Real-time correction of water stage forecast using combination of forecasted errors by time series models and Kalman filter method

Abstract

This study modifies a real-time correction method for water stage forecasts (named the RTEC_TS&KF model) using the time series method developed by Wu et al. (Stoch Environ Res Risk Assess 26:519-531, 2012) (named the RTEC_TS model), by incorporating the Kalman filter (KF) model. The RTEC_TS&KF model adjusts the corrected water stage forecasts resulting from the RTEC_TS model by taking into account the uncertainties in the model structure/inputs as well as the measurement bias. In detail, the water stage forecasts are corrected by separately adding the forecasted errors by the times series model and KF method into the stage forecasts. As compared to the results from the RTEC_TS model using the forecasted and observed water stages for Typhoons Morakot (2009), Saola (2012) and Soulik (2013), the RTEC_TS&KF model not only effectively lessens the uncertainties in regard to the water stage forecasts, but also consistently presents high correction performance of water level forecasts for various rainstorm events. This reveals that the RTEC_TS&KF model is superior to the RTEC_TS model in the correction of water stage forecasts. In the future, the RTEC_TS&KF model will be applied in the real-time corrections of other hydrological variates, such as the outflow of a reservoir, in the case of observation being provided on time.

On the criteria of model performance evaluation for real-time flood forecasting

“…2, flows at the watershed outlet exhibit significant persistence and time series of streamflows can be represented by an autoregressive model. In addition, a few studies have also demonstrated that, with real-time error correction, AR(1) and AR(2) can significantly enhance the reliability of the forecasted water stages at the 1-, 2-, and 3-h lead time (Wu et al 2012;Shen et al 2015). Thus, we suggest using the AR(2) model as the benchmark series for flood forecasting model performance evaluation.…”

Integration of a Low-Cost Global Navigation Satellite System to a Single-Board Computer Using Kalman Filtering

“…Kalman filtering is a recursive approach involving ideal linear filtering to estimate the state variables of a system. For system modeling, the initial positioning values were used to predict and adjust parameters in each new measurement, obtaining the estimated error in each data update (Shen et al, 2015).…”

Research on the Application of Echo State Network in Data Prediction

“…According to the historical data of the research object fitting a curve over time which can analyze the changes more intuitive in the rules, and then extend the curve in the future, which formed a prediction of the future of the data [4]. In recent years, various kinds of time series data prediction method is extracted, according to the different programming models can be divided into: prediction method based on support vector machine, prediction method based on self organizing feature map, prediction method based on extended Calman filter, prediction method based on artificial neural network [5][6][7][8].…”