Influence on Runoff: STATSGO vs. SSURGO: Soil Dataset Effect in Small Semiarid Watershed

This paper highlights the impact of the spatial varaiablity of soil data having different resolution on runoff estimated by Soil Conservation Service Curve Number (SCS-CN) methodology by taking an example of small Sub-watershed (16.59 SQ Km) in Cochise County, Arizona, USA. For this purpose watershed was delineated using DEM (Digital Elevation Model). Input data such as soil maps were prepared using GIS software. Data used is satellite imagery of LANDSAT satellite for the study area along with STATSGO and SSURGO soil maps of different resolutions along with observed (gauged) discharge & rainfall data. Anderson –Darling normality test and 1 sample T-Test was used to compare the proximity between observed discharge values(gauged data) and estimated discharge values using soil datasets of different (Higher & lower) Resolution. The normality test and the time series plot indicate that the data meet the t-test’s assumptions of normality and randomness. Results highlight the importance of dataset scale and influence is found comparable; Trend of computed runoff is better reflected using STATSGO data in comparison to SSURGO data (Figure 4, Table 3, 4 & 5). One-sample T-test is applied, and value of P (Table 4 & 5) is supporting that STATSGO data mean of estimated discharge is more close to observed discharge data mean in comparison to SSURGO data. Anderson-Darling normality test shows STATSGO data is following normal distribution (Figure6 & p-value 0.488(Table4).1-Sample T-Test p-value 0.488 is greater than significance level α (alpha) 0.05. While in the case of SSURGO dataset 1-sample t-test is failed (Table (5), P (0.00) <α (0.05). The 95% confidence interval (Table 4)indicates STATSGO data set (Figure 6 & Table 4). is better reflecting observed discharge trend in comparison to SSURGO dataset (Figure 6 & Table 5) for Study Area (Figure 1). In General, higher resolution datasets are considered better, but it is concluded and depicted that it might not be true always. The model may perform in a better way by accounting for other factors such as AMC conditions and intensities of rainfall. The methodology is easy and can be extended to bigger areas for runoff simulation along with application in all simulation software used in watershed modelling.