Orphaned Oil & Gas Well Proximity Screener (Phase I ESA / Land Development)

Give it a coordinate and a radius. It tells you how many documented orphaned oil and gas wells the USGS national dataset places inside that radius, exactly how far away the nearest one is and in what direction, what the state's own record says about each well, how accurate each well's published location actually is, and a transparent 0-100 screening score.

Built on the USGS US Documented Unplugged Orphaned Oil and Gas Well Dataset (DOI 10.5066/P91PJETI) - 117,672 wells across 27 states, compiled from the state oil and gas regulators themselves. No API key, no scraping, no login.

Who this is for

• Phase I ESA consultants. An undocumented orphaned well on or beside a site is a real recognised environmental condition: a possible vapour and groundwater pathway, and a plugging liability that can land on the current owner. This is a fast screen for that, with the well identifiers you need to pull the state file, and it is honest about the roughly one-third of the dataset where the state does not say whether the well was ever plugged.
• Land developers and homebuilders. A well under a planned foundation is a design problem, not a paperwork problem. Screen a whole land pipeline in one run and rank it.
• Solar, wind, battery and carbon-project developers. Orphaned wells inside a project boundary drive setbacks, and for carbon storage they are a leakage pathway that regulators ask about directly.
• Lenders, insurers and title teams doing pre-close environmental screening.
• AI agents. See "Use as an MCP tool" below.

What makes this different from a map click

1. It is honest about coverage. The dataset holds 27 states. A site in Florida returns zero wells - but that is no data, not no wells. This actor resolves each site's own state (FCC census-block API, Census TIGERweb as fallback) and reports out_of_dataset_coverage instead of publishing a clear. phase1_esa_further_inquiry_flag is only ever false on a completed query inside a covered state.
2. It is honest about positional accuracy. 795 wells in this dataset are located only to a PLSS section centroid - a one-square-mile box, so a reported distance can be wrong by 0.7 mile. 158 are permitted locations that may never have been drilled. 285 the source itself flags as questionable. 7,685 have an unconfirmed horizontal datum, which alone is worth up to about 200 m. Every well ships a positional_class and an uncertainty estimate.
3. It does not guess at codes. USGS republishes each state's own status text verbatim and publishes no codebook for the opaque two-letter codes. Codes are normalized only where the literal English is unambiguous or the dataset documents them; everything else normalizes to null with basis: unmapped_state_code, and the raw value is always emitted. plugged_per_state_record is true for exactly three status values and null - never false - wherever the record does not actually say.
4. A short read is an error, not an answer. The service caps a page at 2,000 features. Before paging, the actor asks the service for its own exact count, then pages with keyset pagination until it has all of them. If it cannot, the site fails loudly rather than reporting "the nearest well" computed off an arbitrary subset.
5. It verifies the upstream on every run before emitting a single billable row: field set, national count floor, live state coverage, status vocabulary, two positive canaries, two negative controls, an explicit projection check and a true-circle fidelity test. Every measured value is printed to the log.

Example input

{
  "assets": [
    { "lat": 34.07, "lon": -118.29, "label": "Los Angeles City oil field - infill parcel, CA", "radiusMiles": 0.5 },
    { "lat": 41.4651, "lon": -79.6504, "label": "Oil City brownfield, Venango County PA", "radiusMiles": 1 },
    { "lat": 39.7392, "lon": -104.9903, "label": "Denver CO downtown control site", "radiusMiles": 1 },
    { "lat": 25.7617, "lon": -80.1918, "label": "Miami FL control site (outside dataset coverage)", "radiusMiles": 1 }
  ],
  "includeWellDetail": true,
  "maxWellsPerAsset": 10
}

Those four sites are the shipped prefill and they demonstrate all four outcomes: the Los Angeles parcel sits on the Los Angeles City oil field (289 documented orphaned wells within half a mile), Oil City PA returns 63 within a mile, Denver returns a verified clear inside a covered state, and Miami returns out of dataset coverage - the same zero, with a completely different meaning.

