EPA Drinking Water Quality Screener - Violations, Lead & PFAS

Give it a coordinate. It tells you which public water system serves that address, what the Safe Drinking Water Act violation record looks like, what the lead and copper 90th-percentile tap results are, and whether PFAS was detected there under EPA's UCMR5 monitoring programme.

Built for Phase I ESA consultants, environmental due-diligence teams, real-estate underwriters and water-utility analysts who need a defensible drinking-water screen per property, in bulk, from authoritative federal sources.

The design rule that matters most: this actor will not tell you a system is clean on a source it never successfully read. null means not checked. false means checked, and the answer is negative. The two are never mixed, and the overall verdict is gated so it can never read LOW while any contributing EPA source was unavailable.

---

Who it is for

Persona	What they get
Phase I ESA consultant	A per-property drinking-water screen with the serving PWSID, health-based violation history, lead action-level exceedances and PFAS results - plus an explicit NO_SERVICE_AREA answer when the site is probably on a private well.
Real-estate / lender due diligence	Bulk screening of a portfolio in one run, with a water_quality_screen_level and a plain-English screen_reason per address.
Environmental / water engineer	Lead and copper 90th percentiles dated to their actual monitoring period, treatment-technique violations, formal enforcement history and per-analyte PFAS concentrations.
PFAS litigation and remediation	UCMR5 detections per system with maximum concentrations, the analytes involved, a Hazard Index screen, and honest non-detect accounting.
AI agents / MCP	A clean, chainable tool: coordinates in, one fully-typed row per location out, with a per-source status on every row so an agent can tell "no finding" from "no data".

---

Data sources

Leg	Source	What it provides
Service area	EPA ORD Community Water System Service Area Boundaries (ArcGIS, ~44,600 polygons)	Point-in-polygon lookup of the serving PWSID
Water system	EPA Envirofacts SDWIS WATER_SYSTEM + GEOGRAPHIC_AREA	System name, type, population served, source water, owner, counties served, utility contact
Violations	EPA Envirofacts SDWIS VIOLATION + ENFORCEMENT_ACTION	Full SDWA violation history, health-based rollup, formal enforcement
Lead and copper	EPA Envirofacts SDWIS LCR_SAMPLE_RESULT joined to LCR_SAMPLE	90th-percentile lead and copper results, dated to their monitoring period
PFAS	EPA UCMR5 occurrence dataset (1,928,117 results, 10,299 systems, 29 PFAS analytes plus lithium)	Detections, maximum concentrations, non-detect counts, Hazard Index screen
Code decoding	EPA Envirofacts SDWIS REF_CODE_VALUE	EPA's own 2,171-row code table, fetched every run so labels cannot go stale

No API key. Every source is public and keyless.

---

Example input

{
  "assets": [
    { "lat": 43.0125, "lon": -83.6875, "label": "Flint MI - lead action level exceedance" },
    { "lat": 40.9793, "lon": -74.1165, "label": "Ridgewood NJ - PFAS detections" },
    { "lat": 39.7392, "lon": -104.9903, "label": "Denver CO - control" }
  ],
  "includeViolations": true,
  "includeLead": true,
  "includePfas": true,
  "violationYears": 10
}

You can also screen systems directly by PWSID, with or without coordinates:

{ "pwsids": ["MI0002310", "NJ0251001"] }

What that example actually returns (verified live 2026-08-04)

Location	System	Verdict	Findings
Flint MI	MI0002310 FLINT, CITY OF	MODERATE	1 health-based violation (Lead and Copper treatment technique), lead 90th 0.0060 mg/L in the period ending 2025-12-31, 0.020 mg/L historically - above the 0.015 action level - PFAS tested and none detected
Ridgewood NJ	NJ0251001 RIDGEWOOD WATER	HIGH	249 PFAS detections, PFOA 0.0314 ug/L and PFOS both above the 2024 NPDWR MCL screen, Hazard Index 0.494, 1 health-based violation
Denver CO	CO0116001 DENVER WATER BOARD	LOW	128 violations, none health-based, lead 0.0031 mg/L, PFAS tested and none detected, lithium 10.8 ug/L (lithium is not a PFAS)

---

Three things that are easy to get wrong, and how this actor handles them

1. A non-detect is not zero, and neither is "not tested"

97.1% of UCMR5 results (1,872,227 of 1,928,117) are non-detects. EPA writes them as AnalyticalResultsSign = '<' with an empty value: the analyte was looked for and not found above the minimum reporting level. Publishing that as 0 would be a fabricated clean measurement.

