Contents - Previous - Next

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

Problems in climate-agriculture relationships: Rice yields in three areas

Sri Lanka

The case of paddy (rice) production in Sri Lanka is of relevance, because it demonstrates the complexities of climate-agriculture relationships in an area that has several elements in common with South-East Asia. It is discussed with reference to results obtained by Yoshino and Suppiah (1984).

There are two cropping seasons in Sri Lanka corresponding with the northeast monsoon, or Maha season, and the south-west monsoon, or Yala season. Sowing dates in the former season extend over several months, but over only a short period in the latter (Yoshino, 1984b: 95). Most of the dry-zone districts (roughly, the north-eastem part of Sri Lanka) show a significant relationship between harvested area and rainfall in the Maha season, while in the wet-zone districts (roughly, the south-westem part of Sri Lanka), the significant relationship is with rainfall in the Yala season. But there are some districts which show strong relationship in both seasons, as they are affected by both monsoons. Figure 8.3, calculated from data over 20 years (1960 80), relates deviations in harvested or sown area to seasonal rainfall deviations for three places.

In the dry zone, Maha season rainfall which is less than one standard deviation below the mean has caused severe crop losses, extending into the subsequent Yala season and, in some cases, into the following Maha season. The failure of the south-west monsoon (Yale) results in water-deficit conditions in the wet-zone highland paddy lands. Excessive rainfall in either or both seasons causes floods and waterlogged conditions in the lowlands of the country. In the driest part of Sri Lanka, there is a unique relationship in the range of anomalous negative departure of rainfall below -10 centimetres: the relationship is positive; but above -10 centimetres, the rainfall has no effect on the harvested area. The reasons for this relationship are not clear.

FIGURE 8.3 Three Types of Relationships between Rainfall Deviation and Harvest Deviation in Sri Lanka

In Figure 8.4, the secular changes of total paddy area harvested and sown for the whole of Sri Lanka over 20 years, and the average seasonal rainfall of all of Sri Lanka, are presented, using data drawn from Yoshino et al. (1983). From this figure, it can be seen that paddy production is influenced more by rainfall in the Maha season than in the Yala season. In fact, the area sown minus area harvested has the closest relationship to the rainfall. A total product departure from the calculated value, which is an experimentally calculated value in view of the secular increase in area, is significantly correlated for the Maha season, but not for the Yala season. However, absolute values of sown area, harvested area and total product have no significant correlation with rainfall. This is due to the sharp increase in these values during the 20 years under consideration, not to an increase in the rainfall parameters.

On the basis of individual years, a number of other interesting effects can be seen; for instance, a low Yala rainfall coincided with low harvested and sown areas in 1976, but not in 1966. Table 8.1 presents the years of minimum sown and harvested area: 1965, 1969, 1972, 1973, 1975, 1976 and 1979. These are compared with minimum points on the rainfall curves, and the years marked according to perfect fit, relatively good fit and non-fit between rainfall and agriculture. Years for minimum sown and harvested area in the Yala seasons of 1965, 1969, 1972 and 1975 correspond to the rainfall minima in the preceding Maha seasons, in each of which, rainfall was below about 800 millimetres. Table 8.1 also compares conditions of drought in the 1964-5 and 1974-5 Maha seasons. In the 1974 5 Maha season, the drought was most serious, rainfall being less than two standard deviations below the mean. The sown and harvested areas, however, were smaller in 1964-5 than in 1974 5, not because of rainfall but because of development of cultivation techniques and irrigation systems.

FIGURE 8.4 Secular Change of Rainfall, Area Sown and Area Harvested for the Whole of Sri Lanka, 1960/1-1979/80

TABLE 8.1 Fitness of Minimums of Yala Sown Extent (Y.S.) and Yala Harvested Acreage Y.H.) to the Minimums of Yala Rainfall (Y.R.) and Maha Rainfall (M.R.) in Sri Lanka, 1961-1980


Y.R. Minimum


Y.S. Minimum Y.H.. Minimum M.R. Minimum M.R. (mm)
1965 1965 x 741
1969 1969 x 787
(1972) 1972 x . 766
1973 (1973) o x 1 008
(1975) 1975 x . 585
1976 (1976) o 825
1979 1979 o x 1 036
    Average (1961 -80) 977

Source: Yoshino et al. (1983).
Note: = perfect fit: o = relatively good fit; x = non-fit.

