Before the Call — Oso Landslide, Washington

The Oso first call is a near-perfect example of the scene size problem: the caller reported what they could see, which was a fraction of what had actually happened. A roof in the road is a legitimate emergency. It is not the description that triggers a mass casualty response. But a mass casualty event had occurred — and the first minutes of the response were scaled to the call, not the incident.

• Escalating call volume from the same area is a signal, not just volume. A single call about a roof in the road is one thing. A second call. A third. Each one describing something slightly different and slightly worse. That pattern — multiple callers, same general location, escalating descriptions — is a signal that something larger than the first call described is happening. The protocol question is: at what point does that pattern trigger a resource escalation independent of what any single caller has described?
• Geographic isolation raises the stakes on ambiguous initial calls. Oso is 17 miles east of Arlington on a two-lane highway. Darrington is another 16 miles east of Oso. Mutual aid travel times in that terrain are not minutes — they are tens of minutes to an hour. When your service area includes isolated communities with long mutual aid response times, the cost of under-dispatching to an ambiguous initial call is much higher than in urban areas. The calculus for what triggers a heavy initial response should reflect that geography.
• The "roof in the road" call type has a landslide differential diagnosis. In western Washington, a roof in the road during or after heavy rainfall is not just a structure problem — it is a potential landslide indicator. Know what the landslide risk indicators are for your service area: recent heavy rainfall, known slide-prone hillsides, river corridor terrain. A call that might be unremarkable in a flat urban area has a different meaning in Snohomish County in March after 200% of normal rainfall.
• What would have changed the response posture? A second call from a different location describing something larger. An aerial asset already in the area. A dispatcher with local knowledge who recognized the terrain and the rainfall history. None of these are guaranteed. But knowing which of them your center has access to — and which it doesn't — is part of understanding your actual response capability for ambiguous initial calls in isolated terrain.

The Oso split incident is a scenario with almost no training precedent for most dispatch centers. A single event produced two geographically isolated halves, each with its own on-scene command, each unable to reach the other by ground, each calling dispatch for resources that would have to be routed through entirely different access points. Dispatch became the only communications node connecting two incident commands that couldn't talk to each other directly.

• Dispatch is the connective tissue when the scene is split. When two incident commands cannot communicate directly, every piece of information that needs to cross the divide goes through dispatch. That means your documentation has to be precise, your relay has to be accurate, and you have to be tracking both sides of the incident simultaneously without conflating them. "East end reports" and "west end reports" need to stay clearly separated in your CAD notes.
• Resource routing becomes a navigation problem, not just a dispatch problem. When the only road connecting two halves of an incident is destroyed, every resource request has to be evaluated against access. A unit that can reach the east end cannot reach the west end. A helicopter can reach both. Knowing which resources can reach which part of the scene — and routing requests accordingly — becomes a critical dispatch function from the first minutes.
• Air assets are not optional in a split incident with no ground access. At Oso, SnoHawk 10 was the first resource to get a complete picture of the scene — and it arrived about an hour after the slide. Fenstermaker's first request when he realized the scope was "all the helicopters I could get." In terrain where ground access can be cut off, aerial asset availability and activation should be an early-escalation trigger, not a fallback.
• The split incident needs a unified command designation explicitly. Two incident commands working the same event from opposite ends will make contradictory resource requests, give conflicting information, and create accountability gaps unless unified command is established. That designation — and the communications architecture that supports it — needs to happen early and needs to be communicated to dispatch so you know who you're routing information to.

Radio saturation is one of the most predictable and least trained-for problems in large-scale incident response. When dozens of agencies with different radio systems, different protocols, and different urgency levels all try to communicate simultaneously on a limited number of frequencies, the result is a communications environment where critical transmissions are buried in noise. At Oso, this happened within the first hour and persisted throughout the active rescue phase.

