Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Mechanisms of forest influence

Among the most immediate and direct effects of forest clearance are the destruction of the forest microclimate and the associated changes in albedo, aerodynamic surface roughness, energy balance, water balance, and Bowen ratio. At least eight theories have been suggested for how these might affect rainfall.

TABLE 2. Effects of deforestation on rainfall distribution in Uttara Kannada district

  20-year average Forest clearance (ha)
Normal rainfall (mm) (1955-1975) (mm) Change  
Bhatkal 3,500 4,370 Increase 1,210
Honavar 3,500 3,830 Increase 1,130
Kumta 3,500 3,760 Increase 970
Ankola 3,500 3,780 Increase 730
Yellapur 2,900 3,170 Increase 1,210
Karwar 3,500 3,250 Decrease 490
Supa 3,500 2,490 Decrease 8,090
Siddapur 3,250 3,050 Decrease 2,430
Sirsi 2,900 2,620 Decrease 3,240
Haliyal 2,490 1,420 Decrease 7,280
Mundgod 2,490 1,630 Decrease 9,310
Uttara Kannadaas a whole 3,300 2,790 Decrease 40,470

Source: Raju 1981


According to the Unesco report (1978), tropical regions, which represent 40% of the earth's surface, contribute about 60% of the water vapour in the global water cycle. The tropical oceans contribute almost 50% and tropical land area provides about 10%. The tropical forests, which cover about one-third of the total tropical land area, contribute 3%. Water cycling is rapid in the atmosphere, with an average residence time of 9 to 10 days, so the role of the tropical forest in influencing the global water cycle is considered to be around 3%. Zon (1927) maintained that the moisture carried by winds into the interior of vast continents, thousands of kilometres from the ocean, is mostly continental vapour. Thickly wooded areas would evaporate about 8,700 tonnes of water per hectare per year.

The evidence from changes in the concentration in rainfall of 18O (an isotope of oxygen) along an east-west transect across the Amazon Basin suggests that the tropical forest returns to the atmosphere as much as 75% of the moisture that it receives through the processes of evaporation and transpiration (Salati et al. 1979). Runoff accounts for 25% of the rain-water; another 25% of the water is evaporated from intercepted rain and 50% is returned to the air by transpiration. It has been estimated that tree-covered land returns to the air ten times as much moisture as a barren area and twice as much as shrubland and grassland. Thus a large forest may significantly affect local weather patterns.

Nicholson (1929) thought that in East Africa trees and shrubs contribute more moisture to the air than herbaceous vegetation or bare soil and are therefore more likely to influence rainfall. According to him, forests increase the probability and the amount of precipitation under unstable meteorological situations favouring the production of rain, the effect being more marked at lower altitude, and, further, numerous small patches of forest attract as much as or even more rainfall than an extensive forest tract, but any compact forest area above 1,800 ha had a role in attracting rains.

Rama (1980a) pointed out that during the monsoon months June to September, when there is a very humid air mass extending to five to six km altitude over India, even weak convective activity leads to precipitation. The contribution of local evapotranspiration is to raise the average humidity of the prevalent air mass by 15% to 20%, without which the rainfall over the subcontinent would be meagre because the synoptic situations producing rain would be rarer and weaker. Stidd (19?5) and Rama (1980b) advocated the need of increased irrigation to enhance evapotranspiration and thereby precipitation.

Green and Minkowski (1977) suggested that transpiration from the evergreen forests and their understorey in the Western Ghats creates a cool atmosphere above the forest that induces rainfall. Because deciduous tree species cannot maintain the atmospheric humidity to the same extent as the evergreens, the replacement of evergreen forests by deciduous ones may cause a decline in rainfall.

