Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
Part II - Climatic change and variability
Introduction
6. Climate model predictions for the south-east
Asian region
Climatic
change and public policy
7. Enso, drought and flooding rain in south-east
Asia
A successful prediction using unconventional
data
8. Climatic change and agriculture: Problems for
the Asian tropics
Climatic change in Indonesia
THE potential consequences of predicted global climatic change hang over all prospects for sustainable development as a threat that cannot be measured, and worse still it cannot be ignored either. More immediately, rainfall variability is increasingly seen not only as a factor of considerable significance in agricultural production but also as important in forest ecology. South-East Asia is strongly affected by variability arising from extremes in the Southern Oscillation. At the time of the Yogyakarta conference in May 1991, large parts of it, including Java, were entering a new period of drought from this cause. Had the conference been held a few months later, this event and its consequences might have been quite prominent in the discussions.
The two questions are linked, because the future behaviour of the Oscillation through a period of more rapid global warming is an unknown-as important in its potential consequences as the warming itself except, principally, in coastal areas that will be grievously impacted by rising levels of the sea. As shall be seen, the probability is that variability due to the Oscillation will continue, and the possibility is that its intensity may change at both extremes. At a later meeting in Bangkok in 1991, Nicholls developed this argument by logic from what is known about the phenomenon and its history, but it was added in discussion that new climate models suggest an increase in intensity is more likely and, moreover, that it could lead to lower soil moisture in the Indonesian/Australian region.
The three chapters in this part of the book address these issues. In Chapter 6, Henderson-Sellers writes a spirited and informative account of all that can be said about the consequences of global warming for South-East Asia. Using principally the somewhat equivocal results of the General Circulation Models, she attempts to extract elements of a regional consensus. Her firmest conclusion is that the state of knowledge is still too imperfect to draw conclusions at regional level.
She is followed in Chapter 7 by Nicholls on the El Niño-Southern Oscillation extremes and their effects. Nicholls has several times demonstrated the close relationship between variability in the east and north of Australia and that in the Indonesian region. He draws on material from both areas, especially on the biological consequences of the exceptional dry and wet events that arise. He demonstrates the need to make use of unconventional data in establishing a historical and continuing pattern, and in discovering unexpected outcomes.
Chapter 8, by Yoshino, deals specifically with the effects of climatic variability and change on agriculture, drawing on case material from Sri Lanka and Hainan as well as South-East Asia; he offers a general model of the flow of climatic change impacts through resource-use systems.
Four discussants addressed these three papers, and three of them provided substantial additional information of great interest. In edited form, they are reproduced following the substantive chapters.
6. Climate model predictions for the south-east Asian region
Human-induced climatic change
Predicted climatic changes: The global view
Climate model predictions and the south-east
Asian region
Uncertainties
and unknowns
ANN HENDERSON-SELLERS
THE Intergovernmental Panel on Climate Change (IPCC) Reports represent a current and state-of-the-art review of the issue of human-induced climatic change (Houghton, Jenkins and Ephraums, 1990; IPCC, 1990; Tegart, Sheldon and Griffiths, 1990). The Climate Change: The IPCC Scientific Assessment (hereafter cited as the IPCC Scientific Report) asserts: that we are certain that there is a natural greenhouse effect operating on the earth; that we are certain that emissions resulting from human activities are substantially increasing the atmospheric concentrations of the greenhouse gases carbon dioxide, methane, chloro-fluorocarbons and nitrous oxide and that these increases will cause an additional warming of the earth's surface; that current models predict a I °C increase in temperature, above 1990 temperatures, by 2025 and a 3 °C increase in temperature by 2100; and that we can calculate with confidence that immediate reductions of over 60 per cent would achieve atmospheric stabilization at a level which would be achieved by the doubling of carbon dioxide (CO2) over pre-industrial levels by 2100 (IPCC Policymakers' Summary, 1990).
