CHAPTER 4 Precipitation OBJECTIVES The purpose of this chapter is to: @ Define precipitation, discuss its forms, and describe its spatial and temporal attributes @ Illustrate techniques for estimating areal precipitation amounts for specific storm events and for maximum precipitation-generating conditions. Precipitation replenishes surface water bodies, renews soil moisture for plants, and recharges aquifers. Its principal forms are rain and snow. The relative importance of these forms is determined by the climate of the area under consideration. In many parts of the western United States, the extent of the snowpack is a determining fac- tor relative to the amount of water that will be available for the summer growing season. In more humid localities, the timing and distribution of rainfall are of principal concern. Precipitated water follows the paths shown in Figs. 1.1 and 1.2. Some of it may be intercepted, evaporated, and infiltrated and become surface flow. The actual disposi- tion depends on the amount of rainfall, soil moisture conditions, topography. vegetal cover, soil type, and other factors. Hydrologic modeling and water resources assessments depend upon a knowl- edge of the form and amount of precipitation occurring in a region of concern over a time period of interest. WATER VAPOR The fraction of water vapor in the atmosphere is very small compared to quantities of other gases present, but it is exceedingly important to our way of life. Precipitation is derived from this atmospheric water. The moisture content of the air is also a signifi- cant factor in local evaporation processes. Thus it is necessary for a hydrologist to be 9596 Chapter 4 Precipitation monly used measures of atmospheric moisture or humidity are vapor pressure, absolute humidity, specific humidity, mixing ratio, relative humidity, and dew point xpressed in centimeters or inches, x An equation for computing the amount of Precipitable water in the atmosphere can be derived as follows, Consider a column of air having a square base 1 cm ona side. The total water mass contained in this column between elevation zero and some height z would be:4.2 Precipitation 97 Geographic and Temporal Variations The quantity of atmospheric water vapor varies with location and time. These varia- tions may be attributed mainly to temperature and source of supply considerations. The greatest concentrations can be found near the ocean surface in the tropics, the concentrations generally decreasing with latitude, altitude, and distance inland from coastal areas. About half the atmospheric moisture can be found within the first mile above the earth's surface. This is because the vertical transport of vapor is mainly through con- vective action, which is slight at higher altitudes. It is also of interest that there is not necessarily any relation between the amount of atmospheric water vapor over a region and the resulting precipitation. The amount of water vapor contained over dry areas of the Southwest, for example, at times exceeds that over considerably more humid northern regions, even though the latter areas experience precipitation while the for- mer do not. PRECIPITATION Precipitation is the primary input vector of the hydrologic cycle. Its forms are rain. snow, and hail and variations of these such as drizzle and sleet. Precipitation is derived from atmospheric water. its form and quantity thus being influenced by the action of other climatic factors such as wind, temperature, and atmospheric pressure. Atmo- spheric moisture is a necessary but not sufficient condition for precipitation. Continental air masses are usually very dry so that most precipitation is derived from moist maritime air that originates over the oceans. In North America about 50 percent of the evaporated water is taken up by continental air and moves back again to the sea. Formation of Precipitation Two processes are considered to be capable of supporting the growth of droplets of sufficient mass (droplets from about 500 to 4000 4m in diameter) to overcome air resistance and consequently fall to the earth as precipitation. These are known as the ice crystal process and the coalescence process. ‘The coalescence process is one by which the small cloud droplets increase their size due to contact with other droplets through collision. Water droplets may be consid- ered as falling bodies that are subjected to both gravitational and air resistance effects. Fall velocities at equilibrium (terminal velocities) are proportional to the square of the radius of the droplet; thus the larger droplets will descend more quickly than the smaller ones. As a result, smaller droplets are often overtaken by larger droplets, and the resulting collisions tend to unite the drops, producing increasingly larger particles. Very large drops (order of 7 mm in diameter) break up into small droplets that repeat the coalescence process and produce somewhat of a chain effect. In this manner, suffi- ciently large raindrops may be produced to generate significant precipitation. This process is considered to be particularly important in tropical regions or in warm clouds. An important type of growth is known to occur if ice crystals and water droplets are found to exist together at subfreezing temperatures down to about —40°C. Under98 Chapter 4 Precipitation these conditions, certain particles of clay minerals and organic and ordinary ocean salts serve as freezing nuclei so that ice crystals are formed. The vapor pressure under these conditions is higher over the water droplets than over the ice crystals, and thus conden- sation occurs on the surface of the crystals. The ice crystals grow in size, and uneven particle size distributions develop, which further favor growth through contact with other particles. This is considered to be a very important precipitation-producing mechanism, The artificial inducement of Precipitation has been studied extensively, and these studies are continuing. It has been demonstrated that condensation nuclei supplied to clouds can induce precipitation. The ability of humans to ensure the production of pre- cipitation or to control its geographic location or timing has not yet been attained, however. Many legal as well as technological problems are associated with the prospects of “rain-making™ processes. Of interest here is the impact on hydrologic estimates that uncontrolled or only partially controlled artificial precipitation might have. Many natu- rally occurring hydrologic variables are considered as statistical variates that are either randomly distributed or distributed with a random component. If the distribution or time series of the variable can be modeled, an inference as to the frequency of occur- rence of significant hydrologic events of a given magnitude (such as precipitation) can be made. If. however, artificial controls are used and if the effects of these cannot be reliably predicted, frequency analyses may prove to be totally unreliable tools. Precipitation Types Dynamic or adiabatic