Water Supply Lessons from Droughts and Floods
The Influence of Rainfall on Run-Off—Year Divided Into Four Periods—Snow Measurement Results—Types of Floods Considered
THE influence of flood conditions and droughts on the run-off of streams and watersheds and the practical effects on water supplies are treated in an interesting and practical manner in the following paper:
Southern New England, at least, recently has passed through a period of drought (1923) so severe that many water supply systems have been seriously affected. This district also has experienced within a year two periods of excessive watershed runoff following each other consecutively. One of these was the greatest of record at least during the past half a century. During the past ten years the hydrological cycle has passed from crest to hollow and exhibited features which are of more than passing interest to water supply workers.
For the calendar year 1920 the rainfall on the Nepaug River Watershed of the Hartford, Conn., water supply system was 51.78 inches, or 16 per cent, above the station average of eleven years, with a runoff of 33.92 inches which was nearly 32 per cent, above the average of that area.
Influence of Rainfall on Run-off
The calendar year 1922, with a rainfall of 43.87 inches, although only less than the station average by 2 per cent., had an average runoff that was the lowest in eleven years, being only 20.37 inches, or 16 per cent, below the station average. In 1921 the rainfall was less than that of 1922 by 3.52 inches (40.35), yet the runoff was 21.29 inches, or 0.92 inches more than in the following year.
The lowest rainfall was that of 1916, which was 37.68 inches this was 6.92 inches or 15.5 per cent below the watershed average. The runoff of this year was 20.99 inches, which is 3.22 inches below the average but is 0.62 inches in excess of the runoff of 1922. which was the lowest of record for the Nepaug area.
Similar conditions appeared in nearby streams. On the Manhan River, the vear 1920 produced the highest runoff experienced in the twenty-eight years of record, while that for 1921 was among the lowest that have been recorded.
It is the purpose of this paper to bring to attention some of the elements that have contributed to these conditions as they have been observed on the watersheds of the Hartford Water Board and others in that vicinity.
Low annual rainfall by no means is a concomitant of local low runoff conditions. For instance, as stated above, during the summer and early fall of 1923 drought conditions were so severe as to put many water departments on short service and in many localities the drought was said to have been nearly, if not quite, a record breaker. Notwithstanding this, both the rainfall and runoff on the Hartford system, for the year, were greater than those of the previous two years.
Drainage Plan of Nepaug Reservoir
The Nepaug Reservoir has a tributary drainage basin of about thirty-two square miles. The drainage plan is divided as follows:
A very marked example of these conditions are found in two small contiguous drainage areas discharging into the Nepaug Reservoir, Clear Brook and Phelps Brook. Clear Brook takes its rise in extensive deposits of gravel and sand, the water bubbling out from the bank over several acres and flowing rapidly to the reservoir. Not over a mile to the North Phelps Brook also discharges its waters. These are gathered for the most part from slightly rolling meadow country containing small swamps and little or no gravel deposits. The average color of the Clear Brook water is 15, with a maximum in 1923 of 21, while that of Phelps Brook is 36 with a maximum of 50. As to sustained flow in periods of draught on Clear Brook the lowest runoff of record was 0.87 cu. ft. per second per square mile with a mean of 1.78 and a maximum of 4.11. On Phelps Brook the lowest recorded flow (weekly average) was 0.03 cu. ft. per second per square mile with an average of 1.74 and maximum of 9.94. The per cent, of collection to rainfall on Clear Brook is 53.5 and on Phelps Brook 52.0 per cent, for the years of record.
Weather Bureau Rain Gauges
Within the limits of the watershed there are four standard Weather Bureau rain gauges and one automatic Friez rain gauge. There are standard concrete weirs with sharp crests and full contracted sides on the Nepaug River, and Phelps and Clear Brooks, all with automatic registers, and these stations are all kept under constant oversight. On the watershed of the East Branch of the Farmington River above the Compensating Reservoir dam there are four rainfall stations and a gauging station with an automatic Bristol gauge which is checked against a nearby staff gauge. Tt is from the records of the instruments on these watersheds that the following data have been compiled, and no effort has been spared to obtain reliable information.
“Low annual rainfall by no means is a concomitant of local low run-off conditions. For instance, during the summer and early fall of 1923 drought conditions were so severe as to put many water departments on short service and in many localities the drought was said to have been nearly, if not quite, a record breaker. Notwithstanding this, both the rainfall and run-off on the Hartford system, for the year, were greater than those of the previous two years.”
(Continued on page 1135)
(Continued from page 1115)
Rainfall and Runoff. Nepaug Reservoir Watershed
This table is presented only for purpose of comparison of the rainfall and runoff in the several years and to show the difference which may occur if one rain gauge record is used or a number from which representative conditions may be obtained. The rainfall data are from the automatic gauge at the Nepaug dam which is located at the bottom of the watershed.