The timing of cultivation is a serious problem for water management in Sri Lanka. It depends upon water being available at the proper time. The phenomena described above show something of the real nature of this problem. Domroes (1978) wrote that the relationship between crop yields and drought conditions is complicated, and that rainfall conditions may not immediately affect crop production. It is therefore of interest that in the worst cases noted here, when deficient Maha rainfall carried its effect over into the succeeding Yala season, Maha season rainfall was less than a standard deviation below the mean. Perhaps this represents a critical climatic value in this complex relationship.

Indonesia: The Drought of 1982 in Java

Indonesia covers a wider climatic range than Sri Lanka. In the months between November and February-March, the northwesterlies bring humid air, in a season generally termed the north-west monsoon. There are some local differences in rainfall distribution in accordance with topography and position in relation to the prevailing wind direction. During the period from April to October-November, Indonesia is influenced by the relatively dry southeasterlies from across Australia.

Of the total area of Indonesia, around 14.2 million hectares are classified as arable land. This arable land can be divided into several types, among which rice-field (sawah) land and upland are most important. Sawah land, in which rice is grown under flooded conditions, can be subdivided into irrigated, rainfed and swamp land. Upland agricultural areas have greater variety (Oldeman, 1984). Three-fifths of sawah land and more than one-third of the upland areas are located in Java.

Since 1967, the average yield of rice has increased markedly, from a fairly constant 2 000 kilograms of rough rice per hectare, to reach 3 170 kilograms per hectare in 1978. Yields in the dryland areas showed a much smaller improvement, from 1 100 kilograms per hectare in 1968 to 1 300 kilograms per hectare in 1978. Because of the large year-toyear variation of rainfall, and the differing water supply conditions, there is great variety, even in Java, in the seasonal patterns of planting and harvesting. Table 8.2 shows the contrast between different parts of the island. The main harvest peaks are between April and June, which coincides with the end of the wet season, but there are different crop calendars in some areas. In Tuban, East Java, there is a sharp concentration in the harvest period, soon after the end of the short December-to-April rainy season; elsewhere, multiple cropping gives a wider annual spread of production peaks.

Malingreau (1987) has analysed the effects of the severe drought of 1982 in Java. In the main producing areas, the 1982 dry-season crop was planted in April-May, more or less on schedule, but started to suffer from the water shortage as the season progressed. In all parts of Java, the harvested area was reduced more than 10 per cent in 1982, compared to previous dry seasons. Then, because of the late arrival of the monsoon at the end of 1982, wet-season planting was delayed, so that the rice was in the ground only by the end of December instead of early November, as normally. The resulting compression of the 1983 wet-season harvest in April put heavy pressure on harvest labour, and on the milling and storage systems. Moreover, many farmers planted corn rather than rice, because the rainfall was still insufficient at the end of 1982. This led to a further reduction in rice production in 1983, especially in East Java.

TABLE 8.2 Share of Monthly Harvested Area of Wetland Rice in Total Annual Harvested Area in Selected Regions in Java (per cent)

Region J F M A M J J A S O N D
West lava  
Garut 6 8 6 8 14 12 9 7 6 6 7 11
Karawang 0 0 0 7 32 13 1 1 16 24 4 1
Central Java  
Kebumen 0 5 22 17 7 4 15 21 6 1 0 0
Sragen 4 20 10 13 17 15 7 5 1 1 1 0
East Java  
Tuban 1 0 1 15 43 23 2 1 3 5 2 1

Source: Oldeman (1984).

Supplementing production data with innovative use of remote sensing, Malingreau (1987) concluded that the most important effect of the drought was a drastic reduction in the growth rate. A satellite-derived vegetation index for an area in West Java showed a peak value in June 1982 only 60 70 per cent of that during other dry seasons in a threeyear period; there was then no recovery of the vegetation index in the second half of the year. The drought affected both the dry-season crop of 1982 and delayed the planting of the 1982-3 wet-season crop. This is a similar pattern to that found in Sri Lanka, discussed above.

