Application of NRCS-curve number method for runoff estimation in a mountainous watershed

The major problem in the assessment of relationships between rainfall and runoff occurs when a study is carried out in ungauged watersheds in the absence of hydro-climatic data. This study aims to evaluate the applicability of Natural Resources Conservation Service-Curve Number (NRCS-CN) method together with GIS in estimating runoff depth in a mountainous watershed. The study was carried out in the semi-arid Kardeh watershed which lies between 36? 37? 17? to 36? 58? 25? N latitude and 59? 26? 3? to 59? 37? 17? E longitude, about 42 km north of Mashhad, Khorasan Razavi Province, Iran. The hydrologic soil groups, land use and slope maps were generated with GIS tools. The curve number values from NRCS Standard Tables were assigned to the intersected hydrologic soil groups and land use maps to generate CN values map. The curve number method was followed to estimate runoff depth for selected storm events in the watershed. Nash-Sutcliffe efficiency, pair-wise comparison by the t-test, Pearson correlation and percent error were used to assess the accuracy of estimated data and relationship between estimated and observed runoff depth. The results showed relatively low Nash-Sutcliffe efficiency (E = ? 0.835). There was no significant difference between estimated and observed runoff depths (P > 0.05). Fair correlation was detected between estimated and observed runoff depth (r = 0.56; P < 0.01). About 9% of the estimated runoff values were within ?10% of the recorded values and 43% had error percent greater than ?50%. The results indicated that the combined GIS and CN method can be used in semi-arid mountainous watersheds with about 55% accuracy only for management and conservation purposes.