Estimated Flood Frequency and Corresponding Water-Surface Elevations at the Confluence of the Potomac and Shenandoah Rivers, Harpers Ferry, West Virginia

Abstract

The Harpers Ferry National Historical Park is located at the confluence of the Shenandoah and Potomac Rivers in Harpers Ferry, West Virginia, and has historically been affected by flooding from both rivers. Because the drainage areas of both rivers are large and drain geographically separate regions, either river may contribute individually to flooding, or the combination of flows may be the source of flooding. Previous flood-frequency analyses have been conducted at U.S. Geological Survey streamflow-gaging stations upstream on the Potomac River at Shepherdstown, West Virginia, and on the Shenandoah River at Millville, West Virginia; and downstream on the Potomac River at Point of Rocks, Maryland. A flood-frequency analysis for Harpers Ferry can be expected to fall within the streamflows and recurrence intervals predicted at the surrounding stations. The prediction of corresponding water-surface elevations at Harpers Ferry is difficult, however, because of differences in rainfall distribution across the two regions and the timing of the peaks on both rivers.

This report presents updated flood-frequency analyses for the U.S. Geological Survey streamflow-gaging stations on the Potomac River at Shepherdstown, West Virginia, Shenandoah River at Millville, West Virginia, and Potomac River at Point of Rocks, Maryland. The analyses include flood peaks through the 1999 water year. Adjustments also were made to account for historical flood information and high outliers.

Peak gage heights, peak discharges, and recurrence intervals are presented for the 12 largest floods at Harpers Ferry where corresponding water-surface elevations from the 3 U.S. Geological Survey streamflow-gaging stations are available. Peak discharges and recurrence intervals for the 12 floods at Harpers Ferry were estimated from data collected at the stations and are presented along with corresponding peak stages determined by the National Park Service. These estimates of peak discharge and corresponding peak gage heights were then used to estimate a stage-discharge relation for Harpers Ferry, which is also presented along with corresponding recurrence intervals at increments of 1 foot in gage height.

The report also describes statistical relations between recorded peak water-surface elevations at Harpers Ferry and the three surrounding streamflow-gaging stations. Three equations were developed to predict the peak water-surface elevation at Harpers Ferry on the basis of elevations from the individual surrounding streamflow-gaging stations. A fourth equation, which was developed by multiple regression techniques, shows the relation between peak water-surface elevation at Harpers Ferry based on known peak water-surface elevations at both the Shenandoah River at Millville, West Virginia and the Potomac River at Shepherdstown, West Virginia. Analysis of the results indicated that the equation based on peak water-surface elevations from both Millville and Shepherdstown had the strongest statistical relation and the smallest range of prediction error, from +0.91 feet to -1.28 feet.

Introduction

The Harpers Ferry National Historical Park is operated by the U.S. Department of the Interior, National Park Service (NPS). The Park, at the confluence of the Shenandoah and Potomac Rivers in Harpers Ferry, W. Va., has historically been affected by flooding from both rivers. Because the drainage areas of both rivers are large and drain geographically separate regions, either river may contribute individually to flooding, or the combination of flows may be the source of flooding. Streamflow-gaging stations have been operated for many years on both rivers at locations just upstream and downstream of the confluence. A continuous-record streamflow-gaging station has never been operated at Harpers Ferry, however, and a relation between flow rate and water-surface elevation at the Park has never been established.

Previous flood-frequency analyses have been conducted at U.S. Geological Survey (USGS) streamflow-gaging stations upstream on the Potomac River at Shepherdstown, W. Va., and on the Shenandoah River at Millville, W. Va.; and downstream on the Potomac River at Point of Rocks, Md. A flood-frequency analysis for Harpers Ferry can reasonably be expected to fall within the streamflows and recurrence intervals predicted at the surrounding stations. Predicting corresponding water-surface elevations at Harpers Ferry is difficult, however, because of differences in rainfall distribution across the two regions and the timing of the peaks on both rivers. The conventional approach to making such predictions requires a detailed and expensive computer model application.

Flood damage to the Park has emphasized the need for the NPS to begin (1) correlating Park-related flood damage with specific flood events, (2) predicting expected inundation and subsequent damage based on the magnitude and frequency of flood events, and (3) prioritizing Park structures for protection from future flood damage (Doheny, 1997). To assist NPS in meeting these objectives, the USGS, in cooperation with NPS, initiated a study in February 2000 to (1) update flood-frequency distributions for 3 USGS streamflow-gaging stations upstream and downstream of the Park, (2) make estimates of flood frequency compared to water-surface elevation at Harpers Ferry for the 12 largest floods for which corresponding water-surface elevations exist, and (3) determine if a relation exists between flood elevations at Harpers Ferry and the surrounding gaged locations.

Purpose and Scope

This report presents updated flood-frequency analyses for three USGS streamflow-gaging stations upstream and downstream of Harpers Ferry National Historical Park. Peak gage heights, peak discharges, and recurrence intervals are presented for the 12 largest floods at Harpers Ferry for which corresponding water-surface elevations from the USGS streamflow-gaging stations were available. Peak discharges and recurrence intervals for the 12 floods were estimated from data collected at the USGS streamflow-gaging stations and presented along with corresponding peak stages determined by NPS (William Hebb, National Park Service, written commun., 2000). These estimates of peak discharge and corresponding peak gage heights were subsequently used to estimate a stage-discharge relation for Harpers Ferry. This stage-discharge relation is presented along with corresponding flood recurrence intervals at increments of 1 ft (foot) in gage height.

The report also presents statistical relations between recorded peak water-surface elevations at Harpers Ferry and the three streamflow-gaging stations nearby. Data used in this analysis include the 12 largest floods at Harpers Ferry where corresponding water-surface elevations from the surrounding stations were available. Three equations were developed to predict the peak water-surface elevation at Harpers Ferry on the basis of data from the individual surrounding stream gages. These equations show the relation between peak water-surface elevations at (1) Harpers Ferry based on a known peak water-surface elevation at Shenandoah River at Millville, W. Va., (2) Harpers Ferry based on a known peak water-surface elevation at Potomac River at Shepherdstown, W. Va., and (3) Harpers Ferry based on a known peak water-surface elevation at Potomac River at Point of Rocks, Md. A fourth equation, which was developed by multiple regression techniques, shows the relation between peak water-surface elevation at Harpers Ferry based on known peak water-surface elevations at both the Shenandoah River at Millville, W. Va., and the Potomac River at Shepherdstown, W. Va.