Input reference

Input	What it does
assets (required)	Array of {lat, lon, label, radiusMiles}. radiusMiles is optional per site and overrides the run-level radius. latitude/longitude/lng are accepted as aliases. Omitting assets entirely returns HTTP 400 before the run starts, so nothing is billed.
radiusMiles	Run-level search radius in statute miles, 1-25. Per-site values may be fractional (0.25, 0.5). Defaults to 1 mile.
includeWellDetail	Emit one row per well in addition to the site summary. Off by default.
maxWellsPerAsset	Cap on per-well rows, nearest kept first. Defaults to 100.
maxAssets	Cap on sites per run. Defaults to 50.
resolveAssetState	Resolve each site's state to test dataset coverage. On by default.
skipDriftChecks	Diagnostic. Skips the live upstream integrity checks.
simulateOutage	Diagnostic seam (none/wells/state/both) to prove the failure path on demand.

Note: no input carries a schema default, deliberately. An Apify input-schema default is injected server-side into every run, so a default on a predicate field silently narrows queries the caller never asked for. Defaults live in code.

Output

Two record types share one flat schema, distinguished by record_type:

• asset_summary - exactly one row per screened site, always.
• well - one row per documented orphaned well inside the radius, nearest first. Only emitted when includeWellDetail is on.

Fields that belong to the other record type are null, and every field is nullable.

The null vs false contract

This is the most important thing to understand about the output:

• null means "not checked" - the source did not answer, the state does not publish it, or the band lies outside the radius you searched.
• false means "checked, and the answer is negative."

They are never collapsed. A run where the well layer fails emits orphaned_wells_present: null, not false. A distance band wider than your radius is null, not 0. A state record that does not state a plugging status gives plugged_per_state_record: null, not false. If every site fails, the run fails and nothing is billed.

Output fields

Site identity and run context

Field	Type	Description
record_type	string	asset_summary for the one row per screened site, or well for a per-well detail row (only emitted when includeWellDetail is on).
asset_label	string	The label you supplied for this site, echoed on every row so results can be joined back to your portfolio.
asset_lat	number	Site latitude in WGS84 decimal degrees, as supplied.
asset_lon	number	Site longitude in WGS84 decimal degrees, as supplied.
search_radius_miles	number	The radius actually searched for this site, in statute miles. This is a true circle, not a bounding box.
checked_at	string	UTC timestamp of the run that produced this row.
dataset_source	string	The upstream dataset and layer this answer came from.
dataset_doi	string	Permanent DOI for the USGS data release, for citation in a Phase I report.
dataset_national_well_count	number	Total wells in the live national dataset, measured on this run. 117,672 on 2026-08-04. Null if the drift checks were skipped.
dataset_states_covered_count	number	How many states the live dataset actually contains, measured on this run rather than hard-coded.
dataset_states_covered	string	The states the dataset covers, measured live. A site outside this list cannot be screened by this dataset and is reported as out_of_dataset_coverage, never as clear.
dataset_vintage_earliest	string	Earliest per-state data file date in the whole dataset. The compilation is a snapshot of state records, not a live feed.
dataset_vintage_latest	string	Latest per-state data file date in the whole dataset.
risk_model_version	string	Version of the screening score formula, so a future change to the model is detectable in your own records.
error	string	Null on a healthy row. Populated when this site could not be screened; the row then asserts nothing about wells.

Source health, completeness and coverage

Field	Type	Description
wells_source_status	string	ok when the orphaned-well layer answered completely for this site; unavailable when it did not. Every well-derived field on an unavailable row is null, never false.
wells_source_error	string	Why the orphaned-well layer could not answer for this site. Null on a healthy row.
wells_reported_by_source	number	Exact well count the service itself reports inside the radius, read from its own returnCountOnly answer. This is the truncation guard.
wells_retrieved	number	How many well records were actually retrieved and mapped. If this is ever below wells_reported_by_source the site fails rather than answering from a partial set.
query_complete	boolean	True only when wells_retrieved equals wells_reported_by_source. A silently truncated ArcGIS read is the single most common cause of a wrong proximity answer.
wells_dropped_outside_radius	number	Wells the service returned that a haversine re-check placed outside the requested radius. Expected to be 0; a non-zero value means the service stopped honouring a true-circle query.
asset_state	string	The US state the site itself falls in, resolved independently of the well data. Used only to decide whether the dataset covers this site.
asset_state_source	string	Which resolver answered: fcc_census_block or census_tigerweb.
asset_state_status	string	ok when a resolver answered, unavailable when both failed. When unavailable, in_dataset_coverage stays null and a zero-well result is reported as coverage_unknown rather than as clear.
asset_state_error	string	Why the state could not be resolved, or a note that the coordinate is not inside any US state.
asset_county	string	County containing the site, when the FCC resolver supplied one.
in_dataset_coverage	boolean	True when the site is in one of the states the dataset actually contains, false when it is not, and null when the state could not be resolved. Never guessed.
coverage_basis	string	How in_dataset_coverage was decided: measured_from_live_dataset, asset_state_unresolved, asset_state_lookup_disabled, or covered_state_list_unavailable.

