Contents - Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
6. Hydrologic process models
Role of forests in the hydrologic cycle
W. E. Reifsnyder
Depending upon the user, hydrologic process models may represent water yield (engineers and water managers), water balance (vegetation managers and ecologists), or water circulation (climatologists, etc.). Inputs will vary accordingly: terrestrial evaporation, for instance, will be entered as a simplistic parameter, an estimated or measured quantity, or marginally ignored, respectively. Processes incidental to some models will be vital in others. Outputs from the models will be specifically user oriented. The singular contribution of forest to the water balance is strongly scalerelated. In part, this accords with the limited scale on which forest conversion can take place. Very much of the forest effect is merely local redistribution of precipitation (particularly of snow) and is also dependent on rainfall type (frontal, convectional, etc.). For various reasons it is contended that the effects of forest conversion will prove to be significant only at the mesoscale if at all. A plea is made for investigations at this scale (the large river basin), especially in the tropics. At the global scale, the atmospheric circulation models are too insensitive to respond to inputs representing realistic changes of land use involving forest.
In any discussion of hydrologic process models in relation to forest versus other land uses, we must first be clear as to what our objectives are. What process or processes do we wish to model? What results and output do we wish to obtain? Conventional modelling efforts include the following.
Water Yield Models
Most of the hydrologic modelling in recent years has been directed toward predicting surface runoff and streamflow. In a recent tabulation (Haan, Johnson, and Brakensiek 1982), 52 of the 75 models listed have surface runoff and streamflow as a primary output. Some of these produce predictions of streamflow on an hourly basis; others produce longer-term averages. As indicated by the title of their paper ("Hydrologic modelling of small watersheds"), most apply to relatively small watersheds, ones that can be expected to be affected by single storms. Many of them were devised to permit prediction of storm hydrographs for flood-warning and flood-protection purposes. Others predict total water yield for managing water supply. This general class of hydrologic models is of interest and use to engineers and administrators charged with the responsibility of managing water supplies and protecting the public from floods and their consequences.
In this connection, some of the models include, as outputs, predictions of erosion, sedimentation, water quality, and other processes relevant to the beneficial use (and disbenefits) of water on and near the surface of the land.
Precipitation and Evaporation
In recent years, there has been much interest in modelling the precipitation and evaporation components of the hydrologic cycle. For the long term, the simplest model of the hydrologic cycle on a small watershed states that runoff (i.e. water yield) equals precipitation minus evaporation (including transpiration). Temporal and spatial distribution of both precipitation and evapotranspiration have been modelled with varying degrees of success. Such models have often been developed by agriculturists, foresters, and terrestrial ecologists who have been interested in water as a major component of biological cycles and the functioning of natural and managed ecosystems. Some of this effort has been directed towards improving crop yield.
However, there is also considerable interest in understanding the cycling of water through terrestrial ecosystems. An important and as yet unanswered question is what role recycled water plays in the preservation of an existing ecosystem such as a forest. If we remove the forest from a particular area, do we change the hydrologic cycle in such a way as to inhibit or prevent the re-establishment of the forest? How are deforestation and desertification related?
Hydrologic Models as Inputs to General Circulation Models (GCMs)
Modellers of the general circulation of the atmosphere are increasingly aware of the critical role that surface processes play in establishing atmospheric circulation patterns. Early models avoided the surface completely or included grossly simplified parameterizations of surface characteristics. The history of the development of GCMs is one of progressive improvement in the fineness in detail of such parameterizations. Current concerns over the possibility that anthropogenic pollutants such as carbon dioxide and various particulates may influence the earth's heat balance and general circulation patterns have heightened interest in possible interconnections.
The Choice of Model
Thus the goal of the modeller largely determines the type of model and what the modeller considers as input and output. By and large, the engineering hydrologist considers streamflow as a major output, with precipitation as an input. Evapotranspiration is then a more or less unavoidable loss to the system. On the other hand, the ecologist may emphasize some aspect of the functioning of an ecosystem (such as biomass increment) with runoff as an incidental loss to the system. Modellers of the general circulation may be interested primarily in evapotranspiration in so far as it returns moisture to the atmosphere and mediates the partitioning of energy inputs between sensible and latent heat.