The Climate Change: The IPCC Impacts Assessment (hereafter cited as the IPCC Impacts Report) (Tegart, Sheldon and Griffiths, 1990) contends that natural, terrestrial ecosystems forced to migrate poleward or to higher elevation may be unable to do so, because of lack of available routes and/or because of the speed of the human-induced changes to climate. Global food production can be maintained at the same level as would be possible without greenhouse-induced warming, but the costs of maintaining that food productivity are unclear. The IPCC Impacts Report also indicates that declining food productivity may occur in regions which are already highly vulnerable to climatic, economic and other stresses and, as a result, changes in patterns of world agricultural trade are likely.
The Climate Change: The IPCC Response Strategies (IPCC, Working Group 3, 1991) asserts that the potentially serious consequences of climatic change are sufficient reasons to begin adopting response strategies immediately, especially those that can also be justified on other grounds, even in the face of significant uncertainties. These suggested response strategies include increasing efficiency in energy supply and energy end use, and a review of energy pricing, agricultural practices, sustainable forest management and reforestation programmes, and the increased use of energy sources with lower or no greenhouse gas emissions.
These three reports were debated and their conclusions agreed during the scientific section of the Second World Climate Conference. They also formed the background to the ministerial section of the same conference. The ministerial statement issued at the end of the Second World Climate Conference in November 1990 called upon the United Nations to initiate negotiations on a framework convention on climatic change. Although such a convention was finally signed at the United Nations Conference on Environment and Development, at Rio de Janeiro in June 1992, it lacked the specific emissions target which had been sought, or any agreement on the questions of tradable emission quotas, technology transfer and agricultural and forestry practices.
Predicted climatic changes: The global view
The IPCC Consensus
The IPCC Scientific Report offers the best currently available model predictions, indicating a global average temperature increase of I °C by 2035 and 3 °C by 2100. These predicted increases, based upon the 'business-asusual' emission scenario, are above present-day values, so that by the end of the twenty-first century it is anticipated that global average temperatures will be 4 °C higher than their pre-industrial levels. Current numerical climate models indicate that the global hydrological cycle (evaporation, cloud formation and precipitation) will intensify by very roughly 10 per cent (the IPCC range is 3-15 per cent) of present-day values for a 3 4 °C temperature rise. The IPCC 'business-as-usual' emission scenario foresees a doubling of atmospheric CO2 concentrations over pre-industrial levels by about the middle of the twenty-first century.
It is important to recognize that CO2 is often used as a radiative surrogate for all greenhouse gases. Thus, model predictions determined on the basis of 'doubled CO2' are intended to capture the combined warming effects of all the added greenhouse gases. The doubling of 'equivalent CO2' will occur significantly earlier than doubling of CO2 alone. This distinction is unimportant in terms of radiation, but must be recognized because of the different atmospheric lifetimes of the different gases and because CO2 has a direct fertilizing effect on plant growth, while the other greenhouse gases do not.
Globally, then, the concentrations of atmospheric CO2 and other greenhouse gases will rise, mean temperatures will rise, and precipitation and evaporation will increase. Of these increases, only for CO2 is it possible to presume that changes in local and regional average values are directly equivalent to changes in global average values. This is because atmospheric CO2 is well mixed globally (see, for example, Pearman, 1989).
The global changes in average temperatures, evaporation and precipitation subsume significant latitudinal variations and regional characteristics. It is agreed that the northern high-latitude regions will warm substantially more in autumn and winter than the global average (about an 8 °C increase for a 4 °C global mean warming). This result is induced in part by the ice-albedo feedback effect, but is mitigated in the southern hemisphere by the ocean circulation. On the other hand, Equatorial regions will warm relatively little, probably by only 1-2 °C.
