Characteristics and Frequency Analysis of Annual Daily Maximum Rainfall at Surat, Gujarat, India

Abstract

The technical and engineering suitability of infrastructure projects such as nuclear/hydro/thermal power plants, dams, bridges, storm water drains and flood control measures need to be carried out during the planning and formulation stages of such projects. In a hydrological point of view, it is well recognised that whatever extreme the design loading, more sever conditions are likely to be encountered in nature. For the reason, frequency analysis of recorded hydro meteorological data such as flood discharge, rainfall, wind speed and temperature relating to the geographical region where the project is located, is a basic requirement for assessing such phenomena and arriving at structural and other design parameters for the project. The annual daily maximum rainfall data recorded at Surat for a period of 38 years (1969–2006) were used in this study. The statistical parameters viz. mean, standard deviation, coefficient of skewness, coefficient of kurtosis and correlation coefficient of original and log transformed rainfall series have been computed. The annual rainfall series has been checked for randomness and homogeneity by parametric and nonparametric test viz. median crossing tests, turning point test, rank difference test, run test, Wald-Wolfowitz test, runs above and below median test, Van Neuman ratio test, auto correlation test, Man Kendall test, Spearman’s Rho test, Kendall’s rank correlation test, etc. For long persistence evolution, Hurst coefficient test has been used. To explore the forms of relationships between time and rainfall, linear regression test has been used. The magnitudes of daily maximum rainfall corresponding to 10, 20, 25, 50 and 100 years return period were estimated using various distributions viz. Gumbel, Frechit, normal, log normal, GEV, Wakeby, logistic, G. logistic, G. Pareto, etc. Analytical procedures such as order statistics, probability weighted moments, L-moments and method of least square are applied for determination of estimators of the distribution. The adequacy of fitting of probability distribution to the recorded rainfall data is assessed by D-index goodness-of-fit test. Based on the identified frequency distribution rainfall amounts for different return periods were estimated. The analysis of annual daily maximum rainfall data and results are discussed in detail in this paper. The results of this study could be beneficial for design engineers and hydrologists for planning and design of small and medium hydrologic structures at Surat, Gujarat, India.