About

Andrea Silverman is an Associate Professor of environmental engineering at the NYU Tandon School of Engineering, affiliated with the Civil, Urban, and Environmental Engineering department and the Center for Urban Science and Progress (CUSP). Her research centers on water quality, wastewater treatment, and urban flooding, with a focus on protecting public health and environmental quality. She specializes in the detection and disinfection of waterborne pathogens, wastewater-based epidemiology, the design of natural wastewater treatment systems such as treatment ponds and constructed wetlands, and the safe reuse of human waste. Dr. Silverman has conducted field research in diverse settings including New York City, California, Accra in Ghana, and Nairobi in Kenya, working in both high- and low-income environments. She co-directs the FloodNet project, which aims to develop low-cost sensors for real-time urban flooding measurement and reporting, and has partnered with the NYC Department of Environmental Protection to develop and implement wastewater surveillance for COVID-19. Her work involves collaboration with government agencies and community organizations to address urban water challenges.

Research topics

• Computer Science
• Artificial Intelligence
• Mathematics
• Geography
• Archaeology
• Remote sensing
• Geology
• Cartography

Selected publications

• FloodSense street sign mounted flood depth sensor

  Zenodo (CERN European Organization for Nuclear Research) · 2022-10-24

  datasetOpen access

  <strong>Flood Depth Data (FDD)</strong> collected by a fleet of sensors deployed across 5 boroughs of New York City with a resolution of half an inch or less. The metadata for the sensors is included in the metadata.csv to identify the deployment coordinates of sensors, each with a unique <strong><em>deployment_id</em></strong>. The depth data is collected at least every five minutes and every minute in some locations depending on the ability to harvest solar energy at that deployment location. The final depth data field is <strong><em>depth_proc_mm</em></strong>, and the raw data is <strong><em>dist_mm</em></strong>. The raw measurement values received from the sensor are distance measurements (dist_mm), which are simply distance measurements collected from a ranging ultrasonic-based sensor. These distance measurements are converted to depths using <strong><em>night_median_dist_mm</em></strong> which is a daily calculated median of nighttime sensor readings. Direct sunlight affects ranging measurements due to high variance in the air column between the sensor and the concrete surface that it is mounted over. Additionally, the housing internally heats up when under direct sunlight, which affects the sensor readings and appears as if the surface dips with the daily increase and decrease in temperature during the daytime. After converting to raw depth values, a simple range filter is applied to the data removing any anomalies that lie below 10 millimeters and above unrealistic depth values (for example a person - between 5ft to 6ft), which is named <strong><em>depth_filt_mm</em></strong>. Further, this filtered depth value is processed through data filters eliminating blips, any pulse chains, or a flat line due to garbage or a car parked underneath the sensor. The output of these filters is labeled <strong><em>depth_proc_mm</em></strong>. This data is intended for use by communities, researchers, and New York City government agencies to better understand the frequency, severity, and impacts of flooding in New York City. Here is the live dashboard for these sensors deployed: FloodNet Data Dashboard More about this project at FloodNet.NYC This is an open-source project and for more information on the sensors and build manuals see the FloodNet FloodSensor GitHub page

  PublisherDOI

• FloodSense street sign mounted flood depth sensor

  Zenodo (CERN European Organization for Nuclear Research) · 2022-10-24 · 1 citations

  datasetOpen access

  <strong>Flood Depth Data (FDD)</strong> collected by a fleet of sensors deployed across 5 boroughs of New York City with a resolution of half an inch or less. The metadata for the sensors is included in the metadata.csv to identify the deployment coordinates of sensors, each with a unique <strong><em>deployment_id</em></strong>. The depth data is collected at least every five minutes and every minute in some locations depending on the ability to harvest solar energy at that deployment location. The final depth data field is <strong><em>depth_proc_mm</em></strong>, and the raw data is <strong><em>dist_mm</em></strong>. The raw measurement values received from the sensor are distance measurements (dist_mm), which are simply distance measurements collected from a ranging ultrasonic-based sensor. These distance measurements are converted to depths using <strong><em>night_median_dist_mm</em></strong> which is a daily calculated median of nighttime sensor readings. Direct sunlight affects ranging measurements due to high variance in the air column between the sensor and the concrete surface that it is mounted over. Additionally, the housing internally heats up when under direct sunlight, which affects the sensor readings and appears as if the surface dips with the daily increase and decrease in temperature during the daytime. After converting to raw depth values, a simple range filter is applied to the data removing any anomalies that lie below 10 millimeters and above unrealistic depth values (for example a person - between 5ft to 6ft), which is named <strong><em>depth_filt_mm</em></strong>. Further, this filtered depth value is processed through data filters eliminating blips, any pulse chains, or a flat line due to garbage or a car parked underneath the sensor. The output of these filters is labeled <strong><em>depth_proc_mm</em></strong>. This data is intended for use by communities, researchers, and New York City government agencies to better understand the frequency, severity, and impacts of flooding in New York City. Here is the live dashboard for these sensors deployed: FloodNet Data Dashboard More about this project at FloodNet.NYC This is an open-source project and for more information on the sensors and build manuals see the FloodNet FloodSensor GitHub page