These outlooks require a different approach when considering the effects of deforestation and afforestation. For example, a model developed for predicting water yield may parameterize evapotranspiration so simplistically as to be virtually useless to the ecologist or land manager concerned with the hydrologic effects of vegetation manipulation. On the other hand, a detailed evapotranspiration model may be too complicated and require too many unmeasured meteorological variables to be of much use to the water-supply manager. In order to put these contrasts in perspective, we must look briefly at the role of forests in the hydrologic cycle.
Role of forests in the hydrologic cycle
Forests cover approximately one-third of the land surface of the earth and approximately 10% of the entire surface of the globe (Baumgartner 1979). About half of the land surface is occupied by other vegetation and the remaining one-sixth by dry deserts or polar surfaces. Considering the small portion of the earth's surface covered by forests, it would seem that even extensive deforestation or conversion to some other form of vegetation would have minimal effects on global circulation. However, one cannot be so certain about regional or local effects. A brief review of the role of forests in the hydrologic cycle follows. A more complete review is given in a recent WMO publication (Reifsnyder 1982).
Forests Redistribute Precipitation
As compared with grass or low vegetation, forests may exert a major influence on what happens to precipitation reaching the ground. This is particularly true with snowfall. By altering wind patterns, forest edges and clearings may accumulate substantial surpluses of snow at the expense of the forest itself (Anderson, Hoover, and Reinhart 1976). Shading effects of the forest may also affect snowmelt, delaying springtime melt and runoff. Rain may also be redistributed to some degree, although this effect is probably more important for local vegetation patterns than it is for any influence on water yield and timing.
Forests also intercept rain and snow that might otherwise reach the ground. Snow caught on branches can melt and its moisture drip to the ground, sublimate directly to the air, or be blown to adjacent forest openings (Miller 1977). The fate of water caught on tree surfaces is equally complicated. With sufficient rainfall intensities, a large portion of total storm precipitation will eventually reach the ground, although it may be redistributed on the microscale through stemflow and branch drip. That portion of the rainfall remaining on vegetation surfaces eventually evaporates, reducing transpiration.
A recent analysis by McNaughton and Jarvis (1983) indicates the complexity of the process in tall tree vegetation as compared with grassland. Their analysis indicates that when the canopy is dry, downward transport of dry air above a rough forest may lead to higher evapotranspiration rates as compared with nearby grassland vegetation. On the other hand, intercepted rain-water may evaporate more slowly from tree canopies than from grass because of lower saturation deficits over the tree canopy. They point out that general statements concerning the relative evapotranspiration from forest and grass are not possible because the two controlling factors, radiation supply and vapor-pressure deficit, vary in ways that are not necessarily controlled by the vegetation but by meteorological factors such as advection.
Forests May Modify the Total Amount of Precipitation
A controversy of long standing, not yet fully resolved, is whether forests increase the total amount of precipitation falling on them as compared with nearby open areas. Early precipitation measurements in forested areas indicated greater amounts of rainfall. However, most of the forest measurements were made in small openings where wind conditions probably resulted in greater accumulations than in the surrounding forest, as indicated above. Thus the measurements were not truly representative of the average rainfall over the forest area.
Effects at Continental Scale. At the continental scale, it is evident that the net continental precipitation (i.e. the total precipitation minus the total evapotranspiration) must equal the runoff as measured by total outflow of all continental rivers, assuming no change in continental storage. Looked at another way, the net influx of moisture along continental boundaries must equal runoff from the continent so long as there is no change in continental water storage. Viewed in this light, it would appear that vegetational changes could not affect total continental runoff unless they produced substantial changes in general circulation patterns of moisture influx. This is not to say that changes in forest cover might not produce changes in the precipitation and runoff patterns on a regional or intra-continental scale. This will be developed subsequently.
A recent modelling effort by Shukla and Mintz (1982) attempted to put some limits on possible changes in general circulation patterns. Using the Goddard Laboratory for Atmospheric Sciences (GLAS) atmospheric general circulation model, they stimulated the precipitation patterns and surface pressures using two extreme conditions: a prefectly dry land surface and a completely wet land surface. As might be expected, the simulated July precipitation amounts differed greatly in the two cases. For example, in the dry case, the continental interiors were nearly devoid of precipitation whereas in the wet case, mid-latitude continental precipitation amounts were as much as five millimetres per day.
Pressure patterns also responded dramatically to these two forcings. In the dry case, giant thermal lows dominated continental regions, whereas they were mostly absent in the wet case. The authors suggest that in the region outside of the tropics, the soil plays a role similar to that of the oceans. They further suggest that summer convective rainfall may be highly dependent on the amount of moisture stored in the soil. It is also clear that there is a potential for modifying the general circulation if changes in land cover and land use are of sufficient magnitude and horizontal area extent. The relevance of these relationships is, of course, highly uncertain for conceivable changes in land use patterns.
Effects at Regional Scale. Changes in precipitation patterns resulting from land use changes may, however, be significant at regional or intra-continental scale. Lettau, Lettau, and Molion (1979) estimate that as much as 88% of the water falling on the western-most regions of the Amazon Basin has evaporated at least once within the basin.
Recent work by Salati and Matsui (1981) on the recycling of moisture in the Amazon Basin indicates that perhaps half of the precipitation falling on the basin has its source in the evapotranspiration within the basin itself. The ultimate source of moisture for precipitation in the Amazon Basin is the Atlantic Ocean. Because of high precipitation amounts and correspondingly high evapotranspiration rates, cells in their simple box model farther and farther from the ocean have increasingly large amounts of their precipitation formed from water evaporated in upwind cells. However, this would only be true if the precipitation were largely or solely from convective showers. Frontal precipitation, on the other hand, would be expected to be derived primarily from ocean sources, with very little derived from ground sources (Holzman 1937). Through the use of appropriate statistical techniques (see, for example, Essenwanger 1976), it may be possible to estimate the relative amounts of precipitation occurring in a region that have their source in convective, frontal, orographic, or tropical storm systems. These precipitation events have different time-intensity distributions that can be discriminated by statistical analysis.
In the case of large land areas with predominantly convective precipitation, such as the Amazon, the role of forests is somewhat problematical. Conventional wisdom is that forests transpire more moisture than grasslands. However, as pointed out by Dickinson (1980) and McNaughton and Jarvis (1983), measurements and theory indicate that forests may evaporate more water or less water than short vegetation de pending on the prevailing meteorological conditions. It is clear that no general conclusion can be made as to whether forest removal or conversion to other plant canopies will increase or decrease evapotranspiration and rainfall.
Ecological effects of recycling of moisture should be expected to be dramatic in situations such as the Amazon Basin. If there were no moisture recycling, it would be expected that precipitation amounts would decrease from the Atlantic Ocean inward. This decrease would be reflected in vegetation types becoming less hygric. However, both precipitation and vegetation in the Amazon do not show such dramatic changes. Indeed, annual precipitation increases as one travels westward from the Atlantic coast (Salati, Barques, and Molion 1978).
Effects at Local Scales. At the local scale forests may affect the distribution of convective cells and consequent precipitation patterns as a result of local variations in the surface temperature and perhaps even by the roughness of the forest canopy. These local effects may find some expression in the patterning of ecosystems but are probably of minor consequence to the hydrologic cycle.
Impact on the Hydrologic Cycle
To assess the hydrologic effects of land use changes involving forests, we should first note that such effects are in more or less inverse relationship to the area of the earth's surface being considered, at least on the scale of current land use changes. That is to say, clear-cutting a small watershed will have dramatic effects on the local hydrologic cycle, measurable impact on basin hydrology, and little or no effect on the regional or globle cycle. Of course we know more about the local effects and run into difficulty when we attempt to extrapolate these to the large scale.
Microhydrology. On the local scale there is much information in infiltration, percolation, erosion, depression storage, surface runoff, local redistribution of snow and rain, and so forth. However, as pointed out previously, the effects on evapotranspiration of forest conversion are relatively uncertain.
Mesohydrology. On the medium-basin scale, the effects of vegetation manipulation in temperate-zone watersheds are relatively well known, although gaps in our knowledge remain. However, little work has been done in tropical and sub-tropical watersheds. Some recent empirical evidence indicates that logging a watershed may decrease water yield (after a brief period of increase) in contrast to what is expected in temperate regions (Dunin 1984).
Macrohydrology. On the regional scale, especially in major tropical basins where moisture recycling may be important, forest conversion may have measureable effects on the hydrologic cycle, at least in portions of the basin. It has been estimated that perhaps 1-3% of the Amazon Basin forests are logged each year (Lovejoy and Salati 1983; Seiler and Crutzen 1980). But most of this reverts eventually to secondary forest, with perhaps 15-20% shifted to permanent agriculture and pasture (Dickinson 1980). It seems probable that these conversions have relatively little hydrologic significance on a long-term or large-scale basis. However, I must admit that this is rather speculative and we need much more hard information on the hydrologic results of forest conversion in tropical regions.
Global Hydrology. At the global scale, it seems unlikely that forest conversion (deforestation and afforestation) at the current scale of operations would have any substantial or measureable effects on the general circulation of atmospheric moisture. I should point out that many of those who predict dire consequences from deforestation tacitly and implicity assume that, once a forest is cut, nothing but bare ground remains. In most forested areas of the world it is almost impossible to keep forest from recolonizing an area once it is cut. Of course, the new forest may not be identical to the one removed and may be disliked for one reason or another; but in the hydrologic sense, it is still an actively transpiring vegetative covering of the earth.
I will conclude with a brief review of the two types of hydrologic process models in general use, along with some comments about their applicability to land use problems.
To most of us the term "hydrologic model" almost certainly implies a deterministic or stochastic basin model. For many years, water-supply engineers have constructed models of varying degrees of complexity with the primary goal of predicting streamflow. The Stanford watershed model (Linsley and Crawford 1960) is a good example of the current generation of such models. This is a deterministic model that uses precipitation and potential evapotranspiration as meteorological inputs and produces time-variant predictions of streamflow and ground-water storage as outputs. In this, as in most engineering models, evapotranspiration is handled highly simplistically. In general, some estimate of potential evapotranspiration (either climatologically or by measurement of pan evaporation) is converted to actual evapotranspiration using a factor based on estimated soil moisture storage. There is usually no vegetation parameter involved.
On the other hand, micrometeorologists have favored energy-balance models. The most enduring expression of these models is that of Penman (1948) and its myriad descendants (see, for example, Monteith 1964). In these, evapotranspiration is a major output, in contrast to a role as an input variable in the basin models. This class of models is of great scientific interest in as much as the processes modelled are close to the physics and physiology of the soil-plant-atmosphere system. Some of these are very elegant and have done much to elucidate the physical and physiological processes involved. There is no reason why these could not be used as inputs or submodels to basin models in order to evaluate the effects of land use conversions on streamflow and other output variables. They have not been generally so used, however, probably because of their substantial requirements for meteorologic data as input, much of which is unavailable, and partly because they are perceived by engineering hydrologists as contributing little to the reduction of the error in their model outputs.
For most land use analysis purposes, models of the Stanford variety are probably quite adequate. Engineering hydrologists will continue to improve them, perhaps by incorporating some of the features of the energy-balance, micro-process models. On the other hand, there are numerous land use problems that require the energy-balance approach. Prime among these is the question of the role of different kinds of vegetation at different stages of ecological succession on the local, regional, and, indeed, the global energy and water balances.
These two approaches to hydrologic process modelling ought to converge, although they seem to be doing so very slowly. Part of the reason is that they are pursued by two different groups of people: engineering hydrologists on the one side and biometeorologists on the other. Certainly more could and should be done to bring these groups together in an atmosphere conducive to effective communication. Perhaps this is a worthy goal for a future workshop.
I have not labelled this section "Conclusions," for much of what I have said is highly speculative. Perhaps "Contentions" is the proper heading. I will recap some of the more important points.
First, it seems to me that many of the doom-sayers amongst the ecologists and climatic modellers have completely misunderstood and misrepresented the consequences of deforestation. In most of the forested regions of the world it is difficult to keep some kind of vegetation from recapturing the land, usually very quickly. It is fatuous to calculate when a basin will be completely deforested. "Timber famines" have been predicted periodically for North America (for example) at least since the middle of the last century. Yet the area covered by forests today is greater than it was a hundred years ago.
Second, I contend that timber harvesting is unlikely to modify the precipitation component of the hydrologic cycle in a way that will inhibit recapture of the area by natural revegetation of some kind. Cutting may modify the microclimate in undesirable ways and this may prevent or make more difficult the establishment of desirable species. Nevertheless, some kind of vegetation will reoccupy most logged areas, the hydrologic effects of which will be similar to those of the replaced vegetation, at least as far as recycling moisture through the atmosphere is concerned.
Third, I contend that forest harvesting, even on a large scale, is unlikely to have measurable effects on the general atmospheric circulation. Nevertheless, in critical or marginal areas, such harvesting may well have local and possibly regional effects.
Last, and perhaps this is not very contentious, the hydrologic role of recycled precipitation in large river basins in both tropical and temperate regions is worth vigorous study.
In conclusion, it seems to me that the biggest challenge facing us today is the understanding of hydrologic processes in relatively large basins. Small watersheds have been studied to death, although there is need for some additional work in small tropical and subtropical watersheds. At the other end of the scale, it appears to me that society's continuing efforts at modifying the landscape have had and will have relatively little, if any, effect on global circulation patterns. Certainly at present parameterizations of surface conditions in current general circulation models are so gross that the input of detail that could be made with current knowledge is mean ingless. Of course as GCMs are refined and elaborated increasing fineness of surface detail will be appropriate.
Anderson H. W., M. D. Hoover, and K. G. Reinhart. 1976. Forests and water: Effects of forest management on floods, sedimentation and wafer supply USDA For. Gen. Tech. Rep. PSW18, Pacific Southwest Forest and Range Experiment Station, Berkeley, Cal, 15pp.
Baumgartner, A. S. 1979. "Climate variability in forestry." Overview paper presented at the WMO World Climate Conference, Geneva, 23 February 1979, 30 pp.
Dickinson, R. E. 1980. "Effects of tropical deforestation on climate. " In Blowing in the wind: Deforestation and long-range implications, pp. 411-441. Studies in Third World Societies, no. 14, College of William and Mary, Dept. of Anthrop., Williamsburg, Va., USA.
Dunin, F. X., I. C. Mcllroy, and E. M. O'Loughlin. 1985. "A Iysimeter characterization of evaporation by eucalypt forest and its representativeness for the local environment." In B. A. Hutchinson and B. B. Hicks, eds., The forest-atmosphere interaction, pp.271-291. Reidel, Dordrecht.
Essenwanger, O. 1976. Applied statistics in atmospheric science. Part A: Frequencies and curve fitting. Elsevier, 412 pp.
Haan, C. T., H. P. Johnson, and D. L. Brakensiek, eds. 1982. Hydrologic modelling of small watersheds. American Society of Agri. Engineers Monograph No. 5, 533 pp.
Holzman, B. 1937. Sources of moisture for precipitation in the United States. USDA Tech. Bull. 589, 41pp.
Lettau, H., K. Lettau, and L. C. B. Molion. 1979. "Amazonia's hydrologic cycle and the role of atmospheric recycling in assessing deforestation effects. " Monthly Weather Review, 107: 227238.
Linsley, R. K., and N. H. Crawford. 1960. Computation of a synthetic streamflow record on a digital computer. International Association of Scientific Hydrology Pub. 51, pp. 526538.
Lovejoy, T. E., and E. Salati. 1983. "Precipitating changes in Amazonia." In F. Moran, ea., The dilemma of Amazonian development, pp. 211-220. Westview Press, Boulder, Co., USA.
McNaughton, K. G., and P. G. Jarvis. 1983. "Predicting effects of vegetation changes on transpiration and evaporation." In Water deficits and plant growth, 7: 1-47. Academic Press, New York.
Miller, D. H. 1977. Water at the surface of the earth: An introduction to ecosystem hydrodynamics. Academic Press, New York. 557 pp.
Monteith, J. L. 1964. "Evaporation and environment." In The state and movement of water in living organisms, pp. 205-234. 19th Symp., Soc. Exp. Biol., Cambridge University Press, Cambridge, UK.
Penman, H. L. 1948. Natural evaporation from open water, bare soil and grass. Proc. Roy. Soc. (Set. A) 194, pp. 120-145.
Reifsynder, W. E: 1982. The role of forests in the global and regional water and energy balances. World Meteorological Organization, GAgM Report No. 8, 33 pp.
Salati, E., J. Marques, and L. C. B. Molion. 1978. "Origem e distribuicão das chuvas na Amazônia." lnterciência, 3: 200 205.
Salati, E., and E. Matsui. 1981. "Isotopic hydrology in the Brazilian Amazon Basin." Inter American Symposium on Isotope Hydrology, Bogota, Colombia, August 1980, Instituto de Asuntos Nucleares, Bogota, pp. 111-120.
Seiler, W., and P. J. Crutzen. 1980. "Estimates of gross and net fluxes of carbon between the biosphere and atmosphere from biomass burning," Climatic Change, 2: 207-247.
Shukla, J., and Y. Mintz. 1982. "Influence of land-surface evapotranspiration on the earth's climate." Science, 215: 14981501.
It is becoming increasingly obvious that assessments of deforestation rates are collected in terms that are not appropriate for many of the uses for which the data are needed. The assessments are ill-suited to detecting the climatic effects of changing vegetative cover. Specifically, the areal extent of replacement or perennial transpiring covers with deciduous covers or intermittent crops is unclear. There is also insufficient information as to the dynamics of the changes, the permanence of the substitute vegetation or bare ground, let alone the temporal and spatial variations in significant physical parameters of the ground cover such as albedo and aerodynamic roughness or the physiological characteristics of the species involved.
In the meantime it has to be admitted that there is a degree of unreality in the experiments with global circulation models: the earth surface conditions used are extreme and are intended to show the theoretical limits of the effects rather than the probable or possible results of man's influences on climate through managing the terrestrial vegetation. In any case the details of the models are so far from being finalized that the simulations should not yet be taken very seriously. Of all the possible outputs of GCMs, the simulation of rainfall is probably the most difficult, yet it is of most practical value. We are only at the beginning of linking micrometeorological models to hydrological process models and general circulation models.
These linkages are in part problems of scale, joining processes at molecular level to those at continental level. The disposition of the earth's forests underlines the importance of the tropics. Similarly, if the drainage basins of the world are classified by discharge or by area, again the tropics are very significant. However, hydrological process models have been constructed mainly from investigations in temperate watersheds within the I to 10 km2 size class or smaller. Extrapolation to the important hydrological systems of the world present many difficulties. Among these are the relatively drastic land use changes carried out in small catchments, such as deforestation, which have massive effects on the microclimate but only marginal influences on the gross hydrological cycle. The scale problem may be illustrated by the evaporation of water intercepted by tree crowns, which has been shown to contribute so largely to excessive forest evaporation. The advective energy for this evaporation ultimately derives chiefly from the lower atmosphere above the surrounding nonforested regions. There is therefore no possibility of a simple extrapolation of interception losses from small woodlands to extensive forest areas. Similarly, the precipitation (and therefore the energy relations of evaporation) is relatively uniform over small basins, but variable over large basins. The relationship between precipitation and runoff cannot simply be transferred from small to large basins because of the detention evaporation in the latter. Many of the relationships in hydrological process models are curvilinear rather than linear and this precludes a simple areal adjustment when extrapolating to large watersheds. Large basin hydrology therefore needs to be actively studied, especially in the tropics, in parallel with the development of methods of transfer between systems with different scales.
Insufficient attention has been paid to feedback mechanisms in hydrological process models originating in the biological milieu. A bewildering variety of plant adaptations ensure that the mix of species in a vegetation relates to the microclimate evolved by the vegetation so as to optimize the internal water balance and other physiological conditions that maximize growth and reproduction. Thus the regulation of transpiration and absorption of water from the soil responds to the water vapor deficit of the air and the soil moisture content, controlling the addition of the moisture to the former and extraction from the latter. In contrast, evaporation from water bodies and the surface of moist soil and intercepted rainfall are relatively simple processes.
Whilst attending to the physical and biological aspects of hydrological processes it is essential to consider the ways in which forest conversion impinges on human societies directly and indirectly through climatic modification. Even imagined or slight actual changes of continental or global climate due to deforestation or reforestation could result in international tensions. Careful appraisal of the evidence from observation, experiments, and simulations suggests that there is no foundation for these reactions with the possible exception of situations where the rainfall is already marginal and the ecosystems so fragile that natural variations of climate could cause degeneration in any case.
Contents - Previous - Next