Graph

SAW Graph

Overview

Safe Agricultural Products and Water Graph (SAWGraph): An Open Knowledge Network to Monitor and Trace PFAS and Other Contaminants in the Nation's Food and Water Systems

A ProtoOKN project

The goal of SAWGraph is to onnect federal and other datasets to more quickly analyze data and make decisions around Per- and polyfluoroalkyl substances (PFAS) sources, testing, impacts, and risks The graph will integrate data data and provide interfaces that support analysis, mapping and communication of the results of testing for PFAS contamination, spatial connections to known and potential sources to support PFAS research and decision making around PFAS testing and mitigation. The primary end users of SAWGraphs are personnel at state and federal agencies who monitor the environment and the safety of drinking water and agricultural land and products. End users include but are not limited to personnel at EPA and state environmental protection agencies (e.g. state environmental toxicologists) and at USDA and state agricultural agencies. Other potential users include experts from FDA, USGS and similar state agencies. We distinguish four types of users based on their primary responsibility: (1) drinking water monitoring; (2) understanding the extent and sources of environmental contamination with PFAS; (3) PFAS researchers; (4) environmental justice. The graph is expected to empower these users to: (1) formulate and implement comprehensive test plans aimed at monitoring elevated PFAS levels in drinking water or agricultural lands, (2) pinpoint the most significant sources of PFAS in a state and how it impacts disadvantaged communities; (3) identify populations at elevated risk and prioritize testing resources and support accordingly; (4) design contamination management and remediation plana; and (5) identify research gaps in knowledge about PFAS fate and transport. Collectively, these applications will expedite comprehension of PFAS contamination and its mitigation, while ensuring the efficient and equitable allocation of testing and remediation resources.

Use Cases

We don’t just want to provide raw data, but actionable knowledge that helps answer these questions

About PFAS Contamination

What are PFAS?

PFAS: A group of thousands of manmade chemicals

  • Highly Persistent due to strong carbon-fluorine bonds
  • Informally named "Forever Chemicals", due to their properties, they do not break down easily

Toxicology and human health impacts of PFAS according to ATSDR (CDC):

  • Exposure and effects are poorly understood and likely underestimated
  • Increased cholesterol levels
  • Decreased vaccine response in children
  • Changes in liver enzymes
  • Small decreases in infant birth weights
  • Increased risk of high blood pressure or pre-eclampsia in pregnant women
  • Increased risk of kidney or testicular cancer

Where are PFAS?

  • They are dispersed via air, soil and water
  • How does PFAs get into the environment?
    • Chemical Plants
    • Landfills
    • Wastewater
    • Biosolids used as fertilizer in agriculture
    • Other Point Sources: Airports, firefighting training sites, etc.
    • Non-point Sources: spills, wet or dry deposition, commercial uses

What's a Knowledge Graph?

  • Establishes connections (links) across datasets
  • Kind of a distributed database that can grow over time

Read more about Geospatial Knowledge Graphs

Some papers on other geospatial knowledge graphs that were developed with the help of some of the team members:

Source data

  • PFAS testing results from drinking water (federal: PFAS Analytics Tools; original source UCMR data and select state data); ground and surface water, soil and tissue samples.
  • Point sources for PFAS based on the Facilities Registry Service using North American Industry Classification System (NAICS) and Standard Industrial Classification (SIC) as well as DoD sites, DoE sites, airports, and superfund sites and landfills
  • EPA’s PFAS release data collected in accordance with the Clean Water and Clean Air Acts
  • Potentially impacted features like private water wells and drinking water protection areas from select states.
  • Data layers on agricultural uses; surface water (stream reaches, watersheds) and groundwater (aquifers) features via connections to geoconnex and KnowWhereGraph.
  • Geospatial relations to connect data: locations and sites are spatially integrated via S2 cells (level 13 or finer) and via administrative regions (geoIds, FIPS codes, ZIP codes, etc.) and hydrologic connections

Competency questions and User queries

  • Testing Coverage and Gaps: Where (locations/towns) should we prioritize environmental/water testing?
  • E.g. which wells are within 5 miles of landfills/airports/biosludge application sites? Which wells near locations with a reported PFOA contamination above 4ppt have not been tested?
  • Assessing Contamination Impacts: Where and who is impacted? (Prevalence, risk/vulnerability mapping)
  • E.g. Which towns or counties have multiple test results with PFOS levels above 20ppt with no known contamination source nearby? Which chemicals show the highest average readings in samples from a particular region? Which towns/counties are most vulnerable based on their proximity to known or potential PFAS sources?
  • Contaminant Tracing: What are potential sources of contamination at a particular location or area?
  • E.g. What potential contamination sources exist 20 miles upstream from the sample result? What wells are hydrologically connected to other wells with a reported contamination of +10ppt

Team

Team from University of Maine, Kansas State University, Northeastern University and EPA brings together computer science, data science, geospatial, and chemistry experts

Torsten

Torsten Hahmann

Principal Investigator

University of Maine

ORCID iD icon 0000-0002-5331-5052

Information

Pascal

Pascal Hitzler

Co-PI

Kansas State University

Information

Ganga

Ganga Hettiarachchi

Co-PI

Kansas State University

ORCID iD icon 0000-0002-6669-2885

Information

Hande

Hande Küçük McGinty

Co-PI

Kansas State University

Information

Onur

Onur Apul

Co-PI

University of Maine

Information

Hari

Hari Palani

Co-PI

Northeastern

Information

Antony

Antony Williams

Federal Partner

EPA

ORCID iD icon 0000-0002-2668-4821

Vasu

Vasu Kilaru

Federal Partner

EPA

Shirly

Shirly Stephen

PostDoc Researcher

University of California Santa Barbara

ORCID iD icon 0000-0003-3547-8058

David

David Kedrowski

Research Assistant

University of Maine

ORCID iD icon 0000-0002-9070-3169

Information

Katrina

Katrina Schweikert

Research Assistant

University of Maine

ORCID iD icon 0000-0003-3271-6700

Information

Sonia

Sonia Moavenzadeh

Research Assistant

University of Maine

.

.

Adrita

Adrita Barua

Research Assistant

Kansas State University

Yinglun

Yinglun Zhang

Research Assistant

Kansas State University

Information

Revanth

Revanth Babu Raavi

Research Assistant

Kansas State University

Information