There are a number of consequences of this latitudinally differentiated warming. The first, and most obvious, is that the Equatorial-to-pole temperature gradient will be decreased in the northern hemisphere and, less certainly, in the southern hemisphere. It is this temperature gradient which, in combination with the rotation of the earth, drives the atmospheric circulation. An important consequence of changing atmospheric circulation patterns is that the roughly latitudinal climatic 'belts' will migrate polewards. This will have the effect of expanding the Equatorial region, which is dominated by the Hadley cell circulations, and shifting the mid-latitude depression belts towards the poles. In general, continents will warm faster than the oceans and the mid and high latitudes (especially in the northern hemisphere) will warm more than the Tropics.
Regional changes in climate are virtually impossible to determine with the present generation of numerical climate models, although a few broad statements can be made about the regional implications of the latitudinal and lan/locean shifts described above; for example, most global model simulations indicate that the Asian summer monsoon will intensify. The impacts of the ocean warming itself include rises in the sea level and the possible poleward expansion of the area affected by tropical cyclones. These single-variable regional climatic effects are poorly understood and, as a consequence, combined effects-such as likely changes in soil moisture which depend upon changes in temperature, near-surface humidity, precipitation and (via biospheric feedback) the atmospheric CO2 concentrations-are exceedingly difficult to determine.
The IPCC Scientific Report identified five regions (designated Central North America, Southern Asia, Sahel, Southern Europe and Australia) for which model-based consensus climatic changes for 2030 were constructed. These 'agreed' changes were determined from high-resolution numerical models using the IPCC 'business-as-usual' emission scenario. These scenarios carry the strong caveat that continental averages hide large subcontinental variations. There are no predictions yet available for small areas, such as fractions of Australian states, and territories or countries in the South-East Asian region.
Scientific and other uncertainties underlie the IPCC conclusions. The unknowns include future human behaviour, particularly in regard to rate of use of fossil fuels, forest-management programmes and agricultural practices. In addition, there is considerable uncertainty about the level of impact of technological advances and, still more importantly, of technology transfer from developed to developing nations. The final, and single largest, factor of importance in this discussion is that of human population growth and the relative distribution of that growth among nation-states and between the developed and developing world. Thus, the direct effects of climatic change are not the only causes of impacts and, indeed, may not be as important as the secondary and tertiary disturbances induced by the changes and by the policy responses put in place to counteract them.
Changes in Variability of Temperature and Precipitation
It is not known how variability will change if mean temperatures and mean totals of precipitation increase as indicated in the previous section. Despite this lack of information, it is often asserted that variability will increase as means increase. Since variability is probably more critical (than mean values) to most impact scenarios, it is worth investigating the interrelationship between means and variance. There are a number of ways of considering the changes in variability which are likely to be associated with increases in mean values. These include investigation of statistical relationships, empirical analysis, consideration of the atmospheric circulation which underlies point or regional variations, and the results of numerical climate models.
STATISTICAL RELATIONSHIPS
Arithmetically, there is no change in the variance of a time series of data if the same value is added to all data points, that is, the mean can increase with no change in the variability. If, on the other hand, each data element is multiplied by a factor, the variance will increase by the square of the factor, that is, the same mean change as in the first case can, this time, induce a large increase in variability. Finally, it is important to note that in these two cases, the data series is assumed to be stationary. If a trend in the mean value occurs, the variance must increase. Since the variance is a measure of deviations of individual observations from the mean, when a trend is added to the time series data, later observations will inevitably occur above the mean and earlier observations, below. Thus, moving from a state where all individual observations are equal to the mean (zero variance)-to a state where some observations are below and some above the mean-must increase the variance.
It should be clear that other statistics beyond the mean and variance are also important if the impacts of physical changes of the climate are to be fully assessed; for example, a statistic clearly warranting attention is that of temporal autocorrelation-a measure of the strength of the relationship between adjacent events. If-as seems to be the case for plants and animals, people, energy usage and consumer goods purchases-the occurrence of multiple, similar events (especially extremes) are very much more important than random, individual occurrences, then identification of present-day autocorrelation and estimation of how this statistic may change as means increase would seem to be of considerable importance (Mearns, Katz and Schneider, 1984). Currently there is no statistical information available beyond predictions of mean values and the estimates of variability changes discussed in the following sections.