The screening answer

Field	Type	Description
screening_result	string	wells_found
orphaned_wells_present	boolean	True when documented orphaned wells were found in the radius, false only on a completed query inside a covered state, and null whenever the question was not actually answered.
phase1_esa_further_inquiry_flag	boolean	True when at least one documented orphaned well sits inside the radius, meaning the site warrants further inquiry in a Phase I ESA. False only on a verified clear inside a covered state. Null when coverage or the source is in doubt. Screening only, not a determination.
well_count_within_radius	number	Documented orphaned wells inside the search radius. Null when the query did not complete.
wells_within_500_ft	number	Wells within 500 feet. Null when the search radius is smaller than this band, because a band you did not search is not zero.
wells_within_0_25_mile	number	Wells within 0.25 mile. Null when the search radius is smaller than 0.25 mile.
wells_within_0_5_mile	number	Wells within 0.5 mile. Null when the search radius is smaller than 0.5 mile.
wells_within_1_mile	number	Wells within 1 mile. Null when the search radius is smaller than 1 mile.
risk_score	number	Screening score 0-100. 0 means a verified clear. Null whenever the screen did not complete. The full formula is documented in the README and versioned by risk_model_version.
risk_level	string	none (0), low (1-19), moderate (20-44), elevated (45-69) or high (70-100). Null when the screen did not complete.
risk_reasons	string	Plain-language list of exactly which factors contributed to the score.

The nearest well

Field	Type	Description
nearest_distance_miles	number	Great-circle distance from the site to the nearest documented orphaned well, in statute miles.
nearest_distance_feet	number	The same distance in feet, for site plans and Phase I text.
nearest_bearing_degrees	number	Initial great-circle bearing from the site to the nearest well, 0-360 degrees clockwise from true north.
nearest_bearing_compass	string	The same bearing as a 16-point compass direction.
nearest_well_id	string	Full identifier of the nearest well as published, including its system prefix.
nearest_well_id_system	string	Identifier system: API (110,976 rows), USGSID (6,121) or IGSID (575).
nearest_well_name	string	Lease or well name of the nearest well as the state records it.
nearest_well_number	string	Well number within the lease. Null on the 64,214 rows where the state publishes none.
nearest_well_state	string	State whose records the nearest well came from.
nearest_well_county	string	County of the nearest well.
nearest_well_status_raw	string	The state's own status text for the nearest well, verbatim and unmodified. This is the authoritative value.
nearest_well_status_normalized	string	Normalized bucket: orphan, potential_orphan, abandoned, idle, shut_in, suspended, active, forfeited, canceled, remediation_pending, unknown or plugged. Null when the state uses an opaque code USGS publishes no codebook for.
nearest_well_status_basis	string	How the bucket was derived: literal_text, documented_in_dataset, unmapped_state_code, or blank_sentinel. Nothing is guessed.
nearest_well_plugged_per_state_record	boolean	True only when the state record literally says Plugged, PA or P&A. False when the status unambiguously says otherwise. Null when the record does not say - which is the case for about a third of the dataset.
nearest_well_type_raw	string	The state's own well type text, verbatim. Null on the 59,909 rows where the state publishes none.
nearest_well_type_normalized	string	Coarse bucket: oil, gas, oil_and_gas, injection_disposal, coalbed_methane, storage, exploratory_stratigraphic, dry_hole, observation_monitoring, water or permitted_location_only. Null for opaque state codes.
nearest_well_injection_or_disposal	boolean	True when the type text identifies an injection, disposal, brine, waterflood or EOR well - the types most relevant to a subsurface contamination pathway. Null when the type is not stated.
nearest_well_latitude	number	Latitude of the nearest well as USGS publishes it (NAD83 decimal degrees).
nearest_well_longitude	number	Longitude of the nearest well as USGS publishes it.
nearest_well_positional_class	string	How the coordinate was derived: surveyed_or_reported, datum_converted, datum_unconfirmed, third_party_derived, plss_quarter_quarter, plss_quarter, plss_section, planned_not_actual, questionable, or unspecified.
nearest_well_positional_uncertainty_miles	number	Approximate positional uncertainty implied by that class - about 0.71 mi for a PLSS section centroid, 0.01 mi for a surveyed coordinate. Null when the source states no method, or when the location is flagged questionable and the error is unbounded.
nearest_well_positional_low_confidence	boolean	True when the nearest well is located only to a PLSS section or quarter, is a permitted-not-actual location, or is flagged questionable - so the reported distance carries real uncertainty. Null when the source states no method.
nearest_well_location_note	string	The source's verbatim location note, so the positional class can be audited.
nearest_well_status_date	string	Date the state last set that status, parsed out of the free-text notes where the source buries it. Impossible dates (06/31/2005) and future dates (09/09/2099) present in the source are rejected rather than published.
nearest_well_orphan_declared_date	string	Date the state formally declared the well orphaned, where the note carries one.
nearest_well_priority_class	string	The state's own plugging priority ranking, where it publishes one (Kansas uses Priority 1A/1B/1C/2).
nearest_well_operator_note	string	Source note about the responsible operator. The dataset has NO operator column; where operator information exists at all it is free text such as "Orphan-No Responsible Operator".
nearest_well_plss_location	string	Township / range / section / quarter description of the nearest well, assembled from the nine PLSS columns.
nearest_well_source_agency	string	The state agency whose records USGS compiled this well from.
nearest_well_data_file_date	string	Date of the state data file this well came from. This is the well record's real vintage and it ranges from 2019-07-01 to 2022-12-10 across states.