• Radio saturation is a predictable consequence of multi-agency response — plan for it, don't just react to it. Every large-scale incident that draws resources from multiple agencies will create frequency competition. The question is whether your center has a protocol for frequency assignment and traffic management before saturation hits, or whether you're improvising after it does.
• The incident commander's after-action insight is the training point. The west side IC said he would have "appointed command post aides to speak and receive information on the radio, forwarding only the most important details to me." That is a named role with a named function. Does your center's MCI protocol include a radio traffic management function? Does someone own that job, or does everyone assume someone else is managing it?
• Text as a supplement to radio is documented and effective — and needs a protocol. Navy SAR crews and Snohomish County assets both used text to augment radio at Oso. Text doesn't get stepped on. Text creates a written record. Text doesn't require a clear frequency. But text also doesn't carry urgency the way voice does, and it requires a device that may not be standard issue. If your center's MCI protocol doesn't address when and how text supplements radio, that gap is worth filling before you need it.
• Dispatch's role in frequency saturation is assignment and discipline, not just monitoring. When additional dispatchers were called in at SNOPAC, additional frequencies were also assigned. That's the right move. But assigning frequencies only helps if responders know which frequency to use for which function — and if someone is enforcing discipline about what goes on each channel. That enforcement function lives partly at dispatch.

Documentation suspension at Oso is understandable and still a problem. When every available person is focused on active rescue of survivors, stopping to document feels like the wrong priority. But the incomplete record of the first hours at Oso had consequences: the after-action analysis was limited, accountability for early decisions was murky, and the institutional learning that came out of the event was harder to root in specific decisions and timelines.

• Documentation at dispatch should not require a separate decision to suspend. Your CAD is running whether you're in surge or not. Call entries, unit assignments, and radio traffic logs are created by the act of dispatching. What degrades in surge is the quality and completeness of supplemental notes — the context, the decisions, the information that doesn't automatically capture. That degradation is the cost, and it's worth naming explicitly.
• The FEMA documentation handoff is a model worth understanding. When FEMA arrived at Oso and assumed the documentation function, it was doing something that most incident command structures don't plan for in advance: designating a dedicated documentation role that is explicitly not also a response role. The people documenting are not the people rescuing. That separation is the only way documentation survives a surge event intact.
• Your CAD notes are the institutional memory of an incident. Every decision made during an MCI — which resources were assigned where, what information was relayed to command, when mutual aid was activated, what was known and when — should be traceable through your CAD record. That record is used for after-action review, for legal accountability, for training, and for the next incident. A gap in that record is a gap in all of those things simultaneously.
• In surge, minimum documentation beats no documentation. If complete notes are impossible, name the minimum: time, unit, assignment, and any safety-critical information relayed. Even a stripped-down record is more useful than nothing. Establishing that floor explicitly — so dispatchers know what "good enough" looks like when "complete" isn't possible — is a training decision worth making before the next MCI.

The Oso aerial assessment problem is the defining dispatch challenge of the event. For approximately an hour, every ground resource was responding to a scene whose actual dimensions were unknown. Resources were being staged, assigned, and deployed based on partial information from callers and ground responders who could only see their immediate surroundings. The first complete picture came from the air.

• Aerial assessment is not a luxury in terrain like Snohomish County — it is a primary intelligence tool. When the debris field covers a square mile and the only road is buried, ground-based scene assessment is structurally limited. The question is not whether you need aerial assets — it's whether you're requesting them early enough. Fenstermaker requested all the helicopters he could get as soon as he understood the scale. That request should come earlier in the incident, before the scale is fully understood, in terrain where ground assessment is inherently limited.
• Dispatch to the worst credible interpretation when scene size is unknown. You don't know what you're dealing with. The calls describe a roof in the road, houses out in the highway, people screaming for help. You can't see it. Your ground resources can't see it. In that environment, the right posture is to dispatch as if the worst description is accurate and let first responders downgrade if the scene is less severe. The cost of over-dispatching to an ambiguous scene is manageable. The cost of under-dispatching to a scene that turns out to be Oso is not.
• Know your aerial asset inventory and activation time before you need it. SnoHawk 10 was diverted from training. That means it was available — but not pre-positioned for this event. Know what aerial assets your county or region has, what their activation and response times are, and what the request pathway is. In terrain where ground access can be cut off instantly by a landslide, that knowledge is operational, not administrative.
• CAD notes about scene uncertainty are valuable. "Scene size unknown — aerial assessment pending" in your CAD notes is not an admission of failure. It is an accurate record of the information state at that moment. It tells the next dispatcher, the next supervisor, and the after-action reviewer exactly what was known and when. Document the gaps, not just the facts.