When the vegetal cover is removed Sheet erosion cannot be accurately
from a land surface, the rate of removal measured by observing directly the of the soil material, at least initially, in- gradual lowering of the ground ele- creases rapidly. So well known is this vation as a function of time. The lower- principle that it hardly needs restate- ing is not areally uniform; on a micro- ment. scale, erosion here is offset by deposi- If attention is focused on any indi- tion there. The process is slow in terms vidual drainage basin in its natural of a man's span, even in a badland state, large or small, and inquiry is area (King and Melin, 1955). To judge made as to the rate of denudation, a the amount of erosion in terms of loss quantitative answer is not easily ob- of a certain portion of a complete soil tained. The possible error in any com- profile supposed to have originally putation of rate of sediment produc- existed is crude at best (though widely tion from any given drainage basin is employed) and hardly satisfies the de- considerable. Significant variations are sire for an objective, quantitative found in sediment yields from closely measure. adjacent watersheds which appear to To measure rate of degradation of a be generally similar. To make a quanti- landscape by gully erosion through tative evaluation of the change in the computation of the volume of the gully rate of denudation when the natural network is possible, though few good vegetation is disturbed is, therefore, data exist. But such estimates are even more difficult. Considering the plagued by the importance of local dep- fact that "soil conservation" has been osition (temporary storage) of the promoted to the status of a science, eroded material in fans near the mouth our lack of ability to answer what is of the gully (Hadley, 1954). Further- apparently so simple a question may more, there is no assurance that at seem surprising. Let us look at some of least some of the gullies did not exist the reasons. prior to the beginning of the period under consideration. Leopold and Mil- * Dr. Leopold is a Hydraulic Engineer in ler (1954) have emphasized that many the United States Geological Survey, Wash- gullies in Wyoming which appear to ington 25, D.C. He was formerly Head (1946- have been formed since the opening of 49) of the Department of Meteorology in the Pineapple Research Institute and Experiment the West are in fact at least pre- Station of the Hawaiian Sugar Planters' Asso- Columbian and may be several thou- ciation, Honolulu, Hawaii. His publications sands of years old. include (with Thomas Maddock, Jr.): The It is theoretically possible to estimate Hydraulic Geometry of Stream Chonnek and Some Physiographic Implications, 1953, and net rate of removal of soil material The Flood Control Controversy, 1954. from a watershed on the basis of the 639 640 Mans Role in Changing the Face of the Earth sediment load of the main stream drain- This represents not all, but the major ing the area. But present techniques portion, of such experimentation. A re- are adequate to measure only the sus- cent survey (Leopold and Maddock, pended portion of a streams load, and 1954) showed that investigations by then only if the material is not coarser federal agencies included work on than sand. There is no practical method about 1,700 experimental plots and on at present for measuring that part of some 560 natural watersheds, together the load moving along or close to the comprising 464 experiments. Of the to- bed of the stream. Though the sus- tal, 86 per cent of the experiments pended portion of the sediment load dealt with areas of less than 100 acres may constitute three-quarters or more in size. Such experiments provide a of the total debris in many streams, the quantitative measure of the effect of suspended load is still only a portion. particular vegetal changes on sediment The load of gravelly streams cannot be production only in similar watersheds accurately measured in the c h a p e l at of like size. all. Rate of degradation of a landscape is The rate of sediment deposition in a not measured solely by the movement reservoir provides the best measure- of discrete particles of debris, for the ment of total load and, therefore, of constituents dissolved in the runoff average denudation rates. Though water may be a significant part of the some suspended sediment does not de- whole. Measurements of reservoir posit in the reservoir but passes through sediment deposits do not include the the gates or over the spillway, this spill dissolved fraction. usually can be estimated with an accu- Clark (1924) estimated that the an- racy commensurate with that of other nual rate of chemical denudation in the necessary measurements. Currently, United States is approximately 100 tons some four hundred reservoirs in the per square mile, though this figure United States have been surveyed and needs revision on the basis of new data. have ranges established for resurvey. From fifty representative records of But this number is hardly adequate to sediment yield in the United States, describe the diversity of watersheds in chosen by Glymph (1951, Table l ) , the river basins of the United States. the median value was 900 tons per Moreover, reservoir surveys do not fur- square mile annually. I t appears from nish information on the relative this rough comparison that chemical amounts of debris from various parts degradation may be of the order of 10 of the basin upstream. per cent of the total. In the Wind River Excellent measurements of rates of Basin, Wyoming (Colby et al., 1955), soil loss are available from experimen- the dissolved load of streams constitutes tal plots and watersheds, but the data about 13 per cent of the total dissolved cover only a small fraction of the and sediment load. It is possible that many possible combinations of soil dissolved loads may be more important type, slope, and vegetal cover. More- in landscape reduction than indicated over, it is very difficult to extrapolate by Glymph. from the measurements on small areas Changes in water quality as a result to large natural drainage basins. of successive use by irrigation are well The scope of the available experi- known, and in the Wind River Basin, mental data dealing with the interrela- for example, Colby et al. (1955, p. 192) tion of vegetation, soils, rainfall, run- believe that irrigation is greatly accel- off, and erosion may be judged from a erating the normal processes of erosion review of federally sponsored research. and transport of water-soluble minerals Land Use and Sediment Yield 641 from the Wind River formation, allu- ish were poor observers of natural his- vial terraces, and associated soils. Such tory, and their records are of little use effects of human activity generally ap- in reconstructing original conditions. ply to only portions of the drainage There are but few good accounts of basin. We are forced, however, from vegetation as it affects sediment yield both lack of data and lack of personal in areas essentially untouched until knowledge, to restrict the present dis- white exploration. The Lewis and Clark cussion to landscape degradation prod- journals are among the best. From ucts carried by streams as sediment. them we learn that the Missouri was certainly high in sediment load. But EFFECT OF HUMAN USE even the best expedition accounts do The relative extent to which human not provide a clear picture of where use has increased sediment yield prob- the sediment originated. Bank-cutting ab]r varies inversely with the rate of on the Missouri was described as an the original yield. This is suggested by active source, but bank-cutting is usual- measurements and appears logical from Y a Process Of sediment-trading-ero- general considerations. Brunes (1948) in One place and deposition in estimates of increase for areas in the another. north central states are much larger Even in where vegetation than those of Rosa and Tigerman on the plains areas is generally far more (1951) for areas in the Colorado River dense than that in topog- Basin. The eastern edges of the prairie raphy Of the Lewis and and the hardwood associations of the lark (Coues~ 18937 347) made the upper ~~~~i~~~~~~ were originally char- following observation near the foot of acterized by nearly complete vegetal the Bear Paw Mountains: cover, whereas large areas in the West . A high, level, dry, open plain . . [con- and Southwest included badlands, stitutes] the whole country to the foot of poorly vegetated scarps, and generally the mountains. The soil is dark, rich, and low vegetation density. The well-vege- fertile; Yet the grass is by no means SO tated mountain areas, though conbib- luxuriant as might have been expected, for it is short and scarcely more than suf- most Of the water comprise Only ficient to cover the ground. There are vast a minor part Of the drainage area* quantities of prickly-pears, and myriads of The presettlement sediment yield of drainage basins in the West is particu- grasshoppers.. . . larly difficult to evaluate. The original In the Same Place during a density of vegetation in woodland and Observe (ibid., P. 348) that semidesert shrub association was char- found the bed of a creek 25 yards wide at acteristically low even in presettlement the entrance, with some timber, but no times. However, this low density need water, notwithstanding the rain. It is in- not necessarily be interpreted as coin- deed astonishing to observe the vast quan- cident with high sediment yield. me tities of water absorbed by the soil of the species composition now extant is often plains, which, being opened in large crev- quite different from that originally ices, presents a fine rich loam. found over great areas, even where veg- A thorough review of the methodol- etation density has not changed appre- ogy and of the results of attempts to ciably. Furthermore, in the Southwest determine the total sediment yield from the relatively good observational record natural watersheds would be out of of early American exploration came place in the present discussion. A few only after two centuries of land use by examples will, however, provide some the Spanish ( Leopold, 1951). The Span- picture of the difficulties involved and 642 Mans Role in Changing the Face of the Earth the possible order of magnitude of the two-thirds of the drainage area. It is effect of human activities on land deg- increased by thirty-five times when radation. more than two-thirds of the drainage One technique is illustrated in a area is cultivated or idle. Brune esti- study by Brune ( 1948), using primarily mated that the present rate of sediment rates of accumulation of sediment in production in the Ohio and the Great reservoirs. By modifying these results Lakes drainage basins is roughly fifty with supplemental suspended sediment times the geologic norm. He .stated records and experimental data from further (ibid.,p. 16) that in the upper plots and small watersheds, Brune de- Mississippi River drainage basin where rived figures on the rate of annual sedi- about 42 per cent of the land is now ment movement from some particular cultivated or idle, the present rate of drainage basins of various sizes. It is sediment production and erosion is ap- generally recognized that the sediment proximately seventy-five times the geo- yield is a function of drainage-basin logic norm. size even in an area of relatively uni- Another approach to b e problem is form characteristics. But the figures on illustrated in a study by Gottschalk and sediment yield for basins of a given Brune ( 1950). A multiple correlation size in the Brune study showed a vari- was used to express the relationship ation of approximately a hundred between total sediment accumulation in times between minimum and maximum a reservoir (considered a dependent sediment yield. He attempted to relate variable) as a function of net water- this variation to land use as well as to shed area, age of the watershed in physical characteristics of the individ- years, rate of gross erosion, and the ual basins. The first step was to segre- ratio of reservoir capacity to watershed gate the data in terms of land use. Three area. The regression is greatly influ- categories were used to represent the enced by the value of the parameter percentage of the drainage area which used to represent the rate of gross ero- was in cultivation. An adjustment for sion. Estimates of this factor were ob- effect of soil type, degree of slope, tained by adding results of two kinds length of slope, and type of rotation of measurements. Gully erosion was de- was made on the basis of a somewhat termined by field observations, using subjective classification of the whole rate of gully development measured on area into zones chosen to represent successive aerial photographs. Sheet relative uniformity in respect of these erosion was estimated by an empirical variables. A further step was to apply interrelation among average length of a factor to the sediment yield to repre- slope, average degree of slope, and type sent the mean annual runoff. of cultivation, based principally on the On the basis of such analysis, Brune results of plot and small-watershed ex- showed that on the average, for a perimentation. drainage area of 100 square miles, in The nature of the problem unfortu- north central United States, as an exam- nately necessitates this kind of round- ple, basins within which one-third of about analysis. Any studious attempt to the total area is cultivated or idle correlate the many variables is com- are characterized by a long-term sedi- mendable; nevertheless, we should not ment concentration in runoff equal to gloss over the fact that the results ob- .015 per cent by weight. He concluded tained can be considered nothing bet- that the concentration is increased by ter than general approximations. six and one-half times when cultivated Still another type of methodology is and idle land represents one-third to illustrated by the study of Rosa and Land Use and Sediment Yield 643 Tigerman (1951), who attempted to only one-third to one-fourth the existing estimate the sediment contribution from rate. It should be realized, however, various portions of the Green and Colo- that such a statement can apply only to rado drainage basins. These workers areas of uniform characteristics. began by restricting their attention to Experimental data indicate that surface runoff from storms, separating changes in land use have a greater out base flow. The sediment load ob- effect on sediment yield than on either tained from daily averages of suspend- total runoff or runoff intensity (Leopold ed sediment was correlated with mean and Maddock, 1954,p. 81). Yet it must daily discharge during the passage of be admitted that available data do not individual hydrograph rises. Using forty permit quantitative generalizations such flood occurrences, a relation be- about the effect of human activity on tween sediment load and daily dis- landscape degradation. Both cultivation charge was derived. For a given dis- and grazing have, without question, for charge the sediment load was then cor- a time increased sediment yield over related with vegetal cover types on the that obtaining in the natural or original watershed to which approximate values condition, but the amount is variable of cover condition had been assigned. and highly dependent on local condi- It was found that there was good agree- tions. ment between the estimates of sediment This cursory description of attempts yield so derived and estimates based to generalize relations of geology, to- on a subjective classification map of pography, vegetation, and climate to erosion conditions compiled from gen- sediment contribution can do no more eral field observation. The same au- than indicate the complexity of the thors studied six small drainage basins problem. All the methods used are, which had different vegetal covers un- basically, forms of correlation between der varied land use. The watersheds observed sediment yields and several were mapped and categorized by sub- controlling factors. In any such corre- jective field observations which at- lations an unexplained variance re- tempted to take into account vegetal mains, and this margin of error may be cover, erosion, soils, slope, and other quite large. It is clear, therefore, that factors. Sediment measures so derived any attempt to estimate the change in were compared with analyses based on sediment yield resultin from a change suspended-load sampling in the Boise % of the controlling varia les depends for River Basin. validity on the relative magnitude of Rosa and Tigerman made further the anticipated consequences of and comparisons with measurements on the the error inherent in describing the amount of sheet erosion from i d l t r o - original condition. meter studies where water is sprinkled The preceding discussion dealt with onto plots varying in size from 12 by the problem of ascertaining the present 30 inches to 6 by 12 feet. They con- rate of sediment production from nat- cluded (ibid.,p. 17) that if all water- ural watersheds. To summarize, one of sheds could be improved from fair to a the most satisfactory methods of meas- good condition [of vegetal cover], sedi- uring sediment yield consists of suc- mentation rates might be expected to cessive measurements of deposition in be reduced to about one-half of the reservoirs adjusted for outflow of sedi- present rate from large drainage basins. ment on the basis of suspended-load .. . If it were possible to restore all measurements. Such measurements are poor watershed areas to a good condi- available on only a small number of tion the future sedimentation would be streams relative to the total number in 644 Mans Role in Changing the Fuce of the Eurth the continent. The values of sediment greater importance than sheet erosion yield may vary markedly even between by reducing channel storage of runoff basins which superficially appear simi- water and by the physical dissection lar. Some of this variation can be quan- of arable land. It is generally believed titatively accounted for by differences that sheet erosion is more important, on in type and condition of plant cover, the average, as a sediment source than soil, slope, and other factors. This vari- is gullying. ability, however, causes most estimates No extensive comment is necessary of sediment yield under virgin condi- on the effects of reservoir sedimenta- tions to be quite imprecise. It is diffi- tion. The recent survey of sediment de- cult, then, to know how much reliance posits in Lake Mead showed that in may be placed on the computed values the first fourteen years of operation of sediment yield under virgin condi- sediment deposits comprised 5 per cent tions. Subject to this error, the magni- of the reservoir capacity below spill- tude of which is unknown, the esti- way-crest elevation. The sediment mates available indicate that in the weight is computed to be about two areas for which studies have been made billion tons (Could, 1951). A particu- human activity has increased sediment larly interesting result of this survey production from as little as twice to as was the information that about half of much as fifty times the original value. the weight of sediment deposit, or 64 These figures are meant only to indi- per cent of the volume, consists of fine- cate orders of magnitude. grained material transported by turbid- ity currents. This indicates the impor- EVALUATION tance of the fine-grained portion of the With this background in mind, let us total load. Again, we can merely specu- examine some of the over-all implica- late on the question of whether soil tions of changes in sediment yield. erosion which results primarily from The first and most obvious economic human use would result in increased reason for an interest in sediment yield or decreased percentage of a particular relates to erosion on the land. So ex- size fraction of the load. tensive is the literature on this subject The Lake Mead survey provides a that no review is attempted here. In specific example of the difficulties in in- the present context the rate of sediment terpretation of reservoir accumulation removal from a watershed should not data. The allocation of the sediment be assumed to be in direct ratio to loss to various portions of the upper Colo- of land productivity. Crop yield as it is rado Basin can be made only roughly, affected by soil removal is also distinct and it is virtually impossible to ascer- from loss of irreplaceable topsoil. tain what percentage of the measured Baver (1950) provided a commendable sediment yield can be attributed to way of thinking about the erosion effects of land use. Methods such as problem when he indicated that some those described earlier represent the topsoil is replaceable. The seriousness only available bases for estimating this of a given amount or rate of erosion de- quantity. pends on the thickness of the regolith, the kind of rock from which it is de- RELATION OF CHANCES IN SEDIMENT rived, and the profile characteristics- LOAD ON RIVER CHANNELS in other words, on many local factors. The Iiterature on rates of reservoir That soil erosion tends to reduce soil sedimentation is