Year Divided into Four Periods
Dividing the years of record into the periods suggested by the late George W. Rafter in Water Supply and Irrigation Paper No. 80, U. S. Geological Survey. “The Relation of Rainfall to Runoff,” the following tabulation presents interesting conditions as to the disposal of precipitation on this watershed. Mr. Rafter gave the following reasons for his divisions, which he is careful to state are more or less arbitrary: The storage period, December to May inclusive, when evaporation and absorption by plants are relatively slight, and a relatively large proportion of the precipitation appears as stream flow; the growing period, June to August inclusive, the period of active vegetation when evaporation and absorption by plants are at a maximum; the replenishing period, September to November inclusive, when with normal rainfall the ground water reservoir tends to refill.
Study of Rainfall and Run-off Records
A study of the records of rainfall and runoff for the years of record on the Nepaug watershed (1912 to 1924) shows that although 1923 was not the minimum yield year, during nineteen summer weeks—June 10 to October 20—it produced the lowest yield for the eleven years of record on the Nepaug watershed, i. e., about 0.35 cubic feet per second per square mile for a period of over four months. With the exception of the 15 and 16 week periods, the year 1923 produced the lowest yields on the Nepaug watershed (1912-1924) for consecutive weekly periods of low flow from the tenth to the twentieth, averaging for these periods flows from about a quarter to a third million gallons per square mile per day (land surface). This is of especial interest when it is recalled that the water loss by evaporation alone during this period averages not far from ten times this amount per square mile of water surface.
The accumulated rainfall was below the station average during the entire period and at the end the growing period for vegetation came on with its demand for water. The total rainfall was helped out by a storm of considerable magnitude in the early part of September which, however, was not sufficient fully to restore the depicted ground storage. The drought of the year 1923, like that of 1922, was due in a measure to depleted ground storage of the previous year, which was somewhat mitigated by the high runoff in March, April and May.
Using Birnie’s* deductions the probable extreme deviation of the mean from the eleven years of record of the Nepaug basin as used here would be about 8 per cent. It is likely that it is much closer than that because of the extremes that have occurred during the period. The outstanding conditions, to be considered in preparing for drought periods during the summer months are, condition of the ground storage at the end of the calendar year, and precipitation during the early months of the year; also continuous precipitation, even if slight, through the summer is much more conducive to normal runoff than are heavy storms after a period of drought.
*“On Mean or Average Rainfall and the Fluctuations to Which it is Subject,” Alex. R. Birnic, Proc. Inst. C. E., Vol. CIX (1892), pp. 89*172.
(Continued on page 1138)
(Continued from page 1135)
Types of Floods to Be Considered
Floods due to storm runoff from New England streams have two principal types: those due to heavy and continuous precipitation during the warm months and those due to occurrence of heavy rainfall in conjunction with a thaw in the winter and spring at a time when the surface of the ground is usuallly sealed with ice, and may be covered with snow.
The flood of October 3 and 4, 1869, was typical of the first kind; those on the Farmington River and neighboring streams in April, 1924, and that on Stony Brook, (Boston), in February, 1886, are representative of the second sort.
The storm and flood of October 2-5, 1869, was apparently a record-breaker not only because of the quantity of rain which fell but also from the wide area over which the storm extended.
The amount of rainfall in this storm and its rate producing the flood was unprecedented in records, nor has it been equalled since. The storm center was near Canton, Conn., about 13 miles northwest of Hartford, and about 2 miles east of Nepaug Reservoir.
Over 150 stations reported this storm, extending from Virginia to Canada and west to about the middle of Pennsylvania. It was therefore a coast storm with two foci, the maximum in Connecticut with 12.35 inches and the second in Eastern Pennsylvania with about 8 inches.
Other Storms and Floods
On December 10, 1878, occurred a storm with runoff that produced a 25.2 gauge height on the Connecticut River, which was 0.7 foot higher than that of October, 1869. While no authentic information is at hand concerning conditions producing this flood, it is very probable that there was an unusual combination of at least considerable depth of snow on the ground together with a warm rain. At any rate, this flood is reported to have been the biggest on the Farmington River above Hartford, carrying away many bridges and threatening several of the power dams on the river. A height of 10.5 feet over the dam is reported at Greenwood’s Mills at New Hartford, and a study of conditions leads to the conclusion that this flood may have produced a runoff at its peak of perhaps 16,000 cu. ft. per sec. or 69.2 cu. ft. per sec. per square mile of drainage area.
On March 1, 1896. according to’ Collins Company records at Collinsville on the Farmington River, there was a flood almost as great as that of 1878. This flood was caused by a hard rain falling on about a foot of snow.
