Robert Stonefield and Jan Jackson


Flash Flood Definition:

Within six hours (often within one hour) of a causative event such as intense rain, dam break, or ice jam formation, one or more of the following occurs:

There are actually three types of floods that the National Weather Service defines. Besides flash flooding, the term "river flooding" applies to any high flow, overflow, or inundation event which is threatening lives and property and can be quantified or indexed at specific locations (forecast points) along main stem rivers. These forecast points on rivers are located where there are established river gages. The term "areal flood" is applied to any high flow, overflow, or inundation in a defined area such as a group of counties or an area along a river or stream which threaten lives and property that is not covered by an river flood, or flash flooding.  The main difference between flash floods and areal floods is that areal flooding applies to longer duration precipitation, (greater than six hours), causing slow rises on rivers and streams.

The data included in this study were compiled from 1994 to 2007. Data for this study were collected from Local Storm Data publications and the National Climatic Data Center (NCDC) Storm Events database. .  For the purposes of this study an event was defined as each report within a county with a unique time and date, and by the type listed in Storm Data (flash flooding or flooding).  Multiple reports with the same date and time were treated as one event. For flood events, the event details page was used to separate the areal small stream floods from floods at river forecast points. This study consists of 1168 documented flood events. Of those, 766 were flash flood events, 277 were river floods, and 125 were areal floods.

Flash floods occurred in every one of the 40 counties in the WFO Blacksburg CWA. There are many factors that would contribute to variability of flash flooding across the WFO Blacksburg CWA, such as steep topography, small drainage basins, land use, soils, and vegetation. Land use, such as in small urban watersheds, can produce a quick flood response from even short duration convective rainfall.   The distribution of flash flood events by county, shown on the map below, indicates that population density, which would be associated with more urban or suburban land use, is one of the main factors. Six of the 10 most densely populated counties are also in the top 10 counties for number of flash flood events.   In addition, the larger urban areas (all independent cities in Virginia), which cover small portions of counties, have had more events than some of the rural counties.  The cities of Roanoke/Salem (14), Danville (11), Lynchburg (7) and Martinsville (7) all had more flash flood events than the counties of Charlotte, Stokes and Yadkin (6 events each). 

The number of flash flood warnings (red) per county, along with the population density (white) are on the map below:

This map shows the number of flash flood warnings (red) per county, along with the population density (white)

A flood severity index was developed based on the impact description for each flood, which often lists monetary damages as well. The table below shows this flood severity index.

Flood severity scale

Flood category




Few road closures, creeks and streams out of their banks. (Little or no damage)



Numerous road closures, numerous creeks and streams flooding, basement flooding, mudslides (Light Damage <$100K)



Some rescues, evacuations, few houses/businesses flooded. (Considerable damage $100K-$500K).



High Threat to Life/Property, several rescues, evacuation of and/or damage to several homes/businesses (Major Damage $500K-$2M).



Very High Threat to Life/Property, numerous rescues, evacuations of and/or damage to homes/businesses, (Catastrophic Damage >$2M).

Ex: Johnstown, PA (7/20/77); Fort Collins, CO (7/28/97); Madison Co, VA (6/27/95)

When the severity of the flash flooding is factored in (Figure below), topography plays a larger role in the significant flash flood events (moderate, severe or catastrophic).  Some of the counties with the highest number of significant flash flood events were also counties with lower population density (Tazewell, Giles, Bath, Rockbridge, Franklin and Smyth).  What these counties all have in common is small drainage basins in steep terrain.

Significant Flash Flooding by County Map

There is a wide range in the number of flood events each year (graph below) across the WFO Blacksburg CWA. Of the 766 flash floods reported during the 14 year period of this study, (1994 to 2007), the annual totals have ranged from lows of 6 in 1994 and 13 in 2005, to highs of 105 in 2003 and 103 in 1996.  There appeared to be no correlation between drought years, (1999 through 2002 in this study period), and frequency of flash flooding.  There also appeared to be no correlation between years active with tropical storms or their remnants, such as 2004, and the frequency of flash flooding.  However, the years with low numbers of areal floods and river floods do correspond to drought years.