Description of Study Area

The study area includes the flood-plain areas of Harpers Ferry National Historical Park along the Shenandoah and Potomac Rivers, and the river reaches between the USGS streamflow-gaging stations on the Potomac River at Shepherdstown, W. Va., the Shenandoah River at Millville, W. Va., and the Potomac River at Point of Rocks, Md. (fig. 1). Basic data compiled for each of these stations and for Harpers Ferry include the station number, station name and location, the latitude and longitude, the period of gage record, the drainage area at the station, and the mean sea level datum of the station in relation to the National Geodetic Vertical Datum (NGVD) of 1929 (Doheny, 1997). This information is presented in table 1.

Figure 1. Location of U.S. Geological Survey streamflow-gaging stations on the Potomac and Shenandoah Rivers near Harpers Ferry National Historical Park, West Virginia. Download a print-quality PDF image

Table 1. Basic data for streamflow-gaging stations in the vicinity of Harpers Ferry National Historical Park, West Virginia
[mi², square miles; ft, feet]

Station no.	Station name and location	Latitude	Longitude	Period of record	Drainage area at station (mi²)	Station datum (ft above sea level)	Miles above mouth
01618000	Potomac River1 at Shepherdstown, W. Va.	39°26'04″	77°48'07″	1928-53, 1964-93	5,936	281.00	185.0
01620000	Potomac River2 at Harpers Ferry, W. Va.	39°19'25″	77°43'44″	1889 to present	9,372	245.53	172.9
01636500	Shenandoah River at Millville, W. Va.	39°16'55″	77°47'22″	1895-1909, 1928 to present	3,022	293.00	4.7
01638500	Potomac River at Point of Rocks, Md.	39°16'25″	77°32'35″	1895 to present	9,651	200.63	160.9

Notes:

1. Annual maximum discharges were recorded from 1954 to 1964. Station operated as a flood-warning gage since August 1998 and currently records only gage heights greater than 5.0 ft.
2. Period of record includes gage-height record only. Some periods have only occasional gage readings. Drainage area includes both Potomac and Shenandoah watershed areas. Gage datum was 247.50 ft above mean sea level prior to 1901, and 245.53 ft from 1901 to present.

Acknowledgments

The authors would like to thank William Hebb of the National Park Service, Harpers Ferry National Historical Park, for planning assistance and providing technical information on historical flooding at Harpers Ferry National Historical Park.

Flood-Frequency Analysis

Flood-frequency analyses for the Potomac River at Shepherdstown, W. Va., Shenandoah River at Millville, W. Va., and the Potomac River at Point of Rocks, Md., were updated to include flood peaks through the 1999 water year. The analysis was performed by standard techniques specified in "Guidelines for Determining Flood Flow Frequency", Bulletin 17B (Interagency Advisory Committee on Water Data, 1982). Adjustments for historical flood information and high outliers as defined in Bulletin 17B were also included in the analysis. The results of the flood-frequency analyses for the three USGS streamflow-gaging stations are shown in table 2.

Table 2. Annual exceedance probabilities, recurrence intervals, and corresponding flood discharges for streamflow-gaging stations upstream and downstream of Harpers Ferry National Historical Park, West Virginia, based on flood peaks up to and including the 1999 water year
[ft³/s, cubic feet per second]

Annual Exceedance Probability	Recurrence Interval1 (years)	Discharge for Station No. 01618000, Potomac River at Shepherdstown, W. Va.2 (ft³/s)	Discharge for Station No. 01636500, Shenandoah River at Millville, W. Va.3 (ft³/s)	Discharge for Station No. 01638500, Potomac River at Point of Rocks, Md.3 (ft³/s)
0.995	1.005	24,630	6,460	31,620
0.990	1.01	26,800	7,450	34,830
0.950	1.05	34,460	11,230	46,350
0.900	1.11	39,890	14,130	54,450
0.800	1.25	48,200	18,850	66,990
0.500	2	71,890	33,830	103,500
0.200	5	112,900	63,410	167,900
0.100	10	146,100	89,660	220,600
0.040	25	195,500	131,500	299,700
0.020	50	238,200	169,700	368,400
0.010	100	286,200	214,600	446,300
0.005	200	340,500	267,200	534,400
0.002	500	423,100	350,400	669,200

Notes:

1. The recurrence interval is the average number of years between floods equal to or greater than the corresponding flood discharges shown. It should be noted that this is an average number of years and does not imply that it will be that many years before another event of that magnitude occurs. The reciprocal of the recurrence interval is the probability of the event occurring in any one year. For instance, a 50-year flood has a 0.02 probability, or 2-percent chance, of occurring in any year.
2. Flood-frequency data adjusted for historical flood information and one high-outlier peak.
3. Flood-frequency data adjusted for historical flood information.

Gage-height and peak-flow records were reviewed for the three streamflow-gaging stations to determine the largest flood events with corresponding peak stages at the three stations and at Harpers Ferry. The 12 largest flood events were selected for analysis to relate the known peak gage height at Harpers Ferry to an approximate recurrence interval. The selected flood events, peak gage heights at Harpers Ferry, and corresponding mean sea-level elevations are presented in table 3.

