excessive precipitation records used in constructing the isopluvial charts. As stated in chapter IV, the excessive precipitation sheets were made out only for rainfall stations having complete records covering not less than 5 consecutive years, and did not include records subsequent to 1914. Therefore, the data shown in figures 38 to 43 does not, in general, include intense rainfalls at stations having less than 5 years of records, nor those which occurred since December 31, 1914. Occasional exceptions were made as in the case of the great storm of Aug. 17-21, 1915, over Texas and Arkansas, which is responsible for the record of 18.6 inches in 24 hours in quadrangle 14-I, figure 39, and for the values in this quadrangle and some neighboring quadrangles on the succeeding charts. Unfortunately, the charts had been completed at the time of occurrence of the great July 1916 storm in North Carolina, and its record breaking precipitation does not, therefore, appear on them. The records used for platting the storm maps were also consulted, and as these included all station records, regardless of length of period covered, it follows that an occasional intense rainfall figure for a short station record has been used. No attempt was made to draw isopluvial lines on the charts of most intense rainfall, because of the nature of the data. A careful consideration of the erratic nature of precipitation and the many difficulties which have attended the observing, recording and publishing of rainfall data, as outlined else where in this report, leads to the conclusion that it will require many additional years of rainfall recording before even an approximation to dependable regularity will be discernible in maximum precipitation data when arranged on charts as in figures 38 to 43. It was largely because of this difficulty that the necessity arose for providing the isopluvial charts, which furnish information less liable to subsequent variation. It is felt, nevertheless, that the charts showing intense rainfall furnish information of considerable value to the practising engineer when interpreted with due regard to the peculiar nature of the data, and when considered in conjunction with the isopluvial index data given in figures 13 to 36. The values given should be looked upon as indicative rather than finite; for they represent essentially temporary maxima, likely to be exceeded in the future. Their applicability is, of course, restricted to small areas. PRACTICAL APPLICATION OF EXCESSIVE PRECIPITATION DATA The isopluvial charts furnish a convenient means of determining the pluvial index for any locality within the eastern United States, 90 100 Average Number of Years Between Periods of Excessive Precipitation QUADRANGLE 3-D. The depth shown corresponding to any frequency period is that which will probably be equaled or exceeded once during that period. Average Number of Years Between Periods of Excessive Precipitation FIG. 45.-FREQUENCY OF EXCESSIVE PRECIPITATION IN QUADRANGLE 9-E. The depth shown corresponding to any frequency period is that which will probably be equaled or exceeded once during that period. Average Number of Years Between Periods of Excessive Precipitation QUADRANGLE 12-J. The depth shown corresponding to any frequency period is that which will probably be equaled or exceeded once during that period. Average Number of Years Between Periods of Excessive Precipitation FIG. 47.—FREQUENCY OF EXCESSIVE PRECIPITATION IN QUADRANGLE 15-E. The depth shown corresponding to any frequency period is that which will probably be equaled or exceeded once during that period. by merely interpolating between the isopluvial lines. In doing this it is necessary to bear in mind the limitations of the charts as above set forth. Detached mountain spurs and other prominent local topographic features, which though important in themselves yet are so small as to be recognized with difficulty on charts of the size here published, had to be ignored in drawing the isopluvial lines, and allowance should be made for them accordingly. In applying the pluvial indices to a particular area, it is perhaps desirable first to construct frequency curves, similar to the four shown in figures 44 to 47, for several of the surrounding quadrangles. The method of constructing these curves from the pluvial indices is readily seen. The six curves shown for quadrangle 3-D, figure 44, correspond to the 1 to 6-day maximum periods of precipitation. The curve for the 1-day period is found by platting, to the proper frequency intervals, the pluvial indices 4.2, 4.8, 5.5, and 6.5, taken from the charts for 15, 25, 50, and 100-year frequencies, respectively. The 2 to 6-day curves are found in a similar way. This enables the investigator to obtain a clear idea of the additional depths which may be expected, above that for the maximum day, for consecutive periods of 2 to 6 days. These curves also show clearly the manner in which the depths increase with the average period of time between occurrences. The four quadrangles for which curves are shown in figures 44 to 47, were selected to show the varying characteristics in different sections of the country. The curves for quadrangle 3-D, figure 44, indicate that in this quadrangle from 65 to 75 per cent of the total rainfall occurs on the maximum day in a 1 to 6-day period of excessive precipitation. In quadrangle 9-E, figure 45, these percentages are reduced to between 50 and 60. Similar percentages are determinable for the other two sets of curves. The curves for quadrangle 12-J, figure 46, indicate the dominance of the storms of long duration to which the Gulf coast is subject. A significant feature of all the curves is their near approach to a horizontal position at a frequency of 100 years, indicating that the maximum depths of rainfall already recorded will probably not be greatly exceeded in the future. The isopluvial charts may be used to advantage in the preparation of frequency curves for the design of sewer systems, bridge and culvert openings for small drainage basins, and in the design of dams, levees, and channel improvements where the watersheds involved are not more than a few square miles in area. Since each pluvial index represents the excessive precipitation at only one station, it will be seen that the rainfall values are applicable to only small areas. For the larger areas the reader is referred to the results of the time-area-depth investigations given in chapter VIII. CHAPTER VI.-SELECTING AND SIZING THE 160 GREAT STORMS The discussion of rainfall statistics in the last two chapters was limited to a consideration of excessive precipitation records at individual stations. This involved the two important rainfall factors of depth and duration, but ignored a third factor, area, which is equally important. The next logical step in the investigation, therefore, is to study a number of large storms as a whole, giving consideration to all three of the factors, time, area, and depth, which determine the size of storms. This chapter is devoted to a description of the methods used in selecting and determining the relative sizes of 160 great storms. These will be used in subsequent chapters in a discussion of seasonal and geographical distribution, frequency, and cyclic variation of storms. Several of the largest and most important of these 160 storms are also later studied and discussed in much greater detail as to their time-area-depth relations, by means of maps and curves. PERIOD COVERED Before proceeding with the actual selection of storms it was necessary to determine what period of years the investigation should cover. It is, of course, highly desirable to have this period as long as possible, and still be sure that the rainfall records during the entire time are sufficiently numerous and well distributed to warrant deductions as to the size, frequency, seasonal distribution, and cyclic variation of the storms which occurred. Prior to 1843 the records are so few and scattered as to be of negligible value for the objects here in view. From 1843 to 1872 there were still very few rainfall gaging stations, and consequently the chance was remote that the center of a storm area would occur near one of these. It was still more unlikely that any but the greatest storms would cause unusual rainfall records at two or more of these widely separated stations. Only the greatest storm of this period, that of October 3-4, 1869, over Connecticut, was therefore selected for further study and comparison with the greatest storms of recent years in the same region. Although for the next nineteen years, 1873-1891, the number of rainfall gaging stations was considerably greater, the same handicap |