All four states are distinguishable on the row:

Situation	pfas_system_tested	pfas_detected	pfas_nondetect_count
Tested, PFAS found	true	true	count of non-detects
Tested, nothing above the reporting level	true	false	count of non-detects
Not a UCMR5 sampling system	false	null	0
The PFAS source failed	null	null	null

pfas_min_reporting_level_ugl tells you what a non-detect actually rules out.

2. Lithium is a UCMR5 analyte and is NOT a PFAS

UCMR5 monitors 30 analytes: 29 PFAS and lithium. Denver detected lithium at 10.8 ug/L and no PFAS at all. Counting lithium as a PFAS detection would report Denver as a PFAS-affected system. Lithium has its own fields (lithium_max_ugl, lithium_detected) and can never enter pfas_detected, pfas_detection_count, the MCL screen or the Hazard Index. A live drift assertion re-checks exactly this against Denver on every run.

3. A lead 90th percentile with no date is worse than no number

LCR_SAMPLE_RESULT has no date column. The monitoring period lives in a separate table, LCR_SAMPLE, joined on sample_id. Without that join you can publish a 2016 result as a system's current lead level. This actor performs the join, reports lead_90th_period_start and lead_90th_period_end, and states the join outcome in lead_period_join_status (complete, partial, failed).

lead_90th_mg_l is the most recent monitoring period. lead_90th_max_ever_mg_l and lead_periods_exceeding_action_level carry the history - Flint's latest period is 0.0060 mg/L while its worst on record is 0.020 mg/L, and both matter to a buyer.

A reported 0 is a real EPA value meaning "at or below the laboratory detection limit at the sampled taps" (26-37% of state LCR rows carry it). It is preserved, never nulled, and flagged with lead_90th_reported_zero so nobody reads it as proof that no lead is present.

---

What each screening level means

Level	Meaning
HIGH	An unresolved health-based violation, a current lead action-level exceedance, a PFAS result above the 2024 NPDWR MCL screen, or a Hazard Index above 1.
MODERATE	A health-based violation on record, a historical lead exceedance, lead at or above the 0.010 mg/L trigger level, a copper exceedance, a Tier 1 public notification, or a PFAS detection below the MCL screen.
LOW	Every requested EPA source answered and none reported any of the above.
NO_SERVICE_AREA	No EPA-mapped community water system service area covers the location. The site is likely served by a private well, or the boundary is not mapped. This is an answer, not a clear: private wells are not regulated under SDWA and warrant their own investigation.
INCONCLUSIVE	At least one requested EPA source did not answer. Real findings are still reported; absence of a finding on a failed source is not evidence of absence.

This is a screening tool, not a regulatory determination. Verify anything material against the primacy agency and the system's Consumer Confidence Report.

About the PFAS MCL comparison

pfas_mcl_screen_exceeded compares the highest UCMR5 detection against the MCL values published in EPA's 2024 PFAS National Primary Drinking Water Regulation (89 FR 32532): PFOA 0.004, PFOS 0.004, PFHxS 0.010, PFNA 0.010, HFPO-DA 0.010 ug/L, plus a Hazard Index over PFHxS, PFNA, HFPO-DA and PFBS.

It is a screening comparison only. UCMR5 reports single entry-point samples; NPDWR compliance is determined by the primacy agency on a running annual average. This actor never states that a system is in violation of a PFAS MCL. The basis is restated on every row in pfas_mcl_screen_basis.

---

Output fields

One row per screened location. Every field below is emitted on every row; an unavailable source yields null, never an absent key and never false.

Location and screening verdict

Field	Type	Description
asset_label	string	Your label for this location, echoed back so results can be joined to your own records.
asset_lat	number	Latitude (WGS84) that was screened. Null for a direct PWSID lookup.
asset_lon	number	Longitude (WGS84) that was screened. Null for a direct PWSID lookup.
asset_index	number	Zero-based position of this location in your input array.
lookup_mode	string	How the water system was identified: 'coordinate' (point-in-polygon against EPA service-area boundaries) or 'pwsid' (you supplied the PWSID directly).
water_quality_screen_level	string	Overall screening level: HIGH, MODERATE, LOW, NO_SERVICE_AREA or INCONCLUSIVE. INCONCLUSIVE whenever any requested EPA source failed - the verdict can never read LOW on data that was never checked. A screening indicator, not a regulatory determination.
screen_flags	array	Every screening flag raised, for example UNRESOLVED_HEALTH_BASED_VIOLATION, LEAD_ACTION_LEVEL_EXCEEDED, PFAS_ABOVE_2024_NPDWR_MCL_SCREEN.
screen_reason	string	Plain-English explanation of the screening level, naming the failed sources when the result is INCONCLUSIVE.
sources_ok	number	How many of the requested EPA sources answered successfully for this location.
sources_requested	number	How many EPA sources were requested for this location (boundary, water system, violations, lead, PFAS).
sources_failed	array	Names of the EPA sources that did not answer. An empty array means every requested source answered.
sources_not_attempted	array	EPA sources that were never attempted because an upstream lookup failed - for example the violation, lead and PFAS legs when the service-area lookup itself went down. Distinct from sources_failed and from a source that genuinely had nothing to look up.
data_complete	boolean	True only when every requested EPA source answered and none was skipped by an upstream failure. False means a finding may be missing.

