Flash Floods: The Role of Science, Forecasting, and Communications in Reducing Loss of Life and Economic Disruptions

An Information Statement of the American Meteorological Society
(Adopted by AMS Council on 17 April 2017)

What is a Flash Flood?

Advances in our understanding of flash floods and their societal impacts warrant a revised definition of a flash flood as:

“Flooding caused by rapidly rising water level in streams, creeks, rivers or other waterways, normally dry stream beds, or in urban areas, usually as a result of intense rainfall over a relatively small area or for moderate to intense rainfall over highly saturated or impervious land surfaces, and generally occurring within minutes to several hours of the rainfall event.”

Steep terrain tends to concentrate runoff into streams very quickly and is often a contributing factor. Changes in soil properties (e.g., burn areas from wildfires), hydrophobic or impervious soils, removal of surface vegetation, and excess runoff from warm rainfall on significant snowpack can also be important contributors.  Additional causes of flash floods include ice jams, and levee and dam failures.


The sciences of meteorology and hydrology have made historic strides in the last few decades. Yet, despite significant advances in the fundamental understanding of weather, increases in high-resolution observations of the atmosphere, land, water in all phases and over all scales, and the integration of supercomputers for timely and increasingly accurate numerical weather prediction (NWP) and hydrologic modeling, the prediction and warning of flash floods remains a challenge, due to their relatively brief and small-scale nature. In 2008, the World Meteorological Organization conducted a survey of national meteorological and hydrologic services, and 105 countries out of the 139 surveyed identified flash floods as one of the two most important forecast hazards among all natural hazards they face. Globally, flash floods take more than 5,000 lives each year and have significant economic, social, and environmental impacts throughout the world. In the United States, the average number of fatalities due to flash flooding is nearly 100 per year [averaged for the period from 2006 through 2015 based on the U.S. National Weather Service (NWS) statistics (http://www.nws.noaa.gov/om/hazstats.shtml)], without a marked reduction in the number of lives lost over the last 20 years. Flash flooding remains one of the deadliest weather-related hazards in the world despite hydrometeorological advances.

Flash flood–causing rainfall is typically associated with convective cells that move slowly or form repeatedly over the same area. Flash floods are produced by a variety of storms ranging from land-falling tropical cyclones and hurricanes, supercell thunderstorms, slow-moving squall lines, mesoscale convective systems, and convective storms near frontal boundaries to persistent rainfall systems with moisture moving upslope in mountainous terrain. Of further concern, research and observational evidence over the past decade and longer has indicated changes in high-intensity rainfall and flooding in many parts of the world (https://www.ametsoc.org/ams/index.cfm/about-ams/ams-statements/statements-of-the-ams-in-force/climate-change/). These changes have been attributed to a number of causes, including greater runoff from land surface changes due to urbanization, wildfires, and human activities. Such alterations may also change large-scale and mesoscale circulation features and modify precipitation microphysics, thus impacting extreme precipitation (https://www.ametsoc.org/ams/index.cfm/about-ams/ams-statements/statements-of-the-ams-in-force/inadvertent-weather-modification/). Additionally, well-intended efforts such as stream channel “improvements” and channelization may have the unintended consequence of increasing the probability of downstream flooding and flash flooding during extreme rain events. With growing populations living in or traveling through flash flood–prone areas, we are witnessing increased risks to life, property, society, and the economy, and these risks further stress local infrastructure and emergency response sectors.