Table 3. Twelve largest floods at Harpers Ferry, West Virginia, where corresponding peak stages were recorded at three surrounding U.S. Geological Survey streamflow-gaging stations
[ft, feet]

Date of flood	Peak gage height at Harpers Ferry (ft)	Mean sea-level elevation (ft)
March 19, 1936	36.5	282.03
October 16, 1942	33.8	279.33
November 6, 1985	29.8	275.33
September 8, 1996	29.8	275.33
June 23, 1972	29.7	275.23
January 20, 1996	29.2	274.73
April 27, 1937	29.0	274.53
May 13, 1924	27.6	273.13
August 20, 1955	23.9	269.43
October 29, 1937	21.5	267.03
March 21, 1975	21.5	267.03
October 10, 1976	21.5	267.03

For these 12 flood events, the peak gage heights, corresponding mean sea level elevations, and peak discharges were compiled for the three streamflow-gaging stations surrounding Harpers Ferry. The date and time of the peaks were also compiled if available. The recurrence interval for each event was calculated at all three stations on the basis of the updated flood-frequency analyses presented earlier. The results are shown in table 4.

Table 4. Peak-flow data and approximate recurrence intervals for U.S. Geological Survey streamflow-gaging stations near Harpers Ferry National Historical Park, West Virginia
[GH, peak gage height; MSL, mean sea level elevation; Q, peak discharge; DT, date and time of peak; RI, approximate recurrence interval; ft, feet; ft³/s, cubic feet per second; ----, time of peak unknown]

Date of peak at Harpers Ferry	Station No. 01618000, Potomac River at Shepherdstown, W. Va.	Station No. 01636500, Shenandoah River at Millville, W. Va.	Station No. 01638500, Potomac River at Point of Rocks, Md.
March 19, 1936	GH = 42.10 ft MSL = 323.10 ft Q = 335,000 ft³/s DT = 03/19/1936, 0600 RI = 185 years	GH = 26.36 ft MSL = 319.36 ft Q = 151,000 ft³/s DT = 03/18/1936, 2030 RI = 35 years	GH = 41.03 ft MSL = 241.66 ft Q = 480,000 ft³/s DT = 03/19/1936, 0930 RI = 125 years
October 16, 1942	GH = 32.68 ft MSL = 313.68 ft Q = 201,000 ft³/s DT = 10/16/1942, 2200 RI = 25 years	GH = 32.40 ft MSL = 325.40 ft Q = 230,000 ft³/s DT = 10/16/1942, 1500 RI = 120 years	GH = 40.43 ft MSL = 241.06 ft Q = 418,000 ft³/s DT = 10/16/1942, 2230 RI = 75 years
November 6, 1985	GH = 31.44 ft MSL = 312.44 ft Q = 187,000 ft³/s DT = 11/07/1985, 0030 RI = 20 years	GH = 25.60 ft MSL = 318.60 ft Q = 142,000 ft³/s DT = 11/06/1985, 1900 RI = 30 years	GH = 36.28 ft MSL = 236.91 ft Q = 309,000 ft³/s DT = 11/07/1985, 0330 RI = 25 years
September 8, 1996	GH = 29.00 ft MSL = 310.00 ft Q = 159,400 ft³/s DT = 09/08/1996, ---- RI = 10 years	GH = 26.82 ft MSL = 319.82 ft Q = 156,000 ft³/s DT = 09/08/1996, 0800 RI = 40 years	GH = 36.32 ft MSL = 236.95 ft Q = 310,000 ft³/s DT = 09/08/1996, 1500 RI = 25 years
June 23, 1972	GH = 31.44 ft MSL = 312.44 ft Q = 187,000 ft³/s DT = 06/23/1972, 2330 RI = 20 years	GH = 21.89 ft MSL = 314.89 ft Q = 103,000 ft³/s DT = 06/23/1972, 1200 RI = 15 years	GH = 37.43 ft MSL = 238.06 ft Q = 347,000 ft³/s DT = 06/23/1972, 2330 RI = 40 years
January 20, 1996	GH = 32.50 ft MSL = 313.50 ft Q = 199,500 ft³/s DT = 01/21/1996, ---- RI = 25 years	GH = 23.61 ft MSL = 316.61 ft Q = 121,000 ft³/s DT = 01/20/1996, 2130 RI = 20 years	GH = 36.54 ft MSL = 237.17 ft Q = 313,000 ft³/s DT = 01/21/1996, 0430 RI = 30 years
April 27, 1937	GH = 33.20 ft MSL = 314.20 ft Q = 207,000 ft³/s DT = 04/27/1937, 1300 RI = 30 years	GH = 20.20 ft MSL = 313.20 ft Q = 87,400 ft³/s DT = 04/27/1937, ---- RI = 9 years	GH = 33.86 ft MSL = 234.49 ft Q = 310,000 ft³/s DT = 04/27/1937, 2000 RI = 25 years
May 13, 1924	GH = 29.80 ft MSL = 310.80 ft Q = 168,000 ft³/s DT = 05/13/1924, ---- RI = 15 years	GH = 21.10 ft MSL = 314.10 ft Q = 119,000 ft³/s DT = 05/13/1924, ---- RI = 20 years	GH = 32.20 ft MSL = 232.83 ft Q = 277,000 ft³/s DT = 05/13/1924, 1500-1800 RI = 20 years
August 20, 1955	GH = 25.26 ft MSL = 306.26 ft Q = 124,000 ft³/s DT = 08/20/1955, 0500 RI = 6 years	GH = 21.45 ft MSL = 314.45 ft Q = 99,000 ft³/s DT = 08/19/1955, 1600 RI = 12 years	GH = 29.08 ft MSL = 229.71 ft Q = 214,000 ft³/s DT = 08/20/1955, 0500 RI = 9 years
October 29, 1937	GH = 26.78 ft MSL = 307.78 ft Q = 138,000 ft³/s DT = 10/29/1937, 2100 RI = 8 years	GH = 12.72 ft MSL = 305.72 ft Q = 34,400 ft³/s DT = 10/29/1937, 2100 RI = 2 years	GH = 24.93 ft MSL = 225.56 ft Q = 175,000 ft³/s DT = 10/30/1937, 0300 RI = 5 years
March 21, 1975	GH = 22.19 ft MSL = 303.19 ft Q = 99,400 ft³/s DT = 03/21/1975, 0300 RI = 4 years	GH = 18.86 ft MSL = 311.86 ft Q = 75,900 ft³/s DT = 03/21/1975, 0400 RI = 7 years	GH = 26.15 ft MSL = 226.78 ft Q = 181,000 ft³/s DT = 03/21/1975, 0930 RI = 6 years
October 10, 1976	GH = 25.31 ft MSL = 306.31 ft Q = 124,000 ft³/s DT = 10/10/1976, 1800 RI = 6 years	GH = 15.30 ft MSL = 308.30 ft Q = 49,400 ft³/s DT = 10/11/1976, 0115 RI = 3 years	GH = 27.25 ft MSL = 227.88 ft Q = 193,000 ft³/s DT = 10/11/1976, 0200 RI = 7 years

