A FIELD GUIDE TO THE EVIDENCE
A measurement is a beginning.
Not the whole story.
Satellites can show methane that our eyes cannot. Plume helps you keep track of what was observed, when it was observed, and where the evidence stops.
Explore the observationsThree quantities. Three different questions.
A colorful plume and an emission-rate estimate can come from the same acquisition. They still measure different things. The map’s legend describes its pixels; it does not describe a facility’s annual emissions.
Excess methane along the atmospheric column. This is what the map colors represent.
The estimated excess mass in a defined plume. A mass, without a time denominator.
An estimate associated with one acquisition. Provider methods and uncertainty matter.
A reported total for a specified year and boundary. It cannot be reconstructed by multiplying one satellite rate by a year.
Plume reproduces published estimates; it does not run a new emissions-detection model. A missing or suppressed rate stays missing. Provider uncertainty appears separately and is not converted into an invented confidence percentage.
Two dates are two moments.
Detections on two separate dates establish recurrence in the selected area. They do not establish continuous leakage, the fraction of time a source emits, or that the same facility caused both observations.
The timeline uses the actual acquisition times. Replay steps between records without inventing intermediate measurements. A gap means there is no record in this snapshot, not that a satellite looked and found nothing.
Seven records can still mean one observation.
The April 2024 Yemen observation has several published products and processing revisions. The shared plume identifier links them. An enhancement image, a mass product, and a revised rate are not independent confirmation.
Plume counts distinct acquisition identifiers and UTC detection dates, with the original records available for inspection. Cross-provider NASA / Carbon Mapper matches have not been verified in this release. We do not advertise cross-provider reconciliation as a demonstrated capability.
Click any finding in an investigation, change the evidence selection, and watch its dates and count recompute. The sentences come from versioned rules, with supporting evidence identifiers in the download.
Sometimes the useful answer is
“we cannot calculate that.”
A few satellite moments cannot establish a full year of methane emissions.
The Newby Island example includes an actual 2023 EPA Subpart HH reporting record: 7,954.5 metric tonnes of methane. Its satellite acquisitions are from 2024. Those records differ in time support, physical quantity, and boundary. Their proximity does not establish attribution.
The assessment exposes source association, spatial boundary, gas, quantity, time, method, uncertainty, and display permissions separately. It reports failed and unknown conditions rather than calculating an unsupported discrepancy.
Nearby infrastructure is context. A registry point is not a facility boundary or proof of an emitting source. Newby Island’s permitted landfill and composting activities illustrate why a place name alone cannot resolve the source process. A reviewer needs source-location evidence and the relevant operational boundaries.
Case evidence and worldwide context.
The regional color layer shows monthly mean atmospheric methane from Copernicus Sentinel-5P, processed by S5P-PAL. It is column-averaged methane mixing ratio in ppbv, not an emission rate or a ground-level exposure measurement. The period selector controls only this layer; NASA plume markers and case observations retain their own dates. The provider filters measurements to qa_value > 0.50. Missing observations remain transparent; absent tiles in the provider’s sparse pyramid are displayed as transparent tiles. Tile colors use bilinear display interpolation, without deriving new numeric measurements or filling unobserved regions. The provider’s original color scale is fixed at 1,680–1,980 ppbv, with end colors including values outside that range. The default display applies a linear RGB tint to the provider’s precolored tiles and the matching legend, using teal, sage, gold, and warm terracotta. This changes presentation only: it preserves transparency and does not recover or alter numeric measurements. Select “Provider colors” to see the original palette. Tiles are regional context, fading out between zoom levels four-and-a-half and six to avoid enlarging coarse regional pixels; zooming further reveals available individual plume observations. This atmospheric layer is not included in investigation evidence exports. Contains modified Copernicus Sentinel data processed by S&T / S5P-PAL. Provider methodology · Layer provenance and periods.