Hydrometeorological forecasting on the very small spatial scales of flash flooding has progressed substantially over the past 20 years. This includes advances in observing systems (in situ, radar- and satellite-based observations), numerical techniques and NWP modeling, and movement toward higher-resolution hydrologic modeling. Communication and understanding across scientific disciplines, and among scientific and disaster and emergency response communities has improved. Outreach such as community education has aimed to improve public understanding of the risks and local impact of flash floods. However, individuals and public authorities often continue to make ill-informed and life-threatening decisions. In many communities the adoption of planning, zoning, building codes, and actions necessary to mitigate fatalities and property loss from flash floods does not occur. People persist in building in and moving into areas vulnerable to flooding, often with little or no oversight or guidance from authorities. With increases in the likelihood of extreme precipitation and with land surface changes (e.g., increased urbanization), the engineering design criteria for stormwater infrastructure may no longer be adequate in many areas. Too often, individual decisions lead to tragic consequences. Even with clear admonitions against the practice of driving vehicles through flooded roadways, motorists continue to traverse inundated roads. Unfortunately, retractable barriers and other devices designed to prevent motorized vehicles from crossing flooded roadways at known problem areas are not commonly found. This results in a large fraction of the fatalities due to flash flooding. Of the 1,075 fatalities from flash flooding recorded from 1996 to 2014 across the United States, over 60% were related to vehicles. Videos and images broadcast on local and network news organizations and social media showing vehicles moving into and through floodwaters tend to desensitize motorists to the dangers of flooding and convey the message that such action is “okay” and not life threatening. Of flood-related fatalities, 21% occur when victims are involved in high-risk-taking behavior such as cleaning out drains and playing in floodwaters.

Current State of the Science

Advances have been made scientifically, operationally, and technologically, resulting in improvements in monitoring and the prediction of heavy rainfall associated with flash floods. Within the United States and globally, there is significant use of weather radar and radar networks. Increasingly equipped with dual-polarization capabilities, such weather radars offer the ability to provide rain rates over continental-sized areas with resolutions down to 1 km and updates as frequent as every 2 minutes. Multisensor quantitative precipitation estimation takes data from multiple radar networks, builds and rapidly updates high-resolution three dimensional analyses, and objectively blends data from radars with public and private-sector surface (including mesoscale rain gage networks) and upper-air observations, lightning and satellite data, and NWP output. In areas with no radar coverage and sparse surface observations, which is the norm globally, satellite-based estimates of precipitation offer the only information on rainfall rates. Although satellite estimates of precipitation have coarser resolution than weather radar, such estimates can provide potentially useful information for heavy rainfall and flash flood potential. The availability of readily configurable and implemented regional NWP models has given hydrometeorological forecasters access to quantitative predictions of rainfall with short lead-times. The development and application of physically based, high-resolution distributed hydrologic models, the associated estimation of flash flood guidance (FFG), and use of integrated FFG systems and decision support tools facilitate rapid assessment of flash flood threats and timely issuance of flash flood warnings to emergency response agencies and the public. Verification and observation of flash flood occurrences have improved with the aid of local observer networks reporting to meteorological services. Through integration with social sciences, including behavioral psychology, advances are being made toward understanding human responses to warnings. This may in turn offer opportunities to improve warnings and yield more effective responses.

Improving Forecasts, Warnings, Decision Support, and Communication

Improvements are needed in the forecast and warning of flash floods, decision support for both public and emergency response, and communication of flash flood risk and impacts. Additional observations of surface rainfall and streamflow are needed. Expanded use of private–public partnerships, such as citizen-based weather observing programs, have effectively improved observations and forecasts.

As of this writing, two national domain, real-time flood and flash flood forecasting systems provide quantitative streamflow, inundation, and other hydrologic information. The first, the NOAA National Water Model (NWM- http://water.noaa.gov/about/nwm), went into official operations in August of 2016.  This system now provides unified national hydrologic predictions for the entire contiguous U.S. landscape with hourly updating analyses and forecasts of streamflow on 2.7 million river channels, and soil moisture and surface inundation at 250-m resolution.  The NOAA NWM automatically integrates meteorological data from operational weather radars and multiple numerical weather prediction systems and produces hydrologic forecasts at a range of timescales.  For higher-frequency analyses and predictions of flash flood events, the Flooded Locations and Simulated Hydrographs (FLASH-  http://blog.nssl.noaa.gov/flash/) system provides updates of flooding conditions every 5 minutes in synch with the national multisensor radar analysis, albeit at a reduced, 1-km spatial resolution. Both systems employ a physics-based, dynamic, data-driven description of existing flood conditions and future flooding potential but both systems are relatively new and require sustained investment and engagement of the hydrometeorological science enterprise to realize their full potential.