Meher-Homji (1982) suggested that, whereas evergreen forests produce an immense total leaf surface, monoculture plantations of trees such as Eucalyptus, growing at low densities, or pines or certain wattles (Acacia spp.) from Australia, planted extensively in the Nilgiris and Palnis, do not. These are xeromorphic plants with special adaptations such as drooping or needle-like leaves or phyllodes (petioles take up the function of leaves) to reduce transpiration. Rakhmanov (1962) stated that the water added to the atmosphere by forests is not higher than from other types of land use. And Brooks (1928) argued that a deep-rooted grassland would transpire more than a forest.

Rejecting the view that the forest does not affect gross precipitation over an area, Molchanov (1960) stressed the role of condensation of water vapour during rain, which may increase the amount by 10%, and Brooks (1928) and Hewlett (1965) emphasized the "occult" precipitation aspect, that is condensation and fog-drip from tree crowns, leaves, and stems. According to Dickinson (1980), however, conclusive evidence on increases or decreases in evapotranspiration due to deforestation is still not available. Much depends on prevailing climatic conditions, such as the frequency and intensity of rainfall and the relative humidity of the canopy air. Generally, small areas of forest scattered amidst drier areas use more water per unit area than do large forest areas because the dry air advected from their surroundings increases the rate of transpiration.

Shukla and Mintz (1982) report that modification of the vegetation cover by way of extensive deforestation does influence precipitation. The hypothesis is based on physical connections between precipitation and evapotranspiration. However, the determining factor is not only the vegetation but, rather, the relationship between the moisture content of the soil, the vegetation, and the solar energy necessary to evaporate water.

The Bowen ratio is the ratio of sensible to latent heat loss for a surface. B=H/E, where B the Bowen ratio, H the sensible heat exchange, and E the latent heat used in evapotranspiration. The forest changes the Bowen ratio relative to low vegetation and barren soil in two ways (Dickinson 1980). Firstly, as roots tap water from greater depths (down to 2 m), more water is available for evapotranspiration. On the other hand, bare soil ceases to evaporate once the top 5 to 20 cm has been desiccated. Since bare soil tends to dry relatively quickly, it will have a higher Bowen ratio than the surrounding forests (i.e. will transfer more sensible heat to the atmosphere). This may increase convection and convection storms. Furthermore, available soil moisture in forests may be greater as the infiltration rate of water into the subsoil is higher. Secondly, forests also increase evapotranspiration by evaporation of the water intercepted by their leaves. The net result is a low value of the Bowen ratio, especially when the forest canopy is wet.

According to Pereira (1973), deforestation reduces transpiration by 10% to 15%, regeneration through secondary growth restores the initial rates. Lettau, Lettau, and Molion (1979) suggested, conversely, that deforestation in the Amazon Basin would increase evapotranspiration. In their model, the sum of sensible and latent heat fluxes was constant as evaporation changed, so the Bowen ratio was dependent on evaporation. Hence, changes in absorbed solar radiation had to be balanced by changes in the net emission of thermal infrared radiation. Assuming increased evapotranspiration in the wake of deforestation and a 35% decrease of forest area over 10 of longitude at the equator, they estimated a 0.03 mean increase of albedo. However, they found that an increase of downward thermal infrared radiation supplied more energy than was lost due to increased albedo. Hence, over and downwind of the deforested region they found a warming of 1 to 2 C and up to a 10% increase of rainfall. This model is oversimplified, with too many arbitrary relationships for its conclusion to be satisfactory in toto. Their results are at variance with those of Potter et al. (1975), who suggested that total deforestation between 5N and 5S would result in reduction of rainfall. Dickinson (1980) concludes that the published studies of both Lettau, Lettau, and Molion (1979) and Potter et al. (1975) on regional effects of tropical deforestation are of limited credibility and arrive at diametrically opposite conclusions. However, both studies support the view that large-scale deforestation in the tropics might bring about a change in the mean temperature by as much as 1 or 2C and in rainfall of up to 10% or so.

