Daily streamflow forecasting using a wavelet transform and artificial neural network hybrid models

In this study, a hybrid factorial stepwise-cluster analysis (HFSA) method is developed for modelling hydrological processes. The HFSA method employs a cluster tree to represent the complex nonlinear relationship between inputs (predictors) and outputs (predictands) in hydrological processes. A real case of streamflow simulation for the Kaidu River basin is applied to demonstrate the efficiency of the HFSA method. After training a total of 24 108 calibration samples, the cluster tree for daily streamflow is generated based on a stepwise-cluster analysis (SCA) approach and is then used to reproduce the daily streamflows for calibration (1995-2005) and validation (2008-2010) periods. The Nash-Sutcliffe coefficients for calibration and validation are 0.68 and 0.65, respectively, and the deviations of volume are 1.68% and 4.11%, respectively. Results show that: (i) the HFSA method can formulate a SCA-based hydrological modelling system for streamflow simulation with a satisfactory fitting; (ii) the variability and peak value of streamflow in the Kaidu River basin can be effectively captured by the SCA-based hydrological modelling system; (iii) results from 2 6 factorial experiments indicate that not only are minimum temperature and precipitation key drivers of system performance, but also the interaction between precipitation and minimum temperature significantly impacts on the streamflow. The findings are useful in indicating that the streamflow of the study basin is a mixture of snowmelt and rainfall water.

Section: Introductionmentioning

confidence: 99%

A hybrid factorial stepwise-cluster analysis method for streamflow simulation – a case study in northwestern China

Zhuang

Huang

et al. 2016

“…high-frequency components of the signal, obtained from the wavelet transform as input for the ANN. However, Santos and Silva (2014) tested the use of approximations (i.e. low-frequency components), or a sum of them, as inputs for a daily streamflow forecasting model with increasing time horizons (1, 3, 5 and 7 days ahead) for the São Francisco River basin in Brazil.…”

Section: Introductionmentioning

confidence: 99%

Monthly streamflow forecasting using neuro-wavelet techniques and input analysis

Honorato

Silva

Santos

2018

Self Cite

Combinations of low-frequency components (also known as approximations) resulting from the wavelet decomposition are tested as inputs to an artificial neural network (ANN) in a hybrid approach, and compared to classical ANN models for flow forecasting for 1, 3, 6 and 12 months ahead. In addition, the inputs are rewritten in terms of the flow, revealing what type of information was being provided to the network, in order to understand the effect of the approximations on the forecasting performance. The results show that the hybrid approach improved the accuracy of all tested models, especially for 1, 3 and 6 months ahead. The input analyses show that high-frequency components are more important for shorter forecast horizons, while for longer horizons, they may worsen the model accuracy.

“…ANNs, and especially their type called Multi-Layer Perceptron (MLP), are now well established, widely applied forecasting tools in hydrology (Maier and Dandy 2000, Maier et al 2010, Adamowski and Chan 2011, Abrahart et al 2012a, Taormina et al 2012. The inter-comparison of different data-driven techniques applied to rainfall-runoff modelling was presented in a number of studies (see, for example, Solomatine and Ostfeld 2008, Wang et al 2009, Wu et al 2009, Elshorbagy et al 2010, Jothiprakash and Magar 2012, Isik et al 2013, Santos and da Silva 2014, Yin et al 2016. These papers frequently show good performance of ANN approaches.…”

Section: Introductionmentioning

confidence: 99%

On the importance of training methods and ensemble aggregation for runoff prediction by means of artificial neural networks

Piotrowski

Napiórkowski

Osuch

et al. 2016

Artificial neural networks (ANNs) become widely used for runoff forecasting in numerous studies. Usually classical gradient-based methods are applied in ANN training and a single ANN model is used. To improve the modelling performance, in some papers ensemble aggregation approaches are used whilst in others, novel training methods are proposed. In this study, the usefulness of both concepts is analysed. First, the applicability of a large number of population-based metaheuristics to ANN training for runoff forecasting is tested on data collected from four catchments, namely upper Annapolis (Nova Scotia, Canada), Biala Tarnowska (Poland), upper Allier (France) and Axe Creek (Victoria, Australia). Then, the importance of the search for novel training methods is compared with the importance of the use of a very simple ANN ensemble aggregation approach. It is shown that although some metaheuristics may slightly outperform the classical gradient-based Levenberg-Marquardt algorithm for a specific catchment, none performs better for the majority of the tested ones. One may also point out a few metaheuristics that do not suit ANN training at all. On the other hand, application of even the simplest ensemble aggregation approach clearly improves the results when the ensemble members are trained by any suitable algorithms.