cooling is the primary cause of condensation and is responsible for most rainfall. Thus it can be seen that vertical transport of air masses is a require- ment for precipitation. Precipitation may be classified according to the conditions that generate vertical air motion. In this respect. the three major categories of precipitation type are convective, orographic, and cyclonic. Convective Precipitation Convective Precipitation is typical of the tropics and is brought about by heating of the air at the interface with the ground. This heated air expands with a resultant reduction in weight. During this period, increasing quantities of water vapor are taken up: the warm moisture-laden air becomes unstable; and Pronounced vertical currents are developed. Dynamic cooling takes place, causing condensation and precipitation. Convective precipitation may be in the form of light showers or storms of extremely high intensity (thunderstorms are a typical example). Orographic Precipitation Orographic precipitation results from the mechanical lifting of moist horizontal air currents over natural barriers such as mountain ranges, This type of precipitation is very common on the West Coast of the United States where moisture-laden air from the Pacific Ocean is intercepted by coastal hills and mountains. Factors that are important in this process include land elevation, local slope. orientation of land slope, and distance from the moisture source. In dealing with orographic precipitation, it is common to divide the region under study into zones for which influences aside from elevation are believed to be4.2. Precipitation 99 reasonably constant. For each of these zones, a relation between rainfall and elevation is developed for use in producing isohyetal maps (see Section 4.5). Cyclonic Precipitation Cyclonic precipitation is associated with the movement of air masses from high-pressure regions to low-pressure regions. These pressure differences are created by the unequal heating of the earth’s surface. Cyclonic precipitation may be classified as frontal or nonfrontal. Any barometric low can produce nonfrontal precipitation as air is lifted through horizontal conver- gence of the inflow into a low-pressure area. Frontal precipitation results from the lift- ing of warm air over cold air at the contact zone between air masses having different characteristics. If the air masses are moving so that warm air replaces colder air, the front is known as a warm front; if, on the other hand, cold air displaces warm air, the front is said to be cold. If the front is not in motion, it is said to be a Stationary front. Figure 4.1 illustrates a vertical section through a frontal surface. Thunderstorms Many areas of the United States are subjected to severe convective storms, which are generally identified as thunderstorms because of their electrical nature. These storms, although usually very local in nature, are often productive of very intense rainfalls that are highly significant when local and urban drainage works are considered. Thunderstorm cells develop from vertical air movements associated with intense surface heating or orographic effects. There are three primary Stages in the life history of a thunderstorm, These are the cumulus stage, the mature stage, and the dissipating stage. Figure 4.2 illustrates each of these stages. All thunderstorms begin as cumulus clouds, although few such clouds ever reach the stage of development needed to produce such a storm. The cumulus stage is char- acterized by strong updrafts that often reach altitudes of over 25,000 ft. Vertical wind speeds at upper levels are often as great as 35 mph. As indicated in Fig. 4.2a, there is considerable horizontal inflow of air (entrainment) during the cumulus stage. This is an important element in the development of the storm, as additional moisture is pro- vided. Air temperatures inside the cell are greater than those outside, as indicated by Warm air mas mass » Ss $s Surface front Earth’s surface FIGURE 4.1 Vertical cross section through a frontal surface.100 Chapter 4 —_ Precipitation LEGEND: Little or no vertical motion, © Rain air spreads horizontally x Snow Elevation = 40,000 ft = Ice crystals -60- -36— Isotherms <— & —15— Elevation = 25,000 ft a £ é FIGURE 4.2 ‘Cumulus, mature. and dissipating stages of a thunderstorm cell. (Department of the Army) the convexity of the isotherms viewed from above. The number and size of the wi droplets increase as the stage progresses. The duration of the cumulus stage is appr mately 10-15 min. The strong updrafts and entrainment support increased condensation and development of water droplets and ice crystals. Finally, when the particles increase in and number so that surface precipitation occurs, the storm is said to be in the mai stage. In this stage strong downdrafts are created as falling rain and ice crystals cool tht below. Updraft velocities at the higher altitudes reach up to 70 mph in the early period the mature stage. Downdraft speeds of over 20 mph are usually above about 5,000 { elevation. At lower levels, frictional resistance tends to decrease the downdraft velo Gusty surface winds move outward from the region of rainfall. Heavy precipitatio often derived during this preiod, which is usually on the order of 15-30 min. In the final or dissipating stage, the downdraft becomes predominant until all air within the cell is descending and being dynamically heated. Since the upd ceases, the mechanism for condensation ends and precipitation tails off and ends. Precipitation Data Considerable data on precipitation are available in publications of the Nati¢ Weather Service [4],{5]. Other sources include various state and federal agen engaged in water resources work. For regional studies it is recommended that all pc ble data be compiled; often the establishment of a gauging network will be necessa Precipitation Variability Precipitation varies geographically, temporally, and seasonally (see Figs. 4.3 and 4.4 should be understood that both regional and temporal variations in precipitationIa19§ uoNDAsasU0 jog ‘summon ‘soyput ut uonendisaid yenuue weayy f ate yoarl \ CANS de CR a re fe ESS CO {0 g WHE OIKE* a G : Bg Ys us 2p(2211ag uowonsesueg os ‘2anynaudly fo mauusedag $1) (-u) wonendisaud djqruow weaw'g *(4,) aimesodwiay qyuoU uvaw 'y :suoNNgunsIp amesadway pue uonendDarg vy auNDId anosvrravnd sf aNosyr favs a anos cHYNsT : aNosVrcHVId f 5 su siuossser UNOSV I TRYNSE sues) MN 4 a © i f B 4 sReERe TMS, angkv craving xL2u2IRV a q vovodsid ues a osv rrNWAS f| 5 ayosvrohynd Sey vor asin seeseee “-S anos rrayays £ 090g Anfang aveypies BRERe OS (OSV TNVRIT fy -[Link] ves UNOS) CEWUNEE 9 i GNOSV LINAS, UN 2ueW 27g anes aNHewse, “LW "eu, fang oor AW pueniod 1024.3 Distribution of Precipitation 103 portant in water resources planning and hydrologic studies. For example, it may be ry important to know that the cycle of minimum precipitation coincides with the ak growing season in a particular area, or that the period of heaviest rainfall