Serving water system (service-area boundary)

Field	Type	Description
pwsid	string	EPA Public Water System Identification number of the system serving this location (2-letter primacy agency code plus 7 characters).
service_area_pws_name	string	Water system name as published on the EPA service-area boundary layer.
service_area_primacy_agency	string	Primacy agency (usually the state) on the boundary record.
service_area_population_served	number	Population served, as published on the boundary record.
service_area_service_connections	number	Service connection count, as published on the boundary record.
service_area_type	string	Type of service area polygon, for example Residential Area or Municipality.
service_area_feature_type	string	How EPA describes the polygon, for example 'Service Area Boundary'.
service_area_data_provider	string	Who supplied the boundary to EPA.
service_area_data_provider_type	string	Category of the boundary data provider, for example Private Company or State Agency.
service_area_data_source_url	string	Source URL EPA cites for this boundary.
service_area_model_method	string	Modelling method used to derive the boundary, when EPA records one.
service_area_modification_method	string	How EPA modified the supplied boundary, when applicable.
service_area_method_details	string	Free-text detail on the boundary derivation method.
service_area_verification_status	string	EPA's verification status for the boundary, when recorded.
service_area_confirmed	string	Whether the boundary has been confirmed, as recorded by EPA.
service_area_km2	number	Area of the service-area polygon in square kilometres. Useful for judging boundary precision.
service_area_vintage	string	Date the boundary was originally created. Boundaries vary widely in age - some date to 2004.
echo_detailed_facility_report_url	string	Link to EPA ECHO's Detailed Facility Report for this system.
boundary_error	string	Error message if the service-area lookup failed. Null on success.
service_area_found	boolean	True if a mapped EPA service area covers this location, false if none does, null if the boundary lookup failed. False and null are different answers.
service_areas_matched	number	How many mapped service areas cover this location. More than one is legitimate (a wholesaler plus its purchaser, or overlapping vintages).
all_pwsids	array	Every PWSID whose mapped service area covers this location, primary first.
all_pws_names	array	Names of every matching water system, in the same order as all_pwsids.
boundary_status	string	Outcome of the service-area lookup: 'ok' or 'unavailable'.
boundary_source	string	Attribution for the service-area boundary layer.
boundary_layer_national_count	number	Total service-area records in EPA's national layer at run time, measured live as a truncation and drift guard.

Water system record (SDWIS)

Field	Type	Description
pws_name	string	Water system name from EPA SDWIS.
pws_type_code	string	SDWIS water system type code (CWS, NTNCWS, TNCWS, NP, U).
pws_type	string	Decoded water system type, for example 'Community water system'.
pws_activity_code	string	SDWIS activity code (A active, I inactive, and others).
pws_activity	string	Decoded activity status, for example 'Active'.
pws_is_active	boolean	True if SDWIS marks the system active, false if not, null if the water system record was not retrieved.
pws_deactivation_date	string	Date the system was deactivated in SDWIS, if applicable.
primacy_agency_code	string	Two-character primacy agency code (usually the state).
primacy_agency	string	Decoded primacy agency name.
primacy_type	string	Whether primacy is held by the State, EPA or a Tribe.
epa_region	string	EPA region number for the system.
state_code	string	State code recorded on the SDWIS water system record.
population_served_count	number	Population served, from SDWIS. May differ from the boundary layer's figure.
service_connections_count	number	Number of service connections, from SDWIS.
owner_type_code	string	SDWIS owner type code (F, L, M, N, P, S).
owner_type	string	Decoded owner type, for example 'Local government' or 'Private'.
primary_source_code	string	SDWIS primary water source code (GW, GWP, SW, SWP, GU, GUP).
primary_source	string	Decoded primary water source, for example 'Surface water purchased'.
gw_sw_code	string	Ground water / surface water code.
gw_sw	string	Decoded ground water or surface water designation.
is_wholesaler	boolean	True if the system sells water to other systems, false if not, null if not recorded.
is_school_or_daycare	boolean	True if the system serves a school or daycare, false if not, null if not recorded.
is_grant_eligible	boolean	True if EPA records the system as grant eligible.
source_water_protection	boolean	Whether a source-water protection programme is recorded for the system.
outstanding_performer	string	EPA 'outstanding performer' designation, when recorded.
pws_address	string	Mailing address of the water system's administrative office.
pws_city	string	City of the water system's administrative office.
pws_zip	string	ZIP code of the water system's administrative office.
pws_admin_name	string	Administrative contact recorded in SDWIS for the utility.
pws_org_name	string	Organisation contact name recorded in SDWIS for the utility.
pws_phone	string	Utility contact phone number published in SDWIS.
pws_email	string	Utility contact email published in SDWIS.
season_begin_date	string	Start of the operating season for seasonal systems.
season_end_date	string	End of the operating season for seasonal systems.
counties_served	array	Counties the system serves, from SDWIS GEOGRAPHIC_AREA.
cities_served	array	Cities the system serves, from SDWIS GEOGRAPHIC_AREA.
zip_codes_served	array	ZIP codes the system serves, from SDWIS GEOGRAPHIC_AREA.
geographic_area_types	array	Kinds of geographic area recorded for the system (County, City, ZIP Code, Tribal, Indian Reservation).
geographic_area_rows	number	Number of GEOGRAPHIC_AREA rows retrieved for the system.
water_system_record_found	boolean	True if SDWIS holds a WATER_SYSTEM record for this PWSID, false if the lookup succeeded and returned none, null if the lookup failed.
water_system_rows_expected	number	Row count EPA's own COUNT endpoint reported for the WATER_SYSTEM query, used as an exact truncation guard.
geographic_area_status	string	Outcome of the GEOGRAPHIC_AREA lookup: 'ok' or 'unavailable'.
system_error	string	Error message if the water system lookup failed. Null on success.
system_status	string	Outcome of the water system leg: 'ok', 'unavailable', 'skipped' or 'not_applicable'.

