Applied Engineering in Agriculture Vol. 17(3): 285–292 E 2001 American Society of Agricultural Engineers ISSN 0883–8542 285 REGIONAL RUNOFF FREQUENCY CURVE FOR SMALL W ATERSHEDS IN THE HIRAKUD CATCHMENT OF EASTERN INDIA S. Ali, R. Singh ABSTRACT. Data from six small watersheds were utilized for developing a regional runoff frequency curve for small watersheds in the Hirakud catchment of eastern India. Five probability distribution functions were tested by comparing the residual sum of square, and a two–parameter log–normal distribution was found to be the best probability distribution function for use in the eastern region of India. A regional runoff frequency curve was generated using the index flood method. The student’ s T–test revealed that the regional runoff frequency curve estimate was not statistically different from a log–normal estimate at the 0.01 significance level. Keywords. Probability distributions, Runoff prediction, Watershed response, Watershed modeling. he watershed has long been recognized as the most effective unit for conservation, development, and optimum utilization of land and water resources for sustained bio–mass production. During the past few decades, the need has arisen for reliable information on watershed runoff along with watershed management programs for economical planning and design of best management practices (BMPs). Estimation of runoff magnitude at different return periods is useful in the design of structural and non–structural measures such as spillways, check dams, pond, etc. At present, several methods such as the Dicken, Ryves, Inglis, Ali Nawab Jung Bahadur, Rational formulae, Envelope Curves of Kanwar Sain, and Karpoce, etc. (CWPC, 1973) for transferring hydrological information have been developed in India as well as abroad including simple regression techniques (Chong and Moore, 1983; Nash and Shaw, 1965), multiple regression (Acreman, 1983; Garde and Kothyari, 1990; Misra, 1987; NERC, 1975.) and index flood method (Dalrymple, 1960). The regional flood frequency approach (index flood method) is often used in Great Britain, United States, Tanzania, and other countries for estimation of flood magnitude at different return periods, when little or no data exist. The present study was conducted in the Hirakud catchment of eastern India to transfer annual runoff from gauged to ungauged watersheds where no data is available, but watershed characteristics are known. Article was submitted for review in March 2000; approved for publication by the Soil & Water Division of ASAE in January 2001. The authors are Shakir Ali, Scientist, Central Soil & Water Conservation Research & Training Institute, Research Centre, Rajasthan, India, and Rajendra Singh, Associate Professor,Department of Agricultural Engineering, Indian Institute of Technology, Kharagpur, West Bengal, India. Corresponding author: S. Ali, Central Soil & Water Conservation Research & Training Institute, Research Centre, Kota 324002. (Rajasthan), India; phone: +0744–441753; fax: +0744–440764; e–mail: CSWCRTI@JP1.dot.net.in. METHOD STUDY AREA Hirakud catchment is located in eastern India between 21° 41’–22°21’ North latitude and 83° 51’–84°–42’ east longitude. The entire catchment is about 84,372 km 2 (32,568 mile 2 ) and divided in 42 sub–watersheds. The climate of the study area is subtropical with an annual temperature of 18³C. Mean annual precipitation is 1200 mm (47.24 in.) and 80% of it occurs in the form of rainfall during June to October. Recently, watershed development programs have been initiated in sub–watersheds of the Hirakud catchment. To date, best management practices (BMPs) have been applied in only a few watersheds, namely Saraswati, Ostali, and Banuslaga, but BMPs will soon be implemented in other watersheds. Information on runoff at different return periods is needed for designing BMPs or extending the runoff database, which is lacking in the catchment. DATA SOURCE Six small watersheds (fig. 1) namely Daku, Tamaga, Ichha, Sapae, Matwali, and Kharla of the Hirakud catchment have been selected, since reliable networks of stream gauging are under operation. Annual runoff data from the selected watersheds was obtained from reports (observed) of the Directorate of Soil Conservation, Bhubneshwar, India (Ali, 1991). The monitoring period varied from watershed to watershed, but spanned the period from 1970–1986. The length of annual runoff ranged from a minimum of 12 years to a maximum of 17 years. The area of these watersheds ranged from 89.50 to 541.25 km 2 (34.55 to 208.92 miles 2 ). The details of the selected watersheds are presented in table 1. TESTING OF DATA SET Regardless of the method used for selecting the data, values must satisfied two important criteria: (1) the events must be independent of a previous or subsequent event, and T