should avoided in scheduling certain construction activities. Precipitation amounts sometimes vary considerably within short distances. ecords have shown differences of 20 percent or more in the catch of rain gauges less an 20 ft apart. Precipitation is usually measured with a rain gauge placed in the open that no obstacle projects within the inverted conical surface having the top of the uge as its apex and a slope of 45°. The catch of a gauge is influenced by the wind, hich usually causes low readings. Various devices such as Nipher and Alter shields we been designed to minimize this error in measurement. Precipitation gauges may of the recording or nonrecording type. The former are required if the time distribu- n of precipitation is to be known. Information about the features of gauges is readily ailable (3). Because precipitation varies spatially, it is usually necessary to use the data from sveral gauges to estimate the average precipitation for an area and to evaluate its reli- lity. This is especially important in forested areas where the variation tends to be ge. ‘Time variations in rainfall intensity are extremely important in the rainfall-runoff ‘ocess, particularly in urban areas (see Fig. 4.5a). The areal distribution is also signifi- nt and highly correlated with the time history of outflow (see Fig. 4.5b). These con- iderations are discussed in greater detail in following chapters. ISTRIBUTION OF PRECIPITATION tal precipitation is distributed in numerous ways. That intercepted by vegetation and es may be equivalent to the total precipitation input for relatively small storms. Once terception storage is filled, raindrops begin falling from leaves and grass, where water red on these surfaces eventually becomes depleted through evaporation. Precipitation t reaches the ground may take several paths. Some water will fill depressions and even- lly evaporate; some will infiltrate the soil. Part of the infiltrated water may strike rela- Wely impervious strata near the soil surface and flow approximately parallel to it as terflow until an outlet is reached. Other portions may replenish soil moisture in the soil zone, and some infiltrated water may reach the groundwater reservoir that sus- ins dry weather streamflow. The component of the precipitation input that exceeds the | infiltration rate will develop a film of water on the surface (surface detention) until erland flow commences. Detention depths varying from } to 1} in. for various condi- ions of slope and surface type have been reported [3]. Overland flow ultimately reaches fined channels and becomes streamflow. Figure 4.6 illustrates in a general way the disposition of a uniform storm input to natural drainage basin. Although such an input is not to be expected in nature, the dicated relations are representative of actual conditions. Modifications resulting rom nonuniform storms will be discussed as they arise, In Fig, 4.6a note that the storm input is distributed uniformly over time ¢, at a rate qual to i (dimensionally equal to LT~). This input is dissected into components é, fhrough i,, the sum of which is equal to i at any time ¢. Figure 4.6b illustrates the104 = Chapter 4 Precipitation o 12 3 4 5 6 7 8 9 0 MW i2 13 14 Time (10? sec) (a) c ie 2.33 B&D. Isohyets—lines of Peevl, 2.0in: equal rainfall depth ai | 3.0 in. | | 3.0in. | | | 2.0in. U u N E \ 1.08 : (b) FIGURE 4.5 (a) Rainfall distribution in a convective storm June 1960, Baltimore, Maryland. (b) Isohyetal pattern, storm of September 10, 1957, Baltimore, Maryland. O, recording rain gauge.4.3 Distribution of Precipitation 105 4 Infiltration Rainfall input iy ; Interception if i=hthtintig Depression storage ig Overland flow supply t (a) f\ t ‘i / Soil moistu f \p-Sotmoinee jolt ‘ Interflow ale ‘ i Groundwater q I (b) Overland flow supply Mechanics Streamflow of surface hydrograph runoff —— © FIGURE 4.6 ‘The runoff process: (a) disposition of precipitation, (b) components of infiltration, and (6) disposition of overland flow supply.106 44 Chapter4 Precipitation manner in which infiltrated water is further subdivided into interflow, groundwater and soil moisture. Figure 4.6c shows the transition from overland flow supply into streamflow. The mechanics of these processes will be treated in detail in later section: The nature of the curves presented depicts the general runoff process, It should by realized, however, that actual graphs of infiltration and/or other factors versus time might appear quite different in form and relative magnitude when compared wit these illustrations because of the effects of nonuniform storm patterns, antecedenj conditions, and other factors. The rate and areal distribution of runoff from a drainage basin are determined b! a combination of physiographic and climatic factors. Important climatic factors includ the form of precipitation (rain, snow, hail), the type of precipitation (convective, oro, graphic, cyclonic), the quantity and time distribution of the precipitation, the characte! of the regional vegetative cover. prevailing evapotranspiration characteristics, and th status of the soil moisture reservoir. Physiographic factors of significance include ge metric properties of the drainage basin, land-use characteristics. soil type, geologi structure, and characteristics of drainage channels (geometry, slope, roughness, ant storage capacity). Large drainage basins often react differently from smaller ones when subjecte: to a precipitation input. This can be explained in part by such factors as geologic ag! relative impact of land-use practices, size differential. variations in storage characte; istics, and other causes. Chow defines a small watershed as a drainage basin whost characteristics do not filter out (1) fluctuations characteristic of high-intensity, short duration storms; or (2) the effects of land management practices [6]. On this basi small basins may vary from less than an acre up to 100 mi*. A large basin is one i which channel storage effectively filters out the high frequencies of imposed precipit tion and effects of land-use practices. POINT PRECIPITATION Precipitation events are recorded by gauges at specific locations. The resulting dat; permit determination of the frequency and character of precipitation events in vicinity of the site. Point precipitation data are used collectively to estimate areal va ability of rain and snow and are also used individually for developing design stor characteristics for small urban or other watersheds. Point rainfall data are used to derive intensity-duration—frequency curves su as those shown in Fig. 4.7. Such curves are used in the rational method for urban stot drainage design (Chapter 12); their construction is discussed in Chapter 7. In applyit the rational method, a rainfall intensity is used which represents the average intensi of a storm of given frequency for a selected duration. The frequency chosen shou reflect the economics of flood damage reduction. Frequencies of up to 100 years al commonly used where residential areas are to be protected. For higher-value distri and critical facilities, up to 500 years or higher return periods are often selected. Loe: conditions and practice normally dictate the selection of these design criter! (Executive Order 11988, Floodplain Management, 1977). It is occasionally necessary to estimate point rainfall at a given location fro recorded values at surrounding sites. This can be done to complete missing records4.4 Point Precipitation 107 10 100-yr frequency 8 > 50-yr frequency $ 20-yr frequency S% 10-yr frequency z 5-yr frequency 5 z ad 2 2 Z FIGURE 4,7 0 ‘Typical intensity-duration-frequency 20 40 60. 