Most communities in American tornado country have outdoor warning sirens mounted on poles, and many residents believe those sirens are supposed to wake them at night. The sirens were never designed to do that. The network grew out of Second World War air raid warning and Cold War civil defense, and its design target was always a person standing outside who would then go find more information. This piece traces that history, explains how coverage is designed and who activates the sirens, and sets out what a siren can and cannot be expected to do in a warning plan today.

Air raid warning: where the outdoor siren started

Fire stations and factories used sirens before the war, but the outdoor siren as a community-wide public warning device took its familiar form in the air raid warning systems of the Second World War. In Britain, sirens were the main way a city learned that enemy aircraft were approaching, and the British signals became widely known. A rising and falling tone meant an alert, and a steady tone meant the all clear. The signal had to reach people where they were during the day, on streets, in markets, on factory floors and in yards, and it had to reach them quickly because warning time against bombers was measured in minutes. A siren mounted high on a building could cover a neighborhood without depending on any receiver in the listener’s hands, and that made it practical in a way that relying on household radios was not.

The United States built its own version once the war made the threat feel real. The Office of Civilian Defense, created by the federal government in 1941, organized volunteer air raid wardens and blackout drills in American cities, and local governments installed or adapted sirens and whistles to sound the alarm. I have not seen a reliable national count of how many sirens went up in those years, and I will not guess at one. What matters for this article is the design logic the war established, because the later systems inherited it largely intact.

That logic was simple. The siren carried one message, which was that danger was coming and the listener should act. The details came from somewhere else, such as an air raid warden on the street, a radio broadcast, or the person in charge at the factory. The siren was built to be heard by people outdoors and in the busy open places of a wartime city, and it assumed that anyone who heard it would go and find out what to do. Every later argument about sirens, including the argument over whether they should be heard indoors, traces back to that original division of labor between a loud cue and the detailed message that followed it.

Civil defense in the Cold War: attack warning, NAWAS and the standard signals

After the war the threat changed from bombers that might reach a coastal city to nuclear attack anywhere in the country. The federal response was the Federal Civil Defense Act of 1950 and the Federal Civil Defense Administration that administered it. Warning was one of the core functions. In the 1950s the federal government set up the National Warning System, usually called NAWAS, as a dedicated landline network that carried attack warnings from federal warning centers to state and local warning points. A county warning point that received an attack warning over NAWAS was expected to activate its local outdoor sirens, which made the siren on the corner the last link in a national chain. NAWAS still operates under FEMA today, and you should confirm its current role and your local connection point with your state emergency management agency rather than assume either.

The siren was paired with broadcast radio from the start. CONELRAD, the Cold War emergency broadcast arrangement, began in 1951 and was replaced by the Emergency Broadcast System in 1963, which was in turn replaced by the Emergency Alert System in the 1990s. The design assumption behind all of them was the one the war had established, in which the siren got people’s attention outdoors and radio told them what was happening and what to do. Federal civil defense guidance defined standard signals for this purpose. As that guidance describes them, a steady tone lasting several minutes was the attention or alert signal and meant that listeners should turn on a radio or television for information. A wavering tone, or a series of short blasts, lasting about three minutes was the attack warning signal and meant that people should take protective action immediately. The steady signal could also be used for peacetime emergencies, and that permission matters for what came later.

Federal civil defense programs also helped local governments pay for warning equipment over the following decades, which is part of the reason so many small towns ended up with sirens at all. Many of the oldest siren poles still standing in American communities date from this period, although I do not have a national inventory I trust and I am not going to put a number on it. The Cold War network was designed against a single hazard with a single response, in which the siren meant take cover now and the radio supplied the rest, and the design criteria written for it assumed a listener who was outdoors.

From attack warning to tornado warning: how the same poles took on weather

Public tornado warning became possible at about the same time the civil defense siren network was being built. In March 1948, Air Force meteorologists Ernest Fawbush and Robert Miller issued a tornado forecast for Tinker Air Force Base in Oklahoma that verified. The National Weather Service’s own historical material describes the Weather Bureau beginning public tornado forecasts in the early 1950s, after a long period of reluctance to use the word in public products. Once there were warnings for a specific tornado threat, communities that already owned sirens had an obvious way to use them, and the steady attention signal had already been described as usable for peacetime emergencies.

