APPLICATION OF MODIFIED SCS-CN METHOD

Abstract

Agricultural experimental field was used for application of the modified (MS 2002) SCS-CN model on the data from a study area located in Toda Kalyanpur, Roorkee, Haridwar, Uttarakhand (latitude: 29°50'5.17"N, longitude: 77°55'15.60"E). For experiment, 9 plots were constructed each having three plots of 8%, 12% and 16% slopes. Each plot had size of 12m x 3m and connected to collection chamber of size 1m x 1m x 1m with approach channel to observe the runoff generated. A five slotted flow divisor made of thin steel plate was installed just upstream of the chamber which allowed runoff to collection chamber from only one slot to reduce the size of chamber. Similarly, an ordinary rain gauge (ORG) was also installed to measure daily rainfall (mm). The degree of complexity of runoff generation is very high, dynamic in nature and influenced by many interrelated physical factors. These physical factors vary temporally as well as spatially and cause uncertainty in runoff prediction. Therefore, predicting the runoff generated by rainfall accurately becomes a more challenging work. At the same time accurate prediction of runoff is prerequisite for the effective management and development of water resources. Thus, there exist a number of methods being used to estimate the runoff. The existing and modified SCS-CN methods are the two methods among many others. These methods are used because of their easy use and based on single input parameter curve number (CN), which incorporates all the physical factors affecting the runoff. CN, the key parameter of SCS-CN method, is a function of hydrological condition, land use/land cover, soil type, soil moisture and hydrological soil group. Therefore, determining the accurate value of CN is the most important work in CN hydrology. Similarly, prediction of accurate initial abstraction ratio (λ) as well as best performing model are other important works for obtaining improved results. After the experimental study, the curve number observed (CNobs) from the existing SCS-CN model employing frequency matching method was higher than that from the modified SCS-CN model. Similarly, the optimized curve number (CNopti) calculated using existing SCS-CN method was higher than modified (MS 2002) SCS-CN model and therefore, the existing SCS-CN model will yield higher runoff than modified SCS-CN model (MS 2002). Initial abstraction ratio (λ) computed by optimization technique using P-Q data has been found 0 as minimum and 0.05 as a maximum for both existing and modified SCS-CN iii models. The value of λ did not match with standard value of λ=0.2, given by SCS (1985) which indicates that use of constant value of λ=0.2 may not give good results. Nash and Sutcliffe efficiency (NSE) is the most appropriate method of evaluating model performance. NSE of two models i.e. modified (MS 2002) and existing SCS-CN model have been computed by two methods, using CN from NEH-4 table and using optimized λ and CN. While using CNs from NEH-4 table, NSE of modified (MS 2002) is found greater than the existing model, except in plot nos. 3 and 6 (Table 4.19). Similarly, using optimization technique, NSE of the modified model is found greater than the existing model in all plots (Table 4.18). Thus, the modified model (MS 2002) is more efficient than the existing model.