80 100 120 curves for Baltimore, Maryland, and Duration (min) Vicinity. to determine a representative precipitation to be used at the point of interest. The National Weather Service has developed a procedure for this which has been verified on both theoretical and empirical bases [7]. Consider that rainfall is to be calculated for point A in Fig. 4.8. Establish a set of axes running through A and determine the absolute coordinates of the nearest sur- rounding points B, C, D, E, and F. These are recorded in columns 3 and 4 of Table 4.1. The estimated precipitation at A is determined as a weighted average of the other five points. The weights are reciprocals of the sums of the squares of AX and AY; that is, D* = AX? + AY?, and W=1/D?. The estimated rainfall at the point of interest is given FIGURE 4.8 Four quadrants surrounding precipitation station A.108 45 Chapter 4 Precipitation TABLE 4.1 Determination of Point Rainfall from Data at Nearby Gauges qa) (2) @) (4) (Ss) (6) @) Point Rainfall (in.) ax ay (D’) wx 10" PXWx10® A - = = = - i B 1.60 4 2 20 50 80.0 € 1.80 1 6 37 27.0 48.6 D 150 3 2 1B 76.9 1154 E 2.00 3 3 18 $5.6 112 F 1.70 2 8 125.0 212.5 ‘Sums = - 334.5 5677 (Pyat A = 567.7/334.5; by = (Px W)/% W. In the special case where rainfall is known in only two adjace quadrants (e.g., | and II), the estimate is given as 5 (PX W). This has the effect reducing estimates to zero as the points move from an area of precipitation to one wi no records. This is considered to be the most logical procedure for handling thi unusual case [7]. The estimated result will always be less than the greatest and great than the smallest surrounding precipitation. For special effects such as mountain inf ences, an adjustment procedure can be applied. AREAL PRECIPITATION For most hydrologic analyses, it is important to know the areal distribution of precip®? tation. Usually, average depths for representative portions of the watershed are det mined and used for this purpose. The most direct approach is to use the arithmety average of gauged quantities. This procedure is satisfactory if gauges are uniformil distributed and the topography is flat. Other commonly used methods are the isohyet method and the Thiessen method. The reliability of rainfall measured at one gauge j representing the average depth over a surrounding area is a function of (1) the dis- tance from the gauge to the center of the representative area, (2) the size of the are (3) the topography, (4) the nature of the rainfall of concern (e.g., storm event vers mean monthly), and (5) the local storm pattern characteristics [8]. For more informa tion on errors of estimation, the reader should consult Refs.7 and 8. Figures 4.9 and 4.10 illustrate how the measured rainfall at a single gauge rela to the average rainfall over a watershed with change in (1) the relative position of t gauge in the watershed and (2) the time period over which the average is calculat In the first case it is clear that the more central the gauge location, the more closely j observations will‘ match the average for a representative area, providing that t region is not too large. Figure 4.10 shows, not surprisingly, that areal averages 01 long time periods, in this case one year, may be expected to conform more closely a single gauge average than those for an individual storm event. This suggest that the design of gauging networks should be tempered with both space and ti considerations.4.5 Areal Precipitation 109 = Yt Storm rainfall, watershed average (in.) ‘Storm rainfall, watershed average (in.) | 1 | |. 1 2 3 4 1 2 3 4 ° 0 Storm rainfall at one gauge (in.) Storm rainfall at one gauge (in.) (a) (b) FIGURE 4.9 Errors resulting from use of a single gauge to estimate watershed average. (a) Watershed area is 0.75 mi* and gauge is near the center. (b) Watershed area is 0.75 mi? and gauge is 4 mi outside the watershed boundary. (Gauge location effect, Soil Conservation Service) Isohyetal Method The two principal methods for determining areal averages of rainfall are the isohyetal method and the Thiessen method. The isohyetal method is based on interpolation between gauges. It closely resembles the calculation of contours in surveying and map- ping. The first step in developing an isohyetal map is to plot the rain gauge locations on 24 340 % 4 §3 % 30 + 3 3 : 3 | 22 3 20 : : = e gi = 10 E 5 | Leonie 0 0 ae0 1 2 3 ASAD 1 2 3 4 Storm rainfall at one gauge (in.) Storm rainfall at one gauge (in.) (a) (b) FIGURE 4.10 Errors resulting from use of a single gauge to estimate watershed average. (a) Watershed area is 5.45 mi? and the gauge is on the boundary. (b) Watershed area is 5.45 mi? and the ‘gauge is on the boundary. (Time period effect, Soil Conservation Service)110 Chapter4 Precipitation Watershed boundary (b) Average precipitation for area Ay is 4.25 in ~ Average precipitation _ £A,P, for entire basin EA, (©) FIGURE 4.11 Construction of an isohyetal map: (a) Locate rain gauges and plot values; (b) interpolate between gauges; and (c) plot isohyets. a suitable map and to record the rainfall amounts (Fig. 4.11). Next, an interpolati between gauges is performed and rainfall amounts at selected increments are plotte Identical depths from each interpolation are then connected to form isohyets (lines equal rainfall depth). The areal average is the weighted average of depths between is hyets, that is, the mean value between the isohyets. The isohyetal method is the m accurate approach for determining average precipitation over an area, but its prope! use requires a skilled analyst and careful attention to topographic and other fact that impact on areal variability. Figure 4.12 illustrates the Tepresentation of a maj storm event in North Carolina by an isohyetal map.