Flash Flood and Flood Events by Year

The combination of steep topography and small basins in areas with heavy land use contribute to numerous flood events during meteorological conditions of deep moisture and strong forcing across the WFO Blacksburg CWA. Flash flood events can occur in widespread areas in the cool season, (October through March), with synoptic scale lift.  Flash flood events occur in both localized and in widespread areas with convection in the warm season, (April through September). Widespread flash flooding may also occur with tropical remnants from July to November.

Flash flooding occurred within the WFO Blacksburg CWA in each of the 12 months (graph below), with a maximum in June, and minimums in October and December.  The monthly distribution of events is represented by a slow increase during spring from March through May, a sharp rise from May to June, and then elevated numbers from July through September. In the cool season, the distribution of events is characterized by low numbers from October through December, a peak in January, and then low numbers again in February.  The peak in January is likely attributable to efficient runoff due to saturated or frozen soils, dormant vegetation and melting snow cover. 

June has the highest number of flash flood events, (187), representing nearly 25% of the total, which is also nearly twice the number of the next highest month. The second, third, and fourth highest months are all in the warm season and include July (104), September (93), and August (90). 

River floods at forecast points occurred in every month of the year (graph below). There were high numbers in each month in the late cool season from January through March. There were other peaks in the number of flood events during June and September. The June peak of 26 events corresponds to the peak month for flash floods, and is likely attributable to more widespread flash floods which evolve into river flooding.  The September peak of 37 events, which tied for the third highest month, corresponds to the peak in tropical remnant activity across the area from 1994 to 2007; five of the seven tropical related flood events occurred in September.

Flash Flood and River Flood Events by Month

Flash flood events occurred at all hours of the day (graph below). The distribution of flash flood events by time shows a gradual rise from a low point in the early morning (300 am to 600 am), to a peak in the late afternoon and early evening (300 pm to 600 pm), and then a sharp decline after 900 pm. Over a third (290 or 38%) of all events occurred between 300 pm and 900 pm. This covers the time of peak heating for convection (300 pm to 600 pm), and a lag period for runoff (600 pm to 900 pm). Nearly two thirds (486 or 64%) of all events occurred during the 12 hour period from 900 am to 900 pm. 

Areal flood events occurred at all hours of the day. Nearly half (45%) of all the areal flood events occurred between 900 am and 300 pm. With the peak frequency occurring before the peak heating of the day, it suggests that areal flooding is most often not a direct result of heavy precipitation from convection. River forecast flood events occurred at all hours of the day (graph below). The distribution of river flood events by time shows less of a trend than flash flooding or areal flooding, but has a relative maximum in the morning with a peak from 600 am to 900 am. This is 12 to 15 hours later than the flash flood peak and may represent the lag time for runoff from creeks and small streams to the mainstem rivers.

Flash and River Flood Events by Time

All of the flood events were classified according the Flood Severity Index. Out of the 766 Flash Flood events (graph below), nearly 47% (356) were FS1. With the addition of all FS2 events (281), the vast majority of flash flood events, over 83%, were either FS1 or FS2. Another 11% (84) of the events were FS3, so a very small number (43 or 6%) of all flash flood events in the WFO Blacksburg CWA from 1994 to 2007 were FS4 or FS5.  Although the significant flash flood events (FS3, FS4 and FS5) were relatively rare, they are critical in terms of forecasts and warnings because of their huge impact with hundreds of thousands to millions of dollars in damage and greater potential loss of life. There were seven flood events with one or more fatalities during the period of study, and all but one of the events were FS3 or higher. 

Flash Flood Events by Severity

Significant flash flood (127) events occurred in every month of the year except December (graph below). Nearly two thirds of them, (81 or 64%), occurred during the following three months: June (32), January (26), and September (23).  Three other months had higher totals of ten or more: May (12), November (12), and July (10).  It is not surprising that June is the peak for significant flash flooding, as it is also the peak for all flash flooding. A second peak in September could also be expected as this month had the third highest of all flash flood events, and would be heavily influenced by remnants of tropical systems which have historically brought many of the area’s significant floods. Twenty of the twenty three significant flash floods in September were associated with two tropical storms or their remnants (Fran, September 4-6, 1996 and Jeanne, September 27-28, 2004). Likewise, the secondary peak for significant flash flooding in January was heavily influenced by just a few synoptic scale storm systems. The twenty six significant flash flood events in January occurred from only four storm systems.  One storm on January 18-19, 1996 accounted for 14 of the significant flash flood events. This storm featured rapid snowmelt of a deep snow pack, in addition to the heavy rainfall (Leathers et al. 1998). The three other storms in January that accounted for the rest of the significant flash flood events (12), did not involve significant snowmelt.

