An activation level is a statement about staffing and delegated authority, not a weather forecast, and jurisdictions get into trouble when they treat the two as the same thing. This piece covers what the levels actually declare, why the numbering runs in opposite directions between neighboring agencies, the trigger points worth writing into the plan before the phone rings, what it costs to activate late against what it costs to activate for nothing, and how to staff a round the clock operation with people who have day jobs and no depth behind them.

What an activation level actually declares

When an emergency manager announces an activation level, the useful content of that announcement is a list of positions that will be filled, a schedule those positions will work, and a set of decisions that have been delegated to the people sitting in them. Everything else is labeling. A jurisdiction that says it is at partial activation and cannot tell you which five seats are occupied and who is authorized to obligate money has announced a mood rather than a posture, and the neighboring agencies calling for help will discover the difference within the hour.

The common structure runs from a steady state, in which nobody is in the room and a duty officer carries a phone, through a monitoring or enhanced watch posture where one or two people are tracking a developing hazard and making notifications, through a partial activation that brings in the emergency support functions relevant to the specific hazard, to a full activation with every function staffed on a defined operational period. Those four postures exist under many different names. Some states use words, some use numbers, some use colors, and a few use all three in different documents produced by the same office.

What matters more than the label is the separation between three things that often get bundled together and should not be. Opening the EOC is an operational decision that an emergency manager can usually make alone. Declaring a local state of emergency is a legal act by the chief elected official or the governing body under state statute, and it is what typically unlocks emergency purchasing, price gouging protections, curfew authority and the path toward a state or federal declaration. Requesting state assistance is a third act with its own form and its own approval chain. You can be fully activated with no declaration and no request in the system, which is a perfectly normal condition for a large planned event, and you can have a declaration signed while the EOC is dark, which is less common but happens in slow onset events like drought and extended flooding.

NIMS, in the third edition published by FEMA in October 2017, describes several ways an EOC team can be organized, including a structure that mirrors ICS, an incident support model that puts situational awareness and resource support at the top of the chart rather than operations, and a departmental structure that keeps agencies intact. The activation level interacts with that choice, because an incident support structure scales differently than an ICS mirror does, and a plan that says “activate the EOC” without saying which chart goes on the wall is leaving an argument for the worst possible moment to have it.

Level 1 or Level 3, and why the number is a trap

FEMA has used a descending scheme at the National Response Coordination Center, where the steady state watch carries the highest number and full activation is Level 1, and many state emergency management agencies adopted the same direction. Plenty of local jurisdictions did the opposite, numbering upward so that Level 1 is a monitoring posture and Level 3 or Level 4 is everything lit up. FEMA’s own terminology at the national level has changed more than once over the years, so confirm the current scheme in current FEMA doctrine rather than in a slide deck somebody saved from a class.

The practical consequence shows up on mutual aid calls and in regional conference bridges. A county telling its neighbor that it has gone to Level 2 has communicated nothing reliable, because the neighbor’s Level 2 may be a single duty officer with a laptop or may be a room with twenty people in it. I have sat on regional calls where two adjacent jurisdictions spent several minutes discovering that they were describing opposite conditions with identical words, and that is several minutes spent during the part of an event when minutes are the only resource nobody can request.

The fix is not to renumber your scheme to match the state, although matching is fine if you can get it done. The fix is to stop leading with the number in any communication that crosses an agency boundary. Say what you have: the EOC is open with logistics, public information, public works and the health department staffed, one twelve hour shift, no overnight coverage yet. That sentence survives translation into any other jurisdiction’s system, and it gives a neighboring emergency manager what they actually need, which is whether there is a human being who will answer a resource request at three in the morning.

Inside your own organization the number still earns its keep, because it is shorthand for a staffing roster, a callback list and a set of delegations that people have rehearsed. Keep that internal utility and give up the illusion that the number travels. Write into the plan that external notifications state the level, the functions staffed, the hours covered and the name and callback number of the person in charge, in that order, and the ambiguity mostly disappears.

Trigger points worth writing down in advance

A trigger is a condition that is observable by somebody who is not in the room and that maps to an action. The value of writing triggers down is that it moves the decision out of the worst hour and into a quiet afternoon when you have time to think about it, and it gives a duty officer at midnight the authority to act without waking a director who may not answer. Triggers should be specific enough that two different people looking at the same information reach the same conclusion.

