Red Flag Warning Thresholds Are Local Numbers, and They Differ by Thirty Percentage Points

A Red Flag Warning arrives with the same three words in every state, and it reads like a national grade for fire danger. It is not one. The National Weather Service issues the product against criteria written office by office, and the humidity number inside those criteria runs from 15 percent in one forecast area to 45 percent in another. Thirty percentage points separate the strictest published threshold from the loosest among the sets collected here.

A threshold is a statement about local climate and local fuels, so a number copied from one area to another would either fire on ordinary days or fail to fire on the days that matter. What follows is the published criteria from five National Weather Service sources, the equation the agency publishes for turning moisture into a humidity reading, and an arithmetic comparison of what a single afternoon would trigger in each area.

The Chain

The path from moisture in the air to a warning on a phone screen runs through six steps, and the last two are administrative rather than atmospheric.

1. The air carries a fixed amount of water vapor. The dewpoint is the measure of that amount. It changes when a different air mass arrives, not when the sun comes out.
2. Relative humidity divides that amount by capacity. Capacity rises steeply with temperature, so an unchanged air mass reads a lower relative humidity every hour the afternoon warms.
3. The criteria attach a fuel or drought term. Humidity is treated as one term in a conjunction, alongside a fuel moisture percentage, a drought index, a computed energy term, or a judgment delegated to fire managers.
4. The criteria attach a wind term. Each set names a wind speed and states whether the number refers to sustained wind or to gusts.
5. The office compares the forecast against a number agreed for that area. Most of the agreed numbers carry a required duration with them, so a momentary reading does not qualify.
6. Timing decides which product is issued. At the Tallahassee office, “A Red Flag Warning is issued when the above conditions are expected to be met within the next 24 hours.” The next line reads, “A Fire Weather Watch is issued when the above conditions are expected to be met during the next 24 to 48 hours.”

What the Threshold Is Trying to Catch

The national glossary entry describes a purpose rather than a measurement. A Red Flag Warning is “A term used by fire-weather forecasters to call attention to limited weather conditions of particular importance that may result in extreme burning conditions.” The entry does list generic figures, among them “a sustained wind average 15 mph or greater”, “relative humidity less than or equal to 25 percent” and “a temperature of greater than 75 degrees F”. They are not the numbers a particular office applies, and none of the eight criteria sets compared below reproduces that combination.

What each office applies is a conjunction, and the second term is about fuel rather than air. The Tallahassee office publishes a fuel moisture condition of “≤ 6%” for its Georgia zones, a Keetch-Byram Drought Index condition of “≥ 300 OR AL Fire Alert” for Alabama, and an Energy Release Component condition of “≥ 26” for Florida. Honolulu requires “Keetch-Byram Drought Index (KBDI) ≥ 500”. Omaha states the term qualitatively as “Fuel status supportive for extreme fire growth.” Denver and Boulder fold the whole judgment into one line: “A combination of weather and fuels conditions (as determined by fire management) for any 3 hours or more in a 12 hour period.”

Two of those are the same drought index set at different thresholds, one is a fuel moisture percentage, one is a computed energy term, and two are judgments delegated to fire managers.

The Denver and Boulder page goes further and lists triggers that do not use its 15 percent humidity line at all. Three of them carry no humidity term: “Dry thunderstorms (15% coverage or more, constituting an LAL 6.)”, “Wind shifts associated with frontal passages.” and “First significant lightning event (wet or dry) after an extended hot and dry period.” A fourth sets a different humidity number for a different time of day: “Poor relative humidity recovery overnight (RH remains at 40% or lower.)”

Relative Humidity Is a Ratio, Not an Amount

The agency publishes the arithmetic behind the reading on the formula sheet for its own weather calculator. Actual vapor pressure and saturation vapor pressure are each written as 6.11 multiplied by 10 raised to the power of 7.5 times the temperature divided by 237.3 plus that temperature, with dewpoint used for the first and air temperature for the second, both in Celsius. Relative humidity is then simply “rh = 100 × e/es”. Two consequences follow from that ratio.