Composition of everything in the radius

Field	Type	Description
wells_plugged_per_state_record	number	How many wells in the radius carry a state status that literally says plugged. The dataset is a compilation of UNPLUGGED orphaned wells, but 323 rows nationally do carry a plugged status, so the count ships rather than being assumed away.
wells_not_plugged_per_state_record	number	How many wells carry a status that unambiguously indicates they are not plugged.
wells_plug_status_not_stated	number	How many wells whose state record does not state a plugging status. These are the ones a Phase I most needs a state file review for.
wells_injection_or_disposal	number	Injection, disposal, brine, waterflood or EOR wells inside the radius.
wells_type_not_stated	number	Wells whose type the state does not publish (about half the national dataset).
wells_low_positional_confidence	number	Wells located only to a PLSS section or quarter, permitted-not-actual, or flagged questionable.
wells_positional_class_not_stated	number	Wells for which the source records no location method at all.
wells_planned_location_only	number	Wells whose published coordinate is a permitted (planned) location, not an as-drilled one.
wells_questionable_location	number	Wells the source itself flags as "Questionable location - Numerous duplicate coordinates".
wells_datum_unconfirmed	number	Wells whose horizontal datum the source could not confirm (NAD27 vs NAD83 can differ by up to about 200 m).
wells_unmapped_status_codes	string	Distinct state status codes found in the radius that USGS publishes no codebook for, listed so you can look them up with the issuing agency. Null when every status was decodable.
count_by_status_raw_json	string	JSON object of verbatim state status text to well count inside the radius.
count_by_status_normalized_json	string	JSON object of normalized status bucket to well count. "(not stated)" covers blanks and unmapped state codes.
count_by_well_type_json	string	JSON object of normalized well type to count.
count_by_source_agency_json	string	JSON object of state agency to count - the provenance of every well in the radius.
count_by_state_json	string	JSON object of state to count, which matters on a radius that crosses a state line.
count_by_county_json	string	JSON object of county to count.
count_by_positional_class_json	string	JSON object of positional-accuracy class to count, so the reliability of the whole set is visible at a glance.
wells_data_vintage_earliest	string	Earliest state data file date among the wells found for this site.
wells_data_vintage_latest	string	Latest state data file date among the wells found for this site.

