Selecting a Typical Period for Predictive Modeling – A conceptual approach and case study for selecting a representative period for collection system and water quality modeling
Abstract:The inherent complexity in modeling environmental systems poses a significant challenge when assessing the performance of wet weather and combined sewer overflow (CSO) control alternatives. Depending on the study system, several environmental factors may impact the effectiveness of controls in reducing CSO discharges and improving ambient water quality in the receiving water. Urban drainage, sewers, and the nearby waterways typically comprise a complex system that is influenced by highly variable environmental forcing conditions that may include lake water levels and lake seiche impacts in addition to rainfall and streamflow.
For cities and municipalities with CSO systems, sufficient amounts of local precipitation can have significant adverse impacts on the collection system, including combined sewer overflows, increases in quantity of wastewater requiring treatment, and possible sanitary sewer overflows. Consequently, storm and snowmelt events may result in significant loading of bacteria and other pollutants to receiving waters. The distribution and dilution of these loadings and the resulting water quality impacts within the receiving water system are strongly dependent on hydraulic conditions in the stream.
The range of possible combinations for these environmental factors is impossible to represent using a set of synthetic “design” or “reference” rainfall and streamflow scenarios. More importantly, a simplified approach to the problem can result in significant over-design of wet weather controls because there is a general tendency to focus on improbable worst-case conditions that are not necessarily realistic when overlapped. When sufficient long-term records are available, continuous model simulations avoid this pitfall by matching rainfall with the actual prevailing river and lake conditions. As a result, continuous simulation is generally acknowledged as a superior approach for modeling wet weather controls and water quality effects (EPA, 1999). This paper describes an innovative, unbiased approach that was used to select a typical period for modeling the response of the Toledo waterways system to control alternatives.
The “typical period” approach, which is used to identify historically representative time periods for collection system and water quality modeling, employs simple statistical analysis, knowledge of the system, and professional engineering judgment. The goal of this approach is to use quantitative metrics and relevant qualitative factors to select a period that is representative of typical environmental conditions in the system based on historical observations. The selected typical period must be long enough to capture seasonal variations and an adequate distribution of storm and flow events, yet short enough to permit continuous simulations of the collection system. The first step in the typical period approach is the identification of quantitative metrics that can be used to evaluate the representativeness of each environmental variable. Metrics can then be computed for the historical period and any combination of discrete single or multi-year periods contained within the historical period. Based on computed metrics, a representative time period can then be selected based on statistical comparability between that period and the historical period. In addition, logistical considerations can factor into the final selection of a typical period.
The City of Toledo releases CSO discharges to the Maumee and Ottawa rivers, which flow through the City and discharge to northwest Lake Erie. Hydraulic conditions in each of these river systems are influenced not only by upstream flow and local precipitation but also by Lake Erie water levels and seiche activity. Evaluation and optimization of long-term control plan (LTCP) alternatives for the City requires that modeling studies be conducted for both the collection system hydraulics and water quality impacts in the Maumee and Ottawa Rivers.
The “typical period” approach is being applied to evaluate system response to LTCP alternatives in an unbiased manner. The environmental variables included in the analysis are precipitation, Maumee River streamflow, Ottawa River streamflow, and Lake Erie water levels and seiche magnitude. A representative one-year period (1998) is nested within the selected five-year period (1997–2001). The one-year period will be used to conduct efficient screening of a large number of potential control alternatives and to evaluate model sensitivity, while the selected five-year period will be used to provide a more robust assessment of a select group of control alternatives. The application of both a one-year and a five-year simulation period provides an ideal combination of efficiency and precision for evaluating control alternatives.
The metrics used to select the representative period were developed for each important environmental factor, including precipitation, streamflow, lake levels, and seiche magnitude. The metrics developed to evaluate the representativeness for precipitation data were based largely on knowledge of the existing relationship between storm events and CSO discharges. These metrics include: 1) the number of storms per year that exceed 0.25 inches, 2) the number of storms exceeding 1.20 inches, and 3) the annual average rainfall volume. The first metric is derived from the minimum rainfall required to trigger CSO discharges in the system. The second metric represents the fifth largest storm that occurs each year on average, and the third metric provides a fundamental indication of the overall rainfall volume observed for a given period.
The metrics used to assess the representativeness of streamflow, lake levels, and seiche magnitude were designed to capture important quantitative measures of the statistical distribution of these environmental variables. The metrics used to evaluate streamflow, water levels, and seiche magnitude were quartile values (25,th, 50th , and 75th percentiles) for the daily flow, annual average level, and average monthly seiche distributions, respectively. For each of these variables, the metrics computed for a given time period were compared to the metrics computed for the historical period to evaluate representativeness. Several qualitative/logistical factors were also considered, including the availability of rainfall data for multiple gages post-1993, the occurrence of extreme events, and computational burden. The potential advantages and disadvantages of these factors for a given time period were addressed on a case-by-case basis.
The use of synthetically generated environmental conditions in predictive collection system and water quality models may result in an inaccurate representation of the physical conditions actually observed in the system. In general, these types of simplified methods will result in an overly conservative assessment of system controls required to meet a given level of compliance. The use of observed historical conditions for modeling system response to various control alternatives provides an unbiased approach evaluation by using observed conditions for each environmental stressor of importance. The selection of a representative period is critically important when using the continuous simulation approach. Identifying and applying quantitative metrics to evaluate similarities in stressor distributions (e.g., daily flow) provides a solid basis for selecting a representative period. Overall, the use of a representative historical period decreases the likelihood of biased evaluations of control scenarios and reduces the potential for over design of control alternatives.
Document Type: Research Article
Publication date: January 1, 2004
More about this publication?
- Proceedings of the Water Environment Federation is an archive of papers published in the proceedings of the annual Water Environment Federation® Technical Exhibition and Conference (WEFTEC® ) and specialty conferences held since the year 2000. These proceedings are not peer reviewed. WEF Members: Sign in (right panel) with your IngentaConnect user name and password to receive complimentary access.
- Subscribe to this Title
- Membership Information
- About WEF Proceedings
- WEFTEC Conference Information
- Ingenta Connect is not responsible for the content or availability of external websites