To determine approximate recurrence intervals for the 12 flood events at Harpers Ferry, a flood-frequency analysis and estimates of peak discharge for each event are required. Because peak discharge has not been measured historically at Harpers Ferry, flood-frequency estimates were made by partially updating a graph of flood magnitudes and frequencies for the Potomac River that was originally presented in Carpenter (1980) and Dillow (1996). This graph allows the user to estimate the 2-year, 5-year, 10-year, 25-year, 50-year, 100-year, and 500-year discharge based on a known drainage area along the Potomac River. The graph was updated for the drainage area interval between the streamflow-gaging stations on the Potomac River at Shepherdstown, W. Va. (drainage area = 5,936 mi², or square miles) and the Potomac River at Point of Rocks, Md. (drainage area = 9,651 mi²). The updated flood-frequency discharges for these stations were used to develop new curves within this range of drainage areas (fig. 2). Using the drainage area of 9,372 mi²/s at the confluence of the Shenandoah and Potomac Rivers from figure 2, flood-frequency estimates for Harpers Ferry were obtained graphically for the 2-, 5-, 10-, 25-, 50-, 100-, and 500-year floods (table 5).

Figure 2. Flood magnitudes and frequencies for drainage areas on the Potomac River between Shepherdstown, West Virginia and Point of Rocks, Maryland. Download a print-quality PDF image

Table 5. Flood-frequency estimates for the Potomac River at Harpers Ferry, West Virginia, based on data up to and including the 1999 water year
[ft³/s, cubic feet per second]

Recurrence Interval (years)	Discharge (ft³/s)
2	102,000
5	162,000
10	212,000
25	293,000
50	360,000
100	435,000
500	650,000

Peak discharges at Harpers Ferry for each of the 12 flood events were then estimated by transferring known discharges from the surrounding stations. Peak discharges at the three streamflow-gaging stations were compared to determine differences in discharge between the areas upstream and downstream of Harpers Ferry. Peak discharges at the respective stations were transferred downstream or upstream to Harpers Ferry on the basis of drainage-area ratios raised to an exponential transfer coefficient. The coefficient was obtained for different flood events by measuring the slope of the appropriate line in figure 2 at a known discharge, drainage area, and recurrence interval. Consideration also was given to differences in the timing of the respective peaks on the Potomac and Shenandoah Rivers, as well as sources of major inflow between Harpers Ferry and Point of Rocks. After estimates of peak discharge were determined for the 12 flood events, recurrence intervals were determined by interpolating between the flood-frequency estimates indicated in table 5. Estimates of peak discharge and recurrence intervals at Harpers Ferry for the 12 flood events are shown in table 6.

Table 6. Peak gage heights, estimated peak discharges, and approximate recurrence intervals for the 12 largest floods at Harpers Ferry, West Virginia, where corresponding peak gage heights were recorded at three surrounding U.S. Geological Survey streamflow-gaging stations
[ft, feet; ft³/s, cubic feet per second; +, slightly greater than; -, slightly less than]

Date of flood event	Peak gage height at Harpers Ferry (ft)	Estimated peak discharge at Harpers Ferry (ft³/s)	Approximate recurrence interval (years)
March 19, 1936	36.5	467,000	125
October 16, 1942	33.8	407,000	75
November 6, 1985	29.8	326,000	35-
September 8, 1996	29.8	326,000	35-
June 23, 1972	29.7	319,500	30
January 20, 1996	29.2	305,000	25
April 27, 1937	29.0	302,000	25
May 13, 1924	27.6	270,000	20-
August 20, 1955	23.9	209,000	10-
October 29, 1937	21.5	178,500	6+
March 21, 1975	21.5	178,500	6+
October 10, 1976	21.5	178,500	6+

The following example indicates how an estimated peak discharge for Harpers Ferry was obtained for the flood of March 19, 1936. From table 4, the peak discharges for station number 01618000 (Potomac River at Shepherdstown, W. Va.) and station number 01636500 (Shenandoah River at Millville, W. Va.) were 335,000 ft³/s (cubic feet per second) at 0600 hours on March 19, 1936, and 151,000 ft³/s at 2030 hours on March 18, 1936, respectively. The sum of these peak discharges upstream of Harpers Ferry is 486,000 ft³/s. From table 4, the peak discharge for station number 01638500 (Potomac River at Point of Rocks, Md.) downstream of Harpers Ferry was determined to be 480,000 ft³/s at 0930 hours on March 19, 1936. Although the cumulative discharges upstream and downstream of Harpers Ferry are very similiar for this event, there is an 8 and one-half hour lag time between the peak on the Shenandoah River and the peak on the Potomac River. As a result, the peak discharge at Harpers Ferry can be expected to be somewhat less than the cumulative 486,000 ft³/s from the stations upstream of Harpers Ferry due to the timing of the peaks. The peak discharge can also be expected to be somewhat less than the 480,000 ft³/s at Point of Rocks because of a slightly smaller drainage area. An estimate of the peak flow at Harpers Ferry was then obtained by using a drainage-area ratio to transfer the discharge from Point of Rocks:

QHF / QPOR = ( DAHF / DAPOR )^x      (1)

where:

QHF = Peak discharge at Harpers Ferry, in ft³/s;

QPOR = Peak discharge at Point of Rocks = 480,000 ft³/s;

DAHF = Drainage area at Harpers Ferry = 9,372 mi²/s;

DAPOR = Drainage area at Point of Rocks = 9,651 mi²/s; and

x = Transfer exponent from figure 2 = approximately 0.93 for discharge at Point of Rocks of 480,000 ft³/s and a recurrence interval of 125 years.

