Investigation of storm durations, antecedent moisture conditions and impervious area measurements for flood discharge estimation

Design flows estimated by flood hydrograph simulation can be reasonably accurate or greatly in error, depending upon the modeling procedures and inputs selected. The objectives of this research project were (1) to determine which combinations of modeling procedures and inputs yield the best discharge estimates under various conditions, (2) to develop specific guidelines for flood hydrograph simulation and (3) to evaluate the potential of remote sensing methods for measuring the proportion of impervious surfaces in urban areas.; In the first part of this research, many different combinations of modeling procedures and inputs were tested in flood simulations for 66 gaged watersheds in Kansas. The key inputs were the duration of the hypothetical frequency-based storm and the antecedent moisture condition (AMC) in the NRCS loss model. Floods with six different recurrence intervals (2, 5, 10, 25, 50 and 100 years) were simulated using four different storm durations (3, 6, 12 and 24 hours), five different antecedent moisture conditions (AMC 2, 2¼, 2½, 2¾ and 3) and the two different unit-hydrograph models (NRCS and Snyder). From these results, combinations of storm durations and AMCs that yield unbiased discharge estimates for each set of conditions were identified.; In the second part of this research, three remote sensing methods were evaluated in terms of their appropriateness for predicting impervious surface proportions in urban areas from Landsat 7 images. The three methods are: (1) Linear Spectral Mixing Analysis (LSMA), (2) Artificial Neural Networks (ANN) and (3) Multivariate Linear Regression (MLR). Severe limitations were encountered while attempting to apply the LSMA method. On the other hand, the ANN and MLR methods proved to be adequate. Several ANN and MLR models were tested.; Impervious proportions were estimated from ANN and MLR models developed from two images, one image representative of early spring (leaf-off) and one image representative of summer (full leaf coverage), and a combined image. Three different calibration sample sizes (1%, 2% and 5%) were also evaluated and compared. The ANN and MLR methods yielded comparable impervious percentages regardless of the image and sample size used.