Regionalization of Non-Stationary Intensity-Duration-Frequency (IDF) Curves for Monsoon Asia
Abstract
Traditionally, water infrastructure designs relied on the assumption of stationarity for modeling extremes precipitation (EP). However, due to climate change, it is expected EP to vary in both space and time. This study majorly focuses on (i) modeling of EP using non-stationary Generalized Extreme Value (NS-GEV); and (ii) identification of regional NS-GEV model parameters for a large-scale study area. A clustering-based approach is adopted to develop regional IDF relationships over the Monsoon Asia region (MAR) using high-resolution gridded Asian Precipitation Highly Resolved Observational Data Integration Towards Evaluation (APHRODITE). The Gaussian Mixture Model (GMM) results found an optimal number of six extreme precipitation zones. The region-wise non-stationary model parameters are obtained by pooling up the in-situ best model parameter from each region selected based on the corrected Akaike Information Criterion. The efficacy of the regional models is evaluated using the Root Mean Square Error (RMSE) measure, and it is observed that the regional NSGEV models performance is comparable to the gridded data in the same region. Finally, the regional IDF relationships are developed, and a relative percentage change in extreme precipitation characteristics is estimated for pre-defined return periods for stationary and non-stationary GEV models. Results indicate an increase in precipitation intensity and reduction in return periods across the coherent regions. Overall, the non-stationary analysis shows that precipitation extremes may occur more frequently with higher intensities. The studys outcomes help to understand the impact of climate change on flood risk assessment and urban water infrastructure design and rehabilitation.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFM.H35ZC.09N