The move from attack warning to weather warning happened gradually and locally, and I am not going to put a date on when weather became the main reason sirens sounded, because that varied by state and by town. In tornado-prone parts of the country it happened early. The federal civil defense office became the Defense Civil Preparedness Agency in 1972 and was folded into FEMA in 1979, and by then the siren had become, for most residents in those regions, the tornado siren. When the Cold War ended and federal attention shifted to all-hazards emergency management, the attack warning mission faded while the poles stayed up, because the weather mission had become the one people relied on.

Other hazards adopted sirens for similar reasons. Communities around commercial nuclear power plants have federally required alert and notification systems for the plume exposure planning zone, with planning criteria set out jointly by the Nuclear Regulatory Commission and FEMA in the document known as NUREG-0654/FEMA-REP-1, and many of those systems are siren networks. Some coastal communities use sirens for tsunami, and some areas downstream of high-hazard dams and near chemical facilities use them as well. What these uses share is a hazard where the people at risk may be outdoors and may have only minutes, so a loud signal across a defined area buys time for the detailed warning to arrive. Anyone working on a nuclear plant planning zone system should read the current NRC and FEMA criteria directly, because they are more specific than general community siren guidance.

Coverage design: decibels, distance and the outdoor listener

Siren coverage is a sound propagation problem, and the basic physics is unforgiving. Sound spreading from a source in open air loses about 6 decibels each time the distance doubles, before anything else is taken into account. Air absorbs higher frequencies more strongly over distance than lower ones, so the pitch of the tone affects reach as well as loudness. Hills and buildings cast acoustic shadows, and dense trees absorb and scatter sound. Wind carries a siren farther downwind and shortens its reach upwind. Temperature layering matters too, because on a calm night with an inversion, sound can bend back toward the ground and carry farther than it does on a hot afternoon, so the same siren has a different footprint at different times of day.

Designers work to a target sound level at the edge of each siren’s coverage area. The figure most often cited in siren planning is around 70 decibels outdoors, with a higher target in noisy places, commonly stated as about 10 decibels above the ambient noise from traffic or industry. Those figures are usually traced to the outdoor warning systems guide FEMA published in 1980, and state programs and vendors have used them since. I treat them as the starting point a designer works from, and I would verify the target your state or your nuclear plant program actually uses before relying on either number, because some programs have set their own.

Siren hardware falls into two broad families. Mechanical sirens use a motor-driven rotor to chop airflow and produce the tone, and many of them rotate so that the loudest part of the sound sweeps around the compass, which is why a listener hears a rotating siren pulse in level as it turns. Electronic sirens use amplifiers driving arrays of horn speakers, which lets them produce several tones and, on many units, recorded or live voice. Vendor coverage maps for either type are predictions from a propagation model that cannot account for every hill, tree line and highway, so the only way to know where a siren can actually be heard at the design level is to go out and measure it. A coverage map drawn when the sirens were installed also says nothing about the subdivisions, schools and parks built since then, which is how many communities end up with sirens placed for the town as it was decades ago.

The indoor misconception: what sirens were never designed to do

The most common misconception about outdoor warning sirens is that they are meant to alert people inside their homes, and in particular to wake people at night. That belief is a myth, and the name of the equipment explains why, because outdoor warning sirens were designed from the air raid era forward to reach people who are outdoors. The National Weather Service and many state and local emergency management agencies say this in their public guidance, and they tell residents to rely on NOAA Weather Radio, wireless emergency alerts on their phones, broadcast warnings and local notification systems when they are indoors. When a resident complains after a night storm that the siren could not be heard in the bedroom, the siren was in most cases working as designed, and the gap was in that household’s indoor warning arrangements.