EMPIRICAL ANALYSIS
The appeal to empirical relationships is uncertain, to say the least. Is it true to assert, for example, that hotter places or seasons exhibit larger variabilities? Mearns, Katz and Schneider ( 1984) show the reverse relationship for temperature, that is, a decrease in standard deviation as mean maximum temperatures increase. This inverse relationship also seems, intuitively, applicable to rainfall in the sense that very arid locations suffer large variability in rainfall in combination with very small rainfall totals, whilst very wet locations generally exhibit smaller variability combined with large rainfall totals. The actual variability of temperatures and precipitation, of course, results from physical phenomena such as the occurrence of fronts, baroclinic instability, anticyclonic blocking, inversions and nearby ocean temperatures. Thus, empirical relationships, which hold for the present day, are probably not transferable spatially or to a future warmed world.
ATMOSPHERIC CIRCULATION
The third method of attempting to identify likely changes in variability is to recognize the importance of physical processes, and appeal to the consensus view that the Equator-to-pole temperature gradients are predicted to decrease significantly. In terms of the atmospheric circulation, this decrease might imply a less energetic global circulation and, in particular, less energy in the Rossby wave circulation which dominates the mid-latitude depression belts. This phenomenon could manifest itself as a weakening of highpressure systems and less deep cyclonic disturbances. There are at least two contrary arguments to this hypothesis. The first depends upon another consensus assertion that the global hydrological cycle will intensify. Such an intensification implies larger latent energy release to the atmosphere, as water vapour condenses into cloud droplets and, hence, presumably, an intensification of some aspects of the global circulation. Secondly, the relatively elevated continental temperatures compared with those of the adjacent oceans might also tend to induce more frequent, and more persistent, anticyclonic blocking than is currently the case.
NUMERICAL CLIMATE MODELS
It is, in principle, possible to determine changes in variability from numerical climate model experiments. However, there are a number of factors which must be borne in mind when deriving changes in standard deviations from numerical simulations. First, a much longer simulation is required to establish that changes in standard deviations, as opposed to changes in means, are statistically significant. Most studies undertaken to date use periods of 15 years (or less), which is barely sufficient to establish that the simulated changes in standard deviation are unlikely to have occurred by chance.
Secondly, although the models exhibit a measure of agreement on the larger scales, there is much less agreement at the regional level, especially in the case of precipitation; hence, only the changes in standard deviations on the larger scales can be considered. Finally, the coupled atmosphere-ocean-mixed layer models that are used to derive currently available results do not reproduce the El Niño-Southern Oscillation (ENSO) phenomenon which is the main source of interannual variability in the Tropics (cf. Mearns et al., 1990).
One chapter of the IPCC Scientific Report (Mitchell et al., 1990) evaluates currently available model estimates of variability changes as follows: slight indications of reductions in interannual, day-to-day and diurnal range in temperatures, but none statistically significant; indications that interannual variability in precipitation increases where mean precipitation increases, and vice versa; and a consistent increase in the frequency of convective precipitation at the expense of large-scale precipitation (Hansen et al., 1989; Rind, Goldberg and Ruedy, 1989; Wilson and Mitchell, 1987). The only conclusion which can be drawn is that, despite the recognized importance of variability (for example, Parry, 1985), it is not known whether it will increase, decrease or remain the same in a greenhouse-warmed world.
Direct Impacts of CO2 Changes
Even the direct effects of increasing CO2 levels, of rising temperatures and of an intensified hydrological cycle are interdependent and complex. If the increases in atmospheric CO2 (rather than the other greenhouse gases) were occurring without the anticipated changes in climate, then the overall consequences for agriculture would probably be beneficial. A doubling of CO2 is observed to increase the photosynthetic rate by between 30 and 100 per cent, depending on other environmental conditions, especially available moisture, temperature and nutrients. Plant species of the C3 photosynthetic pathway (the first product in their biochemical sequence of reactions has three carbon atoms) tend to respond positively to increased CO2 levels because their photorespiration is suppressed. A detailed examination of 51 records of biomass and marketable yield increases for various C3 species under a doubled atmospheric CO2 concentration (660 ppmv) returned values of 40+7 per cent increase in biomass and 26+9 per cent increase in marketable yield (Warrick, Gifford and Parry, 1986; reported in Parry, 1990). However, in C4 plants the enhanced photosynthetic response to increased levels of CO2 is much less marked.
This difference has major implications for terrestrial ecosystems and for world food production. Plants of the C3 species account for at least 95 per cent of the earth's biomass, and 12 of the world's 15 major crops are in this group. Current global food staples, such as wheat, rice and soya bean, are C3 crops, whereas important C4 semiarid, tropical staples include maize, sorghum, sugar-cane and millet. There is therefore the possibility that direct photosynthetic enhancement induced by raised CO2 levels will benefit temperate and humid tropical agriculture, including that in the South-East Asian region, more than that in the semi-arid Tropics. There is also the possibility that C3 crops may have a further advantage since the majority of the world's most troublesome weed species are C4 plants occurring among C3 crops. Although C4 species account for only about 20 per cent of global food production, one such staple alone, maize, accounts for 14 per cent; it represents 75 per cent of all traded grain, and is currently the major component of food aid offered to famine-stricken areas.
Political pressures are likely to arise as a result of the increasing vulnerability of areas where agricultural production is already limited by climate. Most of this limitation is in developing countries which seem likely to undergo large population expansions in the twenty-first century. Overall, 63 per cent of the land area of developing countries is climatically suited to rainfed agriculture, but this percentage varies considerably between regions. The severest climatic limitations to agriculture are currently found in South-West Asia where 17 per cent of the land is too mountainous and cool and 65 per cent too dry, leaving only 18 per cent as potentially productive (FAO, 1984). Predicting a decline in world food stocks, Liverman (1986) observes that aggravated by climatic change, the likely food deficit would be particularly severe in the South and South-East Asian region. Thus, it appears that climatic change may increase political pressure to assist countries in Southern Asia, which may become increasingly dependent upon food aid, and to assist environmental refugees, particularly from the South-West Pacific (cf. Parry, 1990; Tickell, 1989).
Climate model predictions and the south-east Asian region
Obtaining Information on the South-East Asian Region
The definition of the geographical extent of the South-East Asian region varies with usage. Here, the definition in the Mac quarie Dictionary (1987: definition 1619) of South-East Asia is employed: 'the area which includes Brunei, Burma, Indonesia, Cambodia, Laos, Malaysia, the Philippines, Thailand and Vietnam', but the other (eastern) half of the island of New Guinea (that is, Papua New Guinea) is added. The IPCC Scientific Report uses two terms for the geographical region extending from 5 to 30 °N and 70 to 105 °E. In Chapter 5, from which the results reported here are taken, this region is termed 'S. E. Asia' (Mitchell et al., 1990: 156-8); whereas in the IPCC Policymakers' Summary (Houghton, Jenkins and Ephraums, 1990: xxiv), this same region is, more appropriately, termed 'Southern Asia'. In reference to the IPCC Reports, the latter term is used here. Global climate models have very poor spatial resolution, with typical 'grid elements' of a few degrees in latitude and longitude (that is, 'boxes' 300-500 kilometres on a side). At this resolution, most of the countries of the South-East Asian region are very poorly captured or not represented at all. The IPCC Scientific Report includes a small part of South-East Asia in its selected 'Southern Asia' region, while the northern boundary of 'Australia' just reaches the southernmost parts of Indonesia and New Guinea (Figure 6.1).