  PublisherDOI

• NYU FloodSense street sign mounted flood depth sensor

  Zenodo (CERN European Organization for Nuclear Research) · 2021-01-08

  datasetOpen access

  Water depth level in mm from a sensor mounted on a street sign post at the corner of 5th Street and Hoyt, Brooklyn, NY (40.676640, -73.994595). The sensor is designed to detect flood water that fills the street and blocks vehicle and pedestrian traffic, as well as depositing micro-organisms on the street. Ultrasonic technology is used to detect flood water depth. The sensor transmits its data via LoRaWAN and is equipped with a solar panel for continuous operation. Depth data is collected at ~5min intervals. Time fields are in local time (New York). Date format is: 2020-10-04 20:11:45.742594232-04:00 Two flood events have been observed in this dataset between these date ranges: "2020-11-15 19:37:00.000000000-05:00" to "2020-11-16 00:30:00.000000000-05:00" "2020-11-30 10:20:00.000000000-05:00" to "2020-11-30 13:30:00.000000000-05:00" Erroneous data has been observed: There are ~1% decreases in depth measures on days with sun which suggests that the distance sensor is affected by direct sunlight This data is preliminary and is for prototyping purposes. This dataset will be updated when more data is collected. Please see our github org for sensor information and build instructions: github.com/floodsense

  PublisherDOI

• NYU FloodSense Gowanus canal mounted sensor depth

  Zenodo (CERN European Organization for Nuclear Research) · 2021-01-08

  datasetOpen access

  Water depth level in mm from a sensor mounted mounted above the Gowanus Canal, Brooklyn, NY (40.674490, -73.994458) from October 4th 2020 to January 8th 2021. The sensor is designed to detect flood water that fills the street and blocks vehicle and pedestrian traffic, as well as depositing micro-organisms on the street. This one is used for data validation. The sensor transmits its data via LoRaWAN and is equipped with a solar panel for continuous operation. Data is collected at ~5min intervals. Time fields are in local time (New York). Time fields are in local time (New York). Date format is: 2020-10-04 20:11:45.742594232-04:00 Two flood events have been observed in this dataset between these date ranges: "2020-11-15 19:37:00.000000000-05:00" to "2020-11-16 00:30:00.000000000-05:00" "2020-11-30 10:20:00.000000000-05:00" to "2020-11-30 13:30:00.000000000-05:00" One type of erroneous data has been observed: There are ~1% rises in distance measures on days with sun which suggests that the distance sensor is affected by direct sunlight This data is prelimary and is for prototyping purposes. Not to be used as a reliable data source as it is. This dataset will be updated when more data is collected. Please see our github repo for sensor information and build instructions: github.com/floodsense

  PublisherDOI

• NYU FloodSense Gowanus canal mounted sensor depth

  Zenodo (CERN European Organization for Nuclear Research) · 2021-01-08

  datasetOpen access

  Water depth level in mm from a sensor mounted mounted above the Gowanus Canal, Brooklyn, NY (40.674490, -73.994458) from October 4th 2020 to January 8th 2021. The sensor is designed to detect flood water that fills the street and blocks vehicle and pedestrian traffic, as well as depositing micro-organisms on the street. This one is used for data validation. The sensor transmits its data via LoRaWAN and is equipped with a solar panel for continuous operation. Data is collected at ~5min intervals. Time fields are in local time (New York). Time fields are in local time (New York). Date format is: 2020-10-04 20:11:45.742594232-04:00 Two flood events have been observed in this dataset between these date ranges: "2020-11-15 19:37:00.000000000-05:00" to "2020-11-16 00:30:00.000000000-05:00" "2020-11-30 10:20:00.000000000-05:00" to "2020-11-30 13:30:00.000000000-05:00" One type of erroneous data has been observed: There are ~1% rises in distance measures on days with sun which suggests that the distance sensor is affected by direct sunlight This data is prelimary and is for prototyping purposes. Not to be used as a reliable data source as it is. This dataset will be updated when more data is collected. Please see our github repo for sensor information and build instructions: github.com/floodsense