extensive. The eco- productivity is not disputed. Gully ero- nomic aspects of this problem are sion may in many places be of even patent. Land Use and Sediment Yield 645 I wish to direct attention to an aspect picture. The level area bordering a of the effects of sediment yield which stream is built by the stream itself and is less well known and more speculative at such a level that it is overflowed dur- than the problems of accelerated ero- ing high stage. Of greatest interest is sion and reservoir sedimentation. This the concept that the frequency of such is the change in stream channels pro- overbank flow is essentially constant duced by change in sediment yield. The for small rivers and large ones in the river channel is constructed by the same basin and between rivers of dif- river itself. The channel system is the ferent basins (Wolman, 1955; Wolman route by which runoff and erosion and Leopold, 1956). This similarity in products are carried from the land to frequency of overflow of the flood the ocean or to some intermediate plain, which in essence is also the fre- basin. As such, it is logical to suppose quency of the bankfull stage of the that any channel system would be of river, is a consequence of the charac- such configuration and size that it is teristics of sediment load and sediment capable of performing this function. action in flows of various magnitudes. Considerable speculation has been di- Small flows carry small sediment loads rected at the question of how efficient and are essentially ineffective in scour the channel net is for this function. and deposition. The greatest floods are Natural channels generally have a larg- the most effective in shaping the chan- er width-to-depth ratio than a semi- nel and altering existing shape, but circle, which is known to be the most these extreme flows are so infrequent efficient hydraulic cross-section for dis- that, in the long run, they are less charge of water. The fact that natural important than the lesser floods. The channels carry erosion products, as well level of the river flood plain is, there- as water, appears to be the underlying fore, controlled primarily by floods of cause of observed channel shapes. such magnitude that they are capable Increasing attention recently has been of significant erosion and deposition but devoted to the problem of explaining still frequent enough to have cumula- river-channel characteristics. Studies of tive effects of importance. This combi- channels in general led to the conclu- nation appears to characterize 00ws of sion that a quasi-equilibrium tends to that magnitude which recur about exist between the discharge and sedi- twice each year (Wolman, 1955; Wol- ment load emanating from a drainage man and Leopold, 1956). basin and the natural channel which This apparent consistency in the re- carries these products (Leopold and currence interval of bankfull floods in Maddock, 1953). Detailed study of a combination with the concept of a channel system of a single drainage river channel in quasi-equilibrium lead basin confirmed this generalization and to a provocative hypothesis: If a change demonstrated that such quasi-equilib- occurs in the relation of sediment yield rium tends to characterize small to water discharged from a drainage headwater tributaries in youthful to- basin, forces exist which would, over a pography as well as the major stream long period, tend to readjust the height channels ( Wolman, 1955). A generally of the flood plain, so that the frequency similar tendency for quasi-equilibrium of the flood stage would remain con- was shown to typify even ephemeral stant. If activities of man, therefore, headwater channels and rills in a semi- tend to increase markedly the sediment arid area (Leopold and Miller, 1956). yield relative to discharge characteris- The river flood plain is a particularly tics of a drainage basin, the river chan- important feature in the equilibrium nel will, given sufficient time, adjust its 646 Mans Role in Changing thu Face of the Earth channel in such a manner that floods trend can probably be expected to con- over the flood plain will recur at about tinue at least until the best reservoir the same frequency which originally sites have been utilized and for as long prevailed. as there remains economic justification This concept has its first and primary for hydroelectric and irrigation develop- application to the field of flood con- ment. trol through land management. Pro- Projects are considered justifiable, grams for land-use improvement gen- under present laws, if the computed erally anticipate marked reduction in benefits exceed the costs. Most projects sediment yield from a drainage basin. will yield benefits equal to costs during It should be expected that a conse- their economic life, but there will come quence of this reduction of sediment a time when great lengths of major would be a channel readjustment. This river valleys will consist of reservoirs readjustment may be such that over- more or less filled with sediment. When bank floods do not, in the long run, that time comes, the problems of water occur any less seldom than originally. control and of water use will be of a However, mans work directly on distinctly different character from those river channels has been and probably which concern us today, though this will continue to be a far more impor- will not occur until several generations tant determinant of future channel con- hence. ditions than the natural operation of SUMMARY river mechanics in response to mans changes on the watershed. It is prob- In summary, then, we may conclude able that long before the effects of the that mans use of the land can have latter can occur, river conditions will a marked effect on sediment yield. Be- have been so altered by dams that the cause of the difficulties of measurement latter will be the primary factor in con- of the initial conditions, it is extremely trolling river-channel characteristics. difficult to evaluate quantitatively this The degradation of the channel of the effect. Although increased erosion af- Colorado River after the construction fects soil productivity, this effect is in- of Hoover Dam is a well-known exam- fluenced by many variables in the dy- ple of one type of change. There will namics of soil formation. The effects probably be extensive changes of a of high sediment yields on reservoir ca- more subtle nature distributed widely pacity are well known and have obvious over rivers in this country as the dams, economic implications. Less well known already planned, are built. Bank-cut- are the effects on river channels of ting, channel-shifting, and other effects changes in sediment yield. The present not so obviously connected with reser- trend is toward ever increasing num- voir construction as bed degradation bers of dams on the rivers of the United should be expected. In the Mississippi States. The effect of these structures on Basin alone ninety-six new dams are changes in the channels greatly over- contemplated even at this time (Leo- shadows the effects due to varying pro- pold and Maddock, 1954). This figure portions of sediment to water produced indicates the trend in river work. This by mans use of the land. REFERENCES BAVER,L. D. BRUNE, G . M. 1950 How Serious Is Soil Erosion? 1948 Rates of Sediment Production in Proceedings of the Soil Science So- Midwestern United States. (US.Soil ciety of America, 1950, pp. 1-5. Conservation Service, SCS-TP-65.) Land Use and Sediment Yield 647 Washington, D.C.: Government Print- Survey Water Supply Paper NO. ing OfFice. 40 pp. 1373.) (In press.) CLARK, F. W. LEOPOLD, L. B. 1924 The Data of Geochemistry. (US. 1951 Vegetation of Southwestern Wa- Geological Survey Bulletin No. 770.) tersheds in the Nineteenth Century, Washington, D.C.: Government Print- Geographical Reuiew, XLI, 295-316. ing Office. 841 pp. LEOPOLD, L. B., and MADDOCK,THOMAS, COLBY, B. R.; HEMBREE, C. H.; and RAIN- JR. WATER, F. H. 1953 The Hydraulic Geometry of 1956 Sedimentation and Chemical Stream Channels and Some Physio- Quality of Surface Waters in the Wind graphic Implications. (U.S. Geologi- River Basin, Wyoming. (U.S. Geo- cal Survey Professional Paper No. logical Survey Water Supply Paper 252.) Washington, D.C.: Government No. 1373.) Washington, D.C.: Gov- Printing Office. 56 pp. ernment Printing Office. (In press.) 1954 The Flood Control Controversy. COUES, ELLIOT New York: Ronald Press Co. 278 pp. 1893 History of the Expedition under LEOPOLD, L. B., and MILLER,J. P. the Command of Lewis and Clark. 4 1954 A Postglacial Chronology for vols. New York: Harper & Bros. GLYMPEI, L. M., JR. Some Alluvial Valleys in Wyoming. (U.S. Geological Survey Water-Sup- 1951 Relation of Sedimentation to Ac- celerated Erosion in the Missouri Riv- plv Paper No. 1261.) Washington, er Basin. (US. Soil Conservation D.C.: Government Printing Office. Service, SCS-TP-102.) Washington, 90 PP. D.C. : Government Printing Office. 20 1956 Ephemeral Streams: Hydraulic Factors and Their Relation t o the PP. Drainage Net. (U.S. Geological Sur- GOTTSCIIALK,L. C., and BRUNE, G. M. 1950 Sediment Design Criteria for the vey Professional Paper No. 282A.) Missouri Basin Loess Hills. (US. Soil Washington, D.C.: Government Print- Conservation Service, SCS-TP-97.) ing Office. 40 pp. Washington, D.C.: Government Print- ROSA, J. M., and TIGERMAN, M. H. ing Office. 21 pp. 1951 Some Methods for Relating Sedi- GOULD, H. R. ment Production to Watershed Con- 1951 Some Quantitative Aspects of ditions. (U.S. Department of Agricul- Lake Mead Turbidity Currents, So- ture Forest Service, Intermountain ciety of Economic Paleontologists and Forest and Range Experiment Sta- Mineralogists, Special Publication No. tion Research Paper No. 26.) Wash- 2, pp. 34-52. ington, D.C.: Government Printing HADLEY, R. F. Office. 19 pp. 1954 Reconnaissance Investigations on WOLMAN,M. G. Sources of Sediment in Southern Part 1955 The Natural Channel of Brandy- of Cheyenne Basin above Angostura wine Creek, Pennsylvania. ( U S . Geo- Dam. (Unpublished report for the logical Survey Professional Paper No. Bureau of Reclamation, U.S. Geologi- 271.) Washington, D.C.: Government cal Survey, open file.) 31 pp. Printing Office. 56 pp. KING,N. J., and MELIN,K. R. WOLMAN, M. G., and LEOPOLD, L. B. 1956 Sediment Accumulations in Small 1956 River Flood Plains: Some Obser- Reservoirs, in COLBY,B. R., et al., vations on Their Formation. (U.S. Sedimentation and Chemical Quality Geological Survey Professional Paper. of Surface Waters in the Wind River Washington, D.C.: Government Print- Basin, Wyoming. (US. Geological ing Office. (In press.)