By C. R. HURSH
Chief, Division of Forest Influences, Southeastern Forest Experiment Station, Asheville, North Carolina, U.S.A.
Less is understood about the management of water than of any other forest resource. Because man cannot alter climate to any great extent and because climate is associated with water, it is a common belief that he can do nothing about managing natural water supplies. This concept is highly erroneous and misleading.
Through his use of the land man can either facilitate the control and storage of rainfall in the soil, or create soil conditions that cause rapid runoff associated with damaging floods and thus waste the water that would otherwise be available for his use. The aim of water resource management is to control and husband the precipitation that falls on the land.
High-yielding streams of superior water quality are universally associated with undisturbed forested land. This is because the soil structure and porosity are favorable to infiltration, maximum storage of water, and control of erosion. Yet this favorable influence of forests on water yield has repeatedly been questioned by hydrologists and water supply engineers on the ground that trees require more water for transpiration than do other plants, thus reducing the total amount of water that will be available on a watershed for stream flow. Obviously, both concepts are reasonable. One of the practical considerations in resolving this paradox is how to maintain favorable soil conditions and at the same time reduce transpiration draft.
Observational Data Needed
The science of hydrology, dealing with the origin and distribution of water as a natural resource, is still a relatively undeveloped field. The application of the principles of hydrology to the management and use of land is an entirely new field that is scarcely known even to many technicians in forestry and agriculture.
The chief reason, of course, why so little is known about land use-streamflow relations is the lack of experimental data. There are very plausible explanations for this situation. Streamflow studies must be carried out on a drainage area basis, and experimental watersheds are expensive to install and difficult to operate. Many years of continuous, accurate records are required before acceptable results can be obtained. Trained technicians are needed to maintain the necessary recording instruments, and to analyze and interpret the records. Also, the best methods and techniques for watershed research in different climates have not yet been developed or tried out, and much exploration and orientation of the problem will still be required. The same is true in selecting suitable experimental areas that will meet requirements with regard to geology, soils, topography, and rainfall distribution. For these reasons, only a limited number of intensive experimental watershed studies have ever been undertaken throughout the world, notwithstanding the fact that the need for watershed data has long been recognized.
Past Studies
Among the important experiments that have added to the world's knowledge of watershed management, one of the earliest was the Emmen Valley study instituted in 1900 by the Forest Research Institute of Switzerland. In 1909, the Wagon Wheel Gap studies in Colorado, U.S.A., were organized by the U.S. Forest Service in co-operation with the Weather Bureau. Present-day studies to find out the effect of forest vegetation on stream flow are under way in Java and in South Africa.
In the United States of America, watershed studies are being carried out by the U.S. Forest Service in the States of California, Utah, Colorado, Pennsylvania, West Virginia, North Carolina, and elsewhere The Soil Conservation Service has developed an intensive watershed research area at Coshocton, Ohio, and has conducted watershed studies in other sections to find out the effect soil erosion control practices on farm laud. A number of engineering schools have also used experimental watersheds to obtain runoff formulae for use in the design of storm sewers, culverts, and bridges. Much general information has been gained from these experiments as to the basic requirements for studying land use effects on stream flow. They also indicate that certain physiographic regions are more favorable than others for studying fundamental hydrologic processes.
Requirements and Selection
To evaluate the hydrologic principles basic to practical watershed management in the high rainfall belt of the Southern Appalachian Mountains, a watershed laboratory site was located in the southeastern United States by the U.S. Forest Service in 1931. The selection of this field laboratory was made only after careful research to find an area that would meet rigid specifications designated by hydrologists, engineers, and foresters interested in watershed research. In 1933, intensive water resource management research was begun at this 5600-acre* Coweeta Hydrologic Laboratory in western North Carolina.
*Conversions: 1 inch = 2.54 cm.; 1 foot = 0.3048 m.; 1 sq. mile = 2.59 km ²; 1 acre = 0,4047 ha.
Lying at elevations of 2,200-5,200 feet, the topography of the Coweeta Laboratory is particularly suited to the experimental study of small drainage basins that meet all requirements of independent hydrologic units. Slopes are steep, and sharp-crested ridges form natural boundaries for the many small watersheds ranging in size from 25 to 200 acres.