First, one unchanged air mass produces a different humidity reading every hour of the afternoon. Take air with a dewpoint of 45 degrees F. Computed from those equations, that air reads 48 percent relative humidity at 65 degrees, 41 percent at 70, 34 percent at 75, 29 percent at 80, 25 percent at 85, 21 percent at 90 and 18 percent at 95. No moisture entered or left the air. Only the thermometer moved.

Set against the published thresholds, that parcel crosses the Peninsula Florida line of 35 percent at 74 degrees, the North Florida line of 28 percent at 81, the Georgia and Alabama line of 25 percent at 85, the Omaha line of 20 percent at 92, and the Denver and Boulder line of 15 percent at 101. It passes the Honolulu line of 45 percent at 67 degrees, which is to say before most afternoons have begun. Those crossing temperatures are this article's calculation and are not published values.

Second, read in the other direction, a threshold is a statement about how dry the air itself has to be. At 85 degrees, reaching 15 percent requires a dewpoint near 32 degrees F, while reaching 45 percent requires a dewpoint near 61 degrees. The two thresholds are describing moisture contents about 29 degrees of dewpoint apart. Whether such a dewpoint occurs in a given place is a question about that place's climate, and that is the reason the numbers are agreed locally rather than set centrally.

The Published Numbers, Side by Side

Five sources carry the eight criteria sets used here, and each is published on National Weather Service pages.

• Denver and Boulder. The wind and humidity combination is “Frequent gusts of 25 mph or greater – AND relative humidity of 15% or less”, held for “any 3 hours or more in a 12 hour period.”
• Omaha. Two of the criteria listed there pair a wind term with a humidity term. The first is “20 mph sustained winds or frequent gusts to 30 mph for 2 hours or more.” paired with “RH of 20% or less for 2 hours or more (3 hourly observations in a row).” The second is “25 mph sustained winds or frequent gusts to 35 mph for 2 hours or more.” paired with “RH of 25% or less for 2 hours or more (3 hourly observations in a row).” The page carries two further criteria. One is for dry lightning: “When there is a greater than 15% coverage of dry lightning with critical fuels.” The other states that a warning can be issued “in scenarios with strong wind and critical fuels, even if RH values are greater than traditional thresholds”.
• Tallahassee. One table, three states. Georgia takes “≤ 25%” for “1 hr” with wind “≥ 15mph OR frequent gusts ≥ 25mph”. Alabama takes a humidity value below 25 percent for “4 hr” with wind “≥ 15mph”. Florida takes a humidity value below 28 percent with wind “≥ 15mph”.
• The Florida annual operating plan. For the north of the state the criterion is a relative humidity below 28 percent with sustained 20-foot winds above 15 mph, “occurring with a Significant Fire Potential rating of 'Moderate' or higher in the zone(s) of concern.” For the peninsula the same sentence appears with one number changed: a relative humidity below 35 percent with sustained 20-foot winds above 15 mph, “occurring with a Significant Fire Potential rating of 'Moderate' or higher in the zone(s) of concern.”
• Honolulu. Three conditions together: “Keetch-Byram Drought Index (KBDI) ≥ 500”, “Relative Humidity (RH) ≤ 45 %” for “2 or more hours”, and “Wind ≥ 26 mph”, marked “(sustained wind, not wind gusts)”, also for two or more hours.

Laid out together, the humidity terms run from 15 percent to 45 percent: a spread of 30 percentage points and a ratio of three to one. The wind terms do not line up either. The Denver and Boulder condition is written in gusts; the Honolulu condition is written in sustained wind and rules gusts out in the text. A sustained 26 mph is a harder condition to meet than gusts reaching 25 mph, so the distance between those two sets is wider than the humidity numbers alone show.

