A back-fitting algorithm to improve real-time flood forecasting

Abstract Real-time flood forecasting is important for decision-making with regards to flood control and disaster reduction. The conventional approach involves a postprocessor calibration strategy that first calibrates the hydrological model and then estimates errors. This procedure can simulate streamflow consistent with observations, but obtained parameters are not optimal. Joint calibration strategies address this issue by refining hydrological model parameters jointly with the autoregressive (AR) model. In this study, five alternative schemes are used to forecast floods. Scheme I uses only the hydrological model, while scheme II includes an AR model for error correction. In scheme III, differencing is used to remove non-stationarity in the error series. A joint inference strategy employed in scheme IV calibrates the hydrological and AR models simultaneously. The back-fitting algorithm, a basic approach for training an additive model, is adopted in scheme V to alternately recalibrate hydrological and AR model parameters. The performance of the five schemes is compared with a case study of 15 recorded flood events from China’s Baiyunshan reservoir basin. Our results show that (1) schemes IV and V outperform scheme III during the calibration and validation periods and (2) scheme V is inferior to scheme IV in the calibration period, but provides better results in the validation period. Joint calibration strategies can therefore improve the accuracy of flood forecasting. Additionally, the back-fitting recalibration strategy produces weaker overcorrection and a more robust performance compared with the joint inference strategy.

[1]  Thomas C. Pagano,et al.  A dual-pass error-correction technique for forecasting streamflow , 2011 .

[2]  Gwo-Fong Lin,et al.  A non-linear rainfall-runoff model using radial basis function network , 2004 .

[3]  Xinjian Li,et al.  Improvement of hydrological model calibration by selecting multiple parameter ranges , 2016 .

[4]  G. Lannoy,et al.  The importance of parameter resampling for soil moisture data assimilation into hydrologic models using the particle filter , 2011 .

[5]  Wouter Buytaert,et al.  The role of rating curve uncertainty in real‐time flood forecasting , 2016 .

[6]  Pan Liu,et al.  Integrated Hydrologic and Reservoir Routing Model for Real-Time Water Level Forecasts , 2015 .

[7]  Pan Liu,et al.  Optimal design of seasonal flood limited water levels and its application for the Three Gorges Reservoir , 2015 .

[8]  Trevor Hastie,et al.  Generalized Additive Model , 2005 .

[9]  W. C. Lennox,et al.  Groups and neural networks based streamflow data infilling procedures , 2001 .

[10]  Pan Liu,et al.  Non-identical models for seasonal flood frequency analysis , 2007 .

[11]  Peter Reichert,et al.  Bayesian uncertainty analysis in distributed hydrologic modeling: A case study in the Thur River basin (Switzerland) , 2007 .

[12]  Olivier Thual,et al.  Benefits and limitations of data assimilation for discharge forecasting using an event-based rainfall–runoff model , 2013 .

[13]  D. Stephenson,et al.  Criteria for comparison of single event models , 1986 .

[14]  V. Singh,et al.  Streamflow forecast uncertainty evolution and its effect on real-time reservoir operation , 2016 .

[15]  John A. Nelder,et al.  A Simplex Method for Function Minimization , 1965, Comput. J..

[16]  Che-Hao Chang,et al.  Real-time correction of water stage forecast during rainstorm events using combination of forecast errors , 2012, Stochastic Environmental Research and Risk Assessment.

[17]  Gwo-Fong Lin,et al.  A real-time forecasting model for the spatial distribution of typhoon rainfall , 2015 .

[18]  Guang Yang,et al.  Comparative Study of Three Updating Procedures for Real-Time Flood Forecasting , 2016, Water Resources Management.

[19]  Robert Kohn,et al.  Convergence of the backfitting algorithm for additive models , 1994 .

[20]  Mélanie C. Rochoux,et al.  Ensemble-based data assimilation for operational flood forecasting – On the merits of state estimation for 1D hydrodynamic forecasting through the example of the “Adour Maritime” river , 2017 .

[21]  J. Adamowski,et al.  Comparison of multiple linear and nonlinear regression, autoregressive integrated moving average, artificial neural network, and wavelet artificial neural network methods for urban water demand forecasting in Montreal, Canada , 2012 .

[22]  Ioan Nistor,et al.  Data Assimilation for Streamflow Forecasting: State–Parameter Assimilation versus Output Assimilation , 2017 .

[23]  Neil J. Gordon,et al.  A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking , 2002, IEEE Trans. Signal Process..

[24]  Q. J. Wang.  The Genetic Algorithm and Its Application to Calibrating Conceptual Rainfall-Runoff Models , 1991 .

[25]  A. Shamseldin,et al.  A real-time combination method for the outputs of different rainfall-runoff models , 1999 .

[26]  H. H. Rosenbrock,et al.  An Automatic Method for Finding the Greatest or Least Value of a Function , 1960, Comput. J..

[27]  Wade T. Crow,et al.  An integrated error parameter estimation and lag-aware data assimilation scheme for real-time flood forecasting , 2014 .

