ANNOUNCEMENTS
Intensity Duration Frequency (IDF) curves give an idea about how extreme rainfall intensities vary with duration for different return periods. In an uncertain scenario, where temporal variability of rainfall at local scale is not very well understood, the study aims to define a location specific robust model applicable for the megacity of Delhi which will improve the understanding of extreme rainfall pattern with duration. Identification of the best fit statistical distribution for prediction of short duration extreme events and parameter estimation of the chosen model are carried out for different durations. The choice of selection of the extreme value distribution (EVD) is determined using the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC).
Parameters of the EVD emerge as the most influential factor in predicting extremes and this study provides an understanding as to how parameters may be used to characterise short duration extreme rainfall storms. It is found that models which have shape factor as a parameter are more robust in predicting extreme events than those without the shape factor. The 2-parameter gamma distribution predicts extreme intensity through its shape and scale parameters. The generic gamma model is found to be robust and applicable at local scale for short duration storms with return period ≤10 years. It is recommended that the short duration rainfall characteristics determined through the shape and scale factor values may be adopted for smaller cities as well in the data scarce semiarid region.
The 3-parameter GEV model through its shape parameter is a better fit for more intense storms which generally have a larger return period. However, the conventional gamma or GEV model do not provide estimates of uncertainty in the IDF relationships. Other than the length of record and rainfall duration, accuracy in prediction is affected by parameter uncertainty. Studies suggest that future IDF curves will have different levels of uncertainty due to different data sources and different parameter sets. Uncertainty levels can only be reduced at best by using various techniques making the range of magnitude predicted more robust. Since the classical probability models used to build IDF curves do not address uncertainty, Bayesian approach may be used to address the uncertainty in parameters to make the outcomes more robust. It is found that it reduces the uncertainty in predictions of extreme magnitudes at higher return period, for short duration storms. The Bayesian posterior distribution has a calibration effect on the GEV predictions and reduces the upper range of uncertainty in the GEV probability model prediction from a range of 16-31% to 10-28.4% for return period varying from 10-year to 50-year for 1h storms, it has a similar calibrating effect on 3h storms too. In geographically similar areas these extreme intensity estimates may be used to prepare the urban watershed for the rising flash flood risks.
The IPCC prediction of more frequent extreme events of short durations necessitates that the updated IDF curves be applied for designing sustainable urban drainage systems to cater to climate change. Rising urbanization leads to more imperviousness and the urban storm runoff gathers more speed on the impervious surface due to reduced infiltration and evapotranspiration. The blue green spaces (BGS) interventions are advocated in some cities across the world and appear to be a viable alternative for urban runoff capture to reduce flooding episodes. Our study proposes that the probabilistic urban runoff estimation may use the GEV rainfall estimates for 1h duration. The design return period of extreme event considered is 10 and 25 years, and runoff models are simulated for two dissimilar catchments. The predicted urban runoff estimates are different depending on the availability of blue green urban spaces highlighting the need for spatial planning in urban areas for storm mitigation.
