Contents - Previous - Next

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

4. Effects of forests on precipitation in India

The mechanism of rainfall
Observations on the effect of forests on rainfall
Mechanisms of forest influence

M. Meher-Homji


Whether or not forests affect the rain climate is contentious, though differences in scale may explain some difference of opinion. Desertification related to deforestation is an extreme view. From a brief outline of our current understanding of the processes involved in precipitation, with an emphasis on the phenomenon of the Indian monsoons, this review examines the evidence, including that from mathematical simulations, for and against changes of forest area influencing rainfall. A number of mechanisms for the effects of forest on rainfall are described (forest evapotranspiration, albedo, wind relations, carbon dioxide relations, etc.), some of which appear feasible. The conclusion is drawn that tropical storm intensity and numbers of dry days are probably more sensitive to the degree of forest cover than annual precipitation and that these features are very significant for local hydrology and soil conservation. There could even be feedback leading to mutual site and climatic deterioration.


Opinions differ on the effects of forests on rainfall; some maintain that they cause no appreciable increase in precipitation (Walker 1916, among others) while some believe that the disappearance of forests leads to desertification. It has been conjectured that a lighted cigarette dropped in a Siberian or Canadian forest could bring about climate change, leading even to the onset of glaciation (Borisov 1973)! Wadia (1955) pointed out that the forests in Chinese Turkistan are not attracting rainfall but are drying out due to the general dryness of the climate. However, Kaulin (1962) asserted that the precipitation was higher in the forest than in the neighbouring steppe.

Nicholson (1929) contended that because in certain localities and under certain circumstances forests do not induce rain it does not mean that forests cannot induce rain. Harrington (see Luna 1981), on the other hand, stated that the evidence of higher rainfall within the forest does not mean that forests increase the rainfall, while Marsh (see Luna 1981), reviewing the literature, found evidence of decreasing precipitation subsequent to forest removal.

Different values have been attributed to the forest in augmenting rainfall. Thus Kittredge (1948) asserted that forest increased rainfall by 3%, 1% due to the trees of 30 m or higher obstructing air movement and 2% due to the effect of the friction of the canopy. Schubert's (1937) figure for Germany was 6%. Others estimate increases of 10 to 12% in the plains and up to 25% in the hills (Pinchot, see Luna 1981; Nicholson 1929; Hursh and Connaughton 1933; Fedorov and Burov 1967).

Recently the consensus of the hydrological community was that land use changes had no effect on rainfall (Pereira 1973). However, still more recent computer studies have shown significant changes in regional climate, including rainfall, due to surface changes. Dickinson (1980) explains that the differences are due to scale: hydrologists consider small areas and use limited observational data, whereas if the area studied was larger (of the order of several hundred square kilometres), the effects of vegetational change on atmospheric processes would be detected. Also, in the tropics crop yields and water resources over a large area are affected by what appears to be a negligible change of rainfall amounting to a few percentages. Moreover, convective rainfall (thunderstorms) is so variable that it is difficult to detect even moderate changes.

Lockwood (1980) thought that the disappearance of tropical forests would at most modify local climates and any effect on global climate would probably be swamped by natural changes and through an increase in the carbon dioxide content of the atmosphere. The most marked changes would be reflected in the hydrological cycle with increased runoff and an increased tendency to droughts in view of reduced soil water storage. However, he included changes in the vegetation cover among factors responsible for climatic variations such as fluctuations in solar radiation, for volcanic activity, tectonic movements, and changes in sea-level.

Dickinson (1980) considered that tropical deforestation would seriously modify local microclimates and, if sufficiently extensive, could change the climate of large regions in the vicinity of the deforested areas. If huge areas of rain forest disappear, even the global heat balance could be affected significantly. The effects of increases in carbon dioxide on radiation balance would outweigh the effects of albedo increase, at least for a few centuries until much of the carbon dioxide released due to clearance had been absorbed by the oceans. Climatic changes at the global level brought about by forest loss may be of the same magnitude as the natural climatic variability or modification brought about by burning fuels. However, Dickinson (1980) warns that deforestation combined with carbon dioxide released by the combustion of fossil fuels would add so much carbon dioxide to the atmosphere that it would worsen the situation at a global scale and regional changes following deforestation would be of greater magnitude.

