Adverse Impact on Ponds of Organic Debris Deposited by Tornadoes.

Abstract

On May 3, 1999, 58 tornadoes were reported in Oklahoma in one of the largest outbreaks of severe weather in Oklahoma history. The tornadoes killed 44 people and destroyed over 3,100 homes. In addition to the obvious damage inflicted on residential areas, the tornadoes caused extensive damage to natural resources by destroying trees and other vegetation. Several ponds in the path of the largest tornado also were affected. Analysis of water quality revealed that the ponds had become incapable of supporting aquatic life because the decay of wind-deposited organic material had created a reducing atmosphere, which caused levels of dissolved oxygen to sink. Using National Weather Service data from the period 1950-1995, researchers calculated the mean annual area swept by tornadoes in Oklahoma to be approximately 19,000 acres. Combining this information with the estimated number of ponds in Oklahoma suggests that approximately 100 ponds would be in the path of a tornado each year.

Introduction

Tornadoes consist of rapidly rotating winds that circulate around a small area of intense low pressure and almost always begin as a funnel-shaped cloud attached to a large cumulonimbus cloud. The funnel cloud is labeled and recorded as a tornado only after it touches the ground and has been visually identified. Physically, all tornadoes are the same phenomenon, but they differ in size, intensity, and potential for damage. To classify the strength of tornadoes according to wind speed and observed damage, Theodore Fujita developed the F-scale in the late 1960s (Table 1) (1).

Not all tornadoes result in property damage or human injuries or fatalities. Oklahoma, for example, had 2,378 tornadoes between 1950 and 1995, and only 66 of them caused fatalities. Tornadoes typically occur in the spring; however, at least one tornado has occurred in each month in the United States between 1950 and 1999.

One of the crucial aspects of planning a response to tornado outbreaks is knowing the likelihood of occurrence. Rutch et al. formulated a model that predicted the annual probability of a tornado of a certain wind speed hitting a given area (2). This model consisted of two different probabilities--the probability that a tornado would reach a given wind speed and the probability that a tornado would hit a particular area. Tornado probability was found to be greatest in the central portion of the United States, with central Oklahoma and north central Alabama showing the highest probabilities of being struck. Rutch et al. compared the relative frequency of different F-scale tornadoes by dividing the United States into four quadrants (northeast, northwest, southeast, and southwest--latitude and longitude not specified). The wind speeds in each of the four quadrants were found to fit similar Weibull distributions, indicating that the proportion of tornadoes that fall within a given F-scale is fairly uniform among quadrants.

A number of studies have looked at the potential effect that factors such as population density, topography, and global climate patterns may have on tornado formation and frequency. Changnon conducted a five-year study of severe storms in the St. Louis, Missouri, area and concluded that "urban mechanisms tend to intensify rather than initiate convective activity" (3). Elsom and Meaden studied the suppression and dissipation of tornadoes in Greater London (4). They estimated that about 80 tornadoes occur each year in the United Kingdom. Their findings further indicated that, in the case of weak tornadoes, urban factors such as surface roughness and the urban heat island effect appear to suppress the formation of tornadoes and may have dissipated some tornadoes that were formed outside the urban area. Aguirre et al. studied the impact of urbanization and development on tornado occurrence from 1950 to 1990 in the United States and found that in metropolitan areas, the odds of tornado occurrence were higher than in rural counties (5). In a document on the consequences of disasters, the Centers for Disease Control and Prevention (CDC) delineated factors that could have an effect on tornado occurrence (6). Elevation, slope, and terrain roughness were cited as factors affecting tornado incidence and wind speed. Jones found no temporal trend in tornado frequency in Oklahoma during the period 1950-1995 and no relationship to El Nino periods; occurrence was positively correlated with population density but not with topography (i.e., flatness or hilliness made no difference) (7).

Project Background

On May 3, 1999, 58 tornadoes were reported in Oklahoma in one of the largest outbreaks of severe weather in Oklahoma history (Figure 1). Forty-four people were killed, over 3,100 homes were destroyed, 1,729 homes received major damage, and 3,277 homes incurred minor damage (8), Insured losses exceed $1 billion. The largest of the tornadoes traveled from southwest Oklahoma and into the heavily populated areas of Oklahoma County in central Oklahoma, where most of the damage and fatalities occurred. This tornado was over one-half mile wide at some points and was categorized as an F5 tornado on the Fujita scale. It produced winds of 318 miles per hour (mph), the strongest winds ever recorded. The immediate effect of the tornado outbreak was wide-scale destruction of property, death, and the scattering of destroyed materials for many miles. An aircraft wing was reported to have been deposited in Moore Oklahoma, which was the scene of extensive damage, after having traveled from near the town of Chickasha, some 40 miles to the southwest. Estimates have been made that disposal of the debris produced by the central Oklahoma tornado will shorten the life span of the nearest landfill by two years.