The effects of excessive and negligible evapotranspiration were compared by Charney et al. (1977) in their model. The former seemed to stimulate more rainfall in the arid zone because higher rates of evaporation of moisture from the soil resulted in intense convective recycling of water. Under negligible evapotranspiration, rainfall was reduced by a factor of two in the wet zone of Africa. Commenting on this model, Dickinson (1980) points out that short vegetation would generally have greater evapotranspira tion than bare soil. The net change of evapotranspiration between short vegetation and a forest is more questionable since the leaf stomata! resistance has a large effect in retarding water loss. However, the time course, if not the mean of evapotranspiration, is expected to change significantly with deforestation. Charney et al. (1977) imply that disappearance of short vegetation over a large tropical region may drastically reduce mean rainfall; a significant impact of forest removal is also indicated, but the details would depend upon whether the changed land use were to increase or decrease the evapotranspiration and on its effect on soil moisture. Until these questions are resolved, Dickinson (1980) cautions that the simulations may not correspond to any plausible scenario for tropical deforestation.

Dust Loading

Bryson (1974) estimated that 100 to 250 x 106 metric tons of dust particles are being injected annually into the subtropical atmosphere because of soil degradation. The gradual deterioration of the soils of the Thar after the Rangmahal period is largely the result of human activity (Bryson and Baerreis 1967).

Visual observations of the dust layers suggest that the top is at about 7 km over the north-east Arabian Sea, 9 km over the Thar Desert, and 5 km over the Ganga Valley. If this dust is of local origin, strong currents of rising air are needed to lift it the 9 km over the Thar, but then this is incompatible with the fact that there is a large-scale subsidence of the air in that region. Alternatively, the dust is transported by winds from distant regions to the Thar and, therefore, both the origin of the dust and the mechanism of its lift to these heights need further research (Des 1968). If the dust comes from the desert itself, then the arid zone would appear to be a self-maintaining system. The solution would be to create a green belt with suitable trees and grasses to stabilize the soil, prevent the wind erosion, and the lifting of the dust into the atmosphere. This would not only reduce the region of subsiding air but also reduce the cooling rate required to maintain steady monsoon conditions.

Das (1968) concluded that in order to have a steady monsoon circulation the atmosphere should be warmed 3.2 C per day over north-east India and cooled by 2.4 C in the north-west.

Deforestation and the Carbon Cycle

Forests have an indirect role in controlling climate through the carbon cycle. Deforestation adds to the carbon dioxide concentration in the atmosphere, which has been increasing since the beginning of the industrial age. According to Woodwell et al. (1978), 3 x 1012 kg y-1 of carbon is being liberated by tropical deforestation. Other estimates of the loss of carbon from the biosphere vary from 1 to 5 x 1012 kg y-1. This increase in carbon dioxide, by absorbing long-wave radiation from the earth while transmitting short-wave solar radiation, leads to an increase in surface temperature (known as the "greenhouse effect") and a decrease in temperature in the upper atmos phere. The results of these changes are more evaporation, higher water-vapour concentration, and melting of snow and ice. The consequent effect on planetary reflectance is equivocal as it also depends on changes in global cloudiness (Pittock 1983).

There is a consensus of opinion on the warming of the lower atmosphere with an increase of carbon dioxide, but the estimates vary from increases of 0.1 to 4C for a doubling of carbon dioxide concentration. Using a three-dimensional general circulation model, Manabe and Wetherald (1975) suggested that the tropospheric warming is most marked in the lower atmosphere at high latitudes. In the tropics, the warming spreads throughout the entire troposphere due to the intense moist convection, hence it is lesser than the warming in the polar region. The rise in temperature of the equatorial belt is of the order of 2C for a doubling of the atmospheric carbon dioxide content, increasing the intensity of the hydrological cycle by 7% and resulting in a slight increase of rainfall at the equator.

Dickinson (1980) does not predict any great increase in either the carbon dioxide content or the albedo in the wake of total conversion of tropical forests to other land uses. The increase of carbon dioxide in the atmosphere would be of the order of 50 parts per million by volume, that is, about the same increase as that experienced over the last century. In fact, it is feared that the combustion of fossil fuels within the next 50 years will increase carbon dioxide by twice the amount expected from complete destruction of tropical forests (Rotty and Marland 1980).

If forests were converted to grasslands, increasing carbon dioxide would add 0.7 W m-2 to the global heat balance. On the other hand, albedo increase would reduce solar heating. Even the entire destruction of the existing forests in the tropics would lead to an increase of the global surface albedo by only 1.2 x 10-3.

If the ice-cap over the Arctic Ocean were to disappear, the mid-latitude rainy belt would shift 600-800 km northward (Flohn 1958); the USA, Europe, and the USSR would become much drier, whereas Canada might profit by more rainfall, and tropical monsoon regions could face problems of floods and decreased sunshine for crops. On the other hand, with increased reforestation, forests could serve as storehouses of carbon dioxide. They might store 10 to 20 years' carbon dioxide production throughout the world, including that from fossil fuels, and thus delay the change in the climatic pattern.

Tropical forests have effects on other atmospheric trace gases such as methane and nitrous oxide that contribute to the global energy balance. Methane is present in the atmosphere in a concentration of about 1.6 parts per million by volume and, like carbon dioxide, traps heat from the earth's surface. Large-scale increases in the area of paddy fields or decreases in tropical rain forest swamps could change methane concentrations in the atmosphere by possibly as much as 20%, corresponding to a 0.1 W m-2 change in global heat balance. The rate of removal of methane from the atmosphere is declining because of increasing levels of nitrous oxides emitted in vehicle exhausts, affecting the rate at which methane is converted into carbon dioxide. Nitrous oxide concentration could also be changed significantly by large-scale land use change in the areas of tropical forests (Dickinson 1980).

Change in Albedo

Charney et al. (1977) have emphasized the role of lower albedo (i.e. the proportion of radiation reflected from a surface) in increasing rainfall in a forested region.

Radiative energy absorbed by the forest canopy and ground is returned to the atmosphere as sensible and latent heat. Sensible heat transfer is the cooling of the surface by air passing over it. Latent heat transfer refers to the energy the surface loses in water changing from its liquid to vapour phase through either evaporation or transpiration (which is evaporation, but inside the leaf structure).

The sensible heat energy transferred from a surface directly to the atmosphere warms the lower layers of the atmosphere. The latent heat energy warms the atmosphere only after it is released by condensation.

Albedo is one of the factors that governs the energy balance of a surface and therefore the rate of evapotranspiration. The total energy (sensible heat + latent heat + potential energy) imparted to the lower layers of the atmosphere increases as the albedo decreases. The albedo of deserts or dry bare soil is considerably greater than that of vegetated or forested surfaces, so the net energy imparted to the atmosphere over deserts is less. In a desert region, the net radiation is converted largely to sensible heat resulting in high temperature, in contrast to a vegetated region where net radiation is converted into both sensible and latent heat. Thus the air may be cooler over a vegetated region, but its total energy content may be even greater than air over a desert. Eventually, the water evaporated from vegetated regions condenses and falls as rain and the released latent heat warms the atmosphere. So an increase in albedo causes a slight decrease in large-scale atmospheric temperatures. The more complex the vegetation structure, the greater the trapping of radiation by multiple reflection between leaves and the lower the albedo. Tropical forests have a lower albedo (about 10%) than grasslands (about 20%), which in turn have a lower albedo than deserts (about 30%). Thus the clearing of tropical forest causes a change in albedo that could influence both local and regional climates (Lockwood 1980).

Because sensible and latent energy are not lost from the surface-atmosphere system as a whole, global temperatures are much less sensitive to differences in their exchange than to changes in reflected solar radiation. However, changes in the ratio of sensible to latent fluxes can be significant for the hydrological cycle and therefore modify the frequency, amount, and location of tropical cloudiness and rainfall and so change regional climate (Dickinson 1980).

An increase in albedo reduces the absorption of solar radiation by the ground; consequently, there is less transfer of sensible and latent heat from the ground to the atmosphere. Parallel to this the intensity of cloud cover diminishes as does the longwave flux from the clouds to the surface of the earth, so that the net absorption at the ground (short and long wave) is decreased. Thus regions with increased albedo become sinks of energy and the general motion of the atmosphere above such regions is downward. An essential prerequisite for precipitation is upward motion, thus Charney's model provides a basis for deforestation leading to a decrease in rainfall.

Earlier, Otterman (1974) had proposed that the removal of natural vegetation by overgrazing and the exposure of high-albedo (reflective) soils produce "thermal depression" and lesser precipitation, in other words, desertification in regions of marginal rainfall. Klaus (1981) supports this view for the Sahel region of Africa and Gadgil (1981) for India; Jackson and Idso (1975) found that in the Sonoran desert of the USA the consequences of denudation of soils on temperature and climate were just the opposite of those observed by Otterman (1974).

Using a two-dimensional (zonal) atmospheric model to simulate some of the effects of converting all the humid forests between 5N and 5S with an albedo of 7% to a vegetation with an albedo of 25%, enhanced run-off, and lower evaporation, Potter et al. (1975) reported average global cooling of the surface by about 0.2C and a reduction in global precipitation of about 1%. Their computer simulation has led them to suggest the following succession of events:

Deforestation (r)

Increased surface albedo (r)

Reduced surface absorption of solar energy (r)

Surface cooling (r)

Reduced evaporation and sensible heat flux from the surface (r)

Reduced convection activity and rainfall (r)

Reduced release of latent heat, weakened Hadley circulation, and cooling in the middle and upper tropical troposphere (r)

Increased precipitation in the latitude bands 5 to 25N and 5 to 25S and a decrease in the equator-pole temperature gradient (r)

Reduced meridional transport of heat and moisture out of equatorial zones (r)

Global cooling and a decrease of temperature and precipitation at 45 to 85N and 40 to 60S

Manabe and Hahn (1977) attempted a simulation of the July tropical climate for the last ice-age 18,000 years ago. The consequences in the tropics of albedo change may be inferred from the difference between their sea-surface temperature anomaly experiment and their ice-age experiment. The increase in albedo in the ice-age simulation appeared to decrease the rainfall over land by about 10%, which was only half the precipitation change due to sea-surface temperature change between the present and 18,000 years ago. Most significantly, they found that the increased albedo led to a severe weakening of the monsoon circulation of India and the neighbouring countries.

Other scholars such as Sagan, Toon, and Pollack (1979) have considered the influence of anthropogenic albedo changes on the earth's climate. They consider that in the tropics man has caused climatic modifications by both desertification and the clearance of tropical forest. The change from grassland to desert causes an increase in albedo from about 16% to 35%, a change from tropical forest to cultivated fields or grasslands causes a change in albedo from about 10% to about 16%.

For the Sahel region, Charney et al. (1977) partly blamed overgrazing as a cause of the recent droughts. Ripley (1976) offered certain criticisms to this, pointing out that the effect of vegetation on evapotranspiration was ignored. The grazed area, despite its higher albedo and lower net radiation, would be warmer during the day compared to the ungrazed area, where part of the net radiation would be used in transpiration. This, he argued, may reduce convection and rainfall in the area protected from overgrazing rather than increase them.

Dickinson (1980) has pointed out that several studies have used erroneous estimates of the albedo change resulting from deforestation. For example, Rockwood and Cox (1978) give the following mean values for north-west Africa as measured from a lowflying plane:

dense forest : 0.15
moderate forest : 0.17
savanna : 0.28
desert : 0.43

However, these data do not indicate diurnal and seasonal variations. Kung, Bryson, and Lenschow (1964) found no difference in albedo between a forested tract, bare fields, farmlands, and grasslands.

Reasonable estimates of albedos according to Dickinson (1980) would be 0.12 to 0.14 for tropical forests, which, when clear-felled and succeeded be secondary vegetation, would increase by 0.02 to 0.04; conversion of this to barren land or grassland would lead to a further increase by 0.01 to 0.04. The increase due to deforestation is unlikely to exceed .04.

Albedos of vegetated areas depend on solar elevation and atmospheric composition, since molecules in the atmosphere (clouds, aerosols) also influence surface albedo by forward scattering of the solar beam. It is generally not possible to estimate albedo changes due to vegetation changes from data sets taken at different locations and times.

Two studies ideally suited to examining effects of tropical deforestation on surface albedo are those of Oguntoyinbo (1970) for Nigeria and Pinker, Thompson, and Eck (1980) for Thailand. Both reported an average albedo of 0.13 for the forest and 0.15 to 0.16 for the forest clearing, fallow, or derived savanna. Values range between 0.16 and 0.24 for dry types of savanna.

The albedos of soils vary from 0.1 as in the case of black soil to 0.5 for dry, quartzsand deserts. As deforestation results in drier soils with less organic matter, an increase in albedo is to be expected, but it is unlikely to exceed 0.2. For sandy desert regions, Otterman (1981) reported an albedo range of 0.28 to 0.45, depending on the degree of degradation of the vegetation.

Thermal Mountain Effect

A so-called "thermal mountain" concept has been proposed, in which a column of hot air builds over a forest canopy due to the higher temperature. Accordingly, a 10-km strip of forest, say, 6 C warmer than its surroundings, could be responsible for building up a thermal mountain of up to 1,000 m. This would enable the moisture-bearing air to ascend to a height sufficient to condense rain over the forest. The objection to this hypothesis, however, is that forests tend to decrease rather than increase the air temperature.

Orographic Rainfall

Forests increase the effective height of a land surface as an obstruction to air movement. It is estimated that rainfall increases 1% for an increase in height of 30 m. If this is so, why do the tall buildings in hill resorts such as Udhagamandalam, Kodaikanal, and Madikeri (Mercara) fail to influence rains in dry years?

Wind Speed and Aerodynamic Drag

With a decrease in wind velocity, air masses are forced to stack and rise. Garrat (1978) shows that eddy diffusivities are considerably greater just above tall vegetation than at corresponding heights above smooth surfaces.

Tall trees with irregular surfaces transpire at rates higher than calculated potential evapotranspiration rates. Therefore the potential evapotranspiration rate for a given net radiation will be lower over vegetation types with smoother canopy surfaces.

Brunig (1971) has estimated the evapotranspiration for a 45-m high, dense, mixed forest (near Kuching) of Dipterocarpaceae with irregular surface to be around 2,000 mm. Forests with simpler structure have smoother canopies and their transpiration rate is lower. Similarly, conversion of a forest into a rubber plantation with smoother surface would lower the evapotranspiration and alter the character of rainfall.

Rain-Gauging Errors

Certain difficulties in accurately measuring rainfall inside and outside forests have been pointed out by meteorologists. The rainfall measured in a forest clearing may be as much as 10% higher than in an open area in its vicinity because the rain gauge in the clearing is shielded from wind. The turbulence caused by the opening of the canopy may decrease rainfall, or, conversely, a downward component of the wind may lead to an increase. Fox (1979) noted that in Sandakan (Malaysia) the average rainfall as calibrated in a forest rain gauge was less than or equal to that of one installed in an open area 200 m away on 94 out of 132 days of observations, but rainfall in the forest tended to be higher with greater falls of rain.

Legris and Blasco (1969) have cited the example of the Kudiraiyar Basin in the Palni Hills (Tamil Nadu). A rain gauge at the border of the Kukkal Forest at 2,000 m recorded rainfall about 50% higher than that of two rain gauges in grasslands at the same altitude and of the same northern slope during the period 1956-1966. However, in the hilly terrain the topography itself could account for the variation in the rainfall over short distances. Besides, the length of the dry period differs between the two locations, as shown in the climatic map in Blasco (1971).

Contents - Previous - Next