Certainly not all arid zones are due to the absence of forest. Generally deserts owe their origin to their geographic location at subtropical latitudes on the western side of continents and are zones of large-scale descending air masses. Semi-arid zones, in certain cases, are located in special topographic situations, such as the areas in the lee of the Western Ghats (Bellary or Coimbatore) or in the shadow of Sri Lanka (Pamban, Tuticorin). Tropical coastal deserts as in Peru are deprived of rainfall by anomalously large annual ranges of sea surface temperature derived from cold advections. The coastal desert of southern Angola owes its existence to the coastal upwelling of cold water (Guilcher 1982), though the heavy rains occasionally recorded at Baia dos Tigres and Momedes in March may be linked to the meandering of the warm current temporarily reaching the coast and suppressing the upwelling.

Delannoy (1982) has demonstrated the influence of the sea surface temperature on coastal rainfall. The rain-producing efficiency of disturbances coming from elsewhere and the pressure field are controlled by the surface temperature of the sea-water close to the continent. The difference in temperature between sea and land is important in breeze formation. Up to a certain distance from the coast changes in direction or velocity of the breezes and their dissipation are greatest where the land surface warms up quickly, as near towns as opposed to forests (Escourrou 1982).

Anomalies in the tropical sea-surface temperature of only a few degrees may induce large changes in circulation, cloudiness, and rainfall (Julian and Chervin 1978; Rowntree 1978). An increase in ocean temperature by several degrees could raise evaporation rates by as much as 20 to 50%. If the change in evaporation rate involves an area of several thousand square kilometres, it can modify climate over and downwind of the perturbed area.

The mechanism of rainfall

Water Vapour in the Atmosphere

The principal source of water vapour is evaporation from the sea; during a typical monsoon day, it is estimated that about 75 x 109 tonnes of water vapour are transported across the west coast of India (Pisharoty, see Das 1968). The second largest source, about one-fifth of the former, is evaporation from water masses on land and transpiration from vegetation.

Warm air containing water vapour rises; with an increase in altitude, pressure decreases resulting in an expansion of the air mass; the rising air cools as it expands at the rate of 1 C for every 100 m rise in elevation. A stage is reached when the air is cooled beyond its saturation point and it can no longer hold all the water vapour in the gaseous phase; the surplus vapour condenses to form raindrops. A saturated cubic metre of air contains 39.4 g of water vapour at 35C; at 20C, 17.3 g; at 10C, 9.4 g; at 0C, 4.8 g; and at - 10C only 2.1 g. The condensation of the excess vapour releases latent heat to the air mass so that it rises still higher and the process continues.

The rising air will continue to ascend only as long as it remains warmer than the surrounding atmosphere at any level. Rising air tends to accumulate at higher altitudes preventing the upward motion of the air from below, unless wind at the higher level carries away the rising air.

The Ascent of Moist Air

Precipitation is classified according to the meteorological phenomena that cause the ascent of moist air and its condensation. Thus there are convectional rains (thunderstorms), mostly in the pre-monsoon months of April to May, cyclonic (depressional) rains, frontal rainfall, and orographic rainfall.

Depressions from the Bay of Bengal provide considerable amounts of rain during the monsoons. When a depression is formed in the bay, the pressure falls over an area of hundreds of square kilometres. Heavy rainfall follows the path of the cyclone (dictated by the pressure gradient) across the bay to the coast. The Western Ghats, with their orientation roughly parallel to the west coast, force the moist air currents from the Arabian Sea branch of the south-west monsoon to ascend, cool, and shed rain. Cherrapunji (which is devoid of forest growth) and Mawsynram (a nearby village) are probably the rainiest places in the world (rainfall of about 1,100 mm), mainly due to air being forced to ascend the southern slopes of the Khasi hills.