Hainan Island, South China: The Tropical Margin

A different sort of climate-agriculture relationship is exhibited on Hainan Island, South China. The northern boundary of the true Tropics in South-East Asia might be placed at the southern limit of cold-wave invasion in winter; these waves extend into continental South-East Asia. The frequency and magnitude of cold waves differ according to largescale synoptic events and regional-scale topography (Yoshino, 1988). In the eastern part of South China, cold waves occur quite frequently under the influence of anticyclones over the region. In the most striking cases, the temperature fall at ground level may be more than 20 C. The so-called 'cold-dew winds' (hanlufeng) which blow during the flowering period in the autumn are also important (Yoshino, 1984a, 1986).

Rice production in Hainan has increased substantially since the introduction of International Rice Research Institute (IRRI) varieties in the mid-1970s. With fastermaturing cultivars, there has been a substantial increase in the early rice crop, to now about four-fifths the size of the late rice crop and, under good conditions, giving even better yields. However, the full potential productivity is not realized. The early rice cultivars present the major problem. They are sown from December to May and harvested in 160-170 days. Solar radiation in the ripening months is more than adequate, but the greater part of the growing period of early rice is the dry season. Irrigation is necessary, and winter sweet potatoes are also grown as a dry crop. But the actual crop calendar depends strongly on the incidence of cold waves, winter monsoons and also typhoons in late summer. Combinations of crops would provide better insurance against these hazards.

Flow of impacts of climatic change on agriculture

Summarizing the impacts of global environmental change since the 1960s caused by human activities on agriculture, forestry and fisheries, a tentative flow chart derived from Yoshino (1991) is shown in Figure 8.5. The striking acceleration of human activities, including population increase during those years, is the starting point of the flow. It results in the set of effects labelled 'Environmental Change 1'. Through the warming by greenhouse effects, these lead directly to the second group of consequences. The increase in photosynthesis is an important element here, being a response of plants, and is tentatively given equal weight with changes in the atmosphere, oceans and soils. Further consideration is needed on this point. This second group of changes then leads to impacts on agriculture, forestry and fisheries, which in turn feed back to the beginning stage of the whole process, caused by the expansion of arable lands, destruction of natural vegetation, change of land use and accelerated desertification. There are also feedbacks in the ocean and atmosphere.

FIGURE 8.5 Flow of Impacts of Environmental Change on Agriculture, Forestry and Fisheries

The problems listed in the last box of the figure are particularly important in the Asian Tropics. What is not shown, however, is adjustment or adaptation of the relationships between climate and agriculture. Parry (1990), Parry, Mendzulin and Sinha (1990) and Parry and Zhang (1991) have demonstrated the need to examine possibilities which include changes in land use, crop type and crop location, and changes in technology such as irrigation, fertilizer use, control of pests/diseases, soil management, farm infrastructure, and crop and livestock husbandry. These presently unquantifiable possibilities should be studied region by region in the Asian Tropics, because rice cultivation supports high densities of population and can be greatly affected by both climatic change and such adjustments and adaptations.

Concluding remarks

Climatic change can only be discussed against present expectations and, though the future remains unclear, there can be no doubt that there will be large regional differences in the Asian Tropics (Yoshino and Urushibara, 1981). The rapid rate of population growth makes the potential impacts of climatic change on agriculture very serious in this region. Regional scenarios should be studied in relation to the effects of global warming on agriculture, as well as of monsoonal variations. Different crop-climate relationships in nearby regions in the same season, and also in the same region in different seasons, need to be taken into account. As has been seen, drought may project its influence on to the following season's crop by delaying the next planting. Moreover, crop substitution may aggravate the impact of drought on food supplies. Cold waves in the border region of the Tropics cause damage to tropical crops. and future crop selection needs to be taken into account. Co-operative studies on these problems should be internationally conducted. The International Geosphere-Biosphere Programme (IGBP) and the Human Dimensions of Global Environmental Change Programme (HDGEC) should provide the vehicles for such co-operation.

