Files and links (1)
PDFOpen Access (License Unspecified), Open Access
Thesis
Comparison of a high and low spatial resolution model for an urban block and implications for depicting low impact development
Master of Science (M.S.), Drexel University
Mar 2011
DOI:
https://doi.org/10.17918/etd-3450
Abstract
The complexity of watersheds requires engineers to use computer models for planning purposes. The models are a reflection of the available data, software, and subjective judgment of the designer. Assumptions and simplifications may be made in characterizing hydrologic and hydraulic properties such that subcatchments are aggregated and small conduits are ignored. For this study, two models were developed for an urban watershed with different levels of spatial resolution using EPA SWMM. The high resolution model depicted each surface type (backyard, building, sidewalk, and street) as individual hydrological response units. The low resolution model represented the entire domain as a single hydrological response unit. This comparative study of calibrated models at different spatial resolution was undertaken in order to assess the effect of resolution on representing low impact development strategies for an urban watershed. The study consists of two components: the model development and the evaluation of Low Impact Development (LID) retrofits. Although hydrologic response unit resolution was the focus on the study, uncertainty in the observations and model structure was addressed and incorporated in the methodology for calibrating the models. A multi-objective optimization procedure was used to identify pareto solutions, i.e. sets of model parameters that yield model identical predictions. Statistical measures for comparing predictions to observations (e.g. percent volume difference and root mean square error) indicated that the low resolution model had a slightly better fit than the high resolution model to the observed flow regardless of whether the model was calibrated or uncalibrated. An examination of the pareto parameter sets for the two models revealed another inference; the uncalibrated low resolution model was the most physically representative. The calibrated low resolution model exhibited equifinality in the parameter values while the calibrated high resolution had comparative values to the uncalibrated model. Calibration did not greatly enhance the prediction performance of either the low or high resolution model since the initial parameters were informed from an intensive data review. When simulating LID within the models, the low resolution models predicted greater percent reduction in both peak flow rates and total volume flows than the high resolution model. Furthermore, the outcome is highly dependent on the specific storms events considered. Analysis of LID is considered preliminary until performance measures can be validated from field measurements. Even with physical data, models have inherent uncertainty which limit their predicative ability.
Metrics
28 File views/ downloads
19 Record Views
Details
- Title
- Comparison of a high and low spatial resolution model for an urban block and implications for depicting low impact development
- Creators
- Alisha E. Goldstein - DU
- Contributors
- Franco Montalto (Advisor) - Drexel University (1970-)
- Awarding Institution
- Drexel University
- Degree Awarded
- Master of Science (M.S.)
- Publisher
- Drexel University; Philadelphia, Pennsylvania
- Resource Type
- Thesis
- Language
- English
- Academic Unit
- Civil (and Architectural) Engineering [Historical]; College of Engineering (1970-2026); Drexel University
- Other Identifier
- 3450; 991014632220604721