Non-Stationary Modeling of Annual Flood Peak Heights of Mahanadi River Basin with the q-Generalized Extreme Value Distribution

Abstract

In recent years, due to climate change, catastrophic events are increased largely in India. Hence researchers are forced to consider non-stationary flood frequency analysis as an improved method. In this paper, non-stationarity of annual daily maximum flood heights were studied at 12 sites of Mahanadi River Basin (MRB) by analyzing the flood frequency of a stationary model and 4 non-stationary models using time dependent q-GEV model by considering trend as a linear function of its location and scale parameters. The q-GEV distribution is utilized in this study because of its flexibility and accuracy than GEV distribution in modeling extreme flood heights. The results found that there is strong evidence of a linear trend existence for both the location and scale parameters at the Kesinga site; for the location parameter at Pathardi and Simga sites; for the scale parameter at Dharmajagarh, Kotni and Seorinarayan, and no linear trend exists for both location and scale parameters at Alipingal, Bomnidhi, Manendragarh, Mohana, Rajim and Sundargarh, there may be exists other form of trend at these sites. The findings also indicate that nonstationarity is present in the MRB due to climate change, which help to water practitioner for taking precautions against adverse effect of extreme floods.