Using hybrid weighting‐clustering approach for regional frequency analysis of maximum 24‐hr rainfall based on climatic, geographical, and statistical attributes

Abstract

Estimation of extreme rainfall amounts has great importance, especially in some fields such as the design of water structures, water resources engineering, extreme flood management, and soil erosion conservation. One of the problems, which hydrologists faced, is an acceptable estimation of extreme events in areas with insufficient data. In this case, separation of the study area into homogenous regions and performing regional frequency analysis (RFA) result in greater precision and fewer errors in frequency analysis models, enabling estimation of quantiles for each return period in the region of interest. In this study, the maximum 24‐hr rainfall data related to Lake Urmia Basin (LUB) for 63 selected stations during the period of 1979–2008 are used. Moreover, determining an appropriate weight for each group of attributes is attempted according to the degree of importance and contribution share of each climatic, geographical, and statistical attribute. Then, for regionalization using a clustering approach, the attributes are defined in seven different groups. Subsequently, the performance of different groups associated with weighted attributes of maximum 24‐hr rainfall is evaluated for RFA in the study area. The results showed that the combination of climatic, geographical, and statistical attributes present better results and more reliable estimates of extreme values. The results indicated better performance of weighted models to the attributes compared to non‐weighted frequency analysis models in the estimation of maximum 24‐hr rainfall. The results also showed that by evaluating three to four groups and applying average weights of 0.3–0.85 for the attributes through the use of hybrid weighting‐clustering approach, more accurate estimates of extreme values for different return periods can be obtained.