Consider an afternoon at 85 degrees F with 22 percent relative humidity and sustained winds of 18 mph gusting to 28 mph, holding for three hours. Comparing only the weather terms, and assuming each area's fuel or drought term is satisfied, that afternoon meets the Georgia criteria, the North Florida criteria and the Peninsula Florida criteria. It falls short in Alabama, which asks for four hours. It falls short at both Omaha combinations, the lower of which asks for 20 mph sustained or gusts to 30 mph. It falls short at Denver and Boulder, which asks for 15 percent. It falls short at Honolulu, which asks for 26 mph of sustained wind. Three of the eight sets would be met and five would not, on identical weather.

The Threshold Includes a Clock and a Station

Six of the eight sets state how long the conditions have to hold, and those durations are as varied as the humidity numbers. Georgia asks for “1 hr”. Honolulu asks for two or more hours. Both Omaha combinations ask for “2 hours or more (3 hourly observations in a row)”. Alabama asks for “4 hr”. Denver and Boulder ask for three hours or more inside a twelve-hour window. The other two state none: the Tallahassee table leaves its Florida duration blank, and neither Florida plan criterion names a period.

Georgia's one hour and Alabama's four hours sit in the same table on the same office's page. A reading that satisfies the first falls short of the second by a factor of four in persistence, with no difference in the weather itself.

The Omaha wording shows what the clock is counted from. Three hourly observations in a row is a statement about a record, not about the atmosphere in the abstract, and it ties the criterion to a station that reports on the hour. The Denver and Boulder window is written differently: three hours or more inside a twelve-hour period, with no wording that the hours run in a row.

Who Sets the Number, and How It Changes

National Weather Service Instruction 10-401 states the arrangement directly: “Red Flag criteria are determined by coordination between WFO personnel and land management users within the WFO fire weather service area.” The next sentence sends the result to a document: “Each WFO or group of WFOs should have their criteria well-documented in the interagency AOP (see NWSI 10-404) and Station Duty Manual (SDM).”

The same section puts the issuance decision inside a coordination step and then reserves the final call. “Forecasters should coordinate with local fire and land managers and Predictive Services at the National Interagency Fire Center or Area Coordination Centers prior to the issuance of a Fire Weather Watch or RFW.” The sentence that follows reads, “However, final authority to issue a RFW product resides with the NWS.”

The annual operating plan is where the numbers live, and it moves on an administrative cycle rather than a meteorological one: “The Florida Annual Operating Plan will be reviewed annually and revised as needed.” That structure also explains how one office publishes several thresholds at once. Tallahassee's table carries different values for Georgia, Alabama and Florida because it serves zones in three states with three sets of land management partners, and the Florida plan carries a different value for the peninsula than for the north. The number changes at a state line or a zone boundary because the agreement changes there, not because the air does.

Two areas under warnings on the same day are not being told they face the same conditions, and an area without a warning is not being told its air is wetter than an area with one. Each warning reports that a locally agreed combination was crossed, and the combinations are not the same combination.

Where This Doesn’t Apply

• These are the values published on one date. Every figure here was read on 2026-09-04, and the Florida plan says on its face that it is reviewed annually and revised as needed. A criteria page can carry a different number next season.
• The comparison covers weather terms only. Each set also requires a fuel or drought condition, and two of them hand that judgment to fire managers, so the comparison above is arithmetic on published thresholds rather than a prediction of what any office would issue.
• Forecaster judgment is written into the criteria. Denver and Boulder list “Any combination of weather and fuel moisture conditions which, in the judgment of the forecaster, would cause extensive wildfire occurrences.” No threshold table captures that line.
• Wind statistics are not interchangeable. Sustained wind, gusts and a 20-foot reference height are three different quantities. Comparing the printed numbers across offices is not the same as comparing measured conditions.
• The humidity arithmetic is an approximation. The values computed here follow the equations the agency publishes for its own calculator and are rounded to whole percent and whole degrees. A different vapor pressure formulation shifts the crossing temperatures by a fraction of a degree.
• Eight sets are a sample, not a survey. Weather forecast offices cover the whole country and each maintains its own criteria, so the full national range may be wider or narrower than 30 points.
• A warning is not a forecast that a fire will occur. It marks conditions under which a fire that starts could spread quickly. What follows from that is set by land management agencies and the issuing office, and is outside the scope of this arithmetic.