FIGURE 6.1 The South-East Asian Region as Defined by the IPCC Scientific Assessment Report
It would be wrong to read the guarded predictions for 'Southern Asia' as representing South-East Asian conditions in the future, since the former region is chosen to be representative of a particular climatological regime; the monsoonal reversal in South-East Asia takes a totally different form to that in Southern Asia, and the migration of the Intertropical Convergence Zone (ITCZ) is also important in South-East Asia. Moreover, the spatial resolution of current global climate models is very poor. This can be illustrated in many ways.
In the context of trying to assess likely regional climatic change in the South-East Asian area, one method is to consider the coastline represented in these models and, particularly, the areas of continents and islands that are common to all the models. This means of underlining the problems of poor spatial resolution is illustrated by (forward) reference to Figure 6.4, which shows three different representations of the South-East Asian coastline derived from three 'high-resolution' climate models and to Figure 6.5, which throughout uses a 'coastline' which shows only land areas which are common to all three of these models. The difficulty of poor resolution will be stressed throughout this discussion on likely climatic change in the South-East Asian region.
Modelling Climatic Change for the South-East Asian Region
Table 6.1 lists the scaled assessments of surface air temperature, precipitation and soil moisture changes derived for the IPCC from three high-resolution global climate models: the Canadian Climate Centre (CCC) model, the Geophysical Fluid Dynamics Laboratory (USA) (GFHI) model and the UK Meteorological Office (UKHI) model. Note that the term 'high resolution' is used relative to other current global climate models; these three models offer spatial resolutions of approximately 3 '' latitude by 3 ° longitude. The results in Table 6.1 have been scaled to correspond to the IPCC 'best guess' global mean warming of 1.8 °C in 2030 (itself an underestimate of the final warming of 2-5 °C because of the effect of the thermal inertia of the oceans). This scaling is important because, although the three higher-resolution models give better simulations of the present climate and hence, it is hoped, better estimates of regional climates in the future, they also produce warnings which are larger than the overall model consensus achieved by the IPCC and termed 'best guess'. The scaling (by a factor of 1.8//\Ts where /\Ts is the climate sensitivity of the particular model) is believed to be reasonable, because precipitation and soil moisture changes are proportional to global mean changes in temperatures. These lPCC regional assessments are presented in Table 6.1 for two reasons: (i) to alert South-East Asian nations to the values achieved by these models for the Indian subcontinent; and (ii) in the hope that estimates of climatic change over the countries of the South-East Asian region can be achieved by extrapolating east and north respectively from the IPCC regions of 'Southern Asia' and 'Australia' (cf. Figure 6.1).
TABLE 6.1 IPCC Estimates of Changes in Surface Air Temperature, Precipitation and Soil Moisture from Three High-resolution Global Climate Models with Their Results Scaled to the IPCC 'Best Guess' Scenario
| Region | Temperature (°C) | Precipitation (Percentage Change) | Soil Moisture (Percentage Change) | |||
| DJF | JJA | DJF | JJA | DJF | JJA | |
| Southern Asia (5-30 °N, 70 105 °E) |
||||||
| CCC | 1 | 1 | -5 | 5 | 0 | 5 |
| GFHI | 2 | 1 | 0 | 10 | -5 | 10 |
| UKHI | 2 | 2 | 15 | 15 | 0 | 5 |
| Australia (12-45°S, 110-155°E) |
||||||
| CCC | 1 | 2 | 15 | 0 | 45 | 5 |
| GFHI | 2 | 2 | 5 | 0 | -5 | -10 |
| UKHI | 2 | 2 | 10 | 0 | 5 | 0 |
Source: Compiled from Mitchell et al. (1990).
Note: DJF = December. January, February; JJA = June, July,
August.
a Based on the IPCC definition of the region.
Table 6.1 lists the estimated changes from pre-industrial times to 2030. The only consistencies are in (i) surface air temperature, which rises by between 1 and 2 °C over the IPCC regions termed Southern Asia and Australia; (ii) Australian precipitation, which shows no change in winter (June, July, August (JJA)) and a 5-15 per cent increase in summer (December, January, February (DJF)); (iii) Southern Asian JJA precipitation which increases by 5-15 per cent; and (iv) Southern Asian JJA soil moisture which also increases by 5-10 per cent. The IPCC Scientific Report cautions that for Australia, 'the models do not produce consistent estimates of the changes in soil moisture. The area averages hide large variations at the sub-continental level' (Mitchell et al., 1990: 158).