The ones that work best are tied to products and readings that somebody else publishes. A National Weather Service watch or warning for a specific hazard in a specific zone is a clean trigger because it arrives with a timestamp and an issuing office. A river gauge reaching a named stage at a named site is cleaner still, because the stage can be tied to what floods at that stage in your county, and the forecast crest from the river forecast center gives you lead time. Tropical systems give you an H-hour countdown built from the forecast arrival of tropical storm force winds, which is the point at which outdoor work stops and evacuation traffic must already have cleared, so the decision points count backward from that hour rather than from landfall. Fire weather triggers can combine a red flag warning with local fuel and staffing conditions.

Not all triggers come from the weather. Reasonable ones include a defined number of simultaneous working incidents that exhausts your command staff, a multiple alarm assignment that strips coverage from more than a set fraction of the jurisdiction, a utility outage affecting a named percentage of customers or a named duration, the opening of any shelter, the loss of a public safety answering point or a radio site, a hazardous materials incident requiring protective actions beyond the immediate hot zone, a hospital going on full diversion or losing a critical service, a request for assistance from an adjacent jurisdiction, and a state EOC activation for an event that includes your county. Planned events deserve their own trigger list built from expected attendance and the security posture, because a stadium event or a large demonstration is a known date on the calendar and there is no excuse for improvising.

Write down who may pull each trigger, and make the authority generous on the way up and narrow on the way down. A common and sound arrangement is that any on-call duty officer, the fire chief, the police chief, the 911 director or the emergency manager may activate to the monitoring or partial level without asking anybody, while moving to full activation requires the emergency manager or the chief administrative officer, and deactivation requires the emergency manager alone. Put the callback tree next to the trigger list, confirm the phone numbers on a schedule, and make sure the person on call tonight has the current version in a form that works when the network does not.

The cost of activating late

Activating late costs you lead time on the things that cannot be compressed. Resource requests queue at the state EOC in the order they arrive, and the counties that submitted their requests for generators, water, cots and swiftwater teams on Friday afternoon are ahead of the county that submitted on Saturday night, regardless of relative need, because the state has already committed the inventory. Pre-event contracts for debris monitoring, fuel and equipment are cheapest and most available before the event is certain, and the vendors who have signed with a dozen other jurisdictions are not going to answer your first phone call during the event with a good price or a firm date.

Evacuation is where late activation does the most damage, because clearance time is an arithmetic problem about road capacity and it does not care when the decision was made. Hurricane evacuation studies for coastal regions produce clearance times measured in many hours and, for larger regions with limited outbound capacity, in more than a day, and the decision point sits at the beginning of that window rather than at landfall. The Hurricane Floyd evacuation of the Southeast coast in September 1999 produced gridlock on the interstates out of Savannah and Charleston that was widely reported at the time, and South Carolina subsequently adopted lane reversal plans for I-26 that it has used in later storms. The lesson those states drew was about capacity and about timing together, since a plan that requires eighteen hours of clearance fails identically whether the plan is wrong or the order is twelve hours late.

The Katrina record is the most examined case of decision timing in American emergency management, and I would point readers to the primary documents rather than to summaries. The White House report on the federal response to Hurricane Katrina, published in February 2006, and the House Select Bipartisan Committee report titled A Failure of Initiative, published the same month, both examined the sequence of warnings and decisions before landfall on August 29, 2005, including the unusually direct advisories issued by the National Hurricane Center and the personal calls its director made to state and local officials in the two days beforehand. The mandatory evacuation order for New Orleans came on the morning of Sunday, August 28. Read the findings in those reports rather than taking anyone’s characterization of them, including mine.

Late activation also costs you your own people. Staff who have already worked a full day at their regular jobs before the EOC opens start the operation with a deficit they never recover, and the first operational period ends up staffed by people who should be going home. Opening at noon with four positions when the forecast is bad costs a fraction of what it costs to open at eleven at night with everybody already tired.

The cost of activating for nothing

The opposite failure is real and it is undercounted, because nobody writes an after action report about the activation that turned out to be unnecessary. Every full activation spends overtime, pulls people away from their regular duties, burns goodwill with the private employers who release volunteer staff, and makes the next callback slower. The third time you bring in twenty people at 4 a.m. for a storm that goes north, the fourth callback gets answered by fewer phones, and the people who do answer arrive slower and less convinced. That erosion is gradual and it is invisible until the event that matters.

