whose rainfall is probably smaller than the average for the entire area above Mechanicville, probably averaging about 39 or 40 inches for the long term mean, and probably falling in extreme years as low as 28 or 29 inches, as shown by a study of the short-term records for Greenwich, Easton, Cambridge, Hoosic Falls, Eagle Mills, Lebanon Springs and a study of the isohyetal map. The rainfall at Williamstown, Mass., a fairly representative locality for this drainage area, from 1852-1906, averaged 39.4 inches. The drainage area is in general rugged and steep, tending to rapid discharge. The records of stream flow gagings may be found in reports of New York State Engineer for 1905, page 573, with hydrographs, complete records of daily discharge from the beginning to December 31, 1905, inclusive: U. S. G. S. Water Supply Paper No. 202, page 23. The gaging station was established September 24, 1903. The height is read twice daily, and estimates of flow made from a rating curve established by occasional current meter measurements. In winter, flow is more or less retarded by ice, and although the height-gage may be read daily as usual for much of the time, the estimate rests largely upon assumptions about the degree of obstruction caused by the ice. There are omissions in the published record for February 1906, which have been supplied by interpolation. From December 4, 1906, to March 15, 1907, the daily estimates have also been omitted because of ice. It is understood that some careful studies of the flow when obstructed by ice have been made, corresponding to certain readings of height-gage, and it is expected that additional precise measurements of flow under the ice will be made during the present winter. As a whole, these estimates of the daily and monthly discharge of the Hoosic river at Buskirk appear fairly reliable. The mass curve has therefore been computed and plotted for comparison with the diagrams of Plate 3. IRREGULARITIES DUE TO SPIER'S FALLS POWER PLANT. The development and use of the power station at Spier's Falls has introduced serious disturbances in the flow of the river from hour to hour, according to the varying demand for electrical power at different hours of the day. Not only has this disturbed the operation of the paper mills at Glens Falls, and Fort Edward, but it has possibly interfered somewhat with the accuracy of the estimates of stream flow, because of the height over the dam being measured only twice in each twenty-four hours, and because of the more frequent changes caused in the operation of the turbines at times when no flow is passing over the dam. The storage in the ponded river between Spier's Falls and the feeder dam tends to smooth out this irregularity, and the storage in the smaller pond above the dams at Glens Falls and Sandy Hill tend to make the flow somewhat more regular at Fort Edward, and by the time Mechanicville is reached, this hourly irregularity has doubtless in large measure disappeared, so that error from this cause should not be serious at Mechanicville. MASS CURVE COMPUTATIONS. The best method known to the writer for computing the yield of a watershed from a continuous series of gagings and for at the same time determining the volume of reservoir storage required to store the flood waters for use in season of drought so as to maintain a specified constant rate of flow, is that known as the mass curve method.* 66 Briefly described, the method consists in adding up the totals of the daily or monthly yield from month to month for the whole period of gagings under consideration, then plotting the successive steps of accretion of the mass as an irregular line or mass curve." Any desired rate of draft may then be assumed, and its successive sums plotted to the same scale, and if a uniform rate, this draft curve forms a straight inclined line, and if it is made to start co-incident with some point or summit on the mass curve," the divergence of the two curves at successive points serves to show the volume of storage that would have been required on this date to have maintained the required rate of flow up to that time. 66 Since our river flow problems start with rainfall data given in inches depth over the entire watershed, this form of unit of measurement is commonly most convenient in studying the problems of reservoir depletion. Mass curves for the Hudson river records at Mechanicville and Fort Edward and for the record of flow of its tributaries, the Schroon and the Hoosic, covering the whole period of observations, are presented in diagrams on plate No. 3. A mass curve for the flow of the Croton is also shown during the period of smallest rainfall and greatest storage need that has occurred during the nineteen years covered by the Mechanicville gagings, because of the fact that the Croton records, maintained by the New York