The force of the winds sheared homes from their foundations, denuded grass fields, removed asphalt from roads, tore bark from trees, ripped trees from the ground, and removed almost all material in the path of the tornado (see photo on page 11). The disturbed material was deposited over a wide area; however, it accumulated heavily in spots where it encountered immovable objects. Particular points of accumulation were bodies of water such as farm ponds. It is easy to visualize the fate of materials that blow across hard ground and then encounter a body of water: The water acts as a brake, and the pond can become an accumulation point.

The National Resource Conservation Service (NRCS) maintains ponds for erosion control throughout central Oklahoma. These ponds also serve as a source of drinking water for livestock and wildlife, and they are used recreationally for fishing, swimming, and boating. The NRCS office in Stillwater, Oklahoma, was asked to investigate the condition of several ponds under its jurisdiction. Area residents had expressed concern that some ponds might contain hazardous materials and pose a threat to wildlife. There had been reports of dead fish in several of the ponds. After reviewing the condition of several ponds, NRCS staff felt that for the ponds to remain effective in their intended use as erosion control structures, they would have to be drained and the accumulated material produced by the tornado would have to be removed. Because of the threat to downstream bodies of water, NRCS staff had concerns that further problems could develop if the water pumped from the ponds was toxic; therefore, an investigation was in itiated to determine whether pumping was a feasible alternative.

Methods

Five ponds were visited on June 7, 1999, one month after the tornado outbreak. Four of the ponds were directly in the path of a tornado, and one was approximately 75 meters to the south of the storm track. The initial visual review of the ponds revealed a varied array of materials, including trees, car parts, children's toys, mattresses, lumber, and other unidentifiable debris (see photo above). The four ponds in the path of the tornado obviously had been severely affected by the amount of debris--to the point that their volume had been significantly reduced. The fifth pond showed some sign of debris but appeared to be largely unaffected. The four ponds in the storm path had a slight septic odor, and the water in two of these ponds was gray in color. No dead fish were seen at any location.

Dissolved oxygen and pH were measured on site with a YSI Model 55 dissolved-oxygen meter and a Cole Parmer Model 59002-00 pH meter. Turbidity and specific conductance were measured upon return to the laboratory with a HACH Model 2100P turbidimeter and a Cole Parmer Model 19815-00 specific-conductance meter. Chemical oxygen demand (COD) was determined by spectrophotometric analysis after digestion in a HACH COD block digester. Following a five-day incubation at 20[degrees]C, biochemical oxygen demand (BOD) values were determined with a YSI Model 50B dissolved-oxygen meter.

Results

Dissolved-oxygen levels were low for all ponds, with ponds 1, 2, and 5 showing the greatest effect (Table 2). These were the smallest of the four ponds in the path of the tornado. Pond 4 was approximately 3 surface acres in size. Site 4d was in the most affected arm of Pond 4, and the dissolved-oxygen level was considerably lower at this site: 3.8 milligrams per liter (mg/L) at the surface and 1.7 mg/L at the bottom. No thermal stratification was seen at this site. A similar situation occurred at Pond 5, where the level of dissolved oxygen ranged from 3.5 mg/L at the surface to 2.7 mg/L at the bottom in approximately 4 feet of water, with no thermal stratification.

In all ponds except Pond 2, the water was relatively clear for ponds in this area of Oklahoma. Specific-conductance values were within the normal range for this region. All five-day biochemical-oxygen-demand ([BOD.sub.5]) values were relatively high compared with those of normal farm ponds, and Pond 2 showed a very high [BOD.sub.5] level. Ponds 1 and 2 showed relatively high COD levels for farm ponds.

Discussion

The [BOD.sub.5] of normal streams is very low and averages from 0.5-1.0 mg/L during dry seasons to 1.0-2.0 mg/L or higher during runoff periods (9). There is relatively little information concerning the typical [BOD.sub.5] for surface waters in Oklahoma; however, the 1996 mean values (collected monthly) for three sites along the North Canadian River in central Oklahoma were 4.7, 3.3, and 3.9 mg/L, and this river receives a significant quantity of municipal wastewater effluent (10). These data indicate that typical surface water [BOD.sub.5] values are normally less than 5.0 mg/L. The [BOD.sub.5] of domestic wastewater can be as low as 110 mg/L but typically averages around 220 mg/L,, while septic-tank effluent averages around 140 mg/L (11,12). In other words, BOD values in ponds 1 and 2 were elevated beyond background levels, indicating a significant loading of organic material. COD values also indicated extensive organic loading in the ponds.

