The Generalized Additive Models for Non-stationary Flood Frequency Analysis

Flood control and management is a fundamental issue for hydrology researchers and managers. Regarding the design and construct of different hydrologic structures such as reservoirs and dams as effective techniques for flood control, the appropriate estimation of the dimension and resilience of these structures needs correct estimation of the magnitude and return periods of flood for which the design flood in different return periods or probability level is applied. Design flood estimation is done through frequency analysis with the key stationary assumption. Nowadays the factors such as land use change, inappropriate management and climate change has influenced stationary conditions of flood peaks. Therefore, in the existence of non-stationary, the estimation based on stationary assumption is not confident and may have large errors. In this regard, this study for non-stationary flood frequency analysis the GAMLESS model for location, scale and shape parameter estimation as well as visual inspection of non-stationary are introduced and developed for quantile estimation in non-stationary conditions. Six hydrometery stations from different provinces in the North of Iran were selected. Frequency analysis with and without stationary assumptions was done for each station. Results indicated that location and scale parameters have trend with the linear and quadratic manner. In addition, results indicated that design flood estimated by non-stationary procedure has increased around 3 times higher than stationary estimation in Nodekhormalo station. Results also demonstrated that in a station with increasing non-stationary trend, return period of big floods is decreasing and for the same return periods, flood quantiles has increased.