The entire area is underlain with deeply weathered Archean granite formations, principally of schist and gneiss. There is no indication of any quantity of seepage escaping measurement. The soils are relatively deep and porous. Rainfall averages about 72 inches (1,830 mm.) per year and is fairly evenly distributed throughout all the months of the year. Because of this pattern of precipitation and the large number of storms per year, it is possible to obtain experimental results in a relatively few years, as compared with regions having less rainfall.
The entire area has been little influenced by land use occupancy. A dense mixed hardwood forest is dominant, with scattered pines on old fields or occasionally on the ridges. Some scattered hemlock is found along the streams. Although about 60 percent of the area had been cut over a quarter century or more before the federal government acquired ownership, this part of the Coweeta now supports a second-growth forest, with the remainder still in old growth. Before being killed out by the blight, chestnut (Castanea dentata) was the major species. Now about 80 percent of the Experimental Forest supports oak-hickory stands. Fifteen percent is in cove hardwoods: yellow poplar (Liriodendron tulipifera) and northern red oak (Quercus borealis). Hemlock (Tsuga caroliniana) occurs with this type along the streams. Five percent is in northern hardwoods: sugar maple (Acer saccharum); yellow birch (Betula lutea); beech (Fagus grandifolia). Pitch pine (Pinus rigida) is also among this five percent but occurs only at lower elevations.
The forest is three-storied: large trees form an upper layer, small trees and large shrubs the second, and shrubs or herbs cover the ground. The dense understory of laurel (Kalmia latifolia) and rhododendron (Rhododendron maxima), present on some slopes, may be the result of fires during the last century.
Scope and Objectives
The long-range plan of research for the Coweeta Laboratory is directed toward four related fields:
1. To establish fundamental forest and water relations.2. To furnish coefficients of runoff, infiltration, and water storage for different types of land use such as grazing, logging, and "mountain farming."
3. To develop principles for silvicultural management that will provide for maximum supplies of usable water of highest quality.
4. To work out feasible and practical methods of stabilizing the soil in logging operations, on road banks, and along streams and reservoir shore lines.
Technics and Methods
At present, the water behavior of 28 watersheds is being studied. Of these, 16 are now or are to be treated experimentally, and 12 are held as control units and for future experimentation.
Different types of stream-gaging controls are required for different size watersheds, to measure runoff accurately over a wide range from low flow and the maximum storm discharges. Each stream control is designed to meet the specific requirements of the experiment. Sharp-crested notched weirs, blades with a 45-degree down-stream bevel, are used for most small watersheds of less than 100 acres. Rectangular and Cipolleti blades of several sizes are used for larger drainages. Where the most extreme range between minimum and maximum flow occurs, a deep-notched ogee cross-section cement weir is used. This weir is thought to have the added advantage that it will pass sediment and rock debris without clogging the notch.
The behavior of ground water is being studied by means of 26 observation wells in which movements of the water table are recorded by automatic instruments.
Precipitation is measured at 75 rain gage stations after each significant storm. These stations have been located with regard to aspect and elevation in such a manner that no two adjacent stations will differ from each other by more than a 5-percent catch. Ten rainfall intensity recording instruments are also maintained. Total rainfall distribution over any one watershed is estimated on the basis of the Theissenmean area method. This method calls for accurately locating each rain station on a carefully drawn map. Polygons formed by perpendicular bisectors of lines connecting adjacent stations circumscribe the watershed area best represented by the single station enclosed.
Calibration of Watersheds
The over-all research program followed for the Coweeta Hydrologic Laboratory called first for taking continuous hydrologic measurements of all the major watersheds and their tributaries for a period of five to ten years under the natural forest cover. This was considered a period of calibration and standardization during which the streamflow characteristics of each watershed are established. Changes in the vegetative cover are then carried out under carefully controlled experimental procedures, while the records of runoff and rainfall are continued in the same manner as before. Uninterrupted records of the continuous water economy of the experimental watershed are maintained both before and after changes are made in the vegetation cover. The effects of changes on the watershed can then be statistically compared in terms of storm hydrograph characteristics or in terms of changes in monthly yield and water quality. Annual climatological differences are adjusted for by maintaining one or more control watersheds under natural conditions permanently, to be used as reference cheek areas.