A noteworthy flood of recent years on the Farmington River was on February 13, 1900, and computations from conditions at Greenwood’s dam indicate a maximum runoff of about 11,500 second feet or about 50 cu. ft. per sec. per square mile.
Leaving these historical floods, which are chiefly remembered because of the damage caused by them, it is now proposed to take up the details of several excessive runoff periods in this same locality concerning which reliable data are at hand. It is not proposed to treat them in chronological order, as there seems to be no particular reason for so doing.
In point of magnitude, the flood caused by the storm of April 7, 1924, is by far the most important of recent years and possibly was the greatest of this century, on East Branch of The Farmington River. In comparing it with earlier floods its height at Collinsville was 20 inches below the mark recording high water of the 1878 flood, but whether or not this is due to changes in the overflow works, or other local conditions, it is impossible to state now.
For the storm of April. 1924, the rain fell for about nineteen hours almost without intermission, and at the end of this period about 4.25 inches had been recorded. The ground was frozen, and here and there there were patches of snow, but there was no larger amount of runoff from this cause on the Hartford watersheds. Further north the contrary was the case. P. J. Lucey. engineer of the Holyoke water board stating that the rainfall on the Manhan drainage area was 3.49 inches from beginning of storm to crest of flood and the gTound was covered with from six to eight inches of soft snow. Weather Bureau reports from Amherst, Springfield. Turners Falls and Worcester. Massachusetts, as well as from Bridgeport. Colchester, New Haven. Torrington and Waterbury, Connecticut, all reported no snow on the ground.
Comparison of Nepaug and Manhan Areas
Comparing the rainfall and runoff on the Nepaug and Manhan areas, it appears that the rainfall up to the crest of the flood was nearly an inch less at the Holyoke station than at Nepaug. while the discharge at the latter place was only about one-half that at Holyoke. As the difference in rainfall would hardly account for the great difference in runoff, it would appear probable that the cause might be found in the slope and shapes of the two watersheds as well as the accumulated snow on the ground. The Nepaug watershed is roughly circular or oval, while that of the Manhan River is long and narrow.
The flood, or rather excess runoff in the storm of October 23-24, 1923, was due solely to heavy rainfall. That it was not of great magnitude was due to the condition of the ground and height of ground water taken. The rainfall was about 5 1-3 inches, but less than a quarter of it appeared in the runoff. The storm came after a long period of severe drought which has previously been mentioned, and is well shown on the diagram of accumulated rainfall and runoff for 1923.
Due to the condition of the ground and the avidity with which water was absorbed, only from 15 to 25 per cent, of the precipitation appeared as immediate rainfall after a period of 167 hours, while in the case of the April, 1924, storm, nearly 55 per cent, appeared within 93 hours of the beginning of the storm and about 70 per cent, after 167 hours.
The heaviest rainfall that has been recorded in the eleven years of record on the Nepaug watershed was that of October 24-26, 1913, when an average of about 6.73 inches fell in about 36 hours. 6.7 inches falling between 4:30 p. m. of October 25 and 6 p. m. of October 26. This rainfall produced a flood height of about 52 cu. ft. per sec. per square mile.
Results of Snow Measurements
Snow measurements were made in a number of places, representative of conditions over the different Hartford watersheds on two different dates with the following average results as to water equivalent:
In the report for October, 1913, of the Climatological Service, Dist. No. North Atlantic States, in commenting on heavy rainstorms of 1913 in New York City, July 28, Sept. 4-5, Oct. 1-2 and Oct. 24-25, the statement is made that “as nearly as can be determined from the records available, storms of equal intensity have not occurred in the preceding 45 years (a half century at least at the present time—C.M.S.) but since 1869 storms of nearly equal intensity occurred in 1878, 1902, 1905 and 1913. So it is fair to assume that such storms are liable to occur at least once in each ten years. It should be borne in mind that several such storms may occur in one year, and within those of the last twelve years there have been six such storms.”
The storm of October 1-2, 1913, was comparatively widespread and recorded in the area to the west as follows:
It appears from the foregoing that there are two types of flood producing storms to be considered under New England conditions: first, those of rainfall, usually occurring in the fall of the year and while giving maximum precipitation usually result in less flood flow in the streams due to the depleted condition of the ground storage after the dry summer period, and second, those occurring in the spring when meteorological conditions bring about a combination of warm rains falling on a snow and ice-covered country. The floods resulting from these storms are the most severe, although the actual precipitation may he less than that of the fall storms, because of the unlocking of frozen water collected on the ground during the preceding winter months.
(Excerpts from paper read tiefore the annual convention of the New Emgland Water Works Association.)