Lefkowitz, Jamie

Jamie has 17 years of experience as a civil engineer, managing water resources projects for a diverse set of clients and partners throughout the U.S. Her experience includes remote sensing, advanced analytics, internet-of-things application, watershed planning, integrated water resource management, numerical modeling and analysis, surface water quality, and stormwater management. Jamie leads Brown and Caldwell's Data Science Program, including the development of novel machine learning and AI applications tailored for use in the water/wastewater industry.

Titles from this speaker

Modernizing CSO Notification and Forecasting with Digital Tools

Abstract

The prevention, mitigation, and management of combined sewer overflows (CSO) continues to be a focal point for municipal wastewater utilities throughout the country. Regulatory requirements are constantly evolving and continue to increase emphasis on public alerting. Modern digital solutions can be used to improve the way that utilities notify the public about the occurrence and effect of CSOs in their communities. This presentation will provide examples of enhanced CSO prediction and notification tools that leverage predictive analytics and machine learning to improve transparency in CSO management. In the first example, an alert system was developed to inform community members of the risk of high bacteria levels downstream from CSOs. The alert tool uses publicly available information that is fed into an algorithm developed with historic data and relationships, including time of travel and decay constants, specific to the river. The tool has been validated against several overflow events and predicts downstream bacteria concentrations within a reasonable and useful accuracy. The results of the predictive tool are available on a public website hosted by the regional planning agency. Innovative tools that leverage existing engineering information are effective at informing the public about the real-time contamination risks of combined sewer overflows. The second example demonstrates how surrogate modeling using machine learning can predict collection system flows, and associated CSOs, in real time. This example also maximizes the use of existing available data and information to create a modern prediction tool. A calibrated hydraulic model of a combined collection system was used to train a machine learning model to produce predictive flows based on real-time weather forecasts. The knowledge of when, where, and how much combined sewage is predicted to be released into local waterways can be used to alert the public of impending closures and advisories. This type of tool can also be used by utility managers and operators to plan for upcoming overflows and increased inflow to wastewater treatment facilities.

This paper was presented at the WEF Collection Systems and Stormwater Conference, April 9-12, 2024.

SpeakerLefkowitz, Jamie

Presentation time

16:15:00

16:45:00

Session time

13:30:00

16:45:00

SessionCombined Sewer Overflow

Session number07

Session locationConnecticut Convention Center, Hartford, Connecticut

TopicCollaboration, Combined Sewer Overflow, Combined Sewer System, Construction, Deep Sewer Tunnel, Flow control, Long Term Control Planning, Modeling, Predictive Analytics, Public Education/Information/Communication, Pump station, forcemain, Regulatory Compliance - Collection Systems, Sewer Separation, Wet Weather

Author(s)

Lefkowitz, Jamie

Author(s)J. Lefkowitz1, A. Goldberg

Author affiliation(s)Brown and Caldwell 1

SourceProceedings of the Water Environment Federation

Document typeConference Paper

PublisherWater Environment Federation

Print publication date Apr 2024

DOI10.2175/193864718825159347

Volume / Issue

Content sourceCollection Systems and Stormwater Conference

Word count9

Real Time Predictive Modeling for Public Water Quality Alerts

Abstract

Combined sewer overflows (CSOs) and their associated pollution loadings to water bodies continue to be a primary focus for wastewater utilities across the country. Bacteria loadings from CSOs, as measured by concentration of E. coli for freshwaters and Enterococci for marine waters, have been deemed detrimental for human recreation by the U.S. EPA and draw negative attention from effected communities (U.S. EPA, 2012). In Massachusetts (MA), utilities are required to issue electronic alerts via email and/or websites when a CSO event happens in their community. While this information establishes a basis for public awareness of potential risks, it does not necessarily translate into the duration of potential risks after an overflow, or the locations where elevated bacteria levels might be found. This study's objective was to develop a real time predictive model of bacteria loadings resulting from 5 cities' CSOs to the Merrimack River (Manchester and Nashua, NH; Lowell, Lawrence, and Haverhill, MA). Specifically, this included the implementation of an innovative daily alert program to inform the general public, through virtual means, about the potential swimming and boating hazards in the Merrimack River at Newburyport, MA from bacterial pollutants resulting from CSOs. The locations of the CSOs and alert focus area are shown in Figure 1.

An alert system was developed to inform community members of the risk of high bacteria levels downstream from CSOs. The alert tool uses publicly available information that is fed into an algorithm developed with historic data and relationships, including time of travel and decay constants specific to the river. The tool has been validated against several overflow events and predicts downstream bacteria concentrations within a reasonable and useful accuracy. The results of the predictive tool are available on a public website hosted by one of the stakeholder communities. Innovative tools that leverage existing engineering information are effective at informing the public about the real-time contamination risks of combined sewer overflows.

SpeakerLefkowitz, Jamie

Presentation time

13:45:00

14:00:00

Session time

13:30:00

15:00:00

SessionPathogen Pep-Talk: How We Monitor, Model, and Convey Data

Session number510

TopicDisinfection and Public Health, Public Communication and Outreach, Watershed Management, Water Quality, and Groundwater

Author(s)

Jamie Lefkowitz