Title: Influence of climate type on the predictive capabilities of stochastic models applied to monthly dam inflows

Authors: Leila Benchaiba; Larbi Houichi; Hocine Amarchi

Addresses: Department of Hydraulics, University of Annaba, Annaba, 23000, Algeria ' Department of Hydraulics, University of Batna 2, Batna, 05000, Algeria ' Department of Hydraulics, University of Annaba, Annaba, 23000, Algeria

Abstract: This contribution assesses the predictive capacity of monthly inflow of two stochastic models called autoregressive integrated moving average (ARIMA) and TBATS and highlights the influence of climate type on their performances. These are the inflows to three dams in three distinct climates: semi-arid, subhumid and humid. The actual inflows are deduced from the water balance equation for 132-month period. The first ten corresponding years of each series are used for training of the two models and the last one is then used for test. Model performances are evaluated using three commonly used metrics: the square root of the mean square error (RMSE), the mean of the absolute errors (MAE), and the mean absolute error in percentage (MAPE). The results show that the TBATS model performs better than the ARIMA model and its predictive capabilities decrease depending on whether the climate is semi-arid, sub-humid and humid (MAPE = 50.47%, 34.79% and 29.99%, respectively).

Keywords: ARIMA; autoregressive integrated moving average; TBATS; trigonometric; box-cox transform; ARMA errors; trend and seasonal components; climate type; forecast; monthly dam inflows; stochastic models; time series; predictive capabilities.

DOI: 10.1504/IJW.2021.126820

International Journal of Water, 2021 Vol.14 No.4, pp.256 - 271

Received: 29 Jan 2022
Accepted: 13 Feb 2022
Published online: 08 Nov 2022 *