H.L. TELLERDr. H.L. Teller of FAO's Forestry and Forest Industries Division prepared this position paper for a meeting of the Working Party on Watershed Management (European Forestry Commission) held in Romania in September 1967.
A flood may be defined as the occurrence of a flow of such magnitude that the natural banks of a waterway are overtopped and water flows over low-lying adjacent areas or flood plains. Floods are natural phenomena. Existing stream channels represent only the normal flow conditions and will, sooner or later, be overtopped by the unusual peak flow. By encroaching on the flood plain, man takes a calculated risk. Sometimes he pays the price of his occupancy.
To what extent can land management affect the frequency and magnitude of flooding and what role does the forest play in this process) This paper discusses some of the physical relationships involved and indicates the importance and the limitations of land management. Some experimental evidence is reviewed.
Runoff will occur from a watershed if the precipitation rate exceeds the infiltration rate and/or if the capacity of the soil to hold water is exceeded. The flow at any given point in a channel will therefore depend on the interaction of many factors, the most important of which are: distribution, intensity and duration of precipitation; vegetative or other surface cover; soil nature and depth; geologic structure; topography, including area, slope and channel characteristics.
Only two of these factors can be changed to any significant degree by man: the vegetative cover and, to a lesser extent, the nature of the soil. The others represent physical constraints which may have a completely overriding effect over man's efforts.
The role of forest vegetation in water conservation is well known. Leafy canopies intercept rain, reducing both the amount and the impact of that which does reach the ground. The roots stabilize soils and form channels for rapid infiltration. Organic matter from roots and leaves improves soil structure and increases both infiltration rates and water-holding capacity. Through transpiration, plants remove water from the soil profile, thus creating a greater storage capacity for future precipitation. The density and distribution of forests affect snow accumulation and melt. Through their ameliorating effect on soil temperature, forests reduce soil freezing and hence surface runoff. In general, then, a forest cover minimizes flood peaks by keeping infiltration to a maximum, thereby increasing subsurface flow at the expense of surface flow; sediment movement from catchment to stream channel is minimized.
The efficacy of these functions can be altered for better or for worse by man. Vegetation density or vigor can be increased or reduced, deep-rooted species can be replaced by shallow-rooted species (or vice versa); in general, species which are poorly adapted to the site or to the objectives of management can be replaced by more desirable ones.
Significant changes in the soil are perhaps more difficult to achieve in a reasonable length of time. Not much can be done about soil depth. Structure does change gradually with the addition of organic matter, but usually only the surface soil is affected. Subsoil permeability may sometimes be improved by mechanical means. Eroding slopes can be stabilized and sediment production can be decreased within a few years by the establishment of a plant cover on bare soil, provided there is a sufficient soil depth. Structural measures, such as contour trenching and terracing, are also designed to reduce surface runoff.
Good land management techniques for flood control attempt to achieve the above results by various means. Forest cutting is either prohibited or limited to selection fellings on steep, erosive slopes. Fire control measures are aimed at maintaining a continuous forest cover. Cultivation is carried out on the contour or on terraces. Grasses and trees are grown on denuded slopes. Overgrazing is prevented on mountain pastures. The techniques are, in the main, well known. Under what conditions are they effective in controlling floods?
For the past 50 years or so, a great deal of experimental evidence on the effect of land-use practices on floods has accumulated. Direct evidence of forest manipulation effects is more recent and less plentiful. Many of these studies were carried out on small catchments of only a few hectares, or on still smaller plots. Valid data from large watersheds is rare, due mainly to difficulties in the measurement of independent variables, and to the heterogeneous nature of soils, vegetation and land use. Furthermore, extreme climatic conditions are, of course, not encountered frequently, so that the unusual storm event which causes flooding is not often included in the study period. Some examples of forest cover effects on flood flows may be of interest here.
Harz mountains, Federal Republic of Germany - Runoff from a 15-hectare clearcut compared with that from three forested catchments of 87, 38 and 32 hectares. In some storms flood peaks were higher from the forest, in others from the clearcut. Peak flows from the forest were particularly high in winter, when snow melt was followed by rain. In summer, flood peaks were higher from the clearcut area (Delis et al., 1958).
Allegheny mountains, West Virginia, U.S.A. - Various intensities of cutting on four catchments, all less than 40 hectares. Effect of clearcutting on flood peaks was variable, depending on presence or absence of snow and antecedent soil moisture. Major effect of fully stocked stand on flood control was felt in the growing season and in the autumn recharge period. In dormant seasons after completion of soil recharge, effect of forest usually not significant (Reinhart et al., 1963).
Switzerland - Classical experiment on Sperbelgraben and Rappengraben. Flood peaks from 100 percent forested catchment were one third to one half lower than on 30 percent forested catchment. Area of each catchment less than 60 hectares (Burger, 1943).
Allegheny plateau, Ohio, U.S.A. - Reforestation of a 17-hectare watershed. Flood peaks during a five-year tree stage were compared to those during a six-year seedling stage. Flood peak reductions in the tree stage ranged from 52 percent in flows less than 1.6 m3/sec/km² to 84 percent in flows over 16 m3/sec/km² (Hill, 1960).
U.S.S.R. - A study of data from 212 weirs on 123 rivers, with drainage areas from 50 to 50,000 square kilometers, showed that the greatest influence of forests on flood flow was exerted at about 50 percent forest cover. With higher and lower percentage of forest cover, flood flows were increased (Molchanov, 1963).
Tennessee, U.S.A. - 700 hectares of deserted farmland were afforested. The volume of flood flow was not affected, but peak flows were only 15 percent of those prior to afforestation. Lag period increased from 1.5 to 8 hours. Winter floods were not materially affected. In another 35-hectare catchment in the same vicinity, reforestation and soil conservation measures resulted in 90 percent reduction of sediment after five years (Hoyt and Langbein, 1955).
Wasach mountains, Utah, U.S.A. - During the first 75 years of settlement at the base of several steep catchments, there were no major floods. After only 2 to 10 percent of various watersheds had been severely burnt and overgrazed, sediment deposits were greater than during the previous 20,000 years (Croft, 1962).
Cascade mountains, Oregon, U.S.A. A statistical (multiple regression) analysis of peak flows from 54 watersheds, ranging in area from 5.7 to 7,280 square miles (14.7 to 18,850 square kilometers), indicated that clear-felling one square mile of forest could be expected to increase flood peaks by 100 ft³/sec/mi² (1.1 m³/sec/km²) (Anderson, 1958).
While European land and water authorities have for many years recognized the importance of upper catchment management in flood control, much of the available literature on the subject merely gives an account of the measures adopted without relating these measures to results achieved. Although the effects of the check dams, reforestation and torrent improvement works are, of course, obvious in the localities where they have been carried out, a quantitative evaluation of the effect of these works is rarely available.
The importance which is attached to the protective functions of catchments, particularly by the Alpine countries, is indicated both by the legislation which has been adopted and by the large sums of money which have been expended. In France, for instance, a mountain protection law dates back to 1882 and, until recent years, some $3 million per year were spent on torrent control measures. In Italy, the first mountain watershed law was passed in 1910 and annual expenditure on reforestation and torrent control between 1933 and 1962 amounted to some $22 million per year. Significant reductions in the frequency of flooding have been reported from towns situated at the base of treated catchments.
The headwaters of both the Rhine and the Rhone contribute a large proportion of the runoff and sediment loads of these rivers, and vast sums have been spent there on flood control measures. However, much of the engineering work in the valleys was, and often still is, carried out entirely independently of torrent control and reforestation measures at higher altitudes. Nevertheless, significant improvements have certainly been achieved, although the contribution of such measures as reforestation is difficult to evaluate in terms of benefits obtained for costs incurred. The fact that strict laws are now applied to the protection forests of most European countries indicates that the relationship between watershed management and flood control is now at least recognized, if not always acted upon with the vigor which foresters would like to see.
In general, forest soils have higher infiltration capacities than agricultural or grassland soils. Where the soil profile is very shallow, the nature of the surface cover may have little influence on hydrology, but is still important in reducing sedimentation and stabilizing slopes.
The influence of the forest on flood flows is greatest on deep and permeable soils, where detention and retention capacities can be increased by forest transpiration and where infiltrated water can be held in detention storage.
The establishment and later manipulation of a forest influence soil temperatures, snow accumulation and melting. Winter temperatures are higher than in the open and soil freezing is therefore less prevalent so that surface runoff is thus reduced. Snow accumulation in the forest can be increased and spring melting delayed by several cutting techniques. Under some conditions, this could increase the flood danger rather than reduce it.
The influence of forest cover on flooding is greatest on small catchments, where the distance from the top to the outlet is short. In the United States, about 56 percent of the current average annual floodwater and sediment damage occurs in valleys of headwater streams, so that land use here is definitely significant. The same would apply to many Alpine valleys in Europe, where watercourses are short and steep.
The larger the drainage area, the longer it takes floodwaters to assemble in the lower reaches of the main stream. Hence detention of infiltrated floodwater for any given time has less effect on flood runoff as basin size increases. Great floods overpower all effects of vegetation and bear more marks of the storm than of the land surface.
In large rivers, most sediment comes from stream deposits. Upper watershed management will not check bank cutting, flood plain scouring and valley trenching on the bottom lands. Though the stabilization of catchment slopes must reduce the total sediment load, the significance of this reduction in terms of downstream sediment depends on the characteristics of the river channel and the flood plain. In a severe flood its effect may be positive though insignificant.
The effect of land-use practices on flooding is largest during storms of short duration and relatively small total rainfall. The effect decreases with increasing length and magnitude of storm rainfall. Major floods are frequently associated with long periods of rainfall during which the soil becomes saturated, or with frozen ground, melting snow or a combination of these conditions. In general, the introduction of a forest cover may reduce the magnitude of flood peaks, but is not likely to affect the total volume of flood runoff.
ANDERSON, H.W. 1958, Rain-snow flood sources, meteorologically defined. (Abstract) Bull. Amer. Meteorol. Soc. 39 (3): 174-5.
BURGER, H. 1943, The water economy in the Sperbel and Rappen watersheds from 1927-28 to 1941-42. (In German.) Mitt. Schweiz. Anstalt für Forestl. Versuchsw. 23 (1).
CROFT, A.R. 1962, Some sediment phenomena along the Wasatch mountain front. J. Geophys. Res. 67, 4: 1511-1524.
DELFS, J., W. FRIEDRICH, H. KIESEKAMP and A. WAGENHOFF. 1958, The influence of forest and clear felling on runoff, the water economy and soil erosion. (In German.) Aus dem Walde, Hannover, No. 3, 223 p.
HILL, L.W. 1960, Forest plantation development influences streamflow. Proc. Soc. Amer. For., Washington, D.C. p. 168-171.
HOYT, W.G. and W.B. LANGBEIN 1955, Floods. Princeton, N.J., Princeton University Press. 469 p.
LEOPOLD, L.B. and T. MADDOCK. 1954, The flood control controversy. New York, Roland Press. 278 p.
MOLCHANOV, A.A. 1963, The hydrological role of forests. Jerusalem, Israel Program of Scientific Translations. 407 p.
REINHART, K.G., A.R. ESCHNER and G.R. TRIMBLE, Jr. 1963, Effect on streamflow of four forest practices in the mountains of West Virginia. U.S. Forest Service Res. Paper NE-1, 79 p.