On the average, about 50 significant storm hydrographs suitable for detailed analysis were obtained each year. Furthermore, the pattern of seasonal yield by months soon fell into a rather uniform pattern on the basis of only a few years' records.
As the Coweeta projects developed, it was found that due to particularly favorable rainfall, soil, and topographic conditions for watershed studies, the solution of certain fundamental problems of stream behavior on forest land was possible even before the period of watershed calibration had been completed.
During the period of calibration it was found that in short intense storm periods streams rose rapidly - almost as rapidly as streams from relatively impervious, non-forested watersheds. This phenomenon in the past has been interpreted as indicating that there is inadequate infiltration of rainfall into forest soil and that overland storm flow does take place, despite claims to the contrary, from the soil of heavily forested hardwood areas. Observations in the field, however, during such storms on the Coweeta watersheds indicated that there was no accumulation of storm water at any place on the watershed, and consequently there could have been no overland storm runoff.
Further investigation showed that these rapid rises in streams from forested areas are due almost entirely to rain that falls directly into the stream channel, and are not due to lack of infiltration into the forest soil. By establishing the exact contribution of channel precipitation to runoff on a watershed basis, it then became possible to separate out this component of the storm hydrograph for further analysis and interpretation of stream flow as it relates to watershed conditions
Interpretation of Forest Hydrographs
A similar analysis was made of the storm hydrograph from forested areas during periods of long-continued rains. In this ease the storm hydrograph shows a considerable increase in stream flow that obviously cannot be accounted for as channel precipitation alone. Neither can it he assigned to overland storm runoff caused by lack of infiltration into the soil. On the Coweeta watersheds an analysis of the movement of shallow water tables as recorded in observation wells showed that rainfall first entered the soil and then moved laterally through the porous upper soil layers to the stream channels in time to contribute to the storm hydrograph. Thus it was shown that the storm runoff hydrograph was produced both by channel precipitation and a form of subsurface flow, the nature of which had not, until then, been explained clearly. This latter type of storm runoff is intermediate between overland storm flow and ground water seepage, and has been called subsurface storm flow. It is an important factor in runoff from forest land and takes place only where the surface soil is extremely porous and has many large openings due to decaying roots, earthworms, or insect and rodent burrows. Also, a portion of this flow probably moves through old drainage lines in buried ravine fills. Thus the storage of large amounts of rainfall by the porous forest soil mantle is analogous to the regulation of storm water by conservation-detention dams which impound only the water that would be in excess of the stream channel carrying capacity. For Coweeta this detention storage is estimated at about 7 million cubic feet per square mile (77,000 m³ per km²).
Annual Water Balance
Still another study relative to normal stream flow under control of natural forests relates to the annual balanced water economy of small watersheds. It was found that for the humid Coweeta climate, the precipitation recharge for any one year can be accounted for in measured terms of runoff, soil moisture, ground water stored on the watershed, evaporation, and transpiration. The water balance of outflow and inflow is expressed by the expanded equation: precipitation equals runoff, plus or minus free ground water in storage on the watershed, minus evaporation, minus transpiration, plus or minus current field moisture retention. 1 Balances so far have been made on the basis of monthly invoices, but it is recognized that for certain purposes water accounts should be kept storm by storm, and this will be the ultimate objective of the study.
1 This is made up of the capillary film of water surrounding soil particles and is retained against the force of gravity. Consequently, it does not contribute to stream flow, although a portion of it can be used by plants.
Ordinarily the water year is based on a 12-month period, starting the first of October or November. Past studies have shown it to be difficult to strike a water balance on this basis. The time period most favorable for striking such a balance had been anticipated on theoretical grounds as being the spring, but experimental proof was lacking because accurate, continuous records from suitable small watersheds were not available. An analysis of Coweeta records showed that the approximate 12-month period between two spring dates of maximum field moisture storage does furnish the most favorable interval for making the annual water balance. At this time the soil is at field capacity and the factor of field moisture, which is difficult to measure on a watershed basis, is eliminated from the primary equation of inflow = outflow ± storage.