  PublisherDOI

• NYU FloodSense street sign mounted flood depth sensor

  Zenodo (CERN European Organization for Nuclear Research) · 2021-01-08

  datasetOpen access

  Water depth level in mm from a sensor mounted on a street sign post at the corner of 5th Street and Hoyt, Brooklyn, NY (40.676640, -73.994595). The sensor is designed to detect flood water that fills the street and blocks vehicle and pedestrian traffic, as well as depositing micro-organisms on the street. Ultrasonic technology is used to detect flood water depth. The sensor transmits its data via LoRaWAN and is equipped with a solar panel for continuous operation. Depth data is collected at ~5min intervals. Time fields are in local time (New York). Date format is: 2020-10-04 20:11:45.742594232-04:00 Two flood events have been observed in this dataset between these date ranges: "2020-11-15 19:37:00.000000000-05:00" to "2020-11-16 00:30:00.000000000-05:00" "2020-11-30 10:20:00.000000000-05:00" to "2020-11-30 13:30:00.000000000-05:00" Erroneous data has been observed: There are ~1% decreases in depth measures on days with sun which suggests that the distance sensor is affected by direct sunlight This data is preliminary and is for prototyping purposes. This dataset will be updated when more data is collected. Please see our github org for sensor information and build instructions: github.com/floodsense

  PublisherDOI

• NYU FloodSense Street Sign Mounted Flood Depth Sensor [supporting dataset]

  2021

  • Computer Science
  • Artificial Intelligence
  • Remote sensing
  Publisher

• NYU FloodSense Gowanus canal mounted sensor depth

  Figshare · 2020-12-04

  dataset

  Water depth level in mm from a sensor mounted mounted above the Gowanus Canal, Brooklyn, NY (40.674490, -73.994458). The sensor is designed to detect flood water that fills the street and blocks vehicle and pedestrian traffic, as well as depositing micro-organisms on the street. This one is used for data validation. The sensor transmits its data via LoRaWAN and is equipped with a solar panel for continuous operation. Data is collected at ~5min intervals. Time fields are in local time (New York). Time fields are in local time (New York). Date format is: 2020-10-04 20:11:45.742594232-04:00 Two flood events have been observed in this dataset between these date ranges: "2020-11-15 19:37:00.000000000-05:00" to "2020-11-16 00:30:00.000000000-05:00" "2020-11-30 10:20:00.000000000-05:00" to "2020-11-30 13:30:00.000000000-05:00" One type of erroneous data has been observed: There are ~1% rises in distance measures on days with sun which suggests that the distance sensor is affected by direct sunlight This data is prelimary and is for prototyping purposes. Not to be used as a reliable data source as it is. This dataset will be updated when more data is collected. Please see our github org for sensor information and build instructions: github.com/floodsense

  DOI

• NYU FloodSense street sign mounted distance sensor

  Zenodo (CERN European Organization for Nuclear Research) · 2020-10-30

  datasetOpen access

  Ultrasonic distance data in mm from a sensor mounted on a street sign post at the corner of 5th Street and Hoyt, Brooklyn, NY (40.676640, -73.994595). The sensor is designed to detect flood water that fills the street and blocks vehicle and pedestrian traffic, as well as depositing micro-organisms on the street. The sensor transmits its data via LoRaWAN and is equipped with a solar panel for continuous operation. Data is collected at ~5min intervals. Time fields are in local time (New York). Two types of erroneous data has been observed: Large spikes in distance that always manifest at 5000mm - can be excluded There are ~1% rises in distance measures on days with sun which suggests that the distance sensor is affected by direct sunlight This data is prelimary and is for prototyping purposes. Not to be used as a reliable data source as it is. This dataset will be updated when more data is collected. Please see our github org for sensor information and build instructions: github.com/floodsense

  PublisherDOI

• NYU FloodSense street sign mounted distance sensor

  Zenodo (CERN European Organization for Nuclear Research) · 2020-10-30

  datasetOpen access

  Ultrasonic distance data in mm from a sensor mounted on a street sign post at the corner of 5th Street and Hoyt, Brooklyn, NY (40.676640, -73.994595). The sensor is designed to detect flood water that fills the street and blocks vehicle and pedestrian traffic, as well as depositing micro-organisms on the street. The sensor transmits its data via LoRaWAN and is equipped with a solar panel for continuous operation. Data is collected at ~5min intervals. Time fields are in local time (New York). Two types of erroneous data has been observed: Large spikes in distance that always manifest at 5000mm - can be excluded There are ~1% rises in distance measures on days with sun which suggests that the distance sensor is affected by direct sunlight This data is prelimary and is for prototyping purposes. Not to be used as a reliable data source as it is. This dataset will be updated when more data is collected. Please see our github org for sensor information and build instructions: github.com/floodsense

  PublisherDOI

Frequent coauthors

• Charlie Mydlarz

  17 shared

• Elizabeth Hénaff

  17 shared

• Tega Brain

  17 shared

• Junaid Khan

  Sher-e-Kashmir University of Agricultural Sciences and Technology of Kashmir

  16 shared

• Praneeth Sai Venkat Challagonda

  New York University

  15 shared

• Praneeth Challagonda

  City University of New York

  2 shared

• Arvin S. Glicksman

  Brown University

  1 shared

• Dorina M. Lanza

  1 shared

Labs

• Silverman LaboratoryPI