← The record · All reference data

Ohio water-well-log census (Ohio DNR, Division of Water Resources)

Per-county censuses of logged water wells, pulled from the Ohio DNR, Division of Water Resources well-log database. Under R.C. 1521.05 every water-well contractor files a completion / sealing log with the Division; the (nightly- refreshed) database is published as a public ArcGIS MapServer — one point per well. Every location, depth, aquifer, static level, and reported yield here was returned by the DNR’s own public service — nothing is fabricated, inferred, or backfilled. Regenerate a county with watermark waterwells --county <name> (defaults to the active site’s county).

This is the groundwater peer of the surface-water supply model (watermark.hydrology.supply). The census is the population of private and public wells whose static water levels and reported yields are the empirical basis for an [inference] aquifer-parameter estimate (watermark.hydrology.aquifer) and the well-drawdown (“cone of depression”) thread the data-center cooling withdrawal implicates — the “area well concerns” that surface in the PAAC board record (2026-03-30). It is not a withdrawal registry: for who is licensed to withdraw >100,000 gpd, see the sibling data/reference/ohio-water-withdrawal/.

Source

The DNR publishes the well-log database as a public ArcGIS MapServer:

https://gis2.ohiodnr.gov/arcgis/rest/services/DSW_Services/waterwells/MapServer

with a single well-point layer:

The interactive Water Well Locator at waterwells.ohiodnr.gov is the human front-end over the same well-log database (and links each well to its scanned drilling-report PDF).

Method

fetch_county(county) runs one query: layer 0 filtered to COUNTY = '<COUNTY>' (the service stores the county uppercase), fields selected by name (never index), paged at 1,000 (the hosted maxRecordCount) to completion on exceededTransferLimit. Values are passed through verbatim; a blank field is an empty cell, never a fabricated 0. Wells are OBJECTID-sorted, so an unchanged pull regenerates byte-identical CSV.

Files

A flat well census is naturally tabular, so it lands as CSV (one row per well, provenance in this README) — the EPA-ECHO reference-dataset convention, not the nested-YAML shape the WWFRP registry needs.

One CSV per county, named for the county (allen.csv, shelby.csv), plus any committed campus dewatering wellfield. Which counties are committed, and what each census contains, is in the per-site notes under instances/ — this file is about the source, not about a place.

Known gaps & caveats (read before using)

1. Ohio only. This is Ohio’s well-log service. A non-Ohio watershed point (Fort Wayne, IN) has its own state service — the connector/CLI refuses cleanly rather than query the wrong state.

2. No pumping-water-level column — so no true specific capacity. The service reports a static (rest) level and a test yield (gpm), but not the pumping water level during the yield test. Specific capacity (yield ÷ drawdown) and therefore a per-well transmissivity are not derivable from a single record. The aquifer-parameter layer (watermark.hydrology.aquifer) leans instead on the static-water-level surface, the yield distribution, and cited literature aquifer properties by aquifer type — all [inference], never presented as measured.

3. Self-reported, and locational quality varies. Every figure is the drilling contractor’s own log entry — [verified] for what the log states, which is not the same as field-verified truth (a transcription can be wrong). coord_source records how each point was located: a GEOCODEd address or DIGITIZED point is coarser than a GLOBAL POSITIONING SYSTEM fix, and gates spatial confidence for any drawdown-cone intersection.

4. Owner / name / street columns are not ingested. The service exposes OWNER / LAST_NAME / STREETNAME / HOUSE_NO — private residents’ PII the model does not need. The connector selects only locations and hydraulics; the committed CSV carries no personal identifiers.

5. A census, not a time series. Each row is the well’s completion-log snapshot, not repeat water-level measurements. For monitoring-well water-level history, USGS NWIS groundwater (countyCd=39003) is the complementary source.

6. TOTAL_DEPTH = 0 and blank aquifers appear for some non-standard logs (dry holes, sealings, incomplete transcriptions). Preserved verbatim; downstream code filters on record_type / non-null fields rather than mutating the census.

Field reference

CSV columns (an empty cell = the service returned nothing):

Campus dewatering wellfield (lima-campus-dewatering.csv)

A distinct, curated companion dataset: the 44 construction-dewatering wells the developer (“LMA”) installed along North Cole Street at the BOSC/data-center campus (American Township) to lower the water table for site grading. All shallow (23–41 ft) in the unconfined sand & gravel aquifer, pumping 75–100 gpm each (~4.9 MGD of combined capacity). Installed Dec 16 2025 – Feb 12 2026; 43 were sealed May 26–28, 2026, leaving one active well (record 3027949).

This is the documented groundwater stress behind the residents’ “area well concerns” — distinct from the county census above (the population of wells) and from the hypothetical cooling-makeup drawdown scenario (watermark.hydrology.drawdown). Its cone of impact is modeled by watermark.hydrology.dewatering (per-well Cooper-Jacob cones superimposed).

• Source: Ohio DNR Water Well Locator (waterwells.ohiodnr.gov) — the well-log export (WaterWell_CSV) reconciled with the sealing-report export (SealingReport_CSV), both pulled 2026-07-28. Not connector-regenerable: the live ArcGIS MapServer the ohio_waterwells connector reads now shows only the 1 active DEWATERING WELL in Allen County — the 43 sealed wells have dropped off the active layer, so only this manual export preserves the full operation. Owner is the operator company (“LMA”), not a private resident; no personal PII.
• Method: each well is matched to its sealing record by nearest coordinate (≤ 50 ft); the one well with no sealing within 50 ft (3027949, 58 ft to the nearest) is the active one. sealed_date is that nearest sealing’s date; empty = active.
• Caveats: driller-reported figures are [verified] for what the log states; the cone-of-impact drawdowns derived from them are [inference] (literature K, Cooper-Jacob screening on an unconfined aquifer). test_rate_gpm is the well’s yield-test capacity, an upper bound on the sustained dewatering rate; installs and sealings were staged, so a simultaneous-full-rate composite is a bound on concurrency.
• Columns: record_no, operator, township, aquifer_type, well_use, static_water_level_ft, test_rate_gpm, total_depth_ft, completion_date, sealed_date (empty = active), longitude, latitude (NAD83).

Cataloged datasets — generated from data/catalog/reference/; run watermark catalog render --apply after editing an entry.

lima-campus-dewatering — Lima campus construction-dewatering wellfield (Ohio DNR)

Source: Ohio DNR Water Well Locator (waterwells.ohiodnr.gov) — well-log export reconciled with the sealing-report export, both pulled 2026-07-28. Not connector-regenerable: the live MapServer now shows only the 1 active DEWATERING WELL (the 43 sealed wells dropped off the active layer). · License: Ohio public record · Access: public · Site scope: site:lima · Refresh: on-demand, last 2026-07-28

ohio-waterwells-allen — Ohio DNR water-well-log census — Allen

Source: Ohio DNR, Division of Water Resources — water-well-log database, R.C. 1521.05 (contractor completion/sealing logs), public ArcGIS MapServer (layer 0) · License: Ohio public record · Access: public · Site scope: site:lima · Refresh: annual (ttl 365d)

Regenerate: watermark waterwells --county Allen

ohio-waterwells-shelby — Ohio DNR water-well-log census — Shelby

Source: Ohio DNR, Division of Water Resources — water-well-log database, R.C. 1521.05 (contractor completion/sealing logs), public ArcGIS MapServer (layer 0) · License: Ohio public record · Access: public · Site scope: site:sidney · Refresh: annual (ttl 365d)

Regenerate: watermark waterwells --county Shelby