There is a public dimension too, although I would be careful about how it gets described. The research on whether false alarms degrade public response to later warnings is genuinely mixed rather than settled, and I am not going to assert a finding I cannot point you to. What I am confident about, from watching it happen, is the internal version: the credibility an emergency manager spends on unnecessary activations is drawn from the same account that pays for the hard call later, and an elected official who has watched three expensive activations produce nothing is a harder audience for the fourth request.

The way out of the dilemma is to stop treating activation as a switch with two settings. Most of the cost sits in the jump to full activation with a twenty four hour schedule, and most of the benefit of activating early is available at the monitoring level, which costs one or two people and a conference bridge. A monitoring posture that opens twelve hours before a partial activation gives you notifications made, call lists confirmed, a situation report published, resource requests drafted and shelters put on standby, and it costs almost nothing if the hazard misses. Jurisdictions that use the low levels freely and the high levels carefully tend to have both the lead time and the credibility.

The other half of the answer is honest after action treatment of the activations that produced nothing. If the trigger fired and the hazard did not materialize, that is the trigger working as designed and the record should say so, because a trigger set at the point where you are always right is set too late. If the trigger was wrong, adjust it in writing, date the change, and tell the people who got called out what changed and why. Staff tolerate being called out for nothing far better when somebody explains the arithmetic afterward.

Staffing around the clock with people who have day jobs

Twenty four hour operations run on twelve hour shifts in most EOCs, usually with an overlap of thirty to sixty minutes for the shift change briefing, and that arithmetic means two people minimum for every position you intend to staff overnight. Two is not enough, because people get sick, get called to their own agency’s emergency, have a child at home with nobody else to watch them, or simply cannot do day fourteen of an extended operation. Three deep per position is the working target, and most jurisdictions I have worked with are one deep on at least half their positions and know it.

The depth problem is worse than a roster count suggests, because EOC positions are usually filled by people whose primary job is somewhere else. The public works representative is the same person who is supervising the crews clearing the roads. The health department liaison is running the point of dispensing. The person who knows your resource ordering software is the finance clerk who also processes payroll, and payroll does not stop during a hurricane. When you build the depth chart, mark every name that is also committed to a field or departmental role during the same event, because that name is not actually available and the chart is lying to you.

Solve it in advance with three concrete instruments. The first is a written agreement with the employers of your non-government EOC staff, including the volunteer organizations, the utility and the private sector representatives, stating what release looks like and who calls whom. The second is a named alternate for every position, trained to the same standard and rotated through real activations and exercises rather than existing only on paper, which means giving the alternate the seat during the next planned event even though the primary is available. The third is a position checklist, one or two pages, that lets somebody who has not worked the seat in eight months be useful in the first hour, covering what the position does, who it talks to, what it produces and by when.

Shift discipline is the part that separates operations that hold up for a week from those that fall apart on day three. Set the operational period length and the shift change times in the plan and hold to them, run a proper shift change briefing rather than a hallway handoff, require the outgoing person to leave written notes in the incident management system, and enforce off duty time including sleep arrangements for people who cannot get home. The ICS planning cycle exists partly to make this predictable, since a fixed operational period gives every position a known deadline for input and a known time when the next plan is briefed. Watch the people who volunteer for every shift, because they are the ones who will make the consequential error on hour thirty, and the supervisor who is working a position full time is not watching anybody.

Virtual and partial activation as the middle setting

Remote and hybrid EOC operation moved from an experiment to a normal option during the COVID-19 response, and most jurisdictions now run some functions from a screen. It works well for situational awareness, for liaison positions whose main product is information from a parent agency, for planning and documentation work, and for scheduled briefings. It works poorly for anything that depends on overhearing, which includes most of the informal coordination that makes a physical room valuable, and it works poorly for resource ordering when the people involved have not built the working relationship face to face first.

If virtual operation is in your plan, then the dependencies belong in the plan as well, and they are not trivial. Staff working from home are on residential internet and commercial power, both of which fail in exactly the events that cause you to activate, so the plan needs to say what happens when the platform is unreachable and who has a physical fallback. Access to your incident management platform requires accounts that exist and passwords that have not expired, which is an administrative task somebody has to own on an ongoing schedule rather than discover during an activation. Sensitive information handling and public records obligations do not relax because the meeting is on a screen, and in many states the open meetings and records rules apply to what the EOC produces regardless of where the staff were sitting.

Partial activation is where the hybrid approach pays off most clearly, because you can bring the two or three functions that must be in the room physically together and leave the rest on the bridge, then convert to full physical staffing if the event grows. Decide in advance which positions are always physical, and in most jurisdictions that list includes the EOC manager or director, operations coordination, resource ordering and public information, with the reason being that those four spend the most time talking to each other in fragments too small to schedule.

