Date on Master's Thesis/Doctoral Dissertation


Document Type

Doctoral Dissertation

Degree Name

Ph. D.


Civil and Environmental Engineering

Degree Program

Civil Engineering, PhD

Committee Chair

French, Mark

Committee Co-Chair (if applicable)

Rockaway, Thomas

Committee Member

Rockaway, Thomas

Committee Member

Bhaskar, Nageshwar

Committee Member

Kantardzic, Mehmed

Author's Keywords

spatiotemporal rainfall; weather radar; rainfall optimization; urban hydrology; combined sewer overflow; data mining


In urban areas, a prevalence of combined sewer systems (CSS) exist that carry both storm water runoff and sanitary sewer flows in a single pipe, these system are considered combined sewers. In the absence of rainfall-runoff most of these systems function adequately, however CSS capacity is typically inadequate to carry peak stormwater runoff volume. In order to minimize sewage flooding into streets and backups into homes and businesses, most CSSs (as well as separate sanitary sewer systems) are designed to overflow into surface waters such as streams and rivers, lakes and seas. This occurrence is considered a combined sewer overflow (CSO) event and has a critical impact on urban aquatic environment and degrades downstream water quality. This investigation provides a framework for the application of radar-rainfall data to estimate the characteristics of rainfall events that produce a CSO event. The process addresses an urban sewer-shed, denoted as CSO 130, located in Louisville, Kentucky (USA). The characteristics of each heavy rainfall event; total volume, intensity, duration, continuity, and storm types govern the overflow in the approximately 13-ha (30-ac) sewer-shed. In urban hydrology, accurate fine resolution temporal and spatial rainfall observations is a key factor for managing urban hydrologic systems and forecasting storm water runoff, particularly in the current era of higher variability in recent rainfall events. To study this issue, rain gauge data from a ground based rainfall measurement network operated by the local stormwater management agency, Metropolitan Sewer District (MSD), in Jefferson County Kentucky is studied. Rainfall spatial characteristics are evaluated through correlation spectrum by distance and this reveals a spatial rainfall variation concave relationship. Besides, the event based rainfall classification has been performed to provide a context for identification and description of rain events that may be useful as guidance for urban stormwater management. Based on this study, the limitation of the one dimensional rainfall monitoring system has been revealed by the severe variation of the rainfall characteristics. In order to overcome this issue, the reliable areal rainfall measurement with fine spatiotemporal resolutions is urgently required to investigate the urban hydrologic issues. The radar data utilized in this study are from the weather radar associated with the National Weather Service (NWS) Forecast Office Louisville, Kentucky (denoted by call letters KLVX) and rain gauge data are from a regional network. The study applies fine resolution radar rainfall in this urban hydrologic system to reveal insights for planning CSO control and prevention under a range of rainfall event regimes. Weather radar data from the local NWS site is optimized using support vector classification (SVC) and serves as rainfall input for the urban sewer-shed. The radar-rainfall data were optimized through a comparison with NWS radar rainfall and a gauge network, the local stormwater and sewer agency. The optimized radar rainfall estimation has the highest spatiotemporal correlation in quarter hourly temporal resolution. The rainfall and flow events are defined using the criteria proposed by United States Environmental Protection Agency (USEPA) to define the physical continuity of natural rainfall processes and the corresponding hydrologic response. The optimized rainfall product has applied to the small scale urban watershed, CSO130 to investigate the sewer water overflow. In this setting, the extremity of the rainfall governs the overflow mainly with volumetric rainfall in the event based rainfall and its corresponding overflow with other decisive factors; rainfall intensity, duration, rain type as well as rainfall continuity. Discriminant analysis is introduced to classify these precipitation factors. The objective of this study is that downscaled hydrologic application to the places where the sub-hourly rainfall data is required such as a complex urban watershed in order to investigate the fast inundated floods, overflows in the artificial watersheds or any hydrologic preparation.

Included in

Engineering Commons