(22198 2940244 ouonoN) "Wr L1P J 19qWOAON KepuOW\ OF-ure J 18 p 19qUIARON ‘Keptty OI] poled atp 10} oud TeIOL“LL6I 19qM9AON Jo Pooy 9x} pasned yey) UoHENdIoaud 94p SuPROYS Base eULJOIE. YON ‘TTEAsOTEIS-o|AdYSY Jo dE zp ano soyout tt uaat@ s1 won SE 9 11112 Chapter 4 Precipitation (a) (b) (©) SAP, Average depth over entire watershed = 4 | FIGURE 4.13 Construction of a Thiessen diagram: (a) Connect rain gauge locations; (b) draw perpendicular bisectors; and (c) calculate Thiessen weights (Ay, Ax, Ay). (a) A completed network Thiessen Method Another method of calculating areal rainfall averages is the Thiessen method. In procedure the area is subdivided into polygonal subareas using rain gauges as cen The subareas are used as weights in estimating the watershed average depth. Thie: diagrams are constructed as shown in Fig. 4.13. This procedure is not suitable mountainous areas because of orographic influences. The Thiessen network is fixed a given gauge configuration, and polygons must be reconstructed if any gauges relocated. Accuracy Irrespective of the method used for estimating areal precipitation, the location of gauge used in deriving the estimate relative to the point of application of the esti must be taken into consideration. In mountainous localities, vertical distances be more important than horizontal ones. For gentle landscapes, horizontal spacings the most important. When a precipitation gauging network is to be developed, spacing and arrangement of gauges must be considered. Example 4.1 Given the drainage area of Fig, 4.14 and the rainfall data displayed in column 3 Table 4.2, calculate the average rainfall over the area using (a) the arithmetic and (b) the Thiessen polygon weighting system.Solution. 1 2. 4.5 ArealPrecipitation 113 FIGURE 4.14 ‘Thiessen diagram for Example 4.1. Identify those gauges falling within the area boundary. They include gauges 1, 4 through 6, 8, and 9. Averaging the values for these six gauges yields an estimated mean areal rainfall of 3.20 inches. Following the Thiessen method as described in Section 4.5, construct poly- gons using triangles to connect gauge points. These polygons are shown in Fig. 4.14. Calculate the percent of the total area associated with each gauge TABLE 4.2 Data and Thiessen Polygon Calculation for Example 4.1, @ Q) @) @ Gaugeno. % Area —Precip.(in) (2) x(3) 1 Ss 1.56 0.08 4 2.95 0.12 3 3 3.44 0.10 4 1s 291 0.44 5 i 417 0.46 6 19 421 0.80 7 4 27 O11 8 7 245 0.17 9 a 3.88 0.81 10 6 3.98 0.24 i 5 251 0.13 Total 100 3.45114 46 47 48 Chapter 4 Precipitation and record as in column 2 of Table 4.2. The Thiessen weighted average obtained by multiplying the values in column 2 by the values in column The Thiessen average is computed as 3.45 inches of rainfall. The use of spreadsheet (Table 4,2) facilitates computations and aids in organizing datd PROBABLE MAXIMUM PRECIPITATION The probable maximum precipitation (PMP) is the critical depth-duration-area rai fall relation for a given area and season which would result from a storm containi the most critical meteorological conditions considered probable [9]. Such storm event are used in flood flow estimates by the U.S. Corps of Engineers and other wat @ Fesources agencies. The critical meteorological conditions are based on analyses air-mass properties (effective precipitable water. depth of inflow layer, wind, temper ture, and other factors). synoptic situations during recorded storms in the regio@ topography. season, and location of the area. The rainfall derived is termed probaly maximum precipitation since it is subject to limitations of meteorological theory ant data and is based on the most effective combination of factors controlling rainfs intensity [9]. An earlier designation of maximum possible precipitation is synonymoug The seasonal variation of PMP is important in the design and operation of mul purpose structures and in flooding considerations that may occur in combination snowmelt, In both of these cases, annual probable maximums might be less importa than seasonal maximums, Figures 4.15 and 4.16 display 24-hr PMP for the eastern of the United States for 200-mi? watersheds during the month of August (si figures are available from the National Weather Service). GROSS AND NET PRECIPITATION The net (excess) precipitation that contributes directly to surface runoff is equivale to the gross precipitation minus losses to interception, storm period ceparati depression storage, and infiltration. The relation between excess precipitation P, ai gross precipitation P is thus: e P, = P ~ ¥ losses Ce where the losses include all deductions from the gross storm input. ‘The paths that water precipitated over an area may take can be represented flow diagrams of the type given in Fig. 1.2 and by equations of the form of Eq. Models such as these are the basis for most hydrologic investigations, and much of ti content of this book is devoted to the conceptualization of individual components the various hydrologic processes and to synthesizing these components into holist representations of hydrologic events. PRECIPITATION FREQUENCY ANALYSIS, Point precipitation data described in Section 4.4 are subjected to frequency analyses develop frequency relationships among rainfall depth, intensity, and storm durationwn = (oontos saying jouonen 2120; fo wouedeg S71) “ay $2 ut 1 992 20) (sayout us) uoneNdtoasd wsmuunxeAL 2Iqeqo%d sty 3uNDd116 Chapter 4 Precipitation 6 1224 48 hr 1,000 ‘800 600 Drainage area (mi?) S y 10, 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 Percentage of 200 mi, 24-hr values FIGURE 4.16 Seasonal variation, depth-area-duration relations, Percentage to be applied to 200-mi?, 24-hr probable maximum precipitation values for August in Zone 6, (US. Deparment of Commerce, National Weather Service) ‘These relationships, known as intensity—duration-frequency (IDF) curves, are used i designing storm water management facilities and floodway reservations: Such designs are based on estimates of worst-case scenarios of rainfall intensity and duration durin, a given interval of time. This often requires analyzing several storms of different dura tions to find the most critical event for the selected design frequency. Typical recur, rence intervals for designing storm water structures are given in Chapter 13. The IDF curves for Baltimore, Maryland, shown in Fig. 4.7 are representative o those encountered in practice. Similar curves for other locations are available in the lit, occurrence. The correct interpretation of any point value obtained from Fig. 4.7 is that on th average, for any given time duration, storms having a uniform intensity (i) for tha duration would have a recurrence interval equal to the corresponding curve value. For example, in any time duration of 90 minutes, Baltimore could experience a pea 2.0-in/hr storm once every 20 years. The 20-yr, 90-min design storm for Baltimot would have a depth of D = 3.0 in., where D = rainfall intensity multiplied by the storm, duration. A 20-yr, 30-min design storm would have an intensity of 4.6 in/hr but resul in a depth of only 2.3 in. Although the latter storm produces less depth; its highei intensity could be the governing factor in determining the size of drainage works. The probability of occurrence of both storms would be the same, however.48 Precipitation Frequency Analysis 117 Development of IDF Curves Point rainfall data from weighing or tipping-bucket rain gauges provide the basis for calculating IDF curves for a given location. Rainfall depths occurring over selected time increments are compiled and arrayed by duration and storm. For any specified duration, the rainfall depths that occurred are