Optimizing Neural Network for Monthly Rainfall-Runoff Modeling with Denoised-Jittered Data

Abstract

Successful modeling of hydro-environmental processes widely relies on quantity and quality of accessible data and noisy data might effect on the functioning of the modeling. On the other hand in training phase of any Artificial Intelligence (AI) based model, each training data set is usually a limited sample of possible patterns of the process and hence, might not show the behavior of whole population. Accordingly in the present article first, wavelet-based denoising method was used in order to smooth hydrological time series and then small normally distributed noises with the mean of zero and various standard deviations were generated and added to the smoothed time series to form different denoised-jittered training data sets, for Artificial Neural Network (ANN) modeling of monthly rainfall – runoff process of the Pole Saheb(Anyan) station in Zarrineh River watershed, which is a portion of orumiyeh lake drainage basin, that is located in Iran. To evaluate the modeling performance, the obtained results were compared with those of multi linear regression and Auto Regressive Integrated Moving Average models. Comparison of the obtained results via the trained ANN using denoised- jittered data showed that the proposed data pre-processing approach could improve performance of the ANN based rainfall-runoff modeling of the case study up to 38% in the verification phase.