The worldwide map layer includes all 1,686 unique plume records in the saved NASA/JPL EMIT methane feed, checked on 19 September 2026, with observations from 10 August 2022 to 22 September 2025. Point and polygon representations of the same plume are counted once. Amber markers locate the maximum enhancement pixel; they are not source-origin estimates. Cluster numbers count catalog records, not independent acquisitions or emission rates. Zoom in for published footprints and select a marker for its date, peak column enhancement, and source data. This historical catalog does not provide continuous global coverage. NASA records are separate from Carbon Mapper case selections and are not added as independent corroboration or included in case exports. Catalog provenance · NASA EMIT portal.
Place search includes a separate Santiago regional catalog excerpt: 101 published methane plume records from EMIT, Tanager-1, and AVIRIS-NG, retrieved on 19 September 2026. Gold markers locate detections; cluster numbers count plume records; a selected outline shows the provider’s footprint. Three recent Tanager-1 acquisitions also have prepared methane imagery. Their default contour view uses linear interpolation within fully measured cells, with 64 color bands on a shared 0–2,500 ppm·m inferno scale. No spatial smoothing or filling of missing cells is applied. Switch to sensor pixels to inspect the original samples. Other dates show outlines only. Regional records are separate from reviewed case exports. Processing versions of one plume are counted once; different plumes may share an acquisition. That Carbon Mapper excerpt is regional; the separate NASA context layer extends worldwide. Inspect the regional catalog and provenance.
This release contains five reviewed EMIT observations in two geographic investigations: Yemen and Newby Island, California. Their dates run from 20 April to 23 October 2024. Records were retrieved on 19 September 2026. This is historical evidence, not live monitoring or a global methane survey.
Search looks up major world cities and US ZIP centroids, then checks their distance from the included observations. Finding a place does not promise methane coverage. An empty result means this dataset has no included observations nearby.
Numeric provider rasters are reprojected to Web Mercator with nearest-neighbor sampling, then rendered as labeled, interpolated contour bands by default. The sensor-pixel toggle shows the original sample grid. Reviewed maps use inferno with fixed per-case scales; full acquisition scenes and exported evidence retain their original cividis rendering. Valid zero pixels remain valid; missing pixels remain transparent. Comparisons share one numeric scale within each case. Different colors or rate estimates at different dates do not establish a change in continuous emissions.
These satellite column measurements do not establish ground-level exposure, local health risk, regulatory violations, or a facility’s responsibility.
The context travels with the evidence.
Each PDF includes acquisition dates, measurement types, limits, original source links, and attribution. The ZIP adds structured observations, geometries, claims, relationships, imagery, and a checksum manifest. An evidence fingerprint identifies the snapshot and selection; it is not a certification that a scientific conclusion is true. PDF timestamps may differ between exports.
Notes are stored only in this browser. Public evidence links contain the case and selection, never your notes. Notes enter a download only when you explicitly select that option. This app has no accounts and no server-side notes database.
- Carbon Mapper — data and imagery
Source: Carbon Mapper. Noncommercial use, attribution, and applicable downstream terms remain attached to the data and derivatives. This independent project is not endorsed by Carbon Mapper.
- EPA — Newby Island reporting record
Facility 1006179, reporting year 2023, methane, ghg_quantity. Facility coordinates come from the EPA 2023 data summary workbook.
EPA Subpart HH methodology - Bay Area Air District — permitted operations
November 2022 permit, Table II-A: landfill waste decomposition and composting operations. Used as process context; no invented component coordinates.
- GeoNames — place lookup
Major world cities and US postal data, CC BY 4.0. Centroids are approximate and may not represent an entire postal area.
- OpenFreeMap / OpenStreetMap — basemap
Visible map attribution remains on every interactive map. Locator country outlines use Natural Earth public-domain geography.
The paper landscape is decorative generated artwork. It never represents methane data. All scientific imagery in the app and reports is derived from the identified provider rasters.