ranked in decreasing order of magni- tude and are assigned a rank order number [Link] commonly used formula for com- puting plotting positions is: P=mj(n+1) (43) where 7 is the number of observations and P is the estimated probability of values being greater than or equal to the ranked value. Note that where droughts are being studied, the data would be ranked from the lowest to the highest, and P would be the estimated probability of values being less than or equal to the ranked value. Extreme hydrologic events are commonly estimated using annual series or partial-duration series (Chapter 3) of historical data. An annual series is comprised of a single extreme event (this may be either a maximum or a minimum) for each year of record. A partial-duration series is comprised of all events exceeding a selected base value for each year of record. Such a series is used in cases where more than one event of significance occurs per year. In performing frequency analyses on historical data, at least 10 years of record should be used. The following example illustrates the IDF pro- cedure for both an annual and a partial-duration series for an 11-year record. Example 4.2 Perform a frequency analysis of the 30-min Baltimore rainfall data in Table 4.3 as an annual and a partial-duration series and plot the results. Solution. In Table 4.3 the maximum rainfall depths that occurred for any 30-min pe- riod during excessive rainfalls at Baltimore, Maryland, 1945-1954, are shown in the order of occurrence. The 65 observations represent a complete series. The 11 maximum annual events are underlined and represent the annual series. The greatest 11 events throughout the record are identified by an asterisk and represent the partial-duration series. The solution is shown in Table 4.4, and the data are plotted in Fig. 4.17. Note that the larger numbers occur in both series, and hence recurrence intervals for the less fre- quent events are the same. ‘The preceding example leads to consideration of the frequency analysis of rainfall depth or intensity for various durations of rainfall. Design problems often require the estimation of expected intensities for a critical time period. Frequency analysis of the rainfall record for periods other than the 30-min duration—for example, the maximum 5-, 10-, 20-, and 60-min occurrences—would yield a family of curves similar to those of Fig. 4.18. The usual method of presenting these data is to convert depth in inches to an intensity in in/hr and to summarize the data in intensity-duration—frequency curves as shown in Fig. 4.7. These curves are typical of the point analysis of rainfall data. It should be emphasized that the frequency curves join occurrences that are not necessarily from the same storm; that is, they do not represent a sequence of intensities during a single storm but only the average intensity expected for a specific duration.118 Chapter 4 Precipitation TABLE 4.3 Maximum 30-min Rainfall Depths, Baltimore, MD, 1945-1954 RF RF RF Storm depth Storm depth Storm depth Year number (in.) Year number (in) Year number —_—_(in.) era ci 1945 1 0.38 1948 1 133" | 1953 1 0.40 2 0.47 2 0.65 2 0.45 3 039 3 047 3 0.53 4 0.76 4 034 4 2.50° s 0.56 5 0.68 $ 1.03 6 0.35 6 0.63 6 0.75 7 0.43 7 047 7 0.70 8 0.40 8 L.00* 9 0.36 1949 1 0S2 2 049 | 1954 1 0.42 1946 1 0.62 2 0.70 3 05S 1980 1 0.55 3 0.85 a 0.88 2 0.63 4 0.60 4 047 3 0.69 5 036 4 127 | 1955 1 0.70 6 Lise 5 1.10* 2 0.95 7 0.75 3 102 8 1951 hi 0.88 4 0.50 9 2 0.97 5 0.65 3 0.59 6 0.55 1947 1 4 0.46 7 0.52 2 5 0.50 8 0.45 3 6 oss 9 0.54 4 10 0.60 5 1952 1 0.47 i 0.80 6 2 130 2 0.95 7 3 0.93 8 4 0.70 3 0.57 6 0.46 7 0.48 8 1.30" es ‘Note: Underlined items are the annual series. Asterisks identify the partial-duration series. IDF Data Sources The principal source of precipitation data is the National Weather Service (NWS). The agency has six regional offices and is part of the National Oceanic and Atmospheric Administration (NOAA). Precipitation measurements are made daily at over 20,000 US. locations. State data are compiled by satellite and other means in the Weather Service Forecast Office (WSFO). Results of IDF analyses for most major urban loca- tions have been compiled [13]. For additional information and Web addresses, see also Chapter 2). Maximum average rainfall depths have been published by the U.S. Weath Bureau [13] for durations between 30 min and 24 hr and for recurrence interva'sRainfall depth (in.) 4.8 Precipitation Frequency Analysis 119 —— TABLE 4.4 Annual and Partial Series Rainfall Depths Depth (in.) . Recurrence interval, Order Annualseries Partialseries_ (nm + L)/m 1 2.50 2.50 12 2 2.04 2.04 6 3 153 153 4 4 1.33 133 3 5 1.30 130 24 6 1.27 127 2 7 1.02 1.20 1? 8 0.97 120 1s ¥ 0.85 11s 13 10 0.76 1.10 12 it 0.52 1.07 i —— between 1 and 100 years. Depth-duration-frequency curves can be constructed for any location by plotting successive values from the various rainfall maps, preferably on log- arithmic paper to facilitate fitting flatter curves. Correction factors are given to permit estimates of depths for durations less than 30 min. Formulas for IDF Curves Regression analysis can be used to fit intensity-duration-frequency curves, and the constants can be interpreted as regional characteristics. Many formulas have been used to fit these curves, but most of them are in a form with intensity (i) inversely propor- tional to duration (1). Steel [14] has used a model of the form i= A/(¢+ B) to fit rainfall [| dy \ 25 | ont 20 id 1s Lt Partial duration tithe LU 1 thee T | ns |Annual series | TTT T 1 FIGURE 4.17 LE | | | | Difference in annual and partial- 9 durati Le dof for 41121315. 2 +3 4 5678910 2 ee can maximum 30-min durations, Recurrence interval (yt) Baltimore, Maryland.120 Chapter 4 Precipitation S 3 & < FIGURE 4.18 i I | 0 Depth-area~duration curves for 0 500 1,000 1,500 2.000 24-hr storm. Area (mi) data throughout the United States. The constants A and B therefore serve as charac- teristic features of both the rainfall region and the frequency of occurrence in each area. Example 4.3 Fit the following rainfall data to determine the 10-year intensity-duration-frequency curve, 1 = duration (min) 5 10 15 30 60 120 i= intensity (in/hr) 7.1 59 5.1 38 23 14 Vi 0.14 0.17 0.20 0.26 0.43 071 Solution 1. A model of the form i = A/(¢ + B) can be expressed in linear form as 1fi = t/A + B/A. 2, The regression of 1/i versus ¢ yields 1/i = 0.005 + 0.12, from which A = 200 and B = 24. | 3. Thus the rainfall formula is i = 200/(t + 24). The correlation coefficient is - 0,997. Example 4.4 Table 4.5 catalogs data for numerous storms having a range of intensities and dura- tions. The entire record spans 40 years. By interpolating the values in the table, estimate the time versus intensity values for the 5-year storm.48 Precipitation Frequency Analysis 121 > TABLE 4.5 Intense Rainfalls for 40-Year Record Storm Q) 3 @ (8) © 7 (&) (9) 0) (1) (12) (13) Row duration Row immediately below indicates rainfall intensity in in/hr No. (min) | 101250 «15 175° 20 «250 