Dissolved-oxygen values were low at all sites, and for several sites levels were below the Oklahoma State Water Quality Standard (5.0 mg/L) designated to support fish in surface waters. Ponds 1, 2, and 5 were incapable of supporting fish life. The dissolved-oxygen values at Pond 4 were marginally supportive of fish at the surface in some areas, but at Site 4d fell below the water quality standard. At all sites, dissolved-oxygen values were far below saturation even at a depth of only a few inches. Because turbidity values were relatively low and samples were collected on a warm, sunny day light penetration was fairly deep and was great enough to support attached and planktonic algae. The conclusion can be made that any oxygen diffusing into the water from the atmosphere or being produced by photosynthesis was rapidly being consumed by oxygen-demanding substances.

It is apparent that the high level of loading from wind-blown organic debris created a reducing atmosphere that resulted in the consumption of any oxygen present as well as of any new oxygen entering the system through diffusion or photosynthesis. The net result was ponds that were incapable of supporting aquatic life, The length of time (one month) between the event and the measurements indicates that this situation was obviously more than a shortterm phenomenon and that the conditions could be expected to continue until a significant portion of the organic load was completely decayed or had been removed.

Analysis of Historical Tornado Data

Because the effects of the tornado on ponds were severe, an attempt was made to determine the likelihood or risk of ponds being hit by a tornado. The purpose was to determine how many such events environmental managers could anticipate encountering in an average year. This knowledge should increase the ability of managers to respond to this aspect of a tornado emergency.

Data on Oklahoma tornadoes in the period 1950-1995 were downloaded from the Historical Tornado Data Archive maintained by the Storm Prediction Center of the National Weather Service. A total of 2,378 tornadoes were recorded in Oklahoma in the 46-year period. The magnitude of each tornado was indicated by three scales: the F-scale, which is based on wind speed; the PL scale, which is based on the length of the damage path; and the PW scale, which is based on the average path width. Scale categories are summarized in Table 1 and Table 3, and the number of tornadoes in each scale category is summarized in Table 4.

To estimate the area swept by the tornadoes, the authors multiplied the length of each tornado track by the average width of the tornado path. The track length used in the calculation was the entire distance traveled by the tornado, including segments where the tornado was not on the ground. This distance was reported in the database for 61 percent of F0, 73 percent of F1, 80 percent of F2, 95 percent of F3, 96 percent of F4, 100 percent of F5, and 15 percent of unclassified tornadoes. If the track length was not reported, the length in miles was estimated from the PL scale as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

The average width of the tornado path was reported in the database for 60 percent of F0, 68 percent of Fl, 72 percent of F2, 87 percent of F3 and F4, 100 percent of F5, and 13 percent of unclassified tornadoes. If the track width was not reported, an estimated average width in yards was derived from the PW scale value as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Figure 2 shows the frequency with which different values occurred for annual total estimated area swept by tornadoes. The annual total area swept appears to be lognormally distributed, with an arithmetic mean of approximately 19,000 acres, a geometric mean of about 11,000 acres, and a geometric standard deviation of 3.1. The highest annual total area swept in the 1950-1995 period was 89,000 acres in 1991. The fractions of the area swept by tornadoes of different intensities are presented in Table 5.

The total area of the state of Oklahoma is 43.5 million acres. If the total annual area swept by tornadoes is usually within the range of 3,500 to 34,000 acres (i.e., within one geometric standard deviation of the geometric mean), then the fraction of the state's area swept each year will typically be in the range 0.008 percent to 0.08 percent. The probability of two tornado paths overlapping is assumed to be negligible. To estimate the probability of a tornado path crossing a pond, the authors make the simplifying assumption that the distribution of ponds and the probability of tornadoes are uniform throughout the state. In that case, the number of ponds crossed by a tornado in a year would be given by the following expression:

(fraction of state's area swept)

x

(number of ponds in state)

No enumeration of ponds in Oklahoma was available; to estimate the number of ponds in the state, the authors counted ponds in a random sample of 25 digitized aerial photographs. Each photograph covered an area of about 18.6 square miles. Independent counts were made by two readers with an average inter-reader variability of 26 percent. The average number of ponds per square mile was extrapolated over the entire area of the state to arrive at an estimated total of 270,000 ponds with a standard error of 10 percent. The enumeration of very small ponds was difficult from the aerial photographs used in making this determination; therefore, it is likely that the smallest ponds were undercounted. Mid-range estimates of the area swept (19,000 acres per year) suggest that on average, approximately 100 ponds per year will be traversed by a tornado. This is probably a conservative estimate because the density of both tornadoes and ponds is higher in the center of the state than in the western panhandle, and because sma ller ponds were undercounted.

