Search results

The objective of this study is to propose a functional framework for hydrological applications by treating flood hydrographs as functional data. Discrete flow data are transformed into a smoothing hydrograph curve, which can be analysed at any time interval. The concept of functional data considered the entire curve concerning time as a single observation. This chapter briefly discussed the idea of descriptive statistics, principal components and outliers in a functional framework. These methods were illustrated in the flood study at Sungai Kelantan River Basin, Malaysia. The results showed that five main components accounted for almost 73.8% of the overall flow variance. Based on the results of the factor scores, the hydrograph curves for the years 1988, 1993 and 2014 may be said to have a unique cluster of their own, while the rest of the years which consider having the same pattern. Due to various shapes and magnitudes, the hydrograph curves of 1988 and 2014 are considered outliers. In conclusion, the functional framework has shown that it is capable of representing a wide range of hydrographs and is capable of extracting additional information found in the hydrograph curve that cannot possibly be captured using classical statistical methods.