Recalling that neither of these IPCC regions covers more than a small part of the South-East Asian region, and that the region's climate is affected by the Asian Monsoon, the Australian Monsoon, the seasonal migration of the ITCZ and ENSO events, it is difficult to extrapolate from the IPCC tabulations to the area of interest here, except to say that temperatures might be expected to rise by 1-2 °C. For this reason, maps of surface air temperature, precipitation and soil moisture changes have been constructed for the South-East Asian region from the three sets of high-resolution global model results presented in the IPCC Scientific Report (Houghton, Jenkins and Ephraums, 1990). These model simulations of a greenhouse-warmed South-East Asian climate are shown in Figures 6.2 6.4.
Figure 6.2 shows the equilibrium temperatures achieved for a doubled CO2 climate for the South-East Asian region. Temperatures are higher everywhere than in the 1 × CO2 (present-day) simulations. The South-East Asian countries are shown to have temperatures higher by 0-2 °C (CCC model) and 2-4 °C (GFHI and UKHI models). There is a tendency for temperature changes greater than 4 °C to occur to the north of the region of interest. In considering these predicted temperature changes, note that the depicted coastlines are neither exact nor those used in the models themselves, but those of a generic, equal latitude/longitude map used by the IPCC. The models' generalized land/ocean partitioning can be seen in Figure 6.4.
FIGURE 6.2 Model Predicted Surface Air Temperatures for Doubled CO2 ( 10-year means)
FIGURE 6.3 Model Predicted Precipitation Changes for Doubled CO2 ( 10-year means)
The South-East Asian precipitation changes (Figure 6.3) have also been displayed superimposed upon the generic IPCC coastline. There is much less coherence in these patterns than in the predicted temperature changes. Integrating over the region displayed, it is difficult to discern any overall changes and there is very little consistency, either between models or between seasons. The most important features of these maps are the magnitudes of some of the local changes. This is especially clear in Figure 6.3(a) which shows changes of +5 mm/day in oceanic areas to the east of South-East Asia. A similar pattern is also seen in Figure 6.3(c). These large changes in precipitation in the far western tropical Pacific are among the largest in the world in these model simulations in DJF, while in JJA the large changes in precipitation extend west into the Indian Monsoon region, and east across the Pacific.
FIGURE 6.4 Model Predicted Soil Moisture Changes for Doubled CO2 (10-year means)
Soil moisture changes are exceedingly difficult to predict, as they depend upon both precipitation and evaporation, and more crucially on the (relatively poor) parameterizations employed compared to those used for temperature and precipitation (cf. Pitman, Henderson-Sellers and Yang, 1990). None the less, the distributions in Figure 6.4 do exhibit some weak coherency. There is a general tendency for decreases in soil moisture to dominate increases and, at least in the GFHI and UKHI models, for decreases in the east of the mainland/peninsular region to be complemented by increases in the west. The lack of coherency is further complicated by the unmatched distribution of continental areas among the three models.
Consensus Climate Predictions for the South-East Asian Region
In an attempt to overcome some of the differences between the models' representations of the coastline, and between their predictions of future climatic change for the SouthEast Asian region, an agreed set of predictions has been sought. Achieving this agreement involves two stages: first, an agreed coastline was established, which includes only areas designated as land in all three models; and secondly, agreement among the three model predictions was sought. Figure 6.5 identifies the regions of 'agreed land' and of continental soil moisture changes (a) and (b), continental precipitation changes (c) and (d), and land and ocean temperature changes (e) and (f). In unmarked areas, there was no consensus between the three models. These figures summarize the earlier discussions of predicted climatic change in the South-East Asian region: (i) temperatures on the mainland will increase by 2-4 °C and on the islands by 0-4 °C; (ii) precipitation changes on the mainland are highly variable, but the island regions to the south may see increased precipitation; and (iii) soil moisture is very hard to predict but may decrease in some locations on the mainland.