Safe Drinking Water Act violations

Field	Type	Description
violations_total	number	Total Safe Drinking Water Act violations on record for this system, all years and all categories.
violations_health_based	number	Violations EPA itself flags as health-based (is_health_based_ind = Y): MCL, MRDL and treatment-technique violations. Not derived here.
violations_health_based_in_window	number	Health-based violations whose compliance period falls inside your violationYears window. Null when the window is disabled.
violations_health_based_unresolved	number	Health-based violations SDWIS still marks Open. This is the flag most Phase I reviewers care about.
violations_health_based_known_unreturned	number	Health-based violations with SDWIS compliance status 'Known' - recorded but not returned to compliance. Reported separately from Open rather than lumped in, because EPA distinguishes them.
violations_health_based_returned_to_compliance	number	Health-based violations SDWIS marks Returned to Compliance.
violations_unresolved_total	number	All violations, health-based or not, SDWIS still marks Open.
violations_by_category	object	Violation counts keyed by SDWIS violation_category_code (MCL, MRDL, TT, MR, MON, RPT, Other).
violations_by_compliance_status	object	Violation counts keyed by SDWIS compliance_status_code (O Open, K Known, R Returned to Compliance, I System Inactive).
violations_mcl	number	Maximum Contaminant Level violations.
violations_mrdl	number	Maximum Residual Disinfectant Level violations.
violations_treatment_technique	number	Treatment technique violations.
violations_monitoring_reporting	number	Monitoring and reporting violations (categories MR, MON and RPT combined).
violations_lead_copper_rule	number	Violations under the Lead and Copper Rule family (rule family 350).
violations_public_notification_tier1	number	Violations that carried a Tier 1 public notification - the most urgent tier, requiring notice within 24 hours.
violations_earliest_date	string	Earliest violation compliance-period date on record.
violations_latest_date	string	Most recent violation compliance-period date on record.
health_based_contaminants	array	Distinct contaminants named on the system's health-based violations, decoded using EPA's own REF_CODE_VALUE table. Null when EPA publishes no description for the codes involved - see health_based_contaminant_codes, which is always populated.
health_based_contaminant_codes	array	Raw SDWIS contaminant codes on the system's health-based violations. Always populated when health-based violations exist. EPA's reference table omits several modern codes (5200, the Revised Lead and Copper Rule contaminant, is one), so the raw code is authoritative and the decoded name may be null.
health_based_contaminants_undecoded	number	How many of the health-based contaminant codes EPA publishes no description for. Non-zero is normal and reflects a gap in EPA's own reference table, not a gap in the violation record.
most_recent_health_based_violation	object	The most recent health-based violation as a full object (type, category, contaminant, rule, compliance status, dates, public notification tier).
most_recent_health_based_violation_date	string	Compliance-period date of the most recent health-based violation.
health_based_violation_details	array	Up to maxViolationDetails individual health-based violation records, newest first, each fully decoded.
health_based_violation_details_truncated	boolean	True when more health-based violations exist than were included in the details array. Counts above are never truncated.
enforcement_actions_total	number	Number of formal SDWIS enforcement actions on record for this system.
most_recent_enforcement_date	string	Date of the most recent formal enforcement action.
most_recent_enforcement_type	string	Decoded type of the most recent formal enforcement action.
most_recent_enforcement_type_code	string	Raw SDWIS enforcement action type code for the most recent action.
violations_rows_expected	number	Row count EPA's own COUNT endpoint reported for the violation query, used as an exact truncation guard.
violations_rows_retrieved	number	Violation rows actually retrieved. A shortfall against violations_rows_expected fails the leg rather than publishing a rollup over a subset.
enforcement_status	string	Outcome of the enforcement lookup: 'ok', 'unavailable' or 'skipped'.
violations_error	string	Error message if the violation lookup failed. Null on success.
violations_status	string	Outcome of the violations leg: 'ok', 'unavailable', 'skipped' or 'not_applicable'.