Utilization of ensemble-based high-resolution meteorological forecasts with improved high-resolution hydrological models has facilitated a move toward uncertainty-assessing, probabilistic forecast guidance for flash floods. There is active research underway to improve the NOAA NWM- and FLASH-system ensemble forecasts.Validation of operational forecast models with high-resolution observations of rainfall, streamflow, inundation, and soil moisture is essential for understanding and reducing uncertainty in the forecasts. Storm-scale, rapid-update ensemble forecasts can be used to increase forecast lead time for heavy-to-extreme rainfall from storm events with relatively predictable characteristics, and hence increase forecast lead time and response time for flash floods. Improvements in robust high-resolution hydrologic models and flash flood guidance systems, which integrate such ensemble forecasts, are needed to support decision-making processes for hydrometeorological forecasters. Medium- and long-range forecast guidance of potential flood-causing rainfall may also increase emergency response preparedness.

Accurate forecast of rainfall and flash flood potential must be matched with timely issuance of actionable flash flood warnings and response. Communication of the flash flood risk to emergency response authorities and the public must be concise and location- and time-specific so that the risk is understood and action may be taken. Improved cooperation with and education for disaster risk and emergency response agencies will aid in the co-learning of the unique characteristics of flash floods, understanding of forecast capabilities and limitations, and response requirements. Opportunities exist to expand warning communication through visualization tools using neighborhood-scale impact areas, particularly for urban areas. Communication barriers, such as language, physical needs (e.g., for the hearing impaired), or remote accessibility need to be considered in effective communication of warnings.

The short-fused nature of flash flood events necessitates the preparation of effective, easily implemented emergency action plans. Disaster response planning and real-time flood response will benefit from high-resolution inundation mapping, particularly for high-impact, flash flood–prone areas. Emergency managers have long requested event-based real-time forecast inundation maps linked to infrastructure to enable them to pre-position people and resources to mitigate the impacts of floods more effectively. Complex and changing physiography, uncertainty in both observed and forecast rainfall, and accuracy of high-resolution hydrologic-hydraulic modeling make skillful, real-time inundation mapping a significant challenge, particularly on the short time scales of flash flooding. Planning studies, which leverage the expertise and resources among federal, state, and local agencies and the private sector, should be utilized to prepare disaster response agencies to better understand the impact area and target specific response plans for local communities. Such studies should inform communities about flood-prone areas, and be used to discourage community expansion and building in these zones. Response action plans may include specific local actions, such as the use and rapid engagement of local retractable barriers at flash flood–prone roadway crossings during flash flood events to prevent motorized vehicles from entering flood waters. 

Continued integration between the physical science involved with the development of flash flood warnings and the social science of improving the understanding of decision-making and behavioral response offers an opportunity to provide actionable intelligence to reduce the risk of lives lost to flash flooding. This integration can inform the forecasting community about the need for clarity and effective communication of warning messages and provide insight into education programs that increase understanding of flash floods and safety risks for the general public. Effective communication and decision-making should be viewed as an integral part of a successful forecast.


With changing hydroclimatology due to human activities, many communities are facing increasing vulnerability to flash flooding. Forecasting the time and location of flash floods requires high-resolution modeling of weather and water, assimilation of large data sets from high-resolution observations, and an integrated, coherent approach that allows meteorologists and hydrologists to make rapid assessments and warning decisions. Communication of warnings in meaningful ways is critical. To convert the use of increasingly accurate hydrometeorological forecasts into life- and property-saving actions by individuals and decision-makers, integrated approaches that address the critical elements of forecasting, warning, decision support, communication, planning, and outreach are needed.

[This statement is considered in force until April 2022 unless superseded by a new statement issued by the AMS Council before this date.]