Coverage measurement
Perimeter
Two coverage measurements over California's public wildfire datasets, published as counts. CAL FIRE and FRAP document the limits of these datasets in their own metadata. This project counts those limits, per year and per incident, so the people building on the data can reason about them.
Both datasets carry careful descriptions of what they do and do not contain. What is not published alongside them is the arithmetic: how many records per year, how many carry a federal identifier, how many cells in each field hold a value, how many hold a code meaning the value could not be determined, and how many hold nothing at all. Those are the numbers here. Nothing on these pages models fire risk, tracks an incident, or estimates a loss.
The two measurements
Historical fire perimeter completeness
23,334 perimeter records, counted per fire year, with IRWIN identifier coverage inside each year, surviving records measured against the published collection acreages, and counts of records sharing an identifier.
Damage inspection coverage
132,522 structure records across 451 incidents, with per-field and per-incident completeness, and the difference between a structure that could not be reached and a structure inspected and found undamaged kept visible throughout.
How absence is handled
Every measured cell is counted in one of three states, and the three are never collapsed into each other:
| State | What it means |
|---|---|
| Recorded value | A value the agency wrote down, including a recorded zero and including findings of absence such as No Damage or No Eaves. |
| Recorded as unknown | A published code or marker meaning the value could not be determined. Somebody recorded that determination failed. |
| Empty cell | Nothing was written in the cell. For DINS, CAL FIRE states that attributes with null values could not be determined. |
A null is not a zero. A structure not inspected is not a structure without damage. A field a dataset never collected is not a field full of zeroes. Where a share would have no denominator, these pages print words saying so rather than a number. Where a value that reads like a missing-data marker turns up in a field this project has not reviewed for that marker, the build stops rather than guessing what it meant.
Which judgment calls rest on what
Deciding that a cell holds a marker rather than a value is a judgment, and the two
measurement pages label every field with the basis its markers rest on. Twelve of the
twenty-seven fields that declare any marker or finding are published:
the value is a code in the layer's own coded-value domain, or a sentence in the
publisher's documentation, and it can be re-read at the endpoints listed under Sources
below. The other fifteen are inferred: the field is free text, or the
value is one the published domain does not carry, and this project read it off the
acquired file. Both sets are counted the same way and neither is a defect in either
dataset; both publishers document the domains they constrain and say which fields are
free text. The difference is in how much weight a reader should put on the call, so the
pages print it rather than leaving it to be assumed. Every entry, with its evidence, the
URL it can be checked against, its effect on the published figures and a confidence, is
in docs/MARKERS.md.
Sources
Both datasets are published by the California Department of Forestry and Fire
Protection under a Creative Commons Attribution licence, and both were read from the
GeoServices endpoint listed among the dataset's own published resources. Geometry was not
requested, because no measurement here needs it. Full detail, including byte counts and
file hashes, is in PROVENANCE.md and repeated on each page.