Solving equation 1 for QHF; QHF = 467,082; or approximately 467,000 ft³/s.

Peak discharges for the other 11 flood events at Harpers Ferry were estimated in the same manner. In a few instances, slightly different discharges were obtained for different flood events of the same peak stage. This is most likely due to variations in flooding contributions from the two rivers among different flood events. In these instances, a mean discharge was calculated from the discharges obtained by use of drainage-area ratios.

The gage heights and estimated discharges from table 5 were then plotted to determine an approximate partial stage-discharge rating curve for Harpers Ferry. Discharge was estimated at 1-ft intervals of gage height between 21.0 and 37.0 ft. The results are shown graphically in figure 3 and in tabular format (table 7). Table 7 also presents the approximate recurrence interval associated with each 1-ft interval of gage height and corresponding discharge. A summary table of flood-frequency data for Harpers Ferry is presented in appendix 1.

Figure 3. Approximate stage-discharge relation for Harpers Ferry, West Virginia, for gage heights of 21.0 to 37.0 feet. Download a print-quality PDF image

Table 7. Approximate stage-discharge relation and corresponding recurrence intervals for confluence
of the Potomac and Shenandoah Rivers at
Harpers Ferry, West Virginia
[ft, feet; ft³/s, cubic feet per second; +, slightly greater than;
- slightly less than]

Gage height (ft)	Discharge (ft³/s)	Approximate recurrence interval (years)
21.0	170,000	5+
22.0	184,000	7
23.0	198,000	8+
24.0	214,000	10
25.0	228,000	11+
26.0	245,000	13
27.0	262,000	17
28.0	281,000	22
29.0	298,000	25+
30.0	320,000	30+
31.0	338,000	35+
32.0	357,000	50
33.0	376,000	60-
34.0	398,000	70-
35.0	425,000	95-
36.0	450,000	120-
37.0	475,000	145

Figure 3 and table 7 can be used to determine recurrence intervals for floods at Harpers Ferry other than those listed in table 6. For example, the peak gage height for the flood of June 1, 1889, is recorded as 34.8 ft at Harpers Ferry. From figure 3, a gage height of 34.8 ft corresponds to a discharge of approximately 420,000 ft³/s. From table 7, a gage height of 34.8 ft and a discharge of 420,000 ft³/s indicates a recurrence interval of slightly less than 95 years. A recurrence interval of 90 years would be a reasonable estimate for this example.

Relation Between Peak Water-Surface Elevation at Harpers Ferry, West Virginia, and Surrounding Streamflow-Gaging Stations

Prediction of peak water-surface elevations at Harpers Ferry is complicated because large floods usually include drainage from both the Shenandoah and Potomac Rivers. The drainage areas of both rivers are large and drain geographically separate regions. As a result, either river may contribute individually to flooding, or the combination of flows may be the source of flooding. If predictions of expected inundation during floods at Harpers Ferry can be enhanced, Park personnel may be able to more effectively protect Park resources and prevent loss of life.

A statistical analysis of the 12 largest peaks at Harpers Ferry was performed to determine the relation between recorded peak water-surface elevations at Harpers Ferry and at the three surrounding USGS streamflow-gaging stations. Simple linear regression was used to develop three equations to predict the peak water-surface elevation at Harpers Ferry on the basis of data from the individual surrounding streamflow-gaging stations. These equations show the relation between peak water-surface elevations at (1) Harpers Ferry based on a known peak water-surface elevation at Shenandoah River at Millville, W. Va., (2) Harpers Ferry based on a known peak water-surface elevation at Potomac River at Shepherdstown, W. Va., and (3) Harpers Ferry based on a known peak water-surface elevation at Potomac River at Point of Rocks, Md. A fourth equation, developed by multiple regression techniques (Riggs, 1968), shows the relation between peak water-surface elevation at Harpers Ferry based on known peak water-surface elevations at both the Shenandoah River at Millville, W. Va. and the Potomac River at Shepherdstown, W. Va.

Relation Between Elevations at Harpers Ferry and Shepherdstown, West Virginia

Simple linear regression was used to develop an equation for predicting the peak water-surface elevation at Harpers Ferry based on a known peak water-surface elevation at station number 01618000, Potomac River at Shepherdstown, W. Va. The equation was developed within a gage-height range of 22.19 ft to 42.10 ft for Shepherdstown, and 21.50 ft to 36.50 ft for Harpers Ferry. The equation for this relation is as follows:

WSHF = 0.8590 WSSH + 6.08      (2)

where:

WSHF = Water-surface elevation at Harpers Ferry, in ft above mean sea level

WSSH = Water-surface elevation at station number 01618000, Potomac River at Shepherdstown, W. Va., in ft above mean sea level.

R², Coefficient of Total Variability = 0.82

S, Standard Error of Regression = 2.15 ft

Equation 2 was used to predict water-surface elevations for the 12 historical flood peaks at Harpers Ferry. The predicted water-surface elevations were compared to the actual recorded peaks to determine the difference. The results are presented in table 8.

Table 8. Predicted and actual peak elevations above mean sea level at Harpers Ferry, West Virginia, using a predictive equation based on known water-surface elevations at the streamflow-gaging station on Potomac River at Shepherdstown, West Virginia
[ft, feet]

Date of flood event	Shepherdstown, peak elevation above mean sea level (ft)	Harpers Ferry, predicted peak elevation above mean sea level (ft)	Harpers Ferry, actual peak elevation above mean sea level (ft)	Difference from actual elevation (ft)
March 19, 1936	323.10	283.63	282.03	+1.60
October 16, 1942	313.68	275.53	279.33	-3.80
November 6, 1985	312.44	274.47	275.33	-0.86
September 8, 1996	310.00	272.37	275.33	-2.96
June 23, 1972	312.44	274.47	275.23	-0.76
January 20, 1996	313.50	275.38	274.73	+0.65
April 27, 1937	314.20	275.98	274.53	+1.45
May 13, 1924	310.80	273.06	273.13	-0.07
August 20, 1955	306.26	269.16	269.43	-0.27
October 29, 1937	307.78	270.47	267.03	+3.44
March 21, 1975	303.19	266.52	267.03	-0.51
October 10, 1976	306.31	269.20	267.03	+2.17

The data in table 8 show a range of +3.44 ft to -3.80 ft in predicting the actual peak elevation at Harpers Ferry for these 12 flood events. The actual peak elevations for the October 1937, March 1975, and October 1976 floods indicate a 4.59-ft difference in elevation at Shepherdstown with corresponding peak elevations at Harpers Ferry that are the same. These results indicate that prediction of peak elevation at Harpers Ferry on the basis of upstream conditions should also consider corresponding peak elevations at the streamflow-gaging station on Shenandoah River at Millville, W. Va.