Lead and copper (LCR 90th percentile)

Field	Type	Description
lead_90th_mg_l	number	Lead and Copper Rule 90th-percentile tap result. Value for the MOST RECENT monitoring period, in mg/L. Reported by the water system under 40 CFR 141 Subpart I.
lead_90th_period_start	string	Start of the monitoring period the reported lead 90th percentile belongs to. EPA publishes the result and the period in separate tables; this actor joins them, because an undated 90th percentile can present a decade-old result as current.
lead_90th_period_end	string	End of the monitoring period the reported lead 90th percentile belongs to.
lead_90th_result_sign_code	string	EPA result sign code on the lead result. 'L' means the value is an upper bound (below minimum detection limits), 'E' means exactly equal to the value reported. Null means EPA published no sign, which is the case on every row observed to date.
lead_90th_below_detection	boolean	True when the lead result sign code is 'L' and the reported number is therefore an upper bound rather than a measurement. Null when EPA published no sign code.
lead_90th_reported_zero	boolean	True when EPA published the lead 90th percentile as exactly 0. That is EPA's encoding of 'at or below the laboratory detection limit at the sampled taps' - it is a reported value, not a missing one, and it is not proof that no lead is present. Between 26% and 37% of state LCR rows carry it.
lead_action_level_mg_l	number	The federal lead action level, 0.015 mg/L. Constant, included so a row is self-describing.
lead_action_level_exceeded	boolean	True if the most recent lead 90th percentile exceeds 0.015 mg/L, false if it does not, null if the lead leg did not answer or the system has no reported lead result.
lead_90th_max_ever_mg_l	number	Highest lead 90th percentile on record for this system across all monitoring periods.
lead_90th_max_ever_period_end	string	End of the monitoring period in which the highest lead 90th percentile was reported.
lead_periods_exceeding_action_level	number	How many distinct monitoring periods exceeded the lead action level. A system can be compliant today and have exceeded repeatedly in the past.
lead_monitoring_periods	number	Number of distinct lead monitoring periods on record.
lead_results_reported	number	Number of lead 90th-percentile results retrieved for this system.
copper_90th_mg_l	number	Copper 90th-percentile tap result for the most recent monitoring period, in mg/L.
copper_90th_period_start	string	Start of the monitoring period the reported copper 90th percentile belongs to.
copper_90th_period_end	string	End of the monitoring period the reported copper 90th percentile belongs to.
copper_90th_result_sign_code	string	EPA result sign code on the copper result ('L' below detection, 'E' exactly equal). Null when EPA published no sign.
copper_90th_below_detection	boolean	True when the copper result is an upper bound rather than a measurement. Null when EPA published no sign code.
copper_90th_reported_zero	boolean	True when EPA published the copper 90th percentile as exactly 0 - a reported value meaning at or below detection, not a missing one.
copper_action_level_mg_l	number	The federal copper action level, 1.3 mg/L. Constant, included so a row is self-describing.
copper_action_level_exceeded	boolean	True if the most recent copper 90th percentile exceeds 1.3 mg/L, false if it does not, null if the leg did not answer or no result exists.
copper_90th_max_ever_mg_l	number	Highest copper 90th percentile on record for this system.
copper_90th_max_ever_period_end	string	End of the monitoring period in which the highest copper 90th percentile was reported.
copper_periods_exceeding_action_level	number	How many distinct monitoring periods exceeded the copper action level.
copper_monitoring_periods	number	Number of distinct copper monitoring periods on record.
copper_results_reported	number	Number of copper 90th-percentile results retrieved for this system.
lcr_results_total	number	Total Lead and Copper Rule sample-result rows retrieved for this system (lead and copper combined).
lead_period_join_status	string	How well the lead and copper results joined to their monitoring periods: 'complete', 'partial', 'failed', 'no_results' or 'not_attempted'. Anything but 'complete' means some reported values are undated.
lead_period_join_matched	number	How many result rows were successfully matched to a monitoring period.
lcr_rows_expected	number	Row count EPA's own COUNT endpoint reported for the lead and copper query, used as an exact truncation guard.
lcr_sample_period_error	string	Error message if the monitoring-period lookup failed while the results themselves succeeded.
lead_error	string	Error message if the lead and copper leg failed. Null on success.
lead_status	string	Outcome of the lead and copper leg: 'ok', 'unavailable', 'skipped' or 'not_applicable'.