Per-well detail rows (record_type = "well")

Field	Type	Description
well_rank_by_distance	number	Rank of this well among all wells found for the site, 1 = nearest.
well_object_id	number	The layer OBJECTID, stable within a dataset vintage and useful for de-duplication.
well_id	string	Full published identifier including its system prefix.
well_id_system	string	Identifier system: API, USGSID or IGSID.
well_id_value	string	The identifier with the system prefix stripped - the value to paste into a state well-record search.
well_name	string	Lease or well name as the state records it.
well_number	string	Well number within the lease, where the state publishes one.
well_state	string	State whose records this well came from.
well_county	string	County of the well.
well_latitude	number	Published well latitude (NAD83 decimal degrees).
well_longitude	number	Published well longitude.
well_distance_miles	number	Great-circle distance from the site to this well, in statute miles.
well_distance_feet	number	The same distance in feet.
well_bearing_degrees	number	Initial bearing from the site to this well, 0-360 clockwise from true north.
well_bearing_compass	string	The same bearing as a 16-point compass direction.
well_status_raw	string	The state's own status text, verbatim.
well_status_normalized	string	Normalized status bucket, or null when the state code has no published codebook.
well_status_basis	string	literal_text, documented_in_dataset, unmapped_state_code or blank_sentinel.
well_plugged_per_state_record	boolean	True only when the state record literally says the well is plugged; false when it unambiguously says otherwise; null when it does not say.
well_type_raw	string	The state's own well type text, verbatim.
well_type_normalized	string	Coarse well-type bucket, or null for an opaque state code.
well_type_basis	string	literal_text, unmapped_state_code or blank_sentinel.
well_injection_or_disposal	boolean	True for injection, disposal, brine, waterflood, steamflood or EOR wells. Null when the type is not stated.
well_positional_class	string	How this coordinate was derived (see nearest_well_positional_class for the full vocabulary).
well_positional_uncertainty_miles	number	Approximate positional uncertainty implied by that class, in miles.
well_positional_low_confidence	boolean	True when this coordinate is PLSS-derived, permitted-only or questionable. Null when the source states no method.
well_datum_unconfirmed	boolean	True when the source records that the horizontal datum is unconfirmed and NAD27 is likely.
well_planned_location_only	boolean	True when the published coordinate is a permitted (planned) location rather than an as-drilled one.
well_questionable_location	boolean	True when the source itself flags the coordinate as questionable.
well_location_note	string	The source's verbatim location note.
well_status_date	string	Date the state last set this status, parsed from the free-text notes. Invalid and future dates in the source are rejected.
well_type_date	string	Date associated with the well type in the source notes, where present.
well_orphan_declared_date	string	Date the state formally declared the well orphaned, where the note carries one.
well_priority_class	string	The state's plugging priority ranking, where published.
well_operator_note	string	Free-text operator note. There is no operator column in this dataset.
well_coordinate_note	string	Free-text note about how the coordinate was obtained, where the source records one separately.
well_plss_location	string	Township / range / section / quarter description assembled from the nine PLSS columns.
well_source_agency	string	State agency whose records this well came from.
well_data_file_date	string	Date of the state data file this well came from - the record's real vintage.
well_info_raw	string	The complete verbatim Well_info note, kept so nothing fetched is silently discarded.
well_other_note_raw	string	The complete verbatim Other_note field, kept for the same reason.

The screening score

risk_score is a transparent 0-100 screen, versioned by risk_model_version (currently 1.0). It is not a determination, and every contributing factor is spelled out in risk_reasons on the row.

Contribution	Points
Nearest well within 100 ft	+45
Nearest well within 500 ft	+35
Nearest well within 0.25 mi	+25
Nearest well within 0.5 mi	+15
Nearest well within 1 mi	+8
A well in radius but further than 1 mi	+3
Wells within 0.25 mi	+5 each, capped at +25
More than 10 wells in the radius	+1 per 10, capped at +10
Any injection/disposal well within 0.25 mi	+10
Any well within 0.25 mi whose plugging status the state does not state	+7
Any questionable or permitted-only well location in the radius	+8

Capped at 100. Levels: none = 0, low = 1-19, moderate = 20-44, elevated = 45-69, high = 70-100.