The reasons are physical. A building takes a substantial bite out of any outdoor sound before it reaches the people inside, and modern construction with insulated walls and sealed double-pane windows takes a larger bite than an older frame house with single glazing. How much depends on the building and on which side faces the siren, so I am not going to offer a single figure. Indoors, the siren also competes with air conditioning, a television, a fan and the storm itself, because heavy rain on a roof and thunder are loud at exactly the time a tornado siren is most likely to sound. A sleeping person needs a louder or more distinctive sound to wake than an awake person needs to notice, which stacks the odds further against a siren reaching a bedroom at night.

A siren network loud enough to wake people indoors across a whole community would need far louder sirens or far more of them, and a sound level that reliably penetrated closed houses would be punishing to anyone standing outdoors near a pole. I know of no community that designs to that standard, and I am not aware of any federal guidance that asks them to. A resident who hears a siren faintly indoors should treat it as a cue to check another source. A resident who hears nothing should not conclude that no warning is in effect, because the siren was never meant to be the indoor channel.

Activation: who sounds the siren and on what trigger

Another misconception, often heard after a storm, is that the National Weather Service sounds the sirens, and in most places it does not. The Weather Service issues the warning, and a local government decides whether and when to activate its own sirens, usually through a county or city 911 center or the emergency management agency. Because the arrangement differs from place to place, siren activation is a local decision governed by a local policy. That policy should be written down, adopted by whoever holds the authority to adopt it, and known to every person who might be sitting at the console when a warning comes in.

Trigger policies vary more than most residents realize. Some jurisdictions sound every siren in the county for any tornado warning that includes the county. Since the Weather Service moved to storm-based polygon warnings in 2007, other jurisdictions sound only the sirens inside the warning polygon, and some systems are zoned so that this can be done quickly. Many also activate on a credible report of a tornado from a trained spotter or a law enforcement or fire unit, even before a warning is issued, and some sound for destructive straight-line winds above a stated threshold. Countywide activation reaches everyone but sounds for many people who are not in the path, while polygon activation is more precise but depends on someone selecting the right zones quickly and accurately under pressure.

Control is usually by radio and sometimes by wired circuits, and many modern systems report the status of each siren back to the console, so the operator can see whether a unit rotated, sounded and is running on battery. The control path is part of the warning system and deserves the same attention as any other radio link, including protection against unauthorized activation. In April 2017 every outdoor warning siren in Dallas sounded overnight, and city officials attributed the activation to an unauthorized radio signal. I mention it only as an awareness example and would point readers to the city’s own public statements for the details. The activation policy should also name a backup activation point, because the 911 center that normally sounds the sirens may itself lose power or have to evacuate during the same storm.

Tones, tests and trust: what residents learn from the siren

Although the civil defense agencies defined standard signals, practice today varies widely. Some communities use a steady tone for tornadoes and some use a wail, some use different tones for different hazards, and voice-capable systems may play a recorded message. Visitors from another state, or even another county, may not know what a given tone means locally. Whether a community sounds an all clear also varies, and many do not. The belief that a siren falling silent means the danger has passed is a misconception in most places, because the siren usually runs for a set cycle and stops whether or not the threat has ended. Residents should be told what their own jurisdiction does, and told that the end of the sound is not an all clear unless the local plan says so.

Regular testing is necessary because sirens fail, and a test is the cheapest way to find a dead unit before a storm does. Testing also teaches the public something, and the lesson is not always the intended one. A siren heard every week or every month at a fixed time becomes part of the background, and residents who hear sirens often for warnings that do not affect them personally learn to wait for confirmation. The National Weather Service’s service assessment of the May 22, 2011 tornado in Joplin, Missouri addressed this. As I read it, the assessment found that many residents did not take protective action on the siren alone and looked for additional confirmation of the threat first, and it discussed desensitization linked in part to how often sirens and warnings had been issued before. Readers who want the exact findings should go to the assessment itself rather than rely on my summary.