Relation Between Elevations at Harpers Ferry and Millville, West Virginia

Simple linear regression was used to develop an equation for predicting the peak water-surface elevation at Harpers Ferry based on a known peak water-surface elevation at station number 01636500, Shenandoah River at Millville, W. Va. The equation was developed within a gage-height range of 12.72 ft to 32.40 ft for Millville, and 21.50 ft to 36.50 ft for Harpers Ferry. The equation for this relation is as follows:

WSHF = 0.7717 WSMIL + 30.11      (3)

where:

WSHF = Water-surface elevation at Harpers Ferry, in ft above mean sea level; and

WSMIL = Water-surface elevation at station number 01636500, Shenandoah River at Millville, W. Va., in ft above mean sea level.

R², Coefficient of Total Variability = 0.71

S, Standard Error of Regression = 2.76 ft

Equation 3 was used to predict water-surface elevations for the 12 historical flood peaks at Harpers Ferry. The predicted water-surface elevations were compared to the actual recorded peaks to determine the difference. The results are presented in table 9.

Table 9. Predicted and actual peak elevations above mean sea level at Harpers Ferry, West Virginia, using a predictive equation based on known water-surface elevations at the streamflow-gaging station on Shenandoah River
at Millville, West Virginia
[ft, feet]

Date of flood event	Millville, peak elevation above mean sea level (ft)	Harpers Ferry, predicted peak elevation above mean sea level (ft)	Harpers Ferry, actual peak elevation above mean sea level (ft)	Difference from actual elevation (ft)
March 19, 1936	319.36	276.56	282.03	-5.47
October 16, 1942	325.40	281.22	279.33	+1.89
November 6, 1985	318.60	275.98	275.33	+0.65
September 8, 1996	319.82	276.92	275.33	+1.59
June 23, 1972	314.89	273.11	275.23	-2.12
January 20, 1996	316.61	274.44	274.73	-0.29
April 27, 1937	313.20	271.81	274.53	-2.72
May 13, 1924	314.10	272.50	273.13	-0.63
August 20, 1955	314.45	272.77	269.43	+3.34
October 29, 1937	305.72	266.04	267.03	-0.99
March 21, 1975	311.86	270.78	267.03	+3.75
October 10, 1976	308.30	268.03	267.03	+1.00

The data in table 9 show a range of +3.75 ft to -5.47 ft in predicting the actual peak elevation at Harpers Ferry for these 12 flood events. This range is nearly 2 ft greater than that of the predictions based on the Potomac River at Shepherdstown, W. Va. The actual peak elevations for the October 1937, March 1975, and October 1976 floods indicate a 6.14-ft difference in elevation at Millville with corresponding peak elevations at Harpers Ferry that are exactly the same. These results further indicate that prediction of peak elevation at Harpers Ferry on the basis of upstream conditions should consider both peak elevations at the streamflow-gaging stations on Shenandoah River at Millville, W. Va., and the Potomac River at Shepherdstown, W. Va.

Relation Between Elevations at Harpers Ferry, West Virginia, and Point of Rocks, Maryland

Simple linear regression was also used to develop an equation for predicting the peak water-surface elevation at Harpers Ferry based on a known peak water-surface elevation at station number 01638500, Potomac River at Point of Rocks, Md. The equation was developed within a gage-height range of 24.93 ft to 41.03 ft for Point of Rocks, and 21.50 ft to 36.50 ft for Harpers Ferry. The equation for this relation is as follows:

WSHF = 0.8638 WSPOR + 71.13      (4)

where:

WSHF = Water-surface elevation at Harpers Ferry, in ft above mean sea level; and

WSPOR = Water-surface elevation at station number 01638500, Potomac River at Point of Rocks, Md., in ft above mean sea level.

R², Coefficient of Total Variability = 0.95

S, Standard Error of Regression = 1.12 ft

Equation 4 was used to predict water-surface elevations for the 12 historical flood peaks at Harpers Ferry. The predicted water-surface elevations were compared to the actual recorded peaks to determine the difference. The results are presented in table 10.

Table 10. Predicted and actual peak elevations above mean sea level at Harpers Ferry, West Virginia, using a predictive equation based on known water-surface elevations at the streamflow-gaging station on Potomac River at Point of Rocks, Maryland
[ft, feet]

Date of flood event	Point of Rocks, peak elevation above mean sea level (ft)	Harpers Ferry, predicted peak elevation above mean sea level (ft)	Harpers Ferry, actual peak elevation above mean sea level (ft)	Difference from actual elevation (ft)
March 19, 1936	241.66	279.88	282.03	-2.15
October 16, 1942	241.06	279.36	279.33	+0.03
November 6, 1985	236.91	275.77	275.33	+0.44
September 8, 1996	236.95	275.81	275.33	+0.48
June 23, 1972	238.06	276.77	275.23	+1.54
January 20, 1996	237.17	276.00	274.73	+1.27
April 27, 1937	234.49	273.68	274.53	-0.85
May 13, 1924	232.83	272.25	273.13	-0.88
August 20, 1955	229.71	269.55	269.43	+0.12
October 29, 1937	225.56	265.97	267.03	-1.06
March 21, 1975	226.78	267.02	267.03	-0.01
October 10, 1976	227.88	267.97	267.03	+0.94

The data in table 10 show a range of +1.54 ft to -2.15 ft in predicting the actual peak elevation at Harpers Ferry for these 12 flood events. This range of prediction error is significantly less than that from the equations based on data for the Potomac River at Shepherdstown, W. Va. or for the Shenandoah River at Millville, W. Va. The coefficient of total variability (R²) and the standard error of regression (S) for equation 4 also indicate a much stronger relation than the predictions that would be made by equations 2 or 3. However, because the Point of Rocks station is located downstream of Harpers Ferry, it might be of limited use in predicting peak elevations at Harpers Ferry during a flood.