Measurements were also made to find out the amount of rainfall is lost to the forest floor Because of interception by tree canopies. When corrected for the amount of rain that reaches the forest floor by flowing down the tree stems, this was found to be from 12 to 18 percent of the total annual rainfall.
These studies carried out on natural forest watersheds during the period of calibration provided a sounder basis for evaluating the effects of watershed treatments subsequently carried out at Coweeta.
Forest vegetation creates a porous soil which admits and stores large quantities of water, but at the same time the trees use considerable water in their growth. These two facts have caused a great deal of speculation as to the role of forests in the regulation of stream flow. Some authorities have advocated the clearing of forests to increase water yield by eliminating transpiration, while others have maintained that to remove the forest would lead to immediate disaster. Water would rush off over the soil surface as a flood, leaving little to be stored underground for sustaining dry-weather flow. There is some truth in both these extremes, and the job of the watershed manager is to reach an effective compromise between them. In areas where maximum yields of water are important, the task is to reduce the forest stand enough to minimize transpiration but to leave enough trees to perpetuate a favorable forest soil. On the other hand, where the need is to provide maximum storage in the soil for rains that may otherwise produce floods, it would be more desirable to encourage dense vegetation that will draw heavily on soil moisture.
Because of the limited number of standardized watersheds available, regional importance of specific problems has established the priority of treatments carried out at the Coweeta Laboratory. The following watershed problems were agreed upon by representative land administrators and research directors of the region:
1. Effect on stream flow of cutting forest vegetation.a. Complete removal of all major trees and shrubs, followed by annual sprout-cutting of all woody plants.b. Complete removal of all trees and shrubs, followed by natural regrowth.
c. Effect of cutting only stream bank vegetation.
d. Effect of cutting dense rhododendron and laurel understory only.
Effect of Cutting and Subsequent Annual Removal of Sprouts
This experiment was designed as an orientation study to find out what would be the maximum effect on streamflow yields of cutting all forest vegetation. After the period of calibration, all tree and shrub growth was cut on one watershed of 33 acres. The cutting was done with a minimum disturbance to the soil. The logs and slashings were all left on the ground in order not to disturb the surface. Because of the large amount of cut plant material lying on the ground, evaporation from the soil and from the tree branches remained about the same as before cutting. The first summer after cutting, every sprout v as slashed back, and this sprouting was continued each successive year. The effect on water yield was spectacular. Although there was no increase in storm peaks or stream turbidity, the yield of high quality water increased 65 percent in terms of annual runoff, during the first year after cutting. This is equivalent to a volume of water 17 inches deep over the 33-acre watershed, or approximately 15 million gallons. From one square mile of watershed this increase would be enough to supply a town of about 8,000 people for one year. During the later summer months when water is most valuable, the increase in usable base flow of the stream amounted to 100 percent.
This experiment will be continued by cutting back all annual regrowth until more has been learned about changes that take place in the soil under the present ground cover of herbaceous species, low briars, and vines that have invaded the area. The original additions of slashings and logs naturally increased the amount of organic material on the soil surface at first, but after ten years it is rapidly disappearing. The problem now is whether it will be possible to maintain a soil under the present plant cover that has the same infiltration and storage for rainfall as the original forest soil possessed.
Effect of Cutting and Natural Regrowth
On another watershed of 40 acres a quite similar cutting was carried out, but in this ease the forest was allowed to come back through sprouting and natural regrowth. The increased water yield the first year after cutting was the same as in the above experiment, but subsequently there was an increasing transpiration each year and a relative decrease in water yield. After nine years of regrowth the forest is about 30 feet high, but the increase in yield for the watershed still amounts to about 25 percent annually.
Effect of Cutting Riparian Vegetation
On still another watershed only the trees and shrubs close to the stream channel and its associated high water tables were cut and slashed to the ground. The cut was confined to woody vegetation within 15 vertical feet of the streambed. This was all done during the course of a few days in midsummer, at a time when the stream flow was exhibiting a very definite diurnal fluctuation. The cutting eliminated this fluctuation - indicating that it had been caused by transpiration draft. About 12 percent of the total area of the watershed was cut. The maximum daily increase in yield was about 20 percent. The annual increase was less than 10 percent.
Effect of Cutting Ericaceous Understory
A dense understory of Kalmia latifolia and Rhododendron maxima is frequently associated with mountain hardwood stands. This understory is often 10 to 20 feet (3 to 6 m.) tall and is made up of stems that will sometimes reach a diameter of 9 inches (23 cm.) at breast height. The foliage is evergreen and waxy on the surface. It is believed, however, that these plants transpire considerable quantities of water throughout all seasons of the year. Consequently, in making orientation studies of the amount of water used by different types of vegetation, a watershed experiment was carried out in which a dense understory of Kalmia latifolia and Rhododendron maxima was completely cut to the ground over a drainage of 70 acres. The results show a change in water yield, but the magnitude of this change has not vet been ascertained.
Although the preservation of an undisturbed forest cover is probably the best possible watershed management practice for flood reduction and erosion control all land cannot remain in forest. Naturally, in semi-humid and humid regions land must be cleared of forest growth for cultivated crops and improved pasture. The problem becomes one of how to use the land and at the same time maintain a soil structure that will favor infiltration and water storage The problem of watershed management is to aim at maintaining a soil structure that resembles as nearly as possible the soil under the original natural forest This is not a simple problem. For the most part, the clearing of level land has not created important watershed problems. But the clearing and use of steep land for crops, done largely on a trial and error basis, has created many problems and given rise to wide differences of opinions as to how such land can be safely cultivated. Also, there have been conflicting ideas as to the degree of damage done by woodland grazing, logging, and road construction. The effects of these practices on water yield and quality have been investigated on an experimental watershed basis at Coweeta.
Effect of Steep-Land Farming
Although many millions of acres of steep forest lands have been cleared and used for crops and pastures, there is very little precise experimental data on how stream flow is affected by this type of land use. To find out the effects of one very extensive type of mountain farming, a 23-acre Coweeta watershed was first calibrated under forest for a period of seven years. The forest was then cleared away in 1940, and the land area has since been maintained as a mountain farm. The farming methods were those followed on similar land in the locality of the Coweeta Laboratory. About one-third of the watershed was cultivated in maize and the remainder was kept as an unimproved pasture. Half the pasture was so steep it was used very little by the cattle and grew back rapidly in trees and shrubs. During the first two years of cultivation there was practically no change in the amount of storm runoff from the watershed, and soil losses were practically the same as before clearing. These results were surprising to many persons who had contended that removal of the forest on such steep land would induce landslides and erosion with the first heavy rain. The reason why erosion did not take place was simply that the loose friable forest soil structure which characterized original forest conditions remained unchanged for the first two years, although all the trees had been cleared away. However, beginning with the third year, the soil aggregates began to break down into single grains as the organic material disappeared. The surface began to be sealed over, infiltration and storage were reduced, and storm water began to accumulate and run down the steep slopes. Accelerated erosion has progressively increased since the third year after clearing.
When the watershed was in forest cover, the sediment carried by the stream was insignificant. For the period of 4 May to 8 September, 1943, four years after clearing, the sediment measured in the catchment basin averaged 768 pounds per day. During one storm in 1949, 152,000 pounds (69,000 kg.) of soil and rocks came off this mountain farm in 65 minutes of storm flow.
Maximum flood peaks for average storms increased about ten times after this watershed had been converted to a mountain farm. Before clearing, the storm peak was less than 30 cubic feet per second per square mile (0.33 m³ per second per km²), even for severe storms. The maximum was about 100 c.s.m. (1.1 m³/sec/km²) for exceptionally intense rains. After eight years of use as a farm, storm peaks of 200-300 c.s.m. (2.2-3.3 m³/sec/km²) are of common occurrence and the maximum recorded peak for a severe storm is 1,850 calm. (20.2 m³/sec/km²). Today the heavily trampled pasture portion of the pasture land is the principal source of surface storm runoff. Although making up only 20 percent of the watershed, this trampled pasture produces 60 percent of the storm runoff. It is apparent that the trampling by cattle reduced infiltration of rainfall more than maize cultivation did. Surface storm flow from this portion reaches the stream ahead of the runoff from the cultivated land and so produces a double peak in the storm hydrograph. Obviously, the clearing and cultivation of steep land as it was carried out on this watershed has been definitely unfavorable to water quality and to water control.
The results are generally applicable to average and poor farming practices in other areas of similar topography and soils. A better type of farming would call for rotating the corn with clover and small grains, and the regular use of more fertilizer. On the pasture, fencing and rotation grazing of a more restricted number of cattle per acre would result in less trampling damage. However, it is often pointed out by visitors that more intensive farming practices would be difficult and perhaps unprofitable on such steep land, and the private owner would be better off to allow such areas to remain in forest.
Effect of Woodland Grazing
At one time livestock on forest laud produced the principal sources of cash income in the mountains adjacent to the Coweeta Hydrologic Laboratory. Widespread burning was practiced to reduce the density of the forest cover and favor the development of grass and sprouts. The advent of fence laws, elimination of fires, and decreased economic returns have reduced the importance of open ranging in the mountain hardwood forests. Today, however, grazing of small farm woodlands is still a very common practice. In fact, farm programs encourage more livestock, and although the trend is toward establishing improved pastures on good soil, the aggregate of grazed woodland is still considerable, occupying 20 percent of the land area in this county.
To find out the effects of farm woodland grazing on stream flow, an average of eight head of cattle were kept on a 145-acre watershed during the summer months for nine years. For this study it was necessary to definitely overtrample and overgraze the watershed in a manner comparable to the typical use of other farm woodlands in the locality.
The cattle were inclined to concentrate on the rich cove soils where the browse was most palatable. Comparisons of vegetation changes are made of 17 fenced plots and 17 unfenced plots. After seven years of grazing, au legumes have completely disappeared on all the grazed plots. The reproduction of the most desirable tree species has been eliminated in all the browsable range of plants from 3 inches to 15 feet (8 cm. to 4.6 m.) in height on the cove area. One hundred percent increase took place in the number of woody shrubs not palatable to cattle. After the first two seasons, the cattle were unable to maintain themselves on the browse remaining, and would have starved but for supplemental rations.
Although no observable changes took place in stream flow or stream turbidity during the first eight years of this experiment, there was from the start an obvious trampling effect and destruction of the forest litter wherever the cattle concentrated. It was not until after eight years of summer grazing that significant changes in runoff were discernible. The reason is that the forest litter collected in small ravines and depressions as it was blown by the wind from the sheet-eroded trampled areas. This litter filtered out the silt, enabling clear water to go into the soil before it reached the main watercourse. As the sheet-eroding and trampled areas increased in size, there was a gradual increase in the amount of storm water accumulation on the soil surface. As the amount of surface storm water increased, it finally gained sufficient force to carry away the litter plugs along natural drainage lines and develop uninterrupted drainage channels to the main stream. At the end of the eighth grazing season there existed for the first time an uninterrupted channel from the storm runoff source areas to the permanent stream channel. At once there was a sharp upturn in the amount of sediment carried and in the magnitude of the peaks in the stream hydrograph.
This experiment, though it purposely brought about overtrampling of the farm woodlands, showed the nature of gradual soil compaction and its significance in water control. In the early stages, trampling damage to the watershed may easily be overlooked. The insiduous accumulative effects were not conspicuous until they became of sufficient magnitude to create extremely unfavorable watershed conditions. Further studies would be required to find out the number of cow days of grazing per season that would be possible on this watershed without serious damage to water values. Although the Coweeta studies are too limited to permit sweeping conclusions, the results point to the fact that there is insufficient herbaceous forage at best in dense hardwood stands to make very profitable grazing. Also, it is apparent that cattle damage the young growth of desirable tree species and thus reduce the quality composition of the stand. Where watershed values are at stake, there appears to be every reason for the complete elimination of grazing from hardwood stands.
Effects of Logging
In the United States of America the present tendency in mountain logging is to reduce the amount of ground skidding by taking the hauling trucks to the job. As a result more and more woods roads are being pushed directly to the timber operation by bulldozers, without the usual careful planning given to most road construction. The truck trails are not well located or built to provide for adequate drainage. As a result, active erosion takes place, and considerable sediment is carried into the streams.
At the Coweeta Hydrologic Laboratory a 212-acre watershed was cut and logged by a local contractor using the usual methods with which he was most familiar. At first, logs were skidded down the natural drainage channels by horses, in keeping with common logging practices. Later a 2.3-mile truck road was bulldozed in to the operation in a location chosen by the contractor.