[28]  Lihua Xiong,et al.  Comparison of four updating models for real-time river flow forecasting , 2002 .

[29]  Robert J. Abrahart,et al.  HydroTest: A web-based toolbox of evaluation metrics for the standardised assessment of hydrological forecasts , 2007, Environ. Model. Softw..

[30]  Chong-Yu Xu,et al.  Comparison and evaluation of multiple GCMs, statistical downscaling and hydrological models in the study of climate change impacts on runoff , 2012 .

[31]  Héctor Pomares,et al.  Hybridization of intelligent techniques and ARIMA models for time series prediction , 2008, Fuzzy Sets Syst..

[32]  Bijaya K. Panigrahi,et al.  A Support Vector Machine-Firefly Algorithm based forecasting model to determine malaria transmission , 2014, Neurocomputing.

[33]  Angela Lundberg,et al.  Combination of a Conceptual Model and an Autoregressive Error Model for Improving Short Time Forecasting , 1982 .

[34]  James C. Bennett,et al.  A strategy to overcome adverse effects of autoregressive updating of streamflow forecasts , 2015 .

[35]  Ke Zhang,et al.  Improving the flood prediction capability of the Xinanjiang model in ungauged nested catchments by coupling it with the geomorphologic instantaneous unit hydrograph , 2014 .

[36]  P. Kitanidis,et al.  Real‐time forecasting with a conceptual hydrologic model: 2. Applications and results , 1980 .

[37]  Shenglian Guo,et al.  Real-time error correction method combined with combination flood forecasting technique for improving the accuracy of flood forecasting , 2015 .

[38]  Lloyd H.C. Chua,et al.  The data-driven approach as an operational real-time flood forecasting model , 2012 .

[39]  I. Jolliffe,et al.  Forecast verification : a practitioner's guide in atmospheric science , 2011 .

[40]  George Kuczera,et al.  Pitfalls and improvements in the joint inference of heteroscedasticity and autocorrelation in hydrological model calibration , 2013 .

[41]  Cheng Yao,et al.  Event-based hydrological modeling for detecting dominant hydrological process and suitable model strategy for semi-arid catchments , 2016 .

[42]  Li Qian,et al.  An error updating system for real-time flood forecasting based on robust procedure , 2015 .

[43]  Xavier Litrico,et al.  A web based tool for operational real-time flood forecasting using data assimilation to update hydraulic states , 2016, Environ. Model. Softw..

[44]  Chandranath Chatterjee,et al.  A wavelet-based non-linear autoregressive with exogenous inputs (WNARX) dynamic neural network model for real-time flood forecasting using satellite-based rainfall products , 2016 .

[45]  James C. Bennett,et al.  Reliable long‐range ensemble streamflow forecasts: Combining calibrated climate forecasts with a conceptual runoff model and a staged error model , 2016 .

[46]  Vijay P. Singh,et al.  Xinanjiang model combined with Curve Number to simulate the effect of land use change on environmental flow , 2014 .

[47]  S. Sorooshian,et al.  A Shuffled Complex Evolution Metropolis algorithm for optimization and uncertainty assessment of hydrologic model parameters , 2002 .

[48]  Fi-John Chang,et al.  Systematic impact assessment on inter-basin water transfer projects of the Hanjiang River Basin in China , 2017 .

[49]  Pan Liu,et al.  Dynamic control of flood limited water level for reservoir operation by considering inflow uncertainty , 2010 .

[50]  Zhao Ren-jun,et al.  The Xinanjiang model applied in China , 1992 .

[51]  Luca Brocca,et al.  The multi temporal/multi-model approach to predictive uncertainty assessment in real-time flood forecasting , 2017 .

[52]  Si Wei,et al.  Flow Updating in Real-Time Flood Forecasting Based on Runoff Correction by a Dynamic System Response Curve , 2014 .

[53]  Li Zhijia,et al.  Study of the Xinanjiang Model Parameter Calibration , 2013 .

[54]  Jens Christian Refsgaard,et al.  Validation and Intercomparison of Different Updating Procedures for Real-Time Forecasting , 1997 .

[55]  J. Nash,et al.  River flow forecasting through conceptual models part I — A discussion of principles☆ , 1970 .

[56]  Ashu Jain,et al.  Integrated approach to model decomposed flow hydrograph using artificial neural network and conceptual techniques , 2006 .

[57]  A. Shamseldin,et al.  Assessing the performance of eight real-time updating models and procedures for the brosna river , 2005 .

[58]  Seong Jin Noh,et al.  Advancing data assimilation in operational hydrologic forecasting: progresses, challenges, and emerging opportunities , 2012 .

[59]  W. Briggs.  Statistical Methods in the Atmospheric Sciences , 2007 .

[60]  Henrik Madsen,et al.  Adaptive state updating in real-time river flow forecasting—a combined filtering and error forecasting procedure , 2005 .

[61]  George Kuczera,et al.  Comparison of joint versus postprocessor approaches for hydrological uncertainty estimation accounting for error autocorrelation and heteroscedasticity , 2014 .