The Monsoon Phenomenon

The origin of the low-level monsoon current lies in the Southern Hemisphere, while the Arabian Sea supplies a major part of the moisture to the south-westerly flow. Earlier studies had estimated the contribution from cross-equatorial flow to be less than that from the Arabian Sea. Recent investigation suggests that the water vapour influx from the Southern Hemisphere into the Indian subcontinent is of the order of 30% (Cadet 1979). A later paper (Cadet and Reverdin 1981a) assigns a value as high as 70% to the water vapour coming from the Southern Hemisphere and 30% to that from the Arabian Sea.

The monsoon rainfall is greatly influenced by the Arabian Sea surface temperature. Numerical simulations have shown that a cold sea decreases the evaporation rate, increases the surface pressure, in turn reducing the cross-equatorial flow and the moisture flux. Thus a weak monsoon is linked to a considerable drop in the sea-surface temperature.

The water vapour cross-equatorial flux (40-100E) for the complex monsoon season (May-September) is 4.7 x 1010 t d-1, with maximum input (75% of the total flux) over the Arabian Sea, especially between 45 and 60E (50% of the total flux). The cross-equatorial flux into the Bay of Bengal is not negligible, being about one-third of that over the western Indian Ocean (25% of the total flux; Cadet and Reverdin 1981b).

The role of the Western Ghats in the generation of the rains has sometimes been rather exaggerated. The increase in the depth of the monsoon current from low levels over the Indian Ocean and the southern sectors of the Arabian Sea to about 6 km when 200 km from the west coast of India was erroneously attributed to the presence of the Western Ghats. In reality, the monsoon current begins to rise at a distance of about 50 km from the crest of the Western Ghats. Das (1968) favours the instability of the air, which adds buoyancy to the monsoon as it approaches the Indian landmass, as an important factor. Bannerji (see Das 1968) showed that a westerly stream of air, which the south-west monsoon is, may not be able to climb over the Western Ghats unless maintained by extra energy like the heat of condensation produced by the air losing its moisture as it is raised over the Ghats. Investigation by Sarker (see Das 1968) suggests that the Western Ghats may be responsible for about 60% of the observed rainfall and, on rare occasions, even 80%, by forcing the monsoon air to ascend. The peak rainfall occurs on the windward, coastal side at a distance of 10 to 12 km from the crest of the Western Ghats when the monsoon is in full vigour; the distance increases to 25 km when the monsoon is weak (Des 1968). During the southwest monsoon, most of the moisture gathered by the clouds from the sea falls on the coast rather than on the crest of the Western Ghats. Does this imply that there is a secondary source of moisture providing rainfall on the Western Ghats? If so, the forest cover on this mountain range could have a major role in generating humidity in the air.

FIG. 1. Daily rainfall curve

The opinion is often expressed that rainfall in India is reduced due to a southern shift of the monsoon. However, in the past the movements of ships recording rainfall were limited and so comparisons in time are unreliable. Also, the west coast zone of India (represented by stations such as Bombay, Karwar, Mangalore) is not affected very much; probably the humidity from the sea maintains enough moisture over this coastal belt.

In figure 1, Bombay shows a regular pattern of daily rainfall during the monsoon months June to September, while Agumbe (western Karnataka), a station in the Western Ghats that is still well forested, shows little evidence of dry spells. In contrast, Madikeri (Mercara), also from western Karnataka and in the Western Ghats, is undergoing deforestation and shows sudden high peaks followed by rainless spells. This apparent trend of rainfall at Madikeri is reminiscent of the phenomenon observed at many places in India: dry spells interspersed with torrential rains causing floods.

