# N05_E18 Great Lakes pilot: 1996–2019 plat-book page verification and modern NLDI anchor **Research date:** 2026-08-11 UTC **Sector:** `N05_E18` — canonical 24 × 24 grid cell, 37.5–45.0° N and 270–285° E (90–75° W) **Anchor:** Merrillville, Indiana, approximately 41.4828° N, 87.3327° W **Time band:** `1950-present`, with 1996 and 2019 map anchors **Status:** bounded source and geography crosswalk; no derived score and no `provisional_quadrant` ## Selection and boundary The preceding 2026-08-10 Lake County plat-book note identified a specific next step: retrieve and visually inspect the key, township, watershed, and index pages, then obtain an authoritative modern hydrologic geometry before deriving any overlap. This run completed a bounded page-level check of the available PDF scans and retrieved a machine-readable USGS Network Linked Data Index (NLDI) anchor and downstream linked-site response. This is not a parcel audit, a legal title record, a hydrologic time series, or a cell-wide infrastructure measurement. The plat books are publication-specific map and landownership sources. The NLDI features are modern hydrographic/network records and are not evidence that a modern reach or monitoring site existed in 1996 or earlier. ## Raw observations ### A. 2019 Lake County plat book: page and scan record - **Archive.org item:** *Plat Book Lake County, Indiana (2019)*, identifier `plat-book-lake-county-indiana-2019-images`. - **Derivative inspected:** `Plat BookLakeCountyIndiana2019_images.pdf`, 86 PDF pages; OCR derivative and page-number JSON were also retrieved. - **PDF page 4:** page image is headed `KEY MAP OF LAKE COUNTY, INDIANA`. The printed map includes Lake Michigan and a legend with city boundary, county boundary, page boundaries, and township/range shading. The page is a county-level map, not a georeferenced extract of `N05_E18`. - **PDF pages 30–31:** page images are headed `Hobart Townships 35N & 36N - Ranges 7W & 8W`; the printed map page numbers are 28 and 29. The pages are direct scan observations of township/range map plates. Their publication page scale, map projection, and feature geometry were not digitized here. - **PDF page 53:** page image is headed `WATERSHEDS MAP`. The printed page is 51 and includes Lake Michigan, Thorn Creek, Crooked Creek–Kankakee River, and a legend/explanatory statement that identified watersheds are classified by USGS as HUC-10 and typically cover 62–390 square miles. This is the book’s printed explanation; it is not an independent calculation of area or a substitute for USGS geometry. - **Archive.org item metadata:** exposes PDF, OCR text/XML, scan data, page-number JSON, and JP2 derivatives. The access path and derivative identifiers are retained below. ### B. 1996 Lake County plat book: page and scan record - **Archive.org item:** *Plat Book Lake County, Indiana (1996)*, identifier `plat-book-lake-county-indiana-1996-images`. - **Derivative inspected:** `PlatBookLakeCountyIndiana1996_images.pdf`, 61 PDF pages; OCR derivative and page-number JSON were also retrieved. - **PDF page 35:** page image is headed `HOBART TOWNSHIP 35-36 N. RANGE 7-8 W.` The map plate directly confirms the 1996 publication’s township/range map context for the corridor, but not fixed-cell coverage or feature capacity. - **PDF pages 46–49:** page images are headed `INDEX OF OWNERS • LAKE COUNTY`. The pages include entries for township/section references, Gary-related public or industrial names, rail-company entries, and other named landowners. An index entry is evidence that a label occurs in the publication; it is not proof of ownership, operation, emissions, employment, land area, or continuity. - **Printed disclaimers:** the 1996 street-map pages state that information was obtained from various state, county, and city sources, was not field checked by the publisher, and was not guaranteed. This limits the evidentiary status of map labels and is preserved rather than silently discarded. ### C. Modern USGS NLDI geometry and linked sites - **NLDI feature:** COMID `3399712`, retrieved from the USGS NLDI linked-data endpoint. The response is a GeoJSON `FeatureCollection` with one `LineString` feature. The returned geometry includes coordinates around the Merrillville/Hobart anchor, including points near `[-87.332700115, 41.485146277]`. - **Local basin:** the `/basin` endpoint returns a GeoJSON `FeatureCollection` with one polygon. The returned polygon includes coordinates around the local basin near Merrillville, including the anchor vicinity. No area or intersection percentage was calculated in this run. - **Downstream NWIS response:** `/navigation/DM/nwissite` returned 19 linked feature records. The response includes identifiers such as `USGS-04093500` (Burns Ditch at Gary), `USGS-04092975` (Turkey Creek at SR 53 at Merrillville), `USGS-04092977` (Turkey Creek at 61st Avenue at Merrillville), `USGS-04093000` (Deep River at Lake George outlet at Hobart), and `USGS-04092990` (Lake George at Hobart). The response also includes duplicate or alternate identifiers at some network locations; the count is a response count, not a count of unique stations after deduplication. - **Network interpretation:** the NLDI response establishes a reproducible modern hydrographic linkage from COMID `3399712` to a set of indexed downstream NWIS sites. It does not establish historical route continuity, a 1996/2019 observation, a sector-wide watershed share, or a causal relationship to settlement or industry. ## Source interpretation 1. The page images verify that the prior OCR-based Lake County note was anchored to identifiable scan pages rather than only catalog text. The 2019 key, Hobart township, and watershed pages are now page-located at PDF pages 4, 30–31, and 53; the 1996 Hobart township and owner-index pages are page-located at PDF page 35 and 46–49. 2. The 2019 watershed page’s HUC-10 language is useful as a source statement and search lead, but the authoritative modern geometry must come from USGS/NHDPlus/NLDI rather than from an un-digitized printed map. 3. The NLDI reach and basin provide a modern machine-readable spatial anchor for later intersection work. The source footprint is a point/reach/basin/network relationship, not the fixed 24 × 24 cell. A formal overlap calculation remains a separate reproducible task. 4. The 1996 owner index and printed disclaimer strengthen source criticism: publication labels can identify candidates for primary administrative or land records, but cannot carry operational or quantitative claims by themselves. ## Geography and variable crosswalk | Raw observation | Target variable | Relation | Limitation | |---|---|---|---| | 2019 key map and Hobart township plates, PDF pp. 4, 30–31 | Fixed `N05_E18` geometry | related, not equivalent | Printed map footprint/projection not digitized or clipped | | 2019 watershed plate, PDF p. 53 | HUC-10/NHDPlus/NLDI hydrology | related, not equivalent | Printed HUC-10 statement is not a polygon or area measurement | | 1996 Hobart township plate, PDF p. 35 | Historical local map record | related, not equivalent | No georeferencing, map scale extraction, or continuity assumed | | 1996 owner index, PDF pp. 46–49 | Land use/institutional lead | qualitative lead only | Index entries do not establish status, ownership, capacity, or dates beyond publication context | | NLDI COMID `3399712` LineString | Modern connectivity/hydrology | machine-readable modern observation | No historical continuity and no fixed-cell overlap calculated | | NLDI COMID `3399712` basin polygon | Modern drainage footprint | machine-readable modern observation | No area, cell intersection, or watershed-share score calculated | | NLDI downstream NWIS response, 19 feature records | Modern linked monitoring/network context | machine-readable modern observation | Response count includes duplicate/alternate identifiers; unique-station normalization not applied | ## Derived layer No area, overlap, route length, station density, watershed share, connectivity score, energy value, population value, constraint score, adaptive-capacity indicator, or formula output was calculated. The NLDI coordinate sequence and basin polygon are preserved as retrieved source observations, not converted into a sector-wide metric. ## Model classification - `provisional_quadrant`: `unresolved` - **Reason:** the historical pages are source-footprint evidence, while the NLDI reach/basin are modern network records; no fixed-cell intersection, comparable time series, or complete pilot raw-observation table exists. - **Model boundary:** `provisional_quadrant` is a model output and is not an archival fact. ## Missingness and uncertainty - `fixed_cell_geometry_unresolved`: the canonical cell polygon was not intersected with the NLDI reach or basin in this run. - `map_georeferencing_unresolved`: the 1996/2019 page images were visually checked, but no control points or projected geometry were created. - `historical_continuity_unresolved`: modern NLDI records do not prove that the same route, reach, or monitoring locations had the same configuration in 1996 or earlier. - `unique_station_normalization_pending`: the 19 downstream linked records include alternate/duplicate identifiers at shared COMIDs; no unique-site deduplication was derived. - `source_layer_unretrieved`: the underlying NAIP/NED layers named by the 2019 book were not retrieved or measured. - `facility_status_unresolved`: 1996 index labels do not establish operating status, capacity, ownership, or emissions. - `not_applicable`: no score or quadrant formula was attempted. No OCR omission, archival silence, inaccessible derivative, or absent map feature has been converted into zero. ## Next search path 1. Define and version the fixed `N05_E18` polygon in the project’s canonical longitude convention, then intersect it with the retrieved NLDI COMID `3399712` reach and basin polygon using a reproducible geospatial library or GIS command. 2. Normalize the NLDI downstream response by COMID and station identifier before using it as a modern connectivity or monitoring-source table. 3. Retrieve the USGS/NHDPlus or WBD source geometry corresponding to the printed 2019 HUC-10 map labels and document the relationship between the printed map and the authoritative layer. 4. Continue primary administrative-record retrieval for Gary Water Works, Gary Steel Products, rail companies, pipelines, and remediation; treat 1996 owner-index entries as leads only. 5. Complete comparable raw-observation tables for the nine Great Lakes sectors before assigning any quadrant. ## Provenance register | Evidence | Tier | Item / page / endpoint | Date accessed | URL | |---|---|---|---|---| | 2019 Lake County plat book, key map | C | PDF page 4; printed page 2 | 2026-08-11 | | | 2019 Lake County plat book, Hobart plates | C | PDF pages 30–31; printed pages 28–29 | 2026-08-11 | | | 2019 Lake County plat book, watershed plate | C | PDF page 53; printed page 51 | 2026-08-11 | | | 2019 Archive.org metadata | C | Item metadata and derivative manifest | 2026-08-11 | | | 2019 OCR derivative | C | `Plat BookLakeCountyIndiana2019_images_djvu.txt` | 2026-08-11 | | | 1996 Lake County plat book, Hobart plate | C | PDF page 35; printed township plate | 2026-08-11 | | | 1996 Lake County plat book, owner index | C | PDF pages 46–49 | 2026-08-11 | | | 1996 Archive.org metadata | C | Item metadata and derivative manifest | 2026-08-11 | | | 1996 OCR derivative | C | `PlatBookLakeCountyIndiana1996_images_djvu.txt` | 2026-08-11 | | | USGS NLDI modern reach | A | COMID `3399712`, GeoJSON LineString | 2026-08-11 | | | USGS NLDI local basin | A | COMID `3399712`, GeoJSON basin polygon | 2026-08-11 | | | USGS NLDI downstream linked sites | A | COMID `3399712`, downstream `nwissite` response; 19 returned records | 2026-08-11 | | | NLDI documentation | A | USGS description of linked-data, basin, and navigation services | 2026-08-11 | | ## Bottom line This run closes a narrow provenance gap: selected 1996 and 2019 Lake County scan pages were retrieved and page-located, and a modern USGS NLDI reach, basin polygon, and downstream linked-site response were recorded for the Merrillville/Hobart corridor. The evidence improves reproducibility and defines the next geospatial operation, but it does not justify a fixed-cell overlap, a historical continuity claim, a cell-wide measure, a derived score, or a quadrant. The next defensible step is a versioned `N05_E18` polygon intersection and normalized modern network table, followed by primary administrative-record retrieval.