The score is gated on the verdict. It is only ever computed for wells_found and no_documented_wells_in_covered_state. A site that is out of coverage, whose state could not be resolved, or whose source failed gets risk_score: null and risk_level: null - never 0 / none - with risk_reasons explaining why. Otherwise a buyer filtering on risk_level === "none" would read a site the dataset has no data for as clean.

Data vintage - read this before you cite it

This is a compilation of state records, not a live feed. Each state's slice carries its own File_date, and they range from 2019-07-01 (Colorado) to 2022-12-10 (Louisiana). Every row publishes both the dataset-wide range (dataset_vintage_earliest / dataset_vintage_latest) and the vintage of the specific wells found (wells_data_vintage_earliest / wells_data_vintage_latest).

Practical consequences:

• Wells plugged since a state's file date may still appear here.
• Wells documented as orphaned since then will not appear.
• The dataset is titled unplugged, but 323 rows nationally carry a state status that literally says plugged. Those are counted separately in wells_plugged_per_state_record rather than quietly dropped or quietly included.

Always confirm a hit against the issuing state agency - nearest_well_source_agency and well_source_agency name it, and well_id_value is the identifier to search.

Coverage

27 states, measured live on every run and published on every row in dataset_states_covered: Alabama, Alaska, Arkansas, California, Colorado, Illinois, Indiana, Kansas, Kentucky, Louisiana, Michigan, Mississippi, Missouri, Montana, Nebraska, Nevada, New Mexico, New York, North Dakota, Ohio, Oklahoma, Pennsylvania, Tennessee, Texas, Utah, West Virginia, Wyoming.

Largest state counts: Ohio 20,557 - Pennsylvania 19,160 - Oklahoma 15,971 - Kentucky 12,695 - Illinois 8,480 - New York 7,029 - Texas 5,854.

If a state ever disappears from the live dataset the run fails, because a lost state silently turns every site there into a false clear.

Status normalization, and what is deliberately left null

*_status_raw is always the state's own text, verbatim - that is the authoritative value. *_status_normalized is populated only when the mapping is defensible, and *_status_basis tells you which rule applied:

• literal_text - the English is unambiguous (Orphan, Abandoned Well, Idle, Active, Plugged, Unknown Not Found, ...).
• documented_in_dataset - the dataset's own notes decode the code. Only AB and SI qualify, from the Nebraska note "Shut In to be converted to "AB" or "OR" by source".
• blank_sentinel - the source publishes a single space. All 5,854 Texas rows are in this state.
• unmapped_state_code - an opaque code with no published codebook. These are left null rather than guessed. The affected codes and their issuing states are: OR (Oklahoma 15,971 / Ohio 877), HP (Ohio 15,493), UN (Ohio 3,289), OP (Ohio 802), LU (Ohio 96), TA (Colorado 47 / Utah 12), S (Utah 58), DA (Colorado 56), PR (Colorado 39), AL (Colorado 5), DM (Colorado 4), I (Utah 2), WO (Colorado 2), IJ (Colorado 1), XX (Colorado 1). Any of them appearing in your radius is listed in wells_unmapped_status_codes.

Only Plugged, PA and P&A set plugged_per_state_record: true.

Positional accuracy classes

positional_class	Meaning	Uncertainty
surveyed_or_reported	An actual surveyed or operator-reported coordinate	~0.01 mi
datum_converted	USGS converted the original NAD27 value to NAD83	~0.05 mi
datum_unconfirmed	Datum unconfirmed, NAD27 likely for older wells	~0.15 mi
third_party_derived	Generated from third-party data (all 5,854 Texas rows)	~0.15 mi
plss_quarter_quarter	Located to a 40-acre PLSS quarter-quarter	~0.18 mi
plss_quarter	Located to a 160-acre PLSS quarter	~0.35 mi
plss_section	Located to a 1-square-mile PLSS section centroid	~0.71 mi
planned_not_actual	The permitted location, not an as-drilled one	~0.25 mi
questionable	The source itself flags the coordinate as unreliable	unbounded (null)
unspecified	The source records no location method	null

*_positional_low_confidence is true for plss_section, plss_quarter, planned_not_actual and questionable; it is null - not false - when the source states no method at all.

Failure behaviour

Zero rows is a valid success only when the query was well formed, reached the upstream, and the upstream honestly returned nothing.