The lower numbers of significant flash flood events in July and August stand out since they are in the convective season when the moisture content of the atmosphere is at its climatological peak (precipitable water during these two months averages from 1.25 to 1.50 inches). However, there are several factors that are less conducive to significant flash floods in July and August.  Weaker upper level flow tends to produce less organized convection making very heavy downpours more isolated. There were only two storm systems producing multiple significant flash flood events in July, and none in August.  July and August are also before the tropical season peak in September.  Finally, beginning in July, antecedent soil moisture conditions are typically becoming drier, and remain near their driest point into November.

The majority of significant areal flood events, over 40%, (20), occurred in the month of September (graph below). All but two of the 20 significant areal flood events in September were associated with the remnants of tropical systems (Fran, September 4-6, 1996, Isabel, September 17-19, 2003, Frances, September 5-8, 2004, Ivan, September 15-17, 2004 and Jeanne, September 27-28, 2004).  Twenty one of the remaining twenty eight significant areal flood events occurred in the cool season. All of the 21 cool season events were associated with only three synoptic scale storm systems. Each of these storms brought heavy precipitation, producing significant flooding in several counties.  In January (8 events) all the flooding resulted from widespread warm heavy rainfall and snowmelt on January 19, 1996. In February (6 events), all the flooding resulted from a deepening low pressure in the southeast U.S. on February 22 to 23, 2003, producing heavy overrunning rainfall. In November (7 events), all the flooding was caused by a heavy upslope rainfall on November 19, 2003.  The rest of the significant areal flood events, June (4), March (1) and July (1), all came from less organized systems and were not as widespread.

The severity of river flooding is based on three categories: minor, moderate and severe. These categories correspond to pre-defined ranges of levels above flood stage, which varies at each forecast point.  Of the 277 events, 95 were classified as moderate, and 25 were classified as major, the rest (157) were classified as minor. For the purposes of this paper, we are defining significant river flooding as moderate or major.  Impacts to areas around the river gage are much greater beginning at the moderate flood stage. Significant flooding accounted for 43% (120 of 277) of the river flood events in the Blacksburg HSA.  Significant river flooding occurred in every month of the year (graph below).  January had the highest number of significant river flood events (32), and the three month period of January through March accounted for over 50% (63 of 120) of all significant river flood events.

Significant Flood Events

During the period of study, there were a total of 17 synoptic scale storm systems that produced multiple events of significant flash flooding. Ninety seven of the 127 significant flash flood events (76%) occurred in one of these 17 storms, indicating that significant flash flooding is most often widespread enough to occur in multiple counties.  Seven of the 17 storms occurred in the cool season, (October through March), and ten occurred in the warm season, (April through September). Four of the cool season events were in January, and four of the warm season events were in June. There were two areas, or clusters of counties, (figure below) that emerged in the spatial distribution of significant flash flood events in these storms. One area (Zone 1) that had several (7) storms in common encompassed the region that extends from the southwest mountains of Virginia, (Smyth and Tazewell counties), northeast into southeast West Virginia, (Mercer, Summers and Greenbrier counties).  The other area (Zone 2) that had several (5) storms in common was in the region that extends along the Blue Ridge in southwest Virginia from Franklin and Roanoke counties, northeast through Bedford, Campbell, Botetourt, Rockbridge and Amherst counties.  These areas have topographic similarities, with Zone 1 being predominately western slopes of the Appalachian Mountains, and Zone 2 primarily eastern slopes of the Blue Ridge Mountains. 

Most common flash flood zones

The spring and summer months are the most active for flash floods. In particular, June is our most active month, with nearly 25% of all flash floods in the period from 1994 to 2007. June is also the peak month for significant flash floods, which cause hundreds of thousands to millions of dollars in damage, and have a greater potential loss of life. Each year, more deaths occur from flooding than any other severe weather related cause. In addition, more than half of all flood deaths result from vehicles being swept into flood waters. Remember your flood safety rules with the peak of our flash flood season approaching: