Emergency plans usually treat the communications system as the instrument you use to manage the disaster, which holds up until the disaster lands on top of the instrument. Hurricane Katrina is the American reference case for that condition, with flooded central offices, generators that were either submerged or dry of fuel with no truck able to reach them, dispatch centers abandoned by people whose own houses were gone, and mutual aid arriving on radios that could not reach the agencies already working. The after-action record is unusually thick, and most of the communications doctrine a working agency uses today traces back to it.

What the published reports counted in August and September 2005

Katrina came ashore in southeast Louisiana on the morning of August 29, 2005, crossed into Mississippi, and drove a storm surge into the Mississippi coast that destroyed buildings and the infrastructure inside them outright. The communications damage then arrived in stages rather than all at once, because wind and surge took the first set of facilities, the levee and floodwall failures put much of Orleans and St. Bernard parishes under water over the following days, the regional power outage outlasted the batteries and then the fuel, and the loss of road access combined with security perimeters kept repair and refueling crews away from sites that were otherwise repairable.

The commonly cited outage figures come from the FCC. In testimony to Congress in September 2005 and in the record of the Commission’s Independent Panel Reviewing the Impact of Hurricane Katrina on Communications Networks, which delivered its report in June 2006, the Commission reported that more than three million customer telephone lines went out of service in Louisiana, Mississippi and Alabama, that 38 public safety answering points went out of service, and that more than a thousand cell sites were out. Those counts were assembled from carrier reporting gathered ad hoc, since the Disaster Information Reporting System that carriers now use during declared events did not yet exist and was one of the things created afterward.

The human and economic figures are disputed, and they should be presented that way. The National Hurricane Center’s tropical cyclone report on Katrina puts the toll at roughly 1,833 direct and indirect deaths across the affected states. Louisiana’s state health department published a separate count of victims recovered in Louisiana that runs several hundred lower, and later academic reviews have argued the real total is higher once displacement-related deaths are included. Those numbers disagree because they count different populations over different periods, and I am not going to reconcile them here. For damage, NOAA’s National Centers for Environmental Information carries Katrina at roughly $125 billion in 2005 dollars, with a substantially higher figure once adjusted for inflation.

Generators underwater, and fuel that could not get through

The dominant cause of the sustained outage was loss of commercial power followed by the inability to keep backup power running, along with flooding of wire centers, according to the FCC’s Katrina panel record and the carrier reporting behind it, rather than antennas blown off towers. Several telephone company central offices in the New Orleans area took water, and in a city where much of the ground sits below sea level the engine, the day tank and the transfer switch are frequently at or below grade, which means a facility can be structurally intact and electrically dead because the part that makes power is submerged.

A generator runs for as long as the fuel in its tank lasts, and after that its runtime is whatever the resupply chain delivers. Resupply failed in places during the first week for reasons that had nothing to do with the machinery, since roads through the flooded parishes were impassable and suppliers with limited allocation were fielding calls from customers holding larger contracts. Fuel trucks that did get loaded were in some cases turned back at law enforcement perimeters, and the panel’s recommendations included credentialing arrangements so that telecommunications repair and refueling crews could pass checkpoints, which tells you that access control by agencies doing their jobs correctly was itself part of the outage.

The regulatory answer did not survive. In 2007 the FCC adopted a backup power rule in the Katrina proceeding, commonly described as requiring eight hours of emergency backup power at cell sites and twenty-four hours at central offices. Industry challenged it, the associated information collection did not clear federal review, and the requirement never took effect before the Commission withdrew it. Anyone planning around carrier backup power today should confirm the current state of any obligation directly with the FCC rather than assuming a standard exists, because for most of the two decades since Katrina the answer has been that carriers engineer backup power to their own business judgment.

Fuel on hand plus the next confirmed delivery

For every communications site your agency depends on, runtime equals the fuel on hand plus the time to the next confirmed delivery, and the second number is the one nobody writes down. Find out who holds the fuel contract and what priority your agency has on it when a hundred other customers call the same vendor on the same morning, then find out whether the driver can reach the site if a road is closed and a deputy is controlling access. If any of those answers is a name nobody at the agency can produce, your generator is rated in hours rather than days, whatever the tank size says.