Situation	What happens
assets omitted	HTTP 400 at run creation. Nothing runs, nothing is billed.
The upstream layer has drifted or is unreachable	Run FAILS before any row is pushed.
Every site failed to screen	Run FAILS. The billing gate runs before pushData.
Some sites screened, some failed	Rows are emitted; the failed ones carry wells_source_status: "unavailable" and assert nothing about wells.
Query completed, no wells in radius	SUCCESS with a row saying so - a real answer.

Every failure message ends with "Nothing was emitted and nothing was billed."

A deliberate choice worth stating: on a partial outage the failed sites do emit one billable row each. That row is a genuine per-site answer - it says explicitly that the source was unavailable and leaves every well field null - and a total outage fails the run instead, so no one is ever billed for a dataset of nothing but errors.

Use as an MCP tool

This actor is callable by AI agents through mcp.apify.com. Every input and every output field carries a description, so an agent can pick it and chain it without guessing. It composes well: hand it coordinates produced by a geocoder, then feed phase1_esa_further_inquiry_flag and risk_level into a site-selection or due-diligence workflow. Because the null-vs-false contract is explicit, an agent can tell "no wells here" apart from "we could not check", which is exactly the distinction that makes an automated screen safe to act on.

Pricing

$8 per 1,000 results, billed on the standard Apify result event - one charge per dataset row. A summary-only run costs one result per site: screening 50 sites is 50 results, $0.40. Turning on includeWellDetail bills one extra result per well, so cap it with maxWellsPerAsset on dense sites. Graduated discounts apply on paid Apify plans. A failed run emits nothing and bills nothing.

FAQ

Is there a free API for orphaned oil and gas wells near an address? The underlying USGS layer is free and keyless. What this actor adds is the part that is easy to get wrong: complete paging with a truncation guard, a true-circle radius, coverage gating so a no-data state is not reported as clear, positional accuracy per well, and status normalization that refuses to guess.

How many orphaned wells are there in the US? This dataset documents 117,672 across 27 states. Federal estimates of undocumented orphaned wells run into the hundreds of thousands to millions, and by definition none of those are in any dataset - which is why this is a screen and not a clearance.

Does a zero result mean my site is clean? Only when screening_result is no_documented_wells_in_covered_state. If it is out_of_dataset_coverage, coverage_unknown or source_unavailable, the actor has not answered the question and says so.

Why is the nearest well's status blank or unmapped? Because the state publishes it that way. All 5,854 Texas rows carry a blank status, and about a third of the dataset uses two-letter state codes USGS publishes no codebook for. The raw value is always in *_status_raw; look it up with the agency named in *_source_agency.

Can I search a whole county or state instead of a point? Not directly - this is a point-and-radius screener. Use a large radiusMiles (up to 25) around a centroid, or pass many points. The service's own exact count is published in wells_reported_by_source so you always know the true size.

Does it work for Phase I ESA reports? It is a screening tool, and every row cites the DOI, the issuing state agency and the data vintage so it can be referenced properly. It is not a determination and does not replace a state file review or a site reconnaissance.

Why do some distances have a large uncertainty? Because the state never surveyed the well. A PLSS section-centroid location is accurate to about 0.7 mile at best. *_positional_uncertainty_miles tells you which wells those are instead of hiding it behind a four-decimal distance.

How often does the data update? The USGS release is a periodic compilation, currently spanning 2019-2022 state files. This actor reads it live on every run and fails loudly if the national count collapses or a state disappears.

Related actors

Screening a site properly takes more than one dataset. These pair directly with this one:

• EPA Contaminated Site Screener
  • Superfund/SEMS, RCRA, USTs, ECHO enforcement and institutional controls at the same coordinate. The standard Phase I companion to this one.
• Site Due Diligence Bundle
  • 20 federal and state layers for a single coordinate in one call, with a composite go / caution / no-go verdict.
• Oil & Gas Well Permits
  • the other side of the same story: newly permitted and active wells from state regulators, where this actor covers the abandoned ones.

Source

USGS, US Documented Unplugged Orphaned Oil and Gas Well Dataset, https://doi.org/10.5066/P91PJETI, served from services.arcgis.com/v01gqwM5QqNysAAi FeatureServer layer 5. Compiled from 27 state oil and gas regulators. Public domain.