Relation Between Elevations at Harpers Ferry, Shepherdstown, and Millville, West Virginia

To increase the accuracy of prediction using the stations upstream from Harpers Ferry, multiple regression techniques were used to develop an equation (Riggs, 1968). Data used in developing the relation were the peak water-surface elevations for the 12 flood events listed previously in table 3. The equation shows the relation between peak water-surface elevation at Harpers Ferry based on known peak water-surface elevations at both the Shenandoah River at Millville, W. Va. and the Potomac River at Shepherdstown, W. Va. The range of gage heights used in developing this equation are the same as those used in developing equations 2 and 3. The equation for this relation is as follows:

WSHF = 0.5562 WSSH + 0.4053 WSMIL - 27.45      (5)

where:

WSHF = Water-surface elevation at Harpers Ferry, in ft above mean sea level;

WSSH = Water-surface elevation at station number 01618000, Potomac River at Shepherdstown, W. Va., in ft above mean sea level; and

WSMIL = Water-surface elevation at station number 01636500, Shenandoah River at Millville, W. Va., in ft above mean sea level.

R², Coefficient of Total Variability = 0.98

S, Standard Error of Regression = 0.98 ft.

Equation 5 was used to predict water-surface elevations for the 12 historical flood peaks at Harpers Ferry. The predicted water-surface elevations were compared to the actual recorded peaks to determine the difference. The results are presented in table 11.

Table 11. Predicted and actual peak elevations above mean sea level at Harpers Ferry, West Virginia, using a predictive equation based on known water-surface elevations at streamflow-gaging stations on the Potomac River at Shepherdstown, West Virginia, and the Shenandoah River at Millville, West Virginia
[ft, feet]

Date of flood event	Shepherdstown, peak elevation above mean sea level (ft)	Millville, peak elevation above mean sea level (ft)	Harpers Ferry, predicted peak elevation above mean sea level (ft)	Harpers Ferry, actual peak elevation above mean sea level (ft)	Difference from actual elevation (ft)
March 19, 1936	323.10	319.36	281.69	282.03	-0.34
October 16, 1942	313.68	325.40	278.90	279.33	-0.43
November 6, 1985	312.44	318.60	275.45	275.33	+0.12
September 8, 1996	310.00	319.82	274.59	275.33	-0.74
June 23, 1972	312.44	314.89	273.95	275.23	-1.28
January 20, 1996	313.50	316.61	275.24	274.73	+0.51
April 27, 1937	314.20	313.20	274.25	274.53	-0.28
May 13, 1924	310.80	314.10	272.72	273.13	-0.41
August 20, 1955	306.26	314.45	270.34	269.43	+0.91
October 29, 1937	307.78	305.72	267.64	267.03	+0.61
March 21, 1975	303.19	311.86	267.58	267.03	+0.55
October 10, 1976	306.31	308.30	267.87	267.03	+0.84

The data in table 11 show a range of +0.91 ft to -1.28 ft in predicting the actual peak elevation at Harpers Ferry for these 12 flood events. This range of 2.19 ft in prediction error is significantly less than that of equation 2 for the Potomac River at Shepherdstown, W. Va. (7.24 ft) and equation 3 for the Shenandoah River at Millville, W. Va. (9.22 ft). This range in prediction error is also less than that from equation 4 for the Potomac River at Point of Rocks, Md. (3.69 ft). The coefficient of total variability (R²) and standard error of regression (S) for equation 5 also show a slightly stronger relation than they do for equation 4.

Although the most significant variation in water-surface elevation obtained from equation 5 was -1.28 ft for the June 1972 flood, a search of historical data indicates that the National Weather Service (NWS) uses a peak stage at Harpers Ferry for this flood event that is 2 ft lower than the value used by NPS (National Weather Service, Baltimore/Washington Forecast Office, information accessed May 24, 2000, on the World Wide Web at http://tgsv5.nws.noaa.gov/er/lwx/rivers/hfew.htm). NWS uses a peak gage height of 27.7 ft (273.23 ft above mean sea level) for the June 1972 flood, while NPS uses a peak gage height of 29.7 ft (275.23 ft above mean sea level). The NPS value was used in the development of the equations presented here because the discharges from the surrounding stations seem to support a discharge at Harpers Ferry that is more consistent with the NPS value. Equation 5, however, predicted a peak flood elevation above mean sea level (273.95 ft) that falls between the NPS and NWS peak elevations. The predicted elevation from equation 5 is slightly closer to the NWS elevation. When the NWS elevation is compared to the elevation predicted by use of equation 5, there is a difference of +0.72 ft.

Limitations and Implications for Future Study

The recurrence intervals presented in this report are the average number of years between floods equal to or greater than the corresponding flood discharges based on the available data. It should be noted that this is an average number of years, and does not imply that it will be the same number of years before another event of the same magnitude occurs.

Flood probabilities and recurrence intervals for all streamflow-gaging stations should be updated often. Inclusion of the most current data into the flood-frequency analyses will ensure greater accuracy in estimates of flood probabilities.

Estimates of peak discharge at Harpers Ferry could be enhanced with a more accurate accounting of the time difference between peaks at the four gaging stations included in this analysis. Some peak times at the four stations were unavailable for this investigation. As a result, time differences between peaks could not be considered as part of the data set for developing the equations. A review of old newspaper headlines or other historical records might be helpful in obtaining additional data. The effects of timing differences among peaks might also be analyzed by use of a computer model.