Climatic change in Indonesia

The ongoing indonesian climatic change
Possible impact on rice
Editorial comment



ALTHOUGH modern man has, in his history, experienced climates warmer than the present one (Lamb, 1982), he has no knowledge of the consequences of man's interference with the global climate. Simulation models, such as the General Circulation Model (GCM), are therefore useful tools, which provide us with an idea of the consequences of human tampering with the climate. The differences in the output of the models, as presented by Henderson-Sellers (Chapter 6), merely show that the present understanding of the mechanisms of climate does not match the need to foresee the future. One point to be learned from the models' output is that increasing the earth's atmospheric carbon-dioxide (CO2) content will increase the atmospheric temperature. They further show that the temperature increase will differ between latitudes, as well as between seasons.

The ongoing indonesian climatic change

The Indonesian Committee on Climate Change Monitoring directed that the simulated change should be compared with the ongoing one. This requires simulating climatic change from an initial 286 parts per million of CO2 to an assumed final content of 340 parts per million, as exists in the early 1990s. Since the means to model transient climatic change does not exist, the findings of the observed situation will be presented.

Hidayati (1990) has studied the change of climate in Jakarta and the surrounding areas. She found a very significant change of 0.03 C per year in the maximum temperature. The increase through 1949-87 was smaller than that over 1916-87; while between 1970 and 1987, the change was negative. The change during the east monsoon (June-August) was larger than that during the west monsoon (December-February). The minimum temperature, although also significant, increased only by 0.01 C per year over 1916 87. In contrast to the change in the maximum temperature, the minimum temperature shows a progressively higher rate of increase over time. Seasona] change fluctuated, notably showing a decrease over 1940 70. However, Hidayati (1990) went on to show that the total change is partly, if not totally, due to enhancement of the heatisland effect brought about by Jakarta's increasing population (Figure 8.6).

Rainfall over 1864 1987 increased by a mean 0.5 millimetres per year, with a highly significant positive change during the west monsoon and a nonsignificant negative change during the east monsoon. This suggests that the fluctuating rate of temperature changes could be affected by the change in the heat-balance components, which will camouflage any true global change presumed to be taking place at present.

To resolve the question of whether there is a climatic change anywhere in Indonesia which is not local in character, the Agrometeorology Group (1991) at the Bogor Agricultural University recently undertook another study. It focuses on data from climatological stations which are thought to be little affected by the development taking place in Indonesia. The preliminary result shows that of the 12 stations examined, 8 exhibited a definite change in the last 15 years. The magnitude was between 0.29 and 0.63 C, or about 0.02-0.04 C a year since 1970. Figure 8.7 shows the result at Kenten. It might, therefore, be assumed that there is a real change of climate in Indonesia. The negative changes found by Hidayati, for Jakarta, could perhaps be ascribed to feedbacks in one or more climatic elements, resulting from the development of the city.

Possible impact on rice

Notwithstanding the debatable magnitude of the simulated changes of climate, the study provides a basis for the estimation of potential impacts of climatic change. The Goddard Institute for Space Studies (GISS) model output was used for the changed-climate scenario. The dependent variable was the yield of lowland rice in the northern coastal plain of the Citarum River basin, West lava. Results from the empirical model by Irawati (1988) showed that the yield during the January-June harvest decreased by an average of 3.6 per cent; on the other hand, that of the July-December period increased by 3.0 per cent. The annual change was an increase of 0.1 per cent on the average (Figure 8.8).

FIGURE 8.6 Annual Average Maximum Temperature against Population at Jakarta. 1916-1987

FIGURE 8.7 Monthly Average Temperature at Kenten (South Sumatra). 1975-1990

These results were checked using a basic crop growth simulation mod_l (LID), outlined by Penning de Vries et al. (1988) for the three planting seasons of 1983. It was found that while the yield of the June planting under the changed-climate scenario decreased slightly, the other two either equalled or exceeded the yield under the present climate. In Hokkaido, Japan, Yoshino et al. ( 1987) found increases of up to 25 per cent over the yield under the present climate. Thus, there is no reason to doubt the results of Irawati's model, and it may be concluded that climatic change alone would not alter rice yield seriously.

