The updated members are applied to forecast streamflows for up to 10 days ahead, where forecasts for 1 day, 5 day and 10 day lead times are compared to observations. The results show that updated estimates are similar for all three methods. An evaluation of updated members for multi-step forecasting revealed that EDA had the highest forecast accuracy compared to ParetoEnKF and ParetoPF, which have similar accuracies.
Skip Nav Destination
Article navigation
Research Article|
June 05 2013
Integration of an evolutionary algorithm into the ensemble Kalman filter and the particle filter for hydrologic data assimilation
Gift Dumedah;
1Department of Civil Engineering, Monash University, Building 60, Melbourne, Victoria 3800, Australia
E-mail: [email protected]
Search for other works by this author on:
Paulin Coulibaly
Paulin Coulibaly
2School of Geography and Earth Sciences, and Department of Civil Engineering, McMaster University, 1280 Main Street West, Hamilton, Ontario, Canada L8S4L8
Search for other works by this author on:
Journal of Hydroinformatics (2014) 16 (1): 74–94.
Article history
Received:
May 05 2012
Accepted:
May 10 2013
Citation
Gift Dumedah, Paulin Coulibaly; Integration of an evolutionary algorithm into the ensemble Kalman filter and the particle filter for hydrologic data assimilation. Journal of Hydroinformatics 1 January 2014; 16 (1): 74–94. doi: https://doi.org/10.2166/hydro.2013.088
Download citation file: