Methodology

1. Sources. Only public PFBC and DEP data, listed with versions on Sources. Nothing is called at page-load time; data is imported at refresh.
2. Which waters. Inland lakes, ponds and reservoirs in PFBC's lake data, plus lakes on PFBC's trout schedule, DEP lake advisory rows, and PFBC program, Best Fishing Waters and ice lists. Streams, rivers and the Great Lakes waters are left out.
3. Matching. Sources are joined on PFBC's water ID where they carry it; otherwise by name and county with a reviewed alias table (for example, DEP's “Francis Slocum Lake” is PFBC's Frances Slocum Lake).
4. Location checks. Each lake point must fall inside a county PFBC states for it, or within 500 m of the line; otherwise we use the lake polygon's center, or leave the lake out. Every correction is logged.
5. Pages. A lake gets its own page when it has at least two kinds of data and at least one of: a PFBC report, a trout-schedule date, a lake-specific advisory row, or a PFBC Best Fishing Waters listing. Other waters appear on their county page.
6. Survey findings. For each lake we summarize the latest two PFBC reports. The summaries were drafted with AI help at build time; then a program checked that every number appears on the cited page, every species name appears in the report, and every quote is word for word and 25 words or fewer. Anything that failed is not shown. A person-level spot check of 30 random lakes follows each refresh.
7. Dates and expiry. Each block carries its date. Schedule rows read “Was scheduled for…” after their date; advice and program blocks turn amber after their re-check date; surveys older than 10 years are labelled Historical.
8. What we never do. Rate lakes, publish reviews, copy PFBC's descriptions or photos, re-host PFBC or DEP PDFs, or say a lake was stocked.
9. AI. AI is used only to draft survey summaries at build time. The site itself makes no AI calls.

Data status Sources and versions