380 40 5 670 8D w 5 ur 46198 5 2 @) 10 19-76 4613 8 3 2 1 @) 15 99 8443 19 9 3 1 1 () 20 9% 6 4 #16 12 5 2 1 (5) 30 7 6 53 2B 23 09 se 1 (6) 40 68 51 26 13 12 4 3 1 ”) 50 9 27 19 10 9 3 2 (8) 60 31713 6 403 1 9) 80 2 2 4 1 2 (10) 100 12 3 1 (uy) 120 7 1 SSS Values under columns 2-13 in rows 1-11 are the number of storms of the duration shown in the left column for the intensity given below the column number. Solution. The 5-year storm would be equaled or exceeded 40/5 = 8 times in 40 years. Using this as the reference value, interpolations by row and column can be made. For example, for row 1, the interpolation would be as follows: 7 = [(8 - 5)/(13 — 5)] X 1 = 7 - (3/8) = 6.63 For column 11, the interpolation would be as follows: 10 - (8 - 3)/(13 - 3)] x 5 = 10 - (5/10) x 5=75 Similar operations made on all rows and columns yield the values shown below. Row Interpolation Time 10 15 20 30 40 50 60 80 100 Rainfall 663 5.00 417 357 263 225 208 183 138 111 intensity Column Interpolation Rainfall 6.00 5.00 4.00 3.00 250 2.00 1.75 150 1.25 1.00 intensity Time 7.50 10.00 16.25 25.71 32.00 52.00 55.00 71.11 88.89 116.00 Plotting these data, the IDF curve for the 5-year storm can be obtained as shown in Fig. 4.19.122 Chapter 4 Precipitation 10- T 9 + { 8 7 6 5 4 23 5 2 ed ib FIGURE 4.19 0 _| | l Intensity-duration-frequency 0 10 20 3040-50600 800 curve, 5-year storm Rainfall duration (min) Example 4.5 Nou are given average annual values of precipitation (inches) for the period from 18% 1997 for Tallahassee, Florida (see Table 4.6). Arrange the data in orler of magnitud versus average annual precipitation, Solution. The data are arranged by descending order of magnitude in columns 2,5, § 11, 14, and 27. The probabilities are calculated on the spreadsheet using Eq. 4.3 an multiplying by 100 to convert to percent. Figure 4.20 is a plot relating exceedanci probability to average annual Precipitation. 100 80 |— 40 30 t 20 Exceedance probability (%) 8 FIGURE 4.20 Exceedance probabilities for ol average annual precipitation, 300 40 5060 70—~—=«RSC«SS0 to Tallahassee, Florida Average annual precipitation (in.)“{quaosed ur o1e umous sanqea) papaaaxo 40 pazenbs 2q %'9"¢ suUIMIOD ‘g's suumjoD ‘PE suuNIO. 166 01k GOL 78 OBE | OT ISL HDS tL ror SIL 81 clo «18 | LOT OBL SD EES OL ssl et a 96 «76 TSB: LEED OES OL Sor ee 91 vse €0r = SOL OL RRB L]COSS Fee sl Sve 80r Pol 78 SB 9B BID IS 8D See tI 966 tr Ol CLL C6r SB 6 955 Lo se €1 Lt Vee wl Fe OOS 00D «TS Sse a 16 Ter «LOT SL OS RCS OS ose u 606 Ley 6 OT SbL gS BTS SRI ou on 006 «6H = =66OEL HS BLS LOGO gs 6 16S OSt 8H LTLS LOS BS BMS DF $08 g we Ofse 6 BIL LOS LC SS HSI 718 £ €& Lsy % 60 805 B se TiS 818 9 ros TW 8% «OM «SIS kk ORS OBS SLE sts s rss 79H OTIS LL TS EBS SE oss. % sia SW 866 0 TMCS SLC IS HSS SSE sls £ of8 Ol % £9 1% wk GOS SB 9 668 iz Lt Ol 16 ¥O Ze OOF Co SS EE cel sr ut ol st rl a 1 Ot 6 8 L 9 s y £ z 1 “4661-6881 UoneHdielg jenuUY oSerIAY "TLE OSSCUETCL Sy TIVL 123124 Chapter 4 Precipitation 49 SNOW ACCUMULATION Information on the areal distribution of snow cover is required for water supply fore- casting, flood potential investigations, and other hydrologic analyses. Observations of snow cover are generally obtained by ground and air reconnaissance and photography. Between snow surveys. approximations of the extent of the snow cover are based on available hydrometeorological data. Snow cover depletion patterns within a basin are usually somewhat uniform from year to year; thus snow cover indexes can often be developed from data gathered at a few representative stations. Estimating the areal distribution of snowfall poses some problems. Taking arithmetic averages or using Thiessen polygons does not commonly provide reliable estimates of areal snow distribution from point gaugings. This is because orographic and topographic effects are often pronounced, and gauging networks frequently are not dense enough to permit the straightforward use of normal averaging techniques. However, regional orographic effects are relatively constant from year to year and storm to storm for tracts that are small when compared with the areal extent of gen- eral storms occurring in the region [15]. This circumstance permits many useful approaches in estimating the areal snow distribution once the basic pattern has been found for a region. One method used to estimate basin precipitation from point observations assumes that the ratio of station precipitation to basin precipitation is approximately constant for a storm or storms. This can be stated as [15]: Py _ No P,7N, (4.4) Ni, or Py= Pel 6 (4.5) = the basin precipitation the observed precipitation at a point or group of stations the annual precipitation for the basin N, = the normal annual precipitation for the control station or stations The normal annual precipitation is determined from a map (carefully prepared if it is to be representative) displaying the mean annual isohyets for the region. The precipi- tation is determined by using a planimeter on the areas between the isohyets. If the number of stations used and their distribution adequately depict the basin, Eq. 4.5 can provide a good approximation. For stations not uniformly distributed, weighting coef- ficients based on the percentage of the basin area portrayed by a gauge are sometimes used in determining N, for the group. Another system used in estimating areal snowfall is the isopercental method. In this approach, the storm or annual station precipitation is expressed as a percentage of the normal annual total. Isopercental lines are drawn and can be superimposed on a normal annual precipitation (NAP) map to produce new isohyets representing the storm of interest. A NAP map indicates the general nature of the basin’s topographic126 Chapter 4 Precipitation Basin-Wide Water Equivalent In the final analysis, it is the water equivalent of the snowpack that determines runoff. Basin water equivalent may be given as an index or reported in a quantitative manner such as inches or centimeters of depth for the watershed under investigation. The customary procedure for determining the basin water equivalent is to take observed data from snow course stations and to provide an index of basin conditions. Various procedures employ averages, weighted averages, and other approaches to accomplish this [15]. The important point to remember is that the usefulness of any index is based on how well it represents the overall basin conditions, not on how favor- ably it describes a particular point value. Indexes do not actually provide a quantitative evaluation of the property they cover. Instead, they give relative changes in the factor. By introducing additional data, however, an index can be used in a prediction equa- tion. For example, if the basin water equivalent can be estimated by subtracting the runoff and loss components from the precipitation input, the index can be correlated with actual basin water equivalent in a quantitative manner. SUMMARY Precipitation is the source of fresh water replenishment for