PFAS (UCMR5)

Field	Type	Description
pfas_ucmr5_vintage	string	Publication timestamp of the UCMR5 occurrence file actually used, read from its Last-Modified header. The download URL contains '2023-08' but EPA replaces the file in place, so the path is NOT the vintage.
pfas_ucmr5_source_url	string	URL of the EPA UCMR5 occurrence dataset used for the PFAS screen.
pfas_ucmr5_national_rows	number	Total UCMR5 result rows indexed on this run, measured live as a truncation guard.
pfas_ucmr5_national_systems	number	Total public water systems present in UCMR5 on this run.
pfas_mcl_screen_basis	string	Plain statement of what the PFAS MCL comparison is and is not: a screening comparison against EPA's 2024 PFAS NPDWR values, not a compliance determination, because UCMR5 reports single entry-point samples rather than the running annual average used for compliance.
pfas_system_tested	boolean	True if this system appears in UCMR5 at all, false if it does not (most small systems were not in the UCMR5 sample), null if the PFAS leg failed. False and null are different answers.
pfas_results_count	number	Number of UCMR5 result rows for this system, detections and non-detects combined.
pfas_analytes_tested	number	Number of distinct UCMR5 analytes this system was tested for (up to 30: 29 PFAS plus lithium).
pfas_sample_points	number	Number of distinct entry-point sample locations UCMR5 holds for this system.
pfas_nondetect_count	number	Number of UCMR5 results reported as non-detect - the analyte was looked for and not found above the minimum reporting level. A non-detect is NOT zero and NOT untested.
pfas_min_reporting_level_ugl	number	Lowest minimum reporting level applied to this system's UCMR5 samples, in micrograms per litre. A non-detect only rules out concentrations above this level.
pfas_first_collection_date	string	Earliest UCMR5 sample collection date for this system.
pfas_last_collection_date	string	Most recent UCMR5 sample collection date for this system.
pfas_detected	boolean	True if any PFAS analyte was detected, false if the system was tested and no PFAS was detected, null if the PFAS leg did not answer. Lithium is a UCMR5 analyte but is NOT a PFAS and never sets this true.
pfas_detection_count	number	Number of individual PFAS detections (excluding lithium) across all of this system's UCMR5 samples.
pfas_analytes_detected	array	Names of the PFAS analytes detected at this system, alphabetically. Lithium is deliberately excluded.
pfas_max_analyte	string	PFAS analyte with the highest detected concentration at this system.
pfas_max_ugl	number	Highest single PFAS concentration detected at this system, in micrograms per litre.
pfoa_max_ugl	number	Highest detected PFOA concentration, in micrograms per litre. The 2024 NPDWR MCL is 0.004 ug/L.
pfos_max_ugl	number	Highest detected PFOS concentration, in micrograms per litre. The 2024 NPDWR MCL is 0.004 ug/L.
pfhxs_max_ugl	number	Highest detected PFHxS concentration, in micrograms per litre. The 2024 NPDWR MCL is 0.010 ug/L.
pfna_max_ugl	number	Highest detected PFNA concentration, in micrograms per litre. The 2024 NPDWR MCL is 0.010 ug/L.
hfpo_da_max_ugl	number	Highest detected HFPO-DA (GenX chemicals) concentration, in micrograms per litre. The 2024 NPDWR MCL is 0.010 ug/L.
pfbs_max_ugl	number	Highest detected PFBS concentration, in micrograms per litre. PFBS has no individual MCL; it is one of the four Hazard Index analytes.
pfas_hazard_index_screen	number	Unitless Hazard Index computed over PFHxS, PFNA, HFPO-DA and PFBS against their published health-based water concentrations. Null when the system was not tested for any Hazard Index analyte.
pfas_hazard_index_screen_exceeded	boolean	True if the Hazard Index screen exceeds 1.0, false if it does not, null if it could not be computed.
pfas_mcl_screen_exceedances	array	PFAS analytes whose highest detected concentration exceeds the corresponding 2024 NPDWR MCL. A screening comparison only - see pfas_mcl_screen_basis.
pfas_mcl_screen_exceeded	boolean	True if any detected PFAS exceeds its 2024 NPDWR MCL on this screening comparison, false if none does, null if the PFAS leg did not answer.
lithium_max_ugl	number	Highest detected lithium concentration, in micrograms per litre. Lithium is a UCMR5 analyte and is NOT a PFAS; it is reported separately so it can never inflate a PFAS count.
lithium_detected	boolean	True if lithium was detected, false if the system was tested for lithium and none was detected, null if lithium was not among the analytes tested or the leg did not answer.
pfas_error	string	Error message if the PFAS leg failed. Null on success.
pfas_status	string	Outcome of the PFAS leg: 'ok', 'unavailable', 'skipped' or 'not_applicable'.