The effects of logging were soon perceptible in deteriorated water quality. During logging the stream flow carried an average turbidity of 95 parts per million, compared with 4 parts per million for an uncut control unit. Flow from the logged area ran as high as 6,000 parts per million during the most intensive storms, in contrast to a maximum of 120 parts per million for the control unit.
Before logging, sediment carried by the streams was largely organic material that settled out rapidly and was not damaging to the fish habitat. In contrast, the finely divided silt and clay particles carried in the stream after logging had a very low settling rate, and such sediment can be removed only by costly filtration. This type of sediment is also very injurious to game-fish habitat tending to smother fish eggs and alter feeding conditions at the stream bottom.
Not only was water quality from the 212-acre logged area seriously impaired, but all the water from an 1,880 acre drainage basin into which this small tributary flowed was also damaged. Formerly, the larger stream had run clear; during and after the logging job it would have required filtration for municipal and industrial use. During one storm, the main clear stream had a turbidity of 25 parts per million above the entrance of the logged tributary stream, which was carrying a turbidity of 1,200 p.p.m. Below, with a total drainage of 1,880 acres, the main stream carried a turbidity of 395 p.p.m.
Assuming that the water from the 1,880-acre area were to be processed for domestic use and that a filtration plant were available, the total returns of $2,208 for selling wood products on the 212 acres logged would carry the increased processing costs for only a 490-day period, or 20 percent of the time that active logging was under way.
These results bear out observations in other parts of the country, that logging is a primary cause of erosion and sedimentation of forested mountain streams. Operators most familiar with the problem now believe that the efficient planning of a logging job, better road location, better construction and drainage, and better road maintenance will result in savings of time and equipment to the operator as well as in reduction or elimination of filtration costs to water users. Observations to date indicate that efficient logging practices including road construction will meet the essential requirements for the protection of watershed values during timber operations.
Experimental watershed studies provide basic data that will lead to a better understanding of the ways in which forests function in the control of water and the conservation of water resources. A knowledge of these functions is necessary in order that foresters and land managers may effectively discharge their responsibilities as custodians of water resources. Watershed studies at the Coweeta Hydrologic Laboratory have so far only indicated some of the possibilities that lie ahead in this field of watershed research.
The practical application of watershed research is pertinent to many sections of the world where the local commodity value of water far exceeds the total value of harvested timber and other wood products. Here the management of the forests must be directed essentially toward achieving maximum yields of high-quality water at those seasons of the year when shortages usually occur. In the management of municipal and industrial watersheds there exists a particular opportunity for integrating high-quality, long-rotation timber crops with the production of high yield of quality water. Timber management practices can also be directed toward achieving sustained summer flow from headwater areas supplying water for hydroelectric power production.
Throughout the world the water resource is still the unexplored field in forest management. Practically nothing is known as to the effects of trees on water yields in tropical countries. It is quite possible that the cutting of coastal forests in equatorial latitudes has significantly altered the rainfall and hydrology of the interior. In higher latitudes there are many problems to be answered. For example, a change from hardwood to conifer forest, where this is ecologically feasible, has been suggested as a possibility for increasing water yield. Plant physiologists have not all agreed that this is a reasonable hypothesis. This subject and many more related problems have yet to be studied on a watershed basis.
The practicability of protected forests as a best type of plant cover for erosion control has in general not been questioned. In a few rare instances even this conclusion may be wrong, for there are unusual soil conditions where natural land-slips on steep slopes are said to be more common under large trees than under low herbaceous cover. Thus it is difficult to generalize and to do so may be misleading in the solution of local problems. Much more needs to be found out for specific areas in different parts of the world.
The most pertinent of all facts known to date is that careless abuse and damage of the forest soil starts into motion degenerative processes which may be inconspicuous at first, but which can produce gradually accelerating changes in soil structure that ultimately result in damage to the water resource. The conversion of arable areas to desert is the classic example. How to use the land and still maintain a reasonable equilibrium between deterioration processes and rehabilitation processes is a local problem in various sections of the world. Understanding these physical processes is the first essential.