However, rice production could suffer serious set-backs from secondary causes. First, there would be heavier erosion in the upstream area, which may have to be abandoned and reforested. Secondly, some of the fertile coastal alluvial land would be inundated by the sea-level rise. The three coastal districts of the Citarum River basin would lose a total of more than 20 000 hectares of paddy fields. In the district of Subang alone, more than 25 000 hectares would be inundated, of which almost 12 000 hectares are irrigated farm lands which, with two plantings annually, produce about 110 000 tonnes of rice and almost 4 000 tonnes of maize and soya bean. To maintain the present level of the three districts' production, the yield would have to be increased by 37.5 per cent beyond the current yield.

FIGURE 8.8 Comparison of Yield Ratios under 2 x CO2 and 1 x CO2 Climate, 1974-1984

Yoshino et al. (1987) present data indicating that by planting mid-to-late maturing varieties, the negative impact of the climatic change could be reversed. In fact, with this technology, rice yield in Hokkaido could be boosted under the 2 x CO2 climate, to surpass the present yield by 16 (lowest) to 47 per cent (highest). It might therefore be necessary to switch to late-maturing varieties to balance the loss in production in the inundated coastal plains and the abandoned upstream areas. Whether this would compensate for the total loss in production remains to be tested.

One important aspect of global climatic change remains to be clarified. This is the effect of the increased atmospheric carbon dioxide (CO2) content on the rate of photosynthesis. None of the models considers the increased CO2 concentration. By logic, however, the increased air temperature is expected to speed up the biochemical processes in the photosynthetic chain. In turn, the carbon intermediate sink is kept constantly large. Add to this the higher ambient CO' concentration, and the CO2 gradient would be large at all times. Even if the temperature range of CO2 assimilation, as reported by Uchijima (1975 6), is controlled by stomata! closure, the CO2 assimilation rate can be expected to be higher than at present.

Thus, rice yield could benefit from the atmospheric CO2 increase, assuming there is no reduction in the irradiance of the surface. This hypothesis needs to be tested in an environment of increased temperature and relative humidity.

Editorial comment

De Rozari particularly stressed the serious consequences of a rising sea level for Indonesia, with so much of its best rice land very close to sea level. Even a small rise might be sufficient to cause the country to lose the self-sufficiency in rice production it has attained. Especially for Java, there is a need to identify the most vulnerable areas, and to decide which of them is worth engineering protection. There is also a need to determine which upstream areas need to be abandoned and revegetated.

The problem of sea-level rise was also discussed in relation to Bangkok and southern Thailand. The question is complicated by upstream dams, behind which it is necessary to hold water during the dry season to sustain electricity generation. This has the unfortunate effect that, during the dry season, salt water reaches further inland, adversely affecting tree crops; moreover, pollution can also extend upstream, to reach the water intakes used for domestic supply. In a marginal situation, the effects of sealevel change, rainfall shortage and in terference are all present and not easy to separate. Sensitivity to the consequences of global climatic change in these environments is, however, very clearly needed.

Further comment was provided by Rerkasem of the Multiple Cropping Centre at Chiang Mai University, Thailand. He too emphasized the importance of recognizing urban heat-island effects in evaluating long-term temperature trends. In regard to Yoshino's Sri Lankan data, he confirmed the generality of the relationship between rainfall and the planted area. In the lower part of northern Thailand, the transplanting time of photosensitive rice can often be upset by the unreliable onset of rainfall (CMU/CUSRI, 1983). Wet-rice fields have to be prepared as soon as the amount of soil water is adequate for ploughing. If the onset of the rains is delayed, the time available for land preparation, between ploughing and transplanting, is shortened so that, with inadequate farm labour and machinery, significant areas remain unplanted. In drought years, the reduction in the planted area can be as high as 30 per cent, and serious losses of planted cropland occur in about two years out of every five.

Rerkasem also called attention to some consequences of the thinning of the ozone layer, permitting more Ultraviolet-B (UV-B) radiation to reach the earth's surface. This might be beneficial in altering the competitive balance between crops and weeds (IRRI, 1990). On the other hand, it may be that UV-B radiation plays a vital role in pollen germination in crop plants (Jackson and Linskens, 1979). If this were upset, then pollen failure would, if complete, cause total loss of crop yields.

Contents - Previous - Next