Public Data Demonstration
About & Methodology
This demonstration combines publicly available SF311 sewer odor complaints, historical weather observations, and publicly available wastewater infrastructure.
What This Demonstrates
Sewer odor complaints are spatial, temporal and meteorological events. When a complaint is paired with the wind observed at that place and hour, the search area for a plausible source narrows from the whole city to a cone extending upwind.
This application performs that pairing over the full public complaint history, screens publicly mapped wastewater treatment facilities and pump stations against every complaint, and aggregates the results so recurring patterns become visible.
Data Sources
SF311 service requests · DataSF
Publicly reported sewer odor complaints, including reported emission point, status, address and neighborhood.
City Facilities · DataSF
Publicly listed city facilities, used to locate wastewater treatment facilities and collection-system pump stations.
SFPUC catch-basin GIS · DataSF
Mapped catch basins with asset identifiers, system use and drain type. Used as spatial context only.
Open-Meteo ERA5 historical weather · Open-Meteo
Hourly historical wind direction, wind speed and temperature matched to each complaint location and hour.
How Potential Sources Are Screened
Potential sources are screened using complaint location, observed wind direction, wind speed and distance.
- Take the wind direction observed at the complaint location and hour.
- Compute distance and bearing from the complaint to each wastewater treatment facility and pump station.
- Discard assets farther than 5 miles, or more than 45° away from the direct upwind bearing.
- Score the remainder from alignment (70%) and distance (30%), and rank them. Wind speeds under 2 mph are flagged as low confidence because direction is unreliable when the air is nearly calm.
Screening results identify infrastructure consistent with observed conditions; they do not establish odor-source causation.
Why Pump Stations Matter
Wastewater treatment facilities and pump stations can both be potential odor sources. Pump stations can be potential odor sources because wastewater detention, wet wells, turbulence, force-main operation and ventilation can create or release odorous compounds such as hydrogen sulfide.
This does not mean every pump station produces odor. Many operate without generating complaints; the screening only reports which assets are geometrically consistent with a given complaint.
High-Consistency Matches
A complaint is labeled a High-Consistency Match when its top-ranked wastewater asset is within 1 mile, within 15° of directly upwind, and the reported wind speed is at least 3 mph. These thresholds are a restrictive screening filter applied after scoring; they do not change candidate eligibility, the scoring formula, weights or any screening threshold used elsewhere in the application.
A High-Consistency Match is a stronger geometric and meteorological alignment, not a confirmed source, and does not establish causation.
Wind-Informed Screening vs Nearest-Asset Proximity
As a diagnostic, each complaint's wind-informed #1 asset is compared with the asset that is simply closest to it. The two disagree for the large majority of screenable complaints, which shows that wind materially changes which individual asset is prioritized for investigation.
This comparison is a sensitivity check only. It is not a measure of accuracy, and neither method confirms an odor source. Because pump stations are far more numerous and more widely distributed than treatment facilities, the overall share of complaints whose #1 candidate is a pump station is influenced heavily by asset density and geography rather than by any property of the individual assets.
What Public Data Cannot Tell Us
- Public datasets do not include every relevant sewer asset or operating condition.
- Catch basins are shown for spatial context but are not included in candidate-source scoring.
- Reported complaint locations are approximate and reflect where an odor was noticed, not where it originated.
- Modeled historical wind is a regional estimate and does not capture street-level channeling, building wake effects or vertical mixing.
- Non-sewer odor sources, industrial activity and private-side plumbing are not represented.
How a Utility Deployment Would Improve This
For an actual utility deployment, authoritative GIS, sewer network data, SCADA, H2S monitoring, odor-control equipment, maintenance history and other operational data could substantially improve source investigation. Network topology would replace straight-line geometry, real-time telemetry would replace modeled weather, and work-order history would let screening results be validated against confirmed findings.
Attribution & Independence
SF311 / DataSF · City Facilities / DataSF · SFPUC catch-basin GIS / DataSF · Open-Meteo ERA5 historical weather. This is an independent public-data demonstration and is not affiliated with, endorsed by, or produced by the San Francisco Public Utilities Commission or the City and County of San Francisco.