The equations developed for this study were based on data for 12 historical floods. A more extensive statistical analysis that includes more flood events at the three gaging stations could enhance the accuracy of these equations. Additionally, the equations developed for this study are recommended for use only within the range of respective water-surface elevations that were used in their development.

Use of the equations for predicting expected innundation during floods at Harpers Ferry would most likely be limited to the stations upstream of Harpers Ferry. Although it is possible that a predicted peak elevation for Point of Rocks could become available under certain circumstances prior to the peak occurring at Harpers Ferry, the peak at Harpers Ferry most likely would occur prior to a predicted peak for Point of Rocks becoming available.

The differences in peak gage height reported by NPS and NWS during the June 1972 flood indicate a possible need for further analysis. A detailed review of any high-water marks, other historical data, reports, photographs, or newspaper accounts could be helpful in further calibration of the peak gage height for this event. The peak gage height might also be further calibrated by use of a computer model.

Summary and Conclusions

The Harpers Ferry National Historical Park, West Virginia, has historically been affected by flooding from both the Potomac and Shenandoah Rivers. Flooding and resulting damage to the Park have emphasized a need for the National Park Service to begin (1) correlating Park-related flood damage with specific flood events, (2) predicting expected inundation and subsequent damage based on the magnitude and frequency of flood events, and (3) prioritizing Park structures for protection from future flood damage.

This report presents updated flood-frequency distributions for three U.S. Geological Survey streamflow-gaging stations upstream and downstream of the Park, estimates of flood frequency compared to water-surface elevation at Harpers Ferry for the 12 largest floods for which corresponding water-surface elevations are available, and equations that show the relation between peak water-surface elevations at Harpers Ferry and the surrounding gaged locations.

Flood-frequency analyses for the streamflow-gaging stations at Potomac River at Shepherdstown, West Virginia; Shenandoah River at Millville, West Virginia; and at Potomac River at Point of Rocks, Maryland, were updated to include flood peaks through the 1999 water year. Adjustments in the analyses were made to account for historical flood information and high outliers.

The updated flood-frequency analyses for the streamflow-gaging stations were then used to determine an approximate flood-frequency distribution for Harpers Ferry. Peak discharges at Harpers Ferry for each of the 12 flood events were subsequently estimated on the basis of data from the three stations. Recurrence intervals for these flood events were then determined by use of the estimated flood-frequency distribution for Harpers Ferry. The peak gage heights and estimated peak discharges were plotted and used to determine an approximate stage-discharge relation for Harpers Ferry. This stage-discharge relation is presented along with corresponding recurrence intervals at increments of 1 foot in gage height.

This report also presents preliminary statistical relations between recorded peak water-surface elevations at Harpers Ferry and the three U.S. Geological Survey streamflow-gaging stations. Three equations were developed to predict the peak water-surface elevation at Harpers Ferry based on data from the individual stations. A fourth equation was developed to show the relation between peak water-surface elevation at Harpers Ferry based on known peak water-surface elevations at both the Shenandoah River at Millville, West Virginia and at the Potomac River at Shepherdstown, West Virginia. Analysis of the results indicates that the equation utilizing peak water-surface elevations from both Millville and Shepherdstown had the strongest statistical relation and the smallest range of prediction error, from +0.91 feet to -1.28 feet.

References Cited

Carpenter, D.H., 1980

Technique for estimating magnitude and frequency of floods in Maryland: U.S. Geological Survey Water-Resources Investigations Open-File Report 80-1016, 79 p.

Dillow, J.J.A., 1996

Technique for estimating magnitude and frequency of peak flows in Maryland: U.S. Geological Survey Water-Resources Investigations Report 95-4154, 55 p.

Doheny, E.J., 1997

Flood-hydrology data for the Potomac River and selected tributaries in the vicinity of the Chesapeake and Ohio Canal National Historical Park, Maryland, West Virginia, and the District of Columbia: U.S. Geological Survey Open-File Report 97-200, 33 p.

Interagency Advisory Committee on Water Data, 1982

Guidelines for determining flood flow frequency: Water Resources Council Bulletin 17B, 28 p.

National Weather Service, Baltimore/Washington Forecast Office

Station description for the Potomac River at Harpers Ferry, West Virginia: Information accessed May 24, 2000 on the World Wide Web at http://tgsv5.nws.noaa.gov/er/lwx/rivers/hfew.htm.

Riggs, H.C., 1968

Some statistical tools in hydrology: U.S. Geological Survey Techniques of Water-Resources Investigations, book 4, chap. A1, 39 p.

APPENDIX

Appendix 1. Summary of flood-frequency data for Station 01620000, Potomac River at Harpers Ferry, West Virginia
[ft³/s, cubic feet per second; ft, feet]

Recurrence Interval (years)	Discharge for Station 01620000, Potomac River at Harpers Ferry, W. Va. (ft³/s)	Approximate gage height (ft)	Approximate mean sea level elevation (ft)
2	102,000	15.91	261.43
5	162,000	20.41	265.93
10	212,000	23.8	269.33
25	293,000	28.7	274.23
50	360,000	32.1	277.63
100	435,000	35.6	281.13
500	650,000	43.02	288.53

1. Estimated gage height based on straight-line extension of rating curve below discharge of 178,500 ft³/s and gage height of 21.5 ft.
2. Estimated gage height based on straight-line extension of rating curve above discharge of 475,000 ft³/s and gage height of 37.0 ft.

Appendix 1. Summary of flood-frequency data for Station 01620000, Potomac River at Harpers Ferry, West Virginia -- continued. Download a print-quality PDF image

For further information contact:

District Chief
U.S. Geological Survey
5522 Research Park Drive
Baltimore, MD 21228

or visit the Maryland-Delaware-D.C.
District Home Page on the World Wide Web at http://md.water.usgs.gov/

Citation:

Doheny, E.J. and Fisher, G.T. , 2000, Estimated flood frequency and corresponding water-surface elevations at the confluence of the Potomac and Shenandoah Rivers, Harpers Ferry, West Virginia: U.S. Geological Survey Open-File Report 00-461, http://md.water.usgs.gov/publications/0fr-00-461/