Towers, backhaul, and which part actually broke

Structural loss was real on the Mississippi coast, where surge removed buildings and the equipment shelters with them, and along the Louisiana coastal parishes where wind took antennas and lines. Across the larger footprint, though, the failures that kept systems out of service were power and transport. A cell site with a working antenna and a working radio is out of service if the fiber or copper carrying its traffic runs through a wire center that flooded, and a public safety simulcast subsystem is out of service if the leased circuits or the microwave hops that tie its sites to the prime site are gone, no matter how good the tower steel is.

That distinction matters for how an agency spends money. A tower is the visible asset and the one least likely to be the thing that failed, while the dependencies that actually carry the traffic are usually leased from somebody else and frequently missing from an agency’s own documentation. I have looked at systems where the site diagram stopped at the demarcation point, which meant nobody at the agency could say whether two supposedly redundant paths from two sites were riding the same fiber bundle out of the same building. That question has an answer, and the carrier will usually provide a circuit layout record if the radio administrator asks in writing.

Master site and console location deserve the same scrutiny. If the master or prime site controller for a trunked system, the console electronics, and the logging recorder all sit in one building, that building’s flood elevation, roof and generator become the whole system’s single point of failure regardless of how many RF sites you have. Katrina made that concrete in a coastal setting, and the same geometry applies to a river town, a tornado path, or a single-story communications center with a transformer yard beside it.

Dispatch centers evacuated, and continuity that existed only on paper

Thirty-eight 911 centers out of service across three states, the figure the FCC reported, covers several different mechanisms. Some PSAPs were flooded or destroyed outright, and others were evacuated because the building was untenable or the staff had to leave. A dry, powered and fully staffed center could still receive nothing, because the selective router or the tandem office that delivers 911 calls to it was the facility that went under, and a call cannot reach a console through a switch sitting in salt water. Telecommunicators discovered in real time that a working console does not guarantee that the network is still delivering calls to it.

The staffing side is documented in both congressional reviews and is the part agencies skip in their continuity planning. Telecommunicators and radio technicians lived in the affected parishes and counties, lost houses, evacuated families, and in many cases could not return through closed roads even when the center needed them. A relief roster built on the assumption that off-duty staff are reachable and can travel is a roster that works for a structure fire and not for a regional event, and the honest version of that plan names people who live outside the hazard footprint or comes from a mutual aid agreement with a center that does.

The emergency operations side fared no better. The House select committee’s report, published in February 2006 under the title A Failure of Initiative, and the Senate Homeland Security and Governmental Affairs Committee’s report, Hurricane Katrina: A Nation Still Unprepared, published in May 2006, both document the collapse of communications at the New Orleans emergency operations center and the reliance of officials on a small number of satellite phones and, in stretches, on people carrying messages by hand. Those two reports and the FCC panel report are the primary published record on this, and they are worth reading in the original rather than through summaries, including mine.

The alternate PSAP question worth asking this week

Find out which carrier facility your 911 trunks terminate in, and then find out whether your designated alternate PSAP terminates in the same one. If it does, your continuity plan protects you against a fire in your own building and against nothing else, which may be a reasonable decision as long as it is a written and deliberate one. Your 911 director or your Next Generation 911 service provider can produce that answer, and the fix may be as simple as a documented default routing arrangement to a center in another carrier footprint.

The interoperability failure layered on top of the outage

The interoperability problem was separate from the infrastructure problem and made it worse. Responders arrived from across the country on radios covering VHF low band, VHF high band, UHF and 800 MHz, running conventional and trunked systems from several manufacturers, with fleet programming done by their home agencies for their home jurisdictions. Where local repeaters survived, arriving units frequently could not use them, and where national interoperability channels existed in some radios, they were programmed inconsistently or not at all, which is how a mutual aid channel that is legally available to everybody ends up carrying nothing.

A shared frequency by itself solves very little, because talking requires an assignment, a written plan, and somebody with the authority to make the assignment stick. The reports describe an incident where nobody was filling that function at scale in the first days, which meant that arriving crews had no current ICS-205 to read and no cache of pre-programmed radios to draw from, with nobody serving as the single technical authority to reconcile what each agency had brought. Encryption compounds it, since a key that lives only in a home agency’s key management facility makes a perfectly compatible radio useless on a talkgroup somebody else controls.