Coming back down, and the record you keep

Deactivation criteria deserve the same written treatment as activation triggers and almost never get it. Write down what conditions bring you from full to partial and from partial to monitoring, because otherwise the EOC empties out when people get tired rather than when the work is done, and the functions that remain busy are usually the ones with the fewest staff. Reasonable criteria include the closure of the last shelter, the restoration of utility service above a named threshold, the completion of initial damage assessment, the return of the last mutual aid resource, and the transfer of remaining tasks to a named recovery structure with a named lead.

The transition to recovery is the point where most jurisdictions lose continuity, because the EOC closes and the work moves to a handful of people in departments that were not in the room. Before deactivating, produce a written transition document listing every open action, the responsible department, the due date and the funding source, and brief it to the people who will carry it. The site’s article on recovery covers what that long tail involves, and the connection to this topic is that a clean deactivation with a written handoff is the difference between a recovery that starts on Monday and one that starts whenever somebody notices.

The record you keep during the activation determines what you can prove afterward, and the documents an auditor or an attorney will ask for are the position logs, the timestamped situation reports, the EOC action plan for each operational period, the resource requests with their disposition, and the sign-in and sign-out sheets that establish who was present and for how long. Add a decision log to that list, since it records what was decided, by whom and on what information, and it is both the document most often skipped and the one most valuable two years later, when somebody wants to know why the evacuation order came at the hour it did and the only honest answer available is what was written down at the time.

Finish with an after action review that examines the activation decision itself rather than only the response. Ask when the trigger fired, when the EOC actually opened, what the gap was, what caused the gap, and whether the trigger was set at the right point. That question is separate from whether the response went well, and it is the one that improves the next activation, since an operation can run beautifully from an opening that came four hours too late and the report will say nothing about it unless somebody asks.

What to do at your agency

• Have your emergency manager write the activation level table on one page this month, listing for each level the positions staffed, the hours covered, who may order that level and who may cancel it, and post it in the EOC and in the duty officer packet.
• Call the emergency manager in each adjacent county and at your state EOC, ask which direction their activation numbering runs, and write the answer next to their entry in your notification list so nobody has to ask during an event.
• Build a written trigger list with at least six specific conditions drawn from National Weather Service products, named river gauge stages, utility outage thresholds and incident counts, and have the chief elected official or chief administrative officer initial it so the authority is not in question at midnight.
• Print your EOC roster, mark every name that also has a field or departmental assignment during a major event, and count how many positions are left with fewer than two genuinely available people, because that number is your actual depth.
• Put one item on the agenda of your next regularly scheduled department head meeting: each department names a primary and an alternate for its EOC position and commits to sending the alternate to the next activation or exercise.
• Test the remote access path for your incident management platform with three staff working from home, confirm their accounts and passwords are current, and record the date of the test.
• Write one paragraph into your existing emergency operations plan stating the deactivation criteria and requiring a written transition document listing open actions, responsible departments and due dates before the EOC closes.

Takeaways

• An activation level is a statement about which positions are staffed, for what hours, with what delegated authority, and any level that cannot be described that way is a label without content.
• Opening the EOC, declaring a local state of emergency and requesting state assistance are three separate acts with three different authorities, and bundling them in the plan creates delay at the moment you can least afford it.
• Activation numbering runs in opposite directions between jurisdictions, so external notifications should state the functions staffed and the hours covered in plain words before they state the number.
• Triggers work when they are tied to conditions somebody else publishes with a timestamp, such as a weather product, a river gauge stage or a utility outage threshold, and when the plan names who may act on each one.
• Activating late costs queue position at the state EOC, vendor availability, evacuation clearance time and the rest of your staff’s endurance, and the pre-landfall decision record for Hurricane Katrina is examined in the February 2006 White House lessons learned report and the House Select Bipartisan Committee report A Failure of Initiative.
• Activating unnecessarily costs overtime, employer goodwill and callback speed, and the remedy is a genuine monitoring level that buys lead time with one or two people rather than a plan with only two settings.
• Round the clock operation needs three people per position rather than two, written release agreements with the employers of non-government staff, and alternates who have actually worked the seat.
• Deactivation criteria, a written transition document to recovery, a decision log and a cost record started in the first hour are what let you defend the activation afterward and what make the next one faster.