True interval non-stationary Intensity-Duration-Frequency curves under changing climate and effect of temporal discretisation on rainfall extremes

• Published in: Journal of hydrology (Amsterdam) 2023-03, Vol.618, p.129136, Article 129136
• Main Authors: Ghate, Aarti S., Timbadiya, P.V.
• Format: Article
• Language: English
• Subjects: CHCF; Extreme rainfall; GEV; IDF curves; Non-stationary; Stationary
• ISSN: 0022-1694; 1879-2707
• DOI: 10.1016/j.jhydrol.2023.129136

•The effect of temporal discretisation on the extreme rainfall is examined.•Trends in FW and SW rainfall extremes are examined for all the selected durations.•Influence of global and local atmospheric processes on rainfall extremes is identified.•The 24-h rainfall extremes are influenced by global atmospheric processes and CHCF.•The 1-h rainfall extremes are associated with local atmospheric processes. The extreme rainfall intensity-duration-frequency (IDF) curves are used to design stormwater infrastructure in the urban areas and need to be developed keeping in view of the warming climate. The current study develops the non-stationary IDF relationships for 11 class-I urban cities of India under changing climate. The total 11,528 generalized extreme value distribution (GEV) models are developed to quantify non-stationarity in extreme rainfall by introducing a clock hour correction factor (CHCF) as a covariate. The observed hourly rainfall data of the selected cities located in five different regions of India, namely the western region (Mumbai, Ahmedabad, Jaipur), northern region (New Delhi, Dehradun), north-eastern region (Imphal), eastern region (Bhubaneswar, Kolkata), and southern region (Madurai, Chennai, Bengaluru) are collected and converted in the fixed and sliding interval annual maximum rainfall series of eight different durations. The said data set is used in the computation of CHCF, evaluation of discretisation effect on extreme rainfall events, trend analysis, development of stationary and non-stationary IDF curves. The CHCF values of seven (two) cities are above (below) 1.15 respectively; the other two cities show a value ∼ 1.15. The effect of temporal discretisation on the computation of extreme rainfall for 24-h (FW and SW) maximum rainfall depth was found to be highest in Bhubaneswar (50%) and Imphal (44%). The trend analysis revealed no statistically significant increasing (decreasing) trend in extreme rainfall for all durations for all cities under consideration except in the case of Bengaluru for 2-,3-, and 6-h durations (Chennai for a 48-h duration). The current study concluded that, under a non-stationarity climate, the shorter (1-h) and longer (24-h) duration extreme rainfall is influenced by local and global processes, respectively. Further, the short-duration extreme events are intensifying with the decrease in the return period. The 1-h duration non-stationary rainfall intensity for a 2-year return period and stationary rainfall int