Department of Water Supply, present the most accurate and longest record of gagings of stream flow available within the State of New York and go back to a series of years of low rainfall more severe than has occurred since gagings on the Hudson were begun. A study of the Hudson-Mechanicville mass curve makes it plain that on the Mechanicville basis, without addition for the relatively larger rainfall of the Sacandaga, with the great volume of storage proposed in the Sacandaga reservoir, which permits carrying over the surplus of years of large rainfall for use in years of small rainfall, a run-off of 1,700 cubic feet per second equivalent to 22 inches in depth from over the entire watershed, could have been maintained throughout every one of the twenty years since the beginning of the gagings at Mechanicville, and that the maximum depletion of storage caused at any time by this rate of draft would have amounted to 10.4 inches on the Sacandaga watershed of 1,050 square miles, equivalent to twenty five billion cubic feet depletion of the proposed Sacandaga reser*NOTE. This method is explained in detail in the report on New York's Water Supply made to Bird S. Coler, Comptroller, by John R. Freeman, 1900, page 227. voir, corresponding to the volume contained between elevation 763 and elevation 737.* ALLOWANCE FOR EVAPORATION FROM THE RESERVOIR SURFACE. The best data on evaporation yet available are those determined from the experiments of Mr. Desmond Fitzgerald, made on the Boston water works and published in 1889; but owing to the greater altitude and greater coolness of the Adirondack region, it is probable that the loss from the water surface of the proposed Sacandaga reservoir would be somewhat less than found at the Chestnut Hill reservoir near Boston, where Mr. Fitzgerald's average results gave 39 inches annually for the mean value. Mr. H. K. Barrows, Hydrographer, U. S. Geological Survey, has for three years past conducted experiments upon evaporation at several of the Maine lakes, and finds for the mean annual value, only 29 inches. Taking the mean of this and the Fitzgerald determination, we have 34 inches, and I believe this is about what may be expected from the Sacandaga reservoir, but in estimating the yield we have included the reservoir surface as a part of the drainage area, and therefore we do not have to deduct from the yield so large a quantity as the equivalent of 34 inches depth from the entire surface of the proposed Sacandaga reservoir. The present problem is to determine the loss of water due to substituting a water area, varying from 42 to 26 square miles according to fullness of reservoir, for an equal land area from which there is at present a considerable loss by the evaporation of the rain that falls upon it, and which loss as found by deducting the run-off of 24 inches from the rainfall of 45 inches, averages 21 inches. The total yearly loss by increased evaporation, due to flooding this land, therefore is 34 inches evaporation from a water surface minus 21 inches evaporation from a land surface, which gives 13 inches of increased loss over this area. The area of water surface exposed to evaporation will lessen as the reservoir is lowered. It we call the average area of water surface 37 square miles, the loss of 13 inches depth over this area, due to substituting a water surface for a land surface is equivalent to about 36.0 cubic feet per second of constant flow daily throughout the year, which should be deducted from the yield as computed from the mass curves. In the method of computation adopted above, we have included the area of the reservoir itself as a part of the drainage area and have made due allowance for this water surface being less efficient than the land surface as a gathering ground, because of its greater evaporation. Taking the mean *NOTE. The area of the Sacandaga watershed above its mouth at Hadley is given in the publications of the New York State Engineer and the United States Geological Survey as 1,040 square miles. The topographic maps of the State, made in co-operation with the United States Geological Survey, were not all available at the time of these earlier estimated, and indeed one sheet (Stony Creek) is lacking at the present time. From carefully tracing out the watershed line on all of these topographic maps and planimetering it and taking the Sacandaga area for the missing Stony Creek sheet from the best mans available, we find the area above Conklingville dam site is 1050 square miles. There is, however, some doubt as to where the dividing line should be drawn in the vicin ity of Peck Lake, some five miles northerly from Gloversville, which according to the topcgraphic maps has two outlets. This should be investigated and the drainage area revised whenever the Stony Creek sheet of the topographic survey is available. |