Two workarounds from that period are worth remembering. Text messaging frequently succeeded when cellular voice calls failed, because a short message is small and the network stores and forwards it rather than requiring both ends to hold a circuit at the same moment, and that behavior is still true on modern networks during congestion. Satellite telephones were sent in quantity, and post-Katrina reviews describe the practical limits that showed up, including handsets issued to users who had never been trained on them and a charging problem nobody had planned for, on top of the plain physical requirement for a clear view of sky from somewhere other than the middle of a building. A satellite phone that has sat in a sealed case in a closet since the last grant cycle is an untested capability, and the remedy is a monthly call placed from outside the building by a named person who logs the result.

What the reports recommended and what became doctrine

The institutional response came fast. Congress passed the Post-Katrina Emergency Management Reform Act of 2006, which along with the accompanying emergency communications title established an Office of Emergency Communications inside the Department of Homeland Security and required a National Emergency Communications Plan. The first NECP was published in 2008, the office’s functions now sit within the Cybersecurity and Infrastructure Security Agency, and the plan has been revised more than once, so check the current edition at CISA rather than working from an old copy. The same period brought the Statewide Communication Interoperability Plan and the Statewide Interoperability Coordinator into general practice, which is why your state has a SWIC and a SCIP today.

On the operational side, the durable product was the typed Communications Unit under the Incident Command System. The positions that a large incident staffs now, meaning the Communications Unit Leader, the Communications Technician, the Incident Communications Center Manager and the Radio Operator, along with the All-Hazards Type III COML training that federal emergency communications staff built and states deliver, exist in their present form because Katrina demonstrated what happens when nobody owns incident communications. The National Interoperability Field Operations Guide, published by the same federal program and carried in a pocket, came out of the same effort to make a common channel plan available to a crew that arrived from six states away.

SAFECOM’s Interoperability Continuum, which predates the storm, became the standard way of explaining why buying radios does not buy interoperability, because it treats governance, standard operating procedures, technology, training and exercises, and usage as five separate lanes that all have to advance. The FCC’s own follow-through included the Disaster Information Reporting System in 2007, and years later the Commission moved parts of the voluntary industry resiliency framework, including roaming during disasters and mutual aid between carriers, toward mandatory requirements. Confirm the current wireless resiliency rules with the Commission before you rely on them in a plan, because that set of obligations has changed more than once.

One lane out of five

The most common way grant money gets wasted after a disaster is on technology purchased while the governance and standard operating procedure lanes stay empty, which produces a region full of compatible radios and no agreement about who assigns talkgroups on a multi-agency incident. The SAFECOM Interoperability Continuum exists to make that visible. If you cannot name the person who would write the ICS-205 for a five-agency incident in your county tonight, and point to the document that gives them that authority, the gap is in governance rather than in equipment, and more radios will not close it.

What held, what lapsed, and what the follow-up reviews still found

Some of the doctrine took hold well. Statewide and regional system consolidation moved fast in the Gulf states with federal funding behind it, with Mississippi building its statewide wireless network and Louisiana expanding its own, and both states’ interoperability offices are the place to confirm current coverage and governance. The Communications Unit became a real staffed function rather than a box on a chart, and the practice of publishing an incident radio plan for arriving mutual aid is now normal on large incidents in a way it was not in 2005.

Other pieces quietly went away. The FCC backup power rule never took effect, as described above. The dedicated federal grant program for interoperable emergency communications that ran in the years after the Post-Katrina Act was not sustained as a separate line, which pushed interoperability spending back into general homeland security grants where it competes with everything else. Government Accountability Office reviews in the years following continued to identify gaps in emergency communications planning, coordination and the measurement of progress, and the GAO’s published reports are worth searching by topic because they are specific about which recommendations agencies had and had not implemented at the time of writing.

The turnover problem is the one I would flag hardest for a small agency. Governance built after a disaster depends on named people, and a regional interoperability committee that met monthly in 2009 can be a mailing list with no quorum by 2026, at which point the memoranda of understanding sitting in a binder describe authorities nobody has exercised. Nothing in the federal doctrine prevents that, because the doctrine assumes a state and local structure that keeps meeting. Verify the current status of your regional interoperability governance body with your state interoperability coordinator, and find out when it last met and who your agency’s representative is supposed to be.

Finding your own version of the same failure

Most agencies get more out of working down the dependency chain for their own system than out of rehearsing a hurricane. Write the answers where the plan can find them, starting with commercial power and generator fuel at every site, then the backhaul path from each site and whether two paths share a building, then the location and flood elevation of the master site and the dispatch consoles, then the 911 trunk delivery point, then the staffing assumption behind your relief roster, and then the credential that gets your technician or your fuel truck through a controlled perimeter in your county. Each of those has an owner who can produce a document, and the gaps show up as questions nobody can answer.

