Open data · primary sources
Everything on this site, free to download and check
We do not ask you to take our word for it. Every figure comes from a public document, and every dataset we built from those documents is here, with the method and the known gaps stated plainly.
Datasets
Built by extracting tables from published PDFs. Plain CSV, no registration.
-
Combined sewer overflow records
One row per outfall per quarter: NPDES outfall number, receiving water, overflow count, volume in millions of gallons, and duration statistics. 2005 Q1 to 2026 Q1.
-
Consent decree obligations and status
Every requirement DC Water reported against, from both decree series: decree paragraph cited, the requirement text, due date, and the reported status, quarter by quarter.
-
Blue Plains treated flow
Monthly and weekly average flow through the treatment plant, in million gallons per day, from Discharge Monitoring Reports filed under permit DC0021199.
-
Rainfall-normalised overflow
Annual overflow volume per river alongside rainfall, and overflow expressed per inch of rain — the basis for separating infrastructure improvement from weather.
-
Chart data
The exact aggregated figures the charts on this site read, so any chart can be checked against its own numbers.
Primary documents
Court orders and federal policy, hosted here because they are not easy to find and DC Water does not publish them.
-
Consent Decree — Nine Minimum Controls
The operational decree: pumping stations, inflatable dams, tide gates, public notification.
-
First Amendment to the Consent Decree
Contains the Long Term Control Plan and its numeric targets, and moves final compliance from 2025 to 2030.
-
Plaintiffs' objection to the 2015 amendment
Why the groups who brought the original case opposed the five-year extension.
-
EPA CSO/SSO Compliance and Enforcement Strategy
The national policy behind this and similar cases nationwide.
-
EPA Enforcement Management System, Chapter X
How EPA prioritises sewer overflow enforcement — the framework that made DC a target.
How the data was built
DC Water publishes its overflow figures as tables inside quarterly PDF reports — 83 of them, spread across two document categories in a library of 3,261 files. There is no spreadsheet and no API. The datasets here were produced by parsing those tables directly out of the PDF text.
Two independent checks
Internal reconciliation. Each report prints its own subtotal per river. The extraction sums its output and compares. Every parsed report reconciles — which is how six separate parsing faults were found and fixed, including one that was labelling month-only tables as full quarters.
A second, independent publication. DC Water republishes the same tables inside its Nine Minimum Control annual reports, typeset differently. Parsing those separately and comparing: 19 of 19 overlapping quarters agree, 17 of them exactly. The two differences are 0.03% and 0.05% — rounding in the sources.
Rainfall
Daily precipitation from NOAA GHCN-Daily station USW00013743 (Reagan National), a complete record with no gaps. Overflow is normalised against the rain falling in a quarter's five wettest days, which explains 61% of the variation in overflow volume — against 40% for total rainfall and 1% for the number of rainy days.
Why storm rain, not total rain? A combined sewer has a capacity. Rain arriving slower than that capacity is entirely treated; rain arriving faster spills. So the same three inches produces no overflow spread over a fortnight, and a large one in a single afternoon.
What is missing, and why
Stating the gaps is part of the point. Nothing here has been estimated, interpolated or filled in.
| Gap | Reason | Recoverable? |
|---|---|---|
| 2018 Q4 and 2019 Q1 overflow data | DC Water published these two reports as scans whose embedded text is corrupt. Re-running OCR does not recover the numeric columns. | needs a clean copy |
| Almost everything is modelled, not measured | DC Water meters only a handful of outfalls. In 2025, 5.4 million gallons were metered and 248 million were model output — 98% of the reported total. Before 2019 the reports did not record the distinction at all. | no |
| 2005 Long Term Control Plan decree | Neither DC Water nor EPA publishes the original. Only the 2015 amendment is public. | court docket |
| December 2020 second modification | Filed as a joint stipulation; not published. | court docket |
| Treated flow before 2023 | EPA's ECHO service holds roughly three years of discharge monitoring data. | EPA bulk data |
| Event-level overflow timing | Reports aggregate to the quarter, though post-2019 outfalls are marked as metered — so finer data exists internally. | would need a request |
Where a gap could be closed by asking DC Water directly, that is the route we would take. They publish a great deal and their figures hold up to checking; the aim here is better resolution on data they already collect, not an adversarial records fight.
Full source list
| Source | Used for | Access |
|---|---|---|
| DC Water quarterly CSO operations reports | All overflow volumes, event counts, durations | Publications library |
| DC Water consent decree quarterly status reports | Milestone dates, obligation status | Publications library |
| DC Water Nine Minimum Control annual reports | Independent verification of 2012–2016 figures | Publications library |
| EPA enforcement records | Consent decree, First Amendment, LTCP targets | EPA settlement page |
| EPA ECHO, permit DC0021199 | Blue Plains treated flow and effluent quality | echo.epa.gov |
| NOAA GHCN-Daily, station USW00013743 | Daily rainfall at Reagan National | ncei.noaa.gov |
| Earthjustice, July 2015 | Plaintiffs' objection to the deadline extension | Hosted here |
Corrections are welcome. If a figure here disagrees with a DC Water publication, the published document wins and we will fix it.