A revised rainfall-runoff model for an improved runoff estimate

Cheong, Yu Jian (2024) A revised rainfall-runoff model for an improved runoff estimate. Final Year Project, UTAR.

Preview

PDF Download (16Mb) | Preview

Abstract

Soil Conservation Service Curve Number (SCS CN) method was developed in 1954 to estimate the runoff amount of rainfall events, especially in ungauged catchments. However, its assumptions, including a linear correlation between initial abstraction and catchment potential maximum retention (Ia = λS) and a constant initial abstraction ratio (λ) of 0.2, are lack of thorough scrutiny. The invalid correlation from the observations made on a log-log scale graph leads to its failure to generate a precise runoff estimation. Therefore, this study aims to reformulate the CN rainfall-runoff model with a proposed power correlation (Ia= Sλ). With the help of inferential statistics, the optimum S and λ values were found to be 138.91 mm and 0.3159, respectively, to represent the collective representative CN parameter values for the Malaysia catchment using the DID HP 11 and HP 27 datasets. The optimum λ value of power model was transformed to the Ia/S ratio before the comparison with the λ of conventional model. The optimum Ia/S ratio derived from the revised power CN model is 0.034, lower than both the 0.05 and 0.2 ratio (Ia/S) proposed by the Natural Resources Conservation Service, indicating the necessity to calibrate the model parameter using the local dataset. Subsequently, the equivalent CN0.2 was identified to be 75.77 by substituting the optimum S0.3159 into the S power correlation and equivalent CN equations, transforming the conventional role of CN from input to output. The power CN model was benchmarked against the conventional model, where the adjusted R2, PBIAS, KGE, and NSE scores are better than the conventional model. The power CN model can estimate an exact runoff amount as its residual range spanned across the zero error without violating the SCS predefined constraint (Pmin > Ia). In contrast, the conventional model violated about 3.3% of the total predictions, indicating the model’s inconsistency and instability in producing a good simulation. Hence, this study provides SCS practitioners with a novel CN model, which can be adopted globally with the revised methodology to obtain the local optimum value. Lastly, it is advisable for the SCS practitioners to calibrate the power CN model using a sufficiently large sample size of local rainfall-runoff data to achieve a better simulation.

Actions (login required)