Provenance and source integrity

Field	Type	Description
retrieved_at	string	UTC timestamp when this run queried EPA.
sdwis_source	string	Attribution for the tabular data (EPA Envirofacts SDWIS).
actor_version	string	Version of this actor that produced the row.
drift_status	string	Result of this run's live source-integrity checks: 'ok' or 'degraded'. A 'failed' status fails the run before any row is emitted.
drift_measurements	string	JSON string of every value measured by the live drift checks on this run (national record counts, canary results, UCMR5 vintage and vocabulary), so source health is auditable.

Total: 186 declared fields, every one of them documented above.

---

Fields that are legitimately null on a healthy run

Some fields are null because the answer is genuinely absent, not because something failed. Each has a recorded input that populates it.

Field(s)	Why null, and what populates it
boundary_error, system_error, violations_error, lead_error, pfas_error, lcr_sample_period_error	Null is good news - the source answered. Populate with {"simulateOutage": "lead"} or any other leg name.
copper_90th_*, copper_action_level_exceeded, copper_periods_exceeding_action_level	Most systems report lead only. Populate with {"pwsids": ["RI1583829"]} - 8 copper results to 2 lead.
health_based_contaminants	EPA's own REF_CODE_VALUE table omits several modern codes, so the decoded name can be null while health_based_contaminant_codes is populated. See below.
lead_90th_result_sign_code, lead_90th_below_detection, and the copper equivalents	EPA publishes no result sign on any LCR row observed to date. The vocabulary (E equal, L below detection) is asserted live, so the day it starts appearing the value is carried through rather than ignored.
pws_deactivation_date, season_begin_date, season_end_date, outstanding_performer	Only populated for deactivated, seasonal or EPA-designated systems.
hfpo_da_max_ugl, pfna_max_ugl and the other per-analyte maxima	Null when that analyte was never detected at the system. pfas_analytes_tested confirms it was looked for.
sources_failed, sources_not_attempted	Empty arrays on a healthy run.
zip_codes_served, cities_served	Many states report only the county in SDWIS GEOGRAPHIC_AREA.

A known gap in EPA's own reference table

REF_CODE_VALUE is EPA's authoritative code decoder and this actor fetches it on every run - but it is materially incomplete for newer SDWIS codes. Measured live on 2026-08-04 across Michigan's 63,401 violations: 18 alphanumeric violation codes are absent from it (3A alone appears 22,955 times), along with 8 contaminant codes including 5200 and rule code 351, the Revised Lead and Copper Rule codes.

This actor never invents a label. The raw code is always emitted next to the decoded name, health_based_contaminants_undecoded counts how many labels EPA does not publish, and every undecoded code is listed in the run log. The one documented exception is violation_category_code MON and RPT, which live data carries and EPA's table omits; those two are supplied locally and the substitution is stated in the source.

---

Live source-integrity checks

Offline tests cannot see a silently degraded upstream, because the fixtures were shaped from the degraded data. Every run therefore performs 23 live drift assertions before a single billable row is emitted, and logs every measured value. A critical failure fails the run and bills nothing.

Among them:

• The boundary layer resolves, still exposes all 20 fields this actor requests, holds a national record count in band (44,615 measured), and answers in WGS84.
• Positive canary: the point 43.0125, -83.6875 must resolve to MI0002310.
• Negative control: a point in the open Pacific must match no service area. This is what catches a query that silently matches nothing.
• The legacy SDWIS table names must still answer, with national counts in band (WATER_SYSTEM 434,040, VIOLATION 2,098,792, LCR_SAMPLE_RESULT 279,257).
• MI0002310 must still show a health-based violation and a lead result at or above the action level. NC0392373 must still show PFHxS at or above 0.05 ug/L.
• The lithium trap: CO0116001 must still show lithium detected and zero PFAS detections.
• UCMR5 must hold 30 analytes, all classified, with a closed sign vocabulary and units that decode to the micro sign rather than U+FFFD.
• The UCMR5 vintage must come from the file's Last-Modified header. The download URL contains 2023-08, but EPA replaces the file in place - the copy served on 2026-08-04 was last modified 2026-02-12. The real vintage is published on every row in pfas_ucmr5_vintage.

