Intensity-Duration-Frequency (IDF) Curves Under Changing Climate – A Non-Stationary Modelling Approach

Anthropogenic activities have accelerated the global warming phenomena, causing a rapid change in the weather patterns, especially the extremes. Intensification of magnitude and frequency of extreme events have increased the stress on water infrastructures. Hence design methods have to be updated to build climate-resilient infrastructures. Intensity-Duration-Frequency (IDF) curves play a vital role in flood risk assessment and impact the region's socio-economic structure. In this study, a non-stationary modelling approach is proposed to develop IDF curves under changing climate using Global Climate Models (GCMs). Non-Stationary Generalized Extreme Value Distribution (NS-GEVD) location parameter is modelled as a linear function of GCM outputs.  Data used for analysis is the annual maximum daily precipitation generated at a Hyderabad city station, India using 27 GCMs of Coupled Model Intercomparison Project Phase-5 (CMIP-5).  The analysis is carried out for the baseline period of 1971 to 2005 and the future time-period of 2006 to 2100. Corrected Akaike Information Criterion test statistic is used to identify the best NS-GEVD model. The results indicate that NS-GEVD model could capture the non-stationary behaviour and predicted an average increase of 7% in extreme rainfall intensity for the future. Besides, it is observed that six GCM covariates outperform other GCMs. The outcomes of this study will benefit the city municipality, practitioners and decision-makers in identifying future risk for water infrastructures.