Condensation of Water into Raindrops

In this simple model of rain production there are five constraints:

  1. There must be sufficient moisture in the atmosphere.
  2. Moist air should be able to rise to a sufficient height for water to condense. If there is an inversion in the temperature gradient in the lower troposphere, the density gradient in the atmosphere will be extremely stable. Consequently, the air cannot rise to the point of saturation despite a high moisture content.
  3. Unless condensation nuclei (e.g. aerosols) are present in the atmosphere a very high degree of super-saturation is necessary for the conversion of the water vapour into raindrops. A relative humidity as high as 400% may be ineffective without them, whereas in their presence condensation can occur at less than 100% relative humidity. The most important condensation nuclei are salt particles from the sea. However, organic nuclei deflated from the surface litter aid in seeding cumuliform clouds (Schnell 1975). The loss of litter following deforestation may account for some reduction in the rainfall. Similarly, pollen grains are also thought to be successful cloud-seeders. When large wooded areas are denuded the supply of pollen grain dwindles and precipitation becomes erratic. Trials carried out with maize crops at Ajmer at the edge of the Thar Desert in Rajasthan are reported to substantiate this hypothesis (Ramaswamy 1979).

In rain-making experiments the number of nuclei in a cloud is artificially increased by releasing particles such as silver iodide into the atmosphere from an aircraft. Though the value of such cloud-seeding is still controversial (Sirinanda 1982), it was estimated to have increased rainfall by 15% to 200% in trials in Tamil Nadu (June-Sept. 1983). In Yugoslavia the farmers use rockets, which produce a chemical smoke on reaching a certain altitude, seeding clouds at a much lower cost than by aircraft. In India during temple festivals camphor is burnt to invoke the mercy of the Rain God. The smoke produced may serve as condensation nuclei. A fire in a camphor and wattle bark store-room at Madras resulted in a brief rain-shower, though the heat of the fire may also have intensified local convection (Narayanan 1 980).

  1. Ascent of moisture-bearing air to a sufficient height to produce rain may be prevented by descending air. The air heated over the equator rises and descends over the subtropical belt. Deserts such as those of Rajasthan, Pakistan, and Arabia are zones of large-scale descending air masses. Thus the water vapour in the air cannot condense into rain in these regions since it cannot rise and undergo expansion cooling.
  2. The large amount of dust over the desert increases the subsidence rate by as much as 50%. By cutting off a good portion of short wave solar radiation by scattering, the dust particles prevent it from reaching the ground (Angstrom 1962). Another major effect of dust loading is greater cooling and radiation divergence in the troposphere (Chakravarti 1978).

If the dust in the atmosphere could be eliminated, the radiation cooling would lessen and, in turn, there would be a reduction in subsidence, facilitating the ascent of air. As the air is already moisture laden, it would result in higher rainfall.

Bryson (1974) and Chakravarti (1978) have pointed out that the greater soil erosion resulting from the population explosion in the subtropics during the last four decades and the consequent pressure on grazing ground and the overuse of land in the Sahel and the Thar have not only increased the surface albedo but also the amount of dust loading. Increased dust particles in the atmosphere may lead to greater diabetic cooling and enhanced atmospheric subsidence, desiccation, and drought on the margins of the Sahara. Thus the dust loading theory provides an additional explanation besides the albedo effect on rainfall.

The role of dust in the climatic desertification mechanism has not been universally accepted. Otterman (1974) considers it relevant only in limited cases, for example the Rajasthan desert region. Desai (1969) doubts if the reduction in dust would increase either the rising of the moist layer or the rainfall.

