Methodology

Overview

WaterByCounty layers four free federal datasets into a single per-county view:

• EPA SDWIS — drinking water compliance and the A–F letter grade.
• EPA ATTAINS — watershed health under Clean Water Act §303(d).
• EPA Water Quality Portal — monitoring records over a 5-year rolling window.
• USGS NWIS — streamflow snapshots from a representative gauge per county.

Each source is reported separately. The A–F drinking water grade is the only composite score we compute, and it is based only on SDWIS compliance — it does not fold in watershed impairment, monitoring density, or streamflow.

1. EPA SDWIS — Drinking Water Compliance

Data is pulled from the EPA Safe Drinking Water Information System via the Envirofacts API. SDWIS tracks compliance by every public water system that serves at least 25 people or has 15+ service connections.

• Lookback window — Past 5 years of violations.
• Health-Based Violations — Exceeded a Maximum Contaminant Level (MCL) or failed a Treatment Technique rule.
• Violations per 100K served — Health-based violations normalized by population served, not a percent.
• Water Quality Score — 0–100 percentile rank across all US counties; 100 is best.
• Letter Grade — A (90+), B (75–89), C (60–74), D (40–59), F (<40).

2. EPA ATTAINS — Watershed Health

The Assessment, Total Maximum Daily Load Tracking and Implementation System (ATTAINS) is EPA's repository of state §303(d) assessments under the Clean Water Act. States report whether each assessed water body meets standards for its designated uses (drinking water source, recreation, aquatic life, fish consumption).

• Impaired % — Share of assessed water bodies in the county classified as impaired.
• Top causes — The most-frequently-cited reasons for impairment (mercury, E. coli, nutrients, sediment, PCBs, etc.).
• Reporting cycle — States submit on a two-year cycle. The most recent cycle is shown.

County attribution — About half of states attribute assessments directly to counties. The remaining states (PA, FL, TX, NY, WV, NC, WA, NH, IL, MI, and others) only tag assessments by HUC-8 watershed. For those, we use a USGS-derived HUC-8 → FIPS5 crosswalk to map assessments back to the counties they cover, with a guard that prevents cross-state attribution (a Pennsylvania-tagged HUC8 will not be credited to a New York county).

Caveat: data is biennial. The most recent fully published cycle is 2022 for most states. ATTAINS is not real-time — treat impairment % as a structural signal, not a current snapshot.

3. EPA Water Quality Portal — Monitoring Records

The Water Quality Portal aggregates monitoring records from EPA STORET, USGS NWIS, USDA, and state/tribal programs. Each "result" is one sample analyzed for one characteristic (E. coli, pH, dissolved oxygen, total nitrogen, etc.).

• Monitoring sites — Distinct sampling locations active in the past 5 years.
• Measurements recorded — Total sample results over the 5-year window.
• Most-measured groups — Top characteristic categories by sample count (microbiological, physical, inorganics, etc.).

What this measures. Monitoring footprint, not water quality directly. A county with 10,000 readings has stronger scientific evidence behind whatever signal is reported than a county with 50 readings — but high site counts alone don't mean the water is clean.

4. USGS NWIS — Streamflow Snapshot

The USGS National Water Information System streamgage network publishes river and stream discharge readings nationwide.

• Gauge selection — One representative gauge per county: the one with the largest drainage area among all NWIS sites located in the county.
• Snapshot discharge — The latest reading available in cubic feet per second (cfs) at pipeline run time.
• Long-term mean — The full-record annual mean discharge for the gauge.
• % of typical — Current reading ÷ long-term mean × 100. Buckets: ≥130% well above, 110–130% above, 90–110% near typical, 60–90% below, <60% well below.

Caveat: snapshot, not stream. The "% of typical" is the reading at the moment our pipeline last ran. Many counties have multiple gauges; we summarize only the primary one. Counties with no NWIS gauge inside their boundaries simply do not get a Streamflow zone.

Data Freshness

• SDWIS — Refreshed periodically; violations typically lag real time by 3–6 months.
• ATTAINS — Two-year reporting cycle; most recent fully published cycle is 2022 for most states.
• WQP — Rolling 5-year window from the pipeline run date.
• NWIS — Pipeline snapshot of the most recent gauge reading. Refreshes on the next data sync.

Scoring — A through F Letter Grade

The A–F grade is computed only from SDWIS compliance — it is not a multi-source composite. Counties are ranked by violation history against all 3,100+ US counties using percentile-rank methodology and normalized to 0–100.

• Score 90 = better than 90% of US counties on SDWIS compliance.
• Score 50 = at the national median.
• Score 10 = 90% of counties have a better SDWIS record.

Letter Grade Scale

How Water Quality Scores Are Calculated

Water quality scores use percentile-rank methodology. Every county is ranked against all 3,100+ US counties based on its violation history and compliance record. The score is then normalized to a 0-100 scale:

• A score of 90 means the county has better water quality than 90% of all US counties.
• A score of 50 means the county is at the national median.
• A score of 10 means 90% of counties have better water quality.

Lower violation counts and lower population-normalized violation rates result in higher scores. Counties with zero health-based violations score highest.

Letter Grade Scale

Geographic Coverage

WaterByCounty covers all 3,100+ counties and county-equivalents in the United States, including parishes (Louisiana), boroughs (Alaska), and independent cities (Virginia). Coverage varies by data source — every county sees the zones for which it has data.

• SDWIS: ~97% of counties — limited only where public water systems have not yet reported.
• ATTAINS: ~61% of counties — depends on whether the state has assessed water bodies in the county.
• WQP: ~95% of counties — most counties have at least one historical monitoring sample.
• NWIS: ~73% of counties — depends on whether a USGS streamgage operates inside the county.

AI-Generated Content Disclosure

WaterByCounty uses an LLM to generate descriptive narratives on county and state pages. These narratives are grounded in the underlying federal data and reviewed for accuracy, but they are machine-written summaries — not expert analysis.

All numeric values, tables, scores, grades, impairment percentages, monitoring counts, and streamflow readings come directly from EPA SDWIS, EPA ATTAINS, EPA WQP, and USGS NWIS processed through our pipeline. The AI layer adds context and readability but does not alter underlying data.

If you notice an error in any AI-generated content, please contact us so we can correct it.

Limitations

• Public water systems only. SDWIS tracks public water systems only. The ~13% of US households on private wells need their own testing — our grade does not speak to private-well water.
• ATTAINS reporting lag. §303(d) assessments are biennial and the most recent fully published cycle is 2022 for most states. Treat impairment % as a structural signal, not a current snapshot.
• NWIS represents one gauge. Many counties have multiple streamgages — we summarize only the largest by drainage area. "% of typical" is the latest reading divided by long-term mean, not an average over time.
• WQP density ≠ water quality. A high monitoring count means more scientific evidence behind the reported signal — not that the water is clean.
• SDWIS violations are cumulative. A county with historical violations may have fully compliant systems today. The 5-year lookback captures both active and resolved cases.
• A passing grade isn't perfect water. The grade captures regulatory compliance, not taste, hardness, or unregulated contaminants. It means the public water systems met federal standards over the lookback window.