That finding cuts two ways for anyone who manages sirens. Over-activation erodes trust, which argues for precise triggers and for sounding sirens only where the threat actually is. People who seek confirmation are also doing something reasonable, which argues for making the confirming sources fast and easy to reach, so that a Weather Radio alert, a wireless emergency alert, broadcast coverage and the agency’s own social media all match what the siren just signaled. When a siren sounds while every other channel is still silent, residents go looking for information that does not yet exist, and the time they spend looking comes out of the time they had to take shelter.

Failure modes and the layered plan: where sirens still earn their place

Sirens fail in ordinary ways. Utility power goes out in the same storms that call for activation, so units without healthy backup batteries go silent exactly when they are needed, and batteries age and need scheduled replacement. Motors and rotators wear out, lightning damages electronics and control receivers, and radio control links suffer from the same coverage and interference problems as any other radio path. Status reporting helps, but only if someone reviews the reports after every test and opens a work order for every unit that did not report correctly. A network nobody inspects degrades one unit at a time until the coverage map no longer describes what residents can actually hear.

A siren that works perfectly still serves one audience well, and that audience still matters. People outdoors at a ball field, a pool, a golf course, a lake, a campground or a construction site may not hear a phone in a bag, may be outside cell coverage, or may be visitors who never registered for local notifications. A loud signal across the area tells all of them at once that something is wrong, with no device or subscription required, and for many of them it is the fastest warning they will get. That is the job the siren was designed for in the air raid era and the Cold War, and it still does that job well wherever the coverage is real and maintained.

The siren belongs in a layered warning plan as the outdoor layer, alongside NOAA Weather Radio and wireless emergency alerts for people indoors, broadcast and local media for detail, and local notification systems for registered residents. A plan that names which layer serves which audience can explain its own gaps honestly to elected officials and the public. A plan that treats the siren as the whole warning system will be judged against a promise the equipment was never built to keep, and the shortfall will show up at night, when most people are indoors and many of them are asleep.

What to do at your agency

• The emergency management director pulls the written siren activation policy this month, confirms that it names the trigger conditions, the positions authorized to activate and a backup activation point, and adds any missing element to the agenda of the next scheduled meeting with the 911 center.
• The radio system administrator reviews the status reports from the most recent siren test, lists every unit that failed to report rotation, sound or battery status, and opens a work order for each one before the next test date.
• The public information officer reads the agency’s siren web page and storm-season materials this month and replaces any wording that implies sirens will alert people indoors with a sentence stating that sirens are for people outdoors and naming Weather Radio and wireless emergency alerts as the indoor channels.
• The emergency management coordinator overlays the current siren coverage map on a map of parks, schools, ball fields, campgrounds and subdivisions built since the coverage map was drawn, and writes a list of outdoor gathering places that fall outside the mapped coverage.
• The communications center supervisor asks the person who maintains the siren control equipment what protection the control path has against unauthorized activation and when it was last reviewed, and records the answer in the system documentation.
• The training officer adds a ten-minute review of the activation policy, covering polygon versus countywide activation and the backup activation point, to the next scheduled telecommunicator in-service.

Takeaways

• Outdoor warning sirens took their familiar form in Second World War air raid warning, where the siren carried a single message to people outdoors and the details came from wardens and radio.
• Cold War civil defense built on that design, with NAWAS carrying attack warnings to local warning points and federal guidance defining a steady attention signal and a wavering attack warning signal.
• The same poles took on tornado warning gradually and locally, and the weather mission is the main reason most community sirens survived the end of the Cold War.
• Siren coverage is limited by distance, terrain, trees, wind and ambient noise, and vendor coverage maps are predictions that should be checked by measurement and updated as the community grows.
• The belief that sirens are meant to alert or wake people indoors is a myth, because the systems were designed for people outdoors and indoor warning belongs to Weather Radio, wireless emergency alerts and broadcast.
• The National Weather Service issues warnings while local governments decide whether and when to sound their own sirens, so every jurisdiction needs a written activation policy that names triggers, authority and a backup activation point.
• Frequent activation and routine testing can teach residents to wait for confirmation, a pattern the National Weather Service’s Joplin service assessment discussed, and the confirming channels should be fast and consistent with the siren.