Analysis of May 3, 1999, Tornado Data

Preliminary data on the May 3, 1999, tornadoes were obtained online from the National Severe Storms Laboratory. The relationship between path length, path width, and F-scale is illustrated in Figure 3 for the 34 tornadoes for which these parameters were reported. Path length or path width data were not available for 10 of 19 F0 tornadoes, three of 11 Fl tornadoes, one of eight F2 tornadoes, and 10 of 11 tornadoes that were not classified on the Fujita scale. These tornadoes were omitted from the calculations of area swept. The total area swept by the May 3 tornadoes was estimated to be approximately 70,000 acres, representing 0.2 percent of the area of the state. Under these circumstances, it is not surprising that at least five ponds under the jurisdiction of the NRCS were observed to have suffered adverse impact from one tornado.

Conclusions

Although the probability of a tornado striking any particular pond is quite low, analysis of historical data suggests that each year dozens of ponds in Oklahoma will be traversed by a tornado. Organic debris dumped in a small pond--debris ranging from leaves and tree branches carried by F0 storms to lumber and mattresses carried by E4 or F5 storms--can deplete the dissolved oxygen to the extent that the pond is incapable of supporting aquatic life. Larger bodies of water are probably less susceptible to these effects, although the physical impact of debris accumulation in isolated areas could be significant and might result in fish kills. The potential for damage may be highest in aquaculture ponds since these ponds are shallow and contain dense populations. Pond owners should be aware of this potential for tornado damage. They should be prepared to assess ponds after a tornado and, if necessary, clean out or aerate them.

Acknowledgements: The authors would like to thank Ms. Carol Crouch and Mr. Tony Funderburk of the Natural Resources Conservation Service in Stillwater, Oklahoma, for identifying the problem and helping secure access to the tornado site. Thanks also go to Mr. Bob Springer of the Oklahoma Conservation Commission for his help in accessing and using the digital aerial photographs.

Corresponding Author: Dr. Robert A. Lynch, Assistant Professor, College of Public Health, Department of Occupational and Environmental Health, University of Oklahoma Health Sciences Center, 801 N.E. 13th St., Oklahoma City, OK 73104.

REFERENCES

(1.) Ahrens, C.D. (1988), Meteorology Today: An Introduction to Weather Climate, and the Environment, St. Paul, Minn.: West Publishing Company.

(2.) Rutch, C., K.J. Anselmo, R.E. Linney, and J.H. Sheesley (1992), "Tornado Risk Analysis," Plant/Operations Progress, 11(3):134-139.

(3.) Changnon, S.A., Jr. (1978), "Urban Effects on Severe Local Storms at St. Louis," Journal of Applied Meteorology, 17(5):578-586.

(4.) Elsom, D.M. and G.T, Meaden (1982), "Suppression and Dissipation of Weak Tornadoes in Metropolitan Areas: A Case Study of Greater London," Monthly Weather Review, 110(7):745-756.

(5.) Aguirre, B.E., E. Agramonte, J. Edminston, R. Saenz, and D.L. Stuart (1993), "The Human Ecology of Tornadoes," Demography, 30(4):623-632.

(6.) The Public Health Consequences of Disaster 1989 (1989), Atlanta, Ga.: Centers for Disease Control and Prevention.

(7.) Jones, E.P. (1998), "A Multi-Factorial Analysis of Tornado Risk in Oklahoma," Master's thesis, Oklahoma City, Okla.: University of Oklahoma Health Sciences Center.

(8.) National Resources Conservation Service, Disaster Report Update (1999), Flooding and Tornadoes Hit Oklahoma April 24, May 3-4, May 30, June 1, and June 30, 1999, Stillwater, Okla.: (U.S. Department of Agriculture State Office.

(9.)Velz, C.J. (1970), Applied Stream Sanitation, New York, N.Y.: (John Wiley & Sons.

(10.) Blazs, R.L., D.L. Boyle, T.E. Coffey, J.F. Kerestes, D.M. Walters, and D.K. White (1997), Water Resources Data Oklahoma Water Year 1996, Oklahoma City Okla: U.S. Geological Survey Water Data Report OK-96-1.

(11.) Metcalf and Eddy, Inc. (1979), Wastewater Engineering: (Treatment/Disposal/Reuse, New York: McGraw-Hill.

(12.) Salvato, J.A. (1992), Environmental Engineering and Sanitation, New York: John Wiley & Sons.