Observations on the effect of forests on rainfall

The observation is credited to Christopher Columbus that rainfall in the Canary Islands and Azores declined after their forests had vanished and that the afternoon rain in the West Indies was due to the island's luxuriant forests. Early workers like Weikoff (see Blanford 1888) attributed the higher rainfall of Assam, compared to the intensively cultivated Ganga valley, to the extensive forest mantle of the former, not to the physiographic differences. Ribbenthrop (see Blanford 1888) gave another example from central India, where the ban imposed on shifting cultivation in certain parts in 1875 had supposedly led to an increase in precipitation at 14 stations during the following decade. No such increase was noted where shifting cultivation continued. Blanford (1888) observed that the apparent increase noted in the second decennial period was well within the probable error range of 5% for a 10-year average. The stations showing an increase were south of those showing a decrease and the difference might have been due to the paths of the storms of the south-west monsoon. However, the running combined mean rainfall for the 14 stations showed a progressive increase after 1875 that he claimed to be parallel to the gradual regrowth of forest vegetation. These claims may be questioned on several grounds. Firstly, the decadal means are themselves subject to natural fluctuations as appreciable climatic variation may occur from one 30-year "normal" period to another (Mitchell et al. 1966). Further, the rainfall data analysed by Blanford are not for individual stations but means of 14 stations. Thirdly, it is doubtful that shifting cultivation (recently still prevalent) abruptly stopped in 1875.

The likely deterioration of climate due to deforestation was a major argument used by the nineteenth-century conservationists to protect forests (Thompson 1980). In a report entitled "Improvement of Indian Agriculture" by a Dr. Voelcker, it was pointed out that at Udhagamandalam (Ootacamund) the number of rainy days (excluding those of June, July, and August when the rains are not of local origin) increased during the fiveyear period 1886-1890 to 416 when the area was wooded. During an earlier five-year period, 1870-1874, when the area was treeless, there were only 374 rainy days. Ranganathan (1949) provided figures up to 1922. In table 1 I have compiled selected fiveyear totals of rainy days and rainfall. Despite fluctuations, these always show considerably fewer rainy days than even the nineteenth-century treeless period for the September to May interval.

TABLE 1. Periodic fluctuations in rainy days and rainfall at Udhagamandalam

Period Rainy days Rainfall (mm)
Total Excluding June to August Total Excluding June to August
1902-1906 522 312 7,386 4,226
1918-1922 505 326 6,637 4,483
938-1942 540 303 7,055 4,412
1948-1952 417 242 5,583 3,389
1958-1962 590 351 7,289 4,340
1968-1972a 455 276 6,290 4,400
1978- 1982 448 271 5,637 3,871

aData missing for 5-25 December 1971

Legris and Blasco (1969) also noted a diminishing trend of rainfall for Udhagamandalam from the turn of this century. The frequency of dry years (registering less than 1,300 mm) increased from 8 years between 1902 and 1922 to 12 years between 1954 and 1964, with 6 consecutive years below this mean value. Von Lengerke (1977) states that the below-average rainfall of the years 1967 to 1970 at Udhagamandalam led farmers and planters to suggest a change of climate, but he himself found no clear indication of a large-scale shift in rainfall. All the same, he noted comparatively low averages during the late nineteenth century and above-normal 10-year averages from 1955 to 1965, with a subsequent decrease.

FIG. 2. Climate diagram of Udhagamandalam

For Udhagamandalam the average monthly rainfall and temperature data are presented in figure 2. The yearly fluctuations in annual rainfall and annual number of rainy days as well as 7-year, 10-year, and 20-year moving averages are presented in figures 3, 4, and 5. These reveal cycles of above-normal and below-normal rainfall years (with record lowest rainfall of 800 mm in 1982) as opposed to an increasing trend as reported by Von Lengerke (1977) in his study based on data up to 1970.

Though an expert committee appointed by the government of Tamil Nadu did not find sufficient evidence to support the view that the failure of rains in the Nilgiri district was because of large-scale planting of Eucalyptus globulus (C.C.F. 1971), it would be interesting to seek a correlation between the percentage of forest cover at different periods and the trend of rainfall and rainy days. Figure 6 depicts land use in the Nilgiri district. It shows the area remaining under forest at present as well as the area that was under forest until about 1940 (as shown in the Survey of India toposheet) but has now been converted to degraded vegetation and savanna, arable land, lakes or transformed into plantations for tea, coffee, or forest species (Gaussen et al. 1961; Pascal, Shyam Sunder, and Meher-Homji 1982; Bellan 1987).

FIG. 3 Rainfall at Udhamagamandalam

FIG. 4. Moving averages of rainfall (Data to 1970 from Von Lengerke 1977)

FIG. 5. Number of rainy days at Udhagamandalam

FIG. 6. Nilgiri district land use (after Pascal et al. 1982, Bellan 1987)

The alarm caused by Winstanley (1973) stating that India was getting progressively drier has subsequently been discounted by several studies (Agrawal 1976; Chowdhury and Abhyankar 1976; Legris and Meher-Homji 1976; Mukherjee and Singh 1978; Singh 1978). Meteorologists have shown that precipitation in India does not present any trends, but their investigations are limited to certain cities and towns that have long lost their forested surroundings.

As to the effect of change of vegetation on climate, the following studies provide evidence of a tendency for rainfall and rainy days to decrease with a drastic decrease in forest cover.

  1. Warren (1974) for the Ranchi Plateau,
  2. Sarmah (1976) for Dibrugarh,
  3. Padmavalli (1976) for the Nilgiri district,
  4. Meher-Homji (1980a, b; 1982) for western Karnataka and Kerala,
  5. Biswas (1980) for Andaman and Car Nicobar Islands,
  6. Mishra and Dash (1984) for Sambalpur,
  7. Mukherjee, Pradhan, and Rao (1976) for Santa Cruz, Bombay, and
  8. Singh et al. (pers. com.) for Kumaun.

Raju (1981) confirms the following observations of Meher-Homji (1980a, b) on the effect of deforestation on rainfall for the Uttara Kannada district of Karnataka. The larger the area of deforestation the higher the decline in precipitation. Table 2 reveals that in this district, below the Western Ghats, rainfall has increased in the last 20 years in spite of forest destruction, whereas in the Up-Ghat talukas rainfall has declined with increasing deforestation. Yellapur, in spite of considerable loss of forest, registers an increase in precipitation. This may be because of the ability of the remaining forest to attract rainfall. Nevertheless in other talukas, where deforestation is greater, rainfall shows a tendency to decrease progressively: Mundgod, Haliyal, and Supa are already experiencing aridity; Sirsi, Siddapur, and Karwar are showing signs of becoming drier.

At first, deforestation may not have much effect on rainfall, but in the long run it certainly does. If aridity increases with further loss of forest, the remaining forests may start disappearing, being unable to withstand the drier environment. In coastal areas, despite considerable deforestation, reduction in rainfall is negligible, probably due to the high humidity.

Nicholson (see Ranganathan 1949) observed that the Chota Nagpur region, which was well forested at the turn of the century, received fairly frequent afternoon showers (instability rain) during summer that favoured tea plantations. Following the destruction of private forests, in spite of no apparent reduction in the monsoon rainfall, the instability rain has decreased so much that the tea gardens have disappeared. Warren (1974) attributes the decline in rainfall of the pre-monsoon months May and June over the Ranchi Plateau to the degradation of forests to scrub over an extensive area.

Walter (see Brooks 1928) noted that in Mauritius the reduction of the forested area from 25% to 5% was associated with a decline of 4% in the annual rainfall and a corresponding reduction in the number of rainy days, and thunderstorms and cloud bursts were less frequent. According to Brooks (1928) the changes occurred on the seaward side, whereas the leeward side remained unaffected. Hughes (1949) supported Walter's conclusion that the presence of forests increased the probability of local showers on calm afternoons.

Circumstantial evidence of lasting and profound changes being initiated by forest clearing is provided by long-term observations on rainfall in private rubber plantations in Malaysia (Unesco 1978). After large-scale forest clearing the number of rainfall incidents decreased, the rainfall intensity substantially increased, while the total rainfall appeared to remain unaffected.

Contents - Previous - Next