Multi-Medium Wildfire Detection

Regulatory and Operational Context

Wildfire monitoring is not yet governed by a single statute, but a stack of state and federal instruments now defines the obligations of detection operators. California Public Utilities Commission General Order 165 mandates inspection and recordkeeping for overhead and underground electric supply and communication facilities, and post-2017 amendments tie inspection findings to wildfire-mitigation planning under PUC Rulebook 18-12-005 and SB 901. The FCC Wireless Emergency Alerts (WEA) framework, codified at 47 CFR Part 10, governs the geo-targeted alerting channel that state and local emergency managers use to notify the public of imminent fire threats, with WEA 3.0 enhancements requiring polygon-precision targeting and embedded URLs. NIST Special Publication 1900-200 on the Smart Fire Fighting roadmap, together with NIST IR 8262 on public-safety sensor frameworks, sets the technical baseline for multi-spectral fusion across visible, infrared, smoke chemistry, and RF disturbance modalities.

Operationally, the detection landscape spans the National Interagency Fire Center (NIFC) and its supporting USFS, BLM, and NPS systems; state fire authorities including CAL FIRE, Oregon Department of Forestry, and Arizona DFFM; investor-owned utilities running PSPS programs under SB 901; and private-sector detection networks such as Pano AI, AlertCalifornia (UCSD), and Cornea. Each contributes a different modality at a different cadence: GOES-18 satellite radiometry at multi-minute intervals, ALERTCalifornia mountaintop cameras at 10-second cycles, utility-grade arc-fault detectors on distribution feeders at sub-cycle resolution, and crowdsourced reports through 911 PSAPs at human latency. No single source is sufficient, and no two sources are calibrated against each other.

Architectural Requirement

A wildfire detection architecture that can support both rapid response and post-event accountability must satisfy three properties. First, multi-medium corroboration: an alert that triggers a Public Safety Power Shutoff or a WEA broadcast cannot rest on a single sensor, because the consequences of a false positive (de-energizing a hospital feeder, evacuating a metropolitan area) are themselves harmful. Second, lineage preservation: every observation that contributed to a decision must be traceable to its source, its calibration state, and its temporal context, because PUC, NTSB-equivalent state boards, and post-fire civil litigation will all probe the evidentiary chain. Third, governed active probing: when passive observations are ambiguous, the architecture must be able to task an active sensor (a UAV thermal pass, a utility line-test) under explicit authorization rather than as an ad-hoc operator action.

The architecture must also accommodate the asymmetry between detection modalities. A visible-spectrum camera at dusk sees smoke plumes that an infrared sensor at the same site cannot resolve against a hot rock face; a smoke-chemistry sensor distinguishes biomass combustion from diesel exhaust at concentrations where a particulate sensor cannot; an RF arc-fault detector identifies the ignition event minutes before any optical sensor sees the resulting flame. Fusing these signals is not a matter of voting; it is a matter of weighting each modality by its physical applicability to the candidate fire class.

Why Procedural Compliance Fails

Current cross-agency coordination relies on procedural mechanisms, MOUs between NIFC and state authorities, data-sharing agreements between utilities and CAL FIRE, ad-hoc API integrations between private detection vendors and emergency operations centers, that do not produce a coherent observation record. When a fire is investigated post-event, as with the Camp Fire (2018), the Dixie Fire (2021), or the Maui fires (2023), investigators reconstruct the detection timeline from logs that were never designed to be reconciled, and the question of "who knew what, when" becomes a months-long forensic exercise. PUC General Rate Case proceedings and CPUC Safety and Enforcement Division investigations inherit the same fragmentation.

The procedural regime also struggles to absorb new sensing modalities. When a utility deploys a new generation of distribution-feeder arc-fault detectors, or when a state agency adds a smoke-chemistry sensor network funded under the federal Infrastructure Investment and Jobs Act wildfire provisions, integrating the new feed requires bilateral negotiations with each consuming agency, custom data-mapping work, and case-by-case calibration cross-checks. The integration cost scales with the product of the number of sensor types and the number of consuming agencies, which is the opposite of what the rate of sensor proliferation requires. Pano AI's mountaintop network alone has expanded by an order of magnitude in three years, and the procedural integration regime has not kept pace.

Procedural integration also fails under the time pressure of an active ignition. When a utility's arc-fault detector trips at the same minute that a private camera flags a heat plume two miles downwind, the operator on shift has no structured way to determine whether these are the same event, and the procedural answer, call the other operator, does not scale to the dozens of simultaneous candidate ignitions a red-flag day can produce. The result is either over-alerting, which erodes WEA credibility under 47 CFR 10.520, or under-alerting, which is the failure mode that defines post-fire liability.

