A tornado warning (SAME code: TOR) is a public warning that is issued by weather forecasting agencies to an area in the path of a tornado, or a severe thunderstorm capable of producing one, and advises individuals in that area to take cover. Modern weather surveillance technology such as Doppler weather radar can detect rotation in a thunderstorm, allowing for early warning before a tornado develops. They are also commonly issued based on reported visual sighting of a tornado, funnel cloud, or wall cloud, typically from weather spotters or the public, but also law enforcement or local emergency management. When radar is unavailable or insufficient, such ground truth is crucial. In particular, a tornado can develop in a gap of radar coverage, of which there are several known in the United States.
A warning should not be confused with a tornado watch, issued in the United States by the Storm Prediction Center (SPC) and in other countries by applicable regional forecasting agencies or national severe weather guidance centers, which only indicates that conditions are favorable for the formation of tornadoes. Although a tornado warning is generally a higher alert level than a tornado watch, in the U.S., it can be surpassed by a higher-level alert—structured as wording that can be added to the official warning product—to warn the public of intense tornadoes affecting a densely populated area.
A tornado watch is not required for a warning to be issued; tornado warnings are occasionally issued when a tornado watch is not active (i.e. when a severe thunderstorm watch is active, or when no watches are in effect), if a severe thunderstorm develops and has a confirmed tornado or strong rotation.
In April 1899, the Chicago Tribune wrote to the United States Weather Bureau via a news article posing the question on why tornado warnings are not sent out via telegraphs or even the telephone to warn the local population in the path. Cleveland Abbe responded by saying "it is certain that if any such arrangement were possible, the Weather Bureau would have done this many years ago" along with "we must remember that the destructive areas of tornadoes, and even of thunderstorms, are so small that the chance of being injured is exceedingly slight" and that "we do not attempt to prevent that which is inevitable". They also wrote that the chance of being injured by a tornado is "one in ten thousand years".
In June 1899, U.S. Weather Bureau Oklahoma section director J. I. Widmeyer published that long-range forecasters in Oklahoma were sounding "unnecessary tornado alarms" due to "ignorant predictions" to residents in Oklahoma and that they were causing "frightened men, women, and children" to take shelter, despite no tornadoes occurring. Cleveland Abbe added on to the publication by Widmeyer saying, "it is unnecessary to resort to the caves and cellars, or to stop our ordinary avocations for fear of a tornado, until we see the cloud in the distance, or are positively certain that one is about to pass near us".
In April 1908, the U.S. Weather Bureau published several replies regarding a question posed to the Weather Bureau on: How can we protect against tornadoes?.
Lieutenant John Park Finley responded with "the best we can do is to watch the distant tornado, and if it seems to approach us then move away toward the left; so far as we have learned, this still continues to be the best rule".
The Chief of the Weather Bureau responded with the idea to establish a warning system by surrounding a city at a distance of 4 mi (6.4 km) with wires hooked up with alarms. That way, a warning can be given to the city for an impending tornado. The wire system would detect sudden pressure differences, if wires were twisted, or if wires were short circuited. It was also stated that at a distance of four miles from the city, the tornado "would be unable to reach the city from any direction without giving us an alarm".
Cleveland Abbe responded by saying the idea of a wire-based system around a city is not practical as well as how tornadoes are very infrequent. Abbe ended by saying that "the mere forewarning of a tornado is no protection against its coming" and that it would be wiser to "spend your money to protect yourself against diseases, accidents, lightning, ect…".
In 1938, the Weather Bureau rescinded its ban on the usage of the word "tornado" in weather products disseminated to emergency management personnel. The Bureau would develop a network of volunteer storm spotters in the early 1940s during World War II, to provide warning of tornadoes to workers in munitions plants and strategic factories. The ban on issuing tornado warnings to the general public would not be revoked until Chief of Bureau Francis W. Reichelderfer officially lifted the ban in a Circular Letter issued on July 12, 1950, to all first order stations: "Weather Bureau employees should avoid statements that can be interpreted as a negation of the Bureau's willingness or ability to make tornado forecasts", and that a "good probability of verification" exist when issuing such forecasts due to the difficulty in accurately predicting tornadic activity. The American Meteorological Society agreed to have Miller and Fawbush present their methodology for forecasting tornadoes during the organization's 1950 meeting in St. Louis; after garnering press coverage for their successful prediction of past tornadoes, AMS representatives decided to open the presentation to the public.
The first official tornado forecast was made by United States Air Force Capt. (later Col.) Robert C. Miller and Major Ernest Fawbush, on March 25, 1948. The first such forecast came after the events that occurred five days earlier on March 20, 1948; Miller – a California native who became stationed at Tinker Air Force Base three weeks earlier – was assigned to work the late shift as a forecaster for the base's Air Weather Service office that evening, analyzing U.S. Weather Bureau surface maps and upper-air charts that failed to note atmospheric instability and moisture content present over Oklahoma that would be suitable for producing thunderstorm activity, erroneously forecasting dry conditions for that night. Thunderstorms soon developed southwest of Oklahoma City, and at 9:30 p.m., forecasters from Will Rogers Airport sent a warning to Tinker that the storm encroaching the city was producing wind gusts of 92 miles per hour (148 km/h) and a "Tornado South on Ground Moving NE!" Base personnel received an alert written by the staff sergeant on duty with Miller; the twister struck Tinker several minutes later around 10:00 p.m., damaging several military aircraft (with total damage estimated at $10 million) that could not be secured in time before it crossed the base grounds.
Following an inquiry the next day before a tribunal of five generals who traveled to Tinker from Washington, D.C., who ruled that the March 20 tornado was an "act of God ... not forecastable given the present state of the art", base commander Gen. Fred Borum tasked Miller and Fawbush to follow up on the board's suggestion to consider methods of forecasting tornadic thunderstorms. Over the next three days, Miller and Fawbush studied reports and charts from previous tornado events to determine the atmospheric conditions favorable for the development of tornadic activity, in an effort to predict such events with some degree of accuracy. At the time, there had not been studies on how tornadoes formed; however, military radars were being adapted for forecasting use, allowing forecasters to see the outlines of storms but not their internal attributes such as rotation. Miller and Fawbush's findings on atmospheric phenomenon present in past outbreaks would aid in their initial forecast, as the day's surface and upper-air analysis charts determined the same conditions present on March 20 were present on the 25th, concluded that central Oklahoma would have the highest risk for tornadoes during the late-afternoon and evening.