The plan then has to say what degraded operations look like, because some of this you will not fix. If the trunked system loses its links, you need a documented fallback to conventional repeaters or simplex with a channel plan crews have actually practiced, and if the 911 network stops delivering calls you need a published alternate number, a broadcast and mass notification message written in advance, and a station-based walk-in procedure. The Hawaii false alert record makes the same point about correction messages, and it applies here, since the wording you would need on the worst morning should already exist in a file.

The Coast Guard is the contrast case in the Katrina literature, and both congressional reports treat its performance as a bright spot. Its crews arrived with self-contained aviation and cutter communications, having pre-staged assets outside the impact area and pushed decision authority down to the unit level, and its own published accounts credit the service with rescuing or assisting more than 33,000 people. The transferable lesson for a county agency has nothing to do with helicopters, and everything to do with moving communications capability outside the hazard footprint before landfall and deciding in advance who is allowed to act without asking, which is a governance decision written into a plan rather than an equipment purchase.

What to do at your agency

  • Have your radio system administrator produce a one-page table for every site your agency depends on, listing generator fuel type, tank capacity, the fuel vendor named in the current contract, and who is authorized to order a delivery, then bring the rows with a blank vendor to the next communications committee meeting already on the calendar.
  • Ask your 911 director or NG911 service provider in writing which carrier facility your 911 trunks terminate in and whether your designated alternate PSAP terminates in the same facility, and file the answer in the continuity annex rather than in an inbox.
  • Have the technician or vendor request the circuit layout records for the backhaul serving your two most important radio sites, and confirm whether the supposedly redundant paths leave through the same building or the same fiber bundle.
  • Ask your county emergency management director for the written answer to one question: what credential gets a radio technician, a tower crew or a fuel truck through a law enforcement perimeter in your jurisdiction, and who issues it during an event.
  • Pull the ICS-205 from your most recent multi-agency incident, or write one for the next scheduled drill, and verify that each mutual aid channel is listed by the name it carries in the neighboring agencies’ radios and not only by the name in yours.
  • Assign one named person to place a monthly call on each satellite phone in your cache from outside the building, log the date and the result, and replace or charge whatever fails, so the capability is tested rather than assumed.
  • Write one paragraph into your existing communications annex naming who at your agency may request state or federal deployable communications support through the state EOC, and what information that request has to include.

Takeaways

  • The FCC reported in September 2005 congressional testimony and in the record of its Independent Panel Reviewing the Impact of Hurricane Katrina on Communications Networks, which reported in June 2006, that more than three million customer telephone lines were out in Louisiana, Mississippi and Alabama, that 38 911 centers went out of service, and that more than a thousand cell sites were out.
  • Loss of commercial power followed by the inability to refuel generators, along with flooding of telephone company wire centers, accounted for more of the sustained outage than wind damage to antennas and towers did.
  • Death toll figures for Katrina are genuinely disputed, with the National Hurricane Center’s tropical cyclone report at roughly 1,833 direct and indirect deaths, a lower state-level count published by Louisiana’s health department, and academic arguments for a higher total, because each counts a different population over a different period.
  • A PSAP can be dry, powered and fully staffed and still receive nothing when the carrier facility that delivers its 911 calls is the thing that flooded, which is why the trunk termination point belongs in the continuity plan.
  • The interoperability failure was a separate failure, caused by incompatible bands and systems, inconsistent programming of national channels, no cache of pre-programmed radios, and no Communications Unit assigning anything in the first days.
  • The Post-Katrina Emergency Management Reform Act of 2006 and the accompanying emergency communications legislation created a federal emergency communications office, required a National Emergency Communications Plan first published in 2008, and put Statewide Communication Interoperability Plans and Statewide Interoperability Coordinators into general practice, with current versions available from CISA.
  • The FCC adopted a backup power rule in 2007, commonly described as eight hours at cell sites and twenty-four hours at central offices, and it never took effect and was later withdrawn, so confirm any current carrier backup power obligation with the Commission before planning around one.
  • Both congressional reports on Katrina treat the Coast Guard as the exception, and the transferable reason is that it staged self-contained communications outside the impact area and let units act without asking, which is a decision a county agency can copy at its own scale.
Questions or a different view?

Reach me through the contact page. I read every message.