What Environmental Disruption Provides

The environmental-disruption primitive treats every sensor reading as a credentialed observation carrying its source identity, modality, calibration state, and temporal frame. Multi-source corroboration is structural rather than procedural: a candidate fire event is constructed by fusing observations whose spatial and temporal overlap exceeds a configured threshold, and the fusion weights each modality by its physical applicability to the candidate class. A heat plume corroborated by a smoke-chemistry signature and an upstream arc-fault event clears the corroboration threshold for a high-confidence ignition; the same heat plume alone, against a background of agricultural burns, does not.

Multi-medium sensing is the explicit acknowledgement that no single modality is decisive. The primitive ingests visible, near-infrared, mid-wave infrared, smoke-chemistry, RF arc-fault, and acoustic sources under a common observation schema, with calibration metadata and uncertainty bounds preserved. Governed active probing converts the ad-hoc decision to task a UAV or initiate a line-test into a structured authorization: the request carries the corroboration state that justified it, the expected information gain, and the authorization envelope under which it executes. Signed observation lineage produces the evidentiary chain that survives PUC review and civil discovery, because every fused observation can be reduced to its contributing sensors and their states at observation time.

Compliance Mapping

The primitive maps onto each layer of the regulatory stack. PUC General Order 165 inspection records become signed observations from credentialed inspection sources, fused with ongoing telemetry from the same assets, so that a deferred inspection finding raises the corroboration weight of subsequent arc-fault detections in the same circuit. SB 901 wildfire-mitigation plan obligations bind to the corroboration thresholds: the plan declares which combinations of observations justify a PSPS de-energization, and the architecture executes that declaration with signed lineage rather than operator interpretation.

FCC WEA 3.0 polygon targeting under 47 CFR 10.450 inherits the corroborated event geometry directly: the polygon is constructed from the fused observation footprint, not redrawn by an operator, and the WEA broadcast carries a reference to the lineage that justified it. NIST SP 1900-200 multi-spectral fusion guidance is satisfied by construction, because the observation schema preserves the modality and calibration state that the guidance requires. For post-event review, NTSB-equivalent state boards and civil litigants receive a single authoritative timeline rather than a reconstruction from disparate logs, which compresses investigation cycles from months to days.

Adoption Pathway

Adoption begins at the operational coordination layer, where the immediate value is highest and the integration cost is lowest. A state emergency operations center or a utility wildfire operations center can ingest existing sensor feeds, GOES-18, ALERTCalifornia cameras, in-house arc-fault detectors, into the observation schema without modifying the upstream systems, and produce corroborated event records that immediately improve PSPS and WEA decision quality. Cross-agency federation follows: NIFC, CAL FIRE, and participating utilities exchange credentialed observations under declared trust relationships, without exposing internal sensor infrastructure.

Private-sector detection vendors integrate through credentialing rather than bespoke API work, which is a practical precondition for the rapid expansion of the detection landscape that climate-driven fire behavior is forcing. Governed active probing, the tasking of UAV thermal passes, line-tests, or aircraft-based observation, is the last layer to onboard, because it requires the strongest authorization model, but it is also where the architecture displaces the most operator workload during red-flag conditions. By the time a fire season produces its first major ignition under the architecture, the post-event review has already been pre-structured, and the emerging cross-agency monitoring standards that NIST and the U.S. Fire Administration are developing find a substrate ready to receive them.

Investor-owned utilities operating in High Fire Threat District (HFTD) Tier 2 and Tier 3 territory have the most acute incentive, because their PSPS decisions are directly subject to PUC scrutiny and to the financial consequences of both over-shutoff and under-shutoff. State emergency management agencies follow naturally, with WEA polygon construction and inter-agency coordination as the immediate operational gains. Federal participation through NIFC and USFS adds the satellite radiometry layer and the cross-state coordination that makes the architecture useful for fires that cross jurisdictional boundaries. The pathway is compatible with existing CAD and incident-management systems, including IRWIN and WildCAD-E, so that adoption does not require operations centers to abandon the dispatch tooling their personnel are trained on. Within two fire seasons of initial deployment, the corroborated event record becomes the primary input to PSPS decisions, WEA broadcasts, and post-event review, and the procedural coordination layer recedes to a fallback for cases the architecture has not yet been configured to handle.