the planet Earth. Too much or too little can mean the difference between prosperity and disaster. In between these extremes are the normal precipitation events that are experienced with a frequency and intensity related mainly to geographic position and topographic features. After reading this chapter you should understand that both the timing and amount of precipitation occurring over an area are important and that there is con- siderable geographic variability in precipitation. You should be able to estimate areal precipitation amounts from gauge data and conceptualize simple hydrologic process models. It should be recognized that average values of precipitation for a region shed some light on the quantity of water that might be made available for various uses, while a knowledge of the time-distribution and time-disposition of precipitation are requisites for developing management plans for periods of excess and shortage. PROBLEMS 4.1 Rain gauge X was out of operation for a month during which there was a storm. The rain- fall amounts at three adjacent stations A, B, and C were 4.2, 3.7, and 4.9 in., respectively. The average annual precipitation amounts for the gauges are X = 36.5, A = 42.1. B = 37.1,and C = 39.8 in. The delta x and delta y values respectively for each station are 0, 0; A = 3. 7; B= 4, 6; and C = 5, 9. Using a weighted average. estimate the amount of rainfall for gauge X. 4.2 Compute the rainfall for gauge X in Problem 4.1 if the storm readings at A, B, and C were 3.7, 4.1,and 4.8 in...respectively. 4.3 Compute the mean annual precipitation for the watershed in the following figure using the arithmetic mean, the Thiessen polygon method, and the isohyetal method. The gauge a ill ciaProblems 127 readings for gauges A-K, respectively, are: 29.79, 34.97, 25.6,24.27, 24.6, [Link], 15.51. 39.99, 43.04, and 28.41. E © @n G ® 10 mi 4.4 Compute the mean annual precipitation for the watershed in the figure for Problem 4.3 using the arithmetic mean and the Thiessen polygon method. The gauge readings for gauges A-K, respectively, are: 28.1, 33.7, 25.6, 23.9, 24.6,40.7, 41.3, 37.2, 38.7, 41.1, and 29.3 4.5 The chart from a rain gauge shown in the sketch represents a record that you must inter- pret. Find the average rainfall intensity (rate) between 6 a.m. and noon on August 10. Also find the total precipitation on August 10 and August 11. 0.30 0.25 0.20 0.15 0.10 0.05 0 Aug. 10, 1971 ‘Aug. 11, + 1971 Midnight 6am Noon 6pm Midnight 6am Noon 6 p.m Midnight Time 46 Refer to the chart of Problem 4.5. Calculate the rainfall intensity for the period between 6 a.m. and noon on August 11. Would you consider this to be a period of intense rainfall? 47 Use the map of Fig. 4.5 and from it construct a set of Thiessen polygons. Using these, esti- mate the mean rainfall for the region. 4.8 A mean draft of 100 mgd is produced from a drainage area of 200 mi*. At the flow line the reservoir is estimated to cover 4000 acres. The annual rainfall is 37 in., the mean annual128 Chapter 4 Precipitation runoff is 10 in., and the mean annual lake evaporation is 30 in. Find the net gain or loss} storage. Compute the volume of water evaporated. How significant is this amount? 4.9 A mean draft of 380,000 m‘/day is produced from a drainage area of 330 km?, At the flo line, the reservoir is estimated to cover about 1600 hectares. The annual rainfall is 96.5 cn the mean annual runoff is 22.8 cm, and the mean annual lake evaporation is 77.1 cm. Fin the net gain or loss in storage and compute the volume of water evaporated. Calculate vo umes in cubic meters. 4.10 Drainage areas within each of the isohyetal lines for a storm are tabulated for a water shed. Use the isohyetal method to determine the average precipitation depth within th basin for the storm. Make a conceptual sketch. Isohyetal interval (in.) Area (acres) 0-2 2700 24 1900 46 1000 6-8 0 4.11 Rework Problem 4.10 if the values in column 2 of the table are 2,500, 2,100, 1.200, and 300 respectively. 4.12 Discuss how you would go about collecting data for analysis of the water budget of 2 region. What agencies would you contact? What other sources of information would you seek out? 4.13 For an area of your choice. plot the mean monthly precipitation versus time. Explain how this fits the pattern of seasonal water uses for the area. Will the form of precipitation be an important consideration? 4.14 The following table summarizes the number of occurrences of intensities of various dura- tions for a 34-year record of rainfall. Maximum intensities for the given durations were determined for all excessive storms and a count made of the exceedances, Interpolate for the average number of exceedances expected on a 5-year frequency and plot the 5-year intensity-duration-frequency curve. Number of times stated intensities were equaled or exceeded Intensity (in./ hr) Duration (min) 1.0 2.0 3.0 4.0 5.0 6.0 7.0 5 73 48 21 9 2 10 68 51 26 11 3: 1 15 72 35 23 11 5 1 30 29 17 7 3 1 60 1S 6 2 1 120 8 1 4.15 Fit the formula i = A/(t + B) to the data derived in Problem 4.14 for the S-vear intensity-duration-frequency curve.4.16 417 ENCES ty (21 13] [4 [5] {6} m1 References 129 Recorded maximum depths (in.) of precipitation for a 30-min duration at a single station are: Year Date Depth (in.) 1963 May3 2.0 June 3 1.0 1964 June 7 1.0 1965 June 2 1.0 1966 June 1 10 Aug.3 3.0 1967 July4 10 1968 Aug. 8 4.0 1969 May6 6.0 June 8 5.0 Sept.4 1.0 1970 = May4 1.0 1971 Sept.8 5.0 1977 May9 1.0 (a) Determine the return period (years) for a depth of 2.0 in. using the California method with an annual series. (b) Repeat Part (a) using a partial-duration series. (c) Determine from the partial-duration series the depth of 30-min rain expected to be equaled or exceeded (on the average) once every 8 years. For a 60-year record of precipitation intensities and durations, a 30-min intensity of 2.50 inJhr was equaled or exceeded a total of 85 times. All but 5 of the 60 years experienced one or more 30-min intensities equaling or exceeding the 2.50-in/hr value. Use the Weibull formula to determine the return period of this intensity using (a) a partial series and (b) an annual series. Tennessee Valley Authority, Heat and Mass Transfer Between a Water Surface and the Atmosphere.” Lab. Rep. No. 14, TVA Engineering Lab, Norris, TN, Apr. 1972. ‘A.L. Shands, “Mean Precipitable Water in the United States,” U.S. Weather Bureau, Tech. Paper No. 10, 1949. R.K. Linsley, Jr., M. A. Kohler, and J, L. H. Paulhus, Applied Hydrology, New York: McGraw-Hill, 1949. D. W. Miller, J. J. Geraghty, and R. S. Collins, Water Atlas of the United States, Port Washington, NY: Water Information Center, 1963. U.S. Weather Bureau, Tech. Papers 1-33, Washington, D.C: US. Government Printing Office. Ven Te Chow (ed.), Handbook of Applied Hydrology, New York: McGraw-Hill, 1964. Staff, Hydrologic Research Laboratory, “National Weather Service River Forecast System Forecast Procedures," NOAA Tech. Mem, NWS HYDRO 14, National Weather Service, Silver Spring, MD, Dec. 1972."