---

Failure behaviour

Situation	What happens
No usable assets or pwsids	The run FAILS with a message naming the field and an example value. Nothing emitted, nothing billed.
A coordinate is out of range	That location is rejected, named in the log, and no row is emitted for it - you are not billed for an unusable input. Other locations still run.
One EPA source is down	The row is still emitted with that leg's fields null, <leg>_status set to unavailable, and the verdict forced to INCONCLUSIVE.
Every source is down for every location	Buffered rows are discarded before any push and the run FAILS. Zero rows, zero billable events.
A live drift assertion fails critically	The run fails before the first push. Zero rows, zero billable events.
Upstream answered and there is genuinely nothing	SUCCEEDED with an honest row - for example NO_SERVICE_AREA, or a system with zero violations.

You can prove any of this yourself with the simulateOutage input.

---

Performance and the PFAS cache

UCMR5_All.txt is 308,819,534 bytes uncompressed. It is streamed and filtered rather than materialised, and the per-system index is cached in a named key-value store keyed to the file's Last-Modified vintage. The first run in a new vintage rebuilds the index in about 40 seconds; later runs reuse it. EPA republishing the file invalidates the cache automatically - no manual refresh is needed, though refreshPfasCache: true forces one.

Set includePfas: false to skip that leg entirely when PFAS is out of scope.

---

Use as an MCP tool

This actor is callable by AI agents through mcp.apify.com. Every output field carries a description, and the per-source status fields let an agent distinguish "EPA reports no health-based violations" from "the violation source did not answer" - a distinction most tools collapse, and one that matters a great deal in an environmental screen.

---

Pricing

Pay per result: $12 per 1,000 rows, one row per screened location. Graduated paid-plan discounts apply (Bronze 20%, Silver 30%, Gold 45%, Platinum 60%, Diamond 70%).

A run that fails loudly emits zero rows and bills zero result events.

---

Related actors

Same buyer, same due-diligence workflow:

• EPA Contaminated Site Screener - Superfund, RCRA, USTs, spills and ECHO enforcement within a radius of the same coordinate.
• Site Due Diligence Bundle - a 20-layer federation for one coordinate in one call, covering contamination, wetlands, flood, seismic and habitat.
• EPA Impaired Waters 303(d) Screener - surface-water impairments and TMDLs near the site; the surface-water counterpart to this actor's drinking-water view.

---

FAQ

How do I find out which water utility serves an address? Pass the address coordinates in assets. The actor returns pwsid, pws_name, population_served_count, the counties served, and the utility's published contact details.

How do I check a water system for lead? lead_90th_mg_l is the most recent 90th-percentile tap result with its monitoring period. lead_action_level_exceeded compares it to the 0.015 mg/L federal action level. lead_90th_max_ever_mg_l and lead_periods_exceeding_action_level give the history, which is what a system like Flint makes clear: currently compliant, historically not.

How do I check whether a water system has PFAS? pfas_detected, pfas_analytes_detected and pfas_max_ugl come from EPA's UCMR5 monitoring. pfas_mcl_screen_exceeded compares detections against the 2024 NPDWR MCL values as a screen. If pfas_system_tested is false, the system was simply not in the UCMR5 sample - that is not a clean result.

What if my site is on a private well? You get a NO_SERVICE_AREA row saying so explicitly. SDWA does not regulate private wells, so a drinking-water screen from SDWIS cannot cover the site.

Does this cover every US water system? SDWIS covers all regulated public water systems. The service-area boundary layer maps roughly 44,600 of them, so a coordinate lookup can miss a system whose boundary EPA has not published. Use pwsids to screen a specific system directly when you already know it.

Is a lead result of 0 mg/L proof there is no lead? No. EPA uses 0 to mean "at or below the laboratory detection limit at the sampled taps". The row flags it with lead_90th_reported_zero.

Why does a row say INCONCLUSIVE when most of it looks fine? Because at least one requested EPA source did not answer. The findings you can see are real; what you cannot see was never checked. sources_failed names the legs, and data_complete is false.

Can I use this for a Phase I ESA? It is designed for that screening step and reports its own data completeness on every row. It is a screening tool, not a determination - verify material findings with the primacy agency and the system's Consumer Confidence Report.

Is an API key needed? No. Every source is a public, keyless EPA endpoint.