Unfortunately, these ensemble estimates, derived from the three high-resolution climate models used by the IPCC, are of relatively little value for the region. There is little consistency between models, either in their representation of the land areas themselves or in the predictions of climatic changes. These two 'features' are, to some extent, linked since without land/sea contrasts and orography, local and regional climates cannot be captured.
This difficulty is, however, probably much less important for the climatic simulation of the region than, for example, capturing the effects of tropical cyclones and ENSO events. Neither of these crucial features of the climate system are, as yet, adequately simulated in numerical models. Moreover, day-to-day variability changes are still unknown, and although it is generally believed that the intensity of precipitation features, ranging from monsoons to local convection, will increase, there is no confirmation of this hypothesis and there are no estimates of the magnitude of the changes.
FIGURE 6.5 Model Consensus Climate Predictions for South-East Asia
The Effects of the Oceans. Transient Vs. Instanneous Simulations
Attempts to assemble an agreed, or consensus, view of future climatic changes for the South-East Asian region have so far been discussed solely in terms of global (atmosphere plus mixed-layer ocean) numerical model simulations of the equilibrium response to doubling CO2. The estimates have been drawn from the IPCC Scientific Report (Houghton, Jenkins and Ephraums, 1990), but this report also quantifies the likely effect of coupling ocean models to atmospheric models, and evaluates the evidence for humaninduced climatic change. Figures 6.6(a) and 6.6(b) compare the predictions of surface air temperature changes for the region, derived from models constructed in two ways. An atmospheric model linked to a mixed-layer ocean model, similar to the models used to construct Figures 6.2-6.4, is used in Figure 6.6(a). This is compared with the simulation from a fully coupled ocean-plus-atmosphere model, running a transient (that is, gradually increasing) CO2 experiment, in Figure 6.6(b). Figures 6.6(a) and 6.6(b) show respectively a 10-year average derived from an equilibrium experiment (that is, instantaneously doubled CO,), and a 20-year average centred on 70 years, which is the point at which CO2 doubles in the transient simulation.
The main features of Figures 6.6(a) and 6.6(b) are common to the global means when comparisons are undertaken between transient and equilibrium simulations. For a steadily increasing forcing, the rise in temperature is a (roughly constant) fraction of the equilibrium rise. Temperatures simulated in the transient coupled atmosphere-ocean models correspond approximately to those derived in the instantaneous experiment, but for a time that is earlier by a fixed offset period. Bretherton, Bryan and Woods (1990) assert that regional patterns of temperature and precipitation change generally resemble those of an equilibrium simulation for an atmospheric model, but are uniformly reduced in magnitude. These results are consistent with the current understanding of ocean circulation and sequestration of heat.
Thus, coupling a fully three-dimensional ocean model to the atmospheric model-used to generate the results considered in the preceding section- dampens the warming. Examining the results from a 'snapshot' taken from a 100-year transient experiment increases confidence in the equilibrium (instantaneous doubling) data assembled in discussing the model predictions for the region.
Confidence in Estimates and Observations
Numerical climate models predict that the human-induced greenhouse effect will cause: (i) the lower atmosphere and the earth's surface to warm; and (ii) the stratosphere to cool. The surface warming and its seasonal variation are expected to be smallest in the Tropics. The models also predict that surface air will warm faster over land than over oceans, and that a minimum of warming will occur around Antarctica and in the northern North Atlantic. The IPCC Scientific Report cautions, however, that 'our confidence in the prediction of the detail of regional changes is low.... There are less consistent predictions for the tropics and the Southern Hemisphere' (Houghton, Jenkins and Ephraums, 1990: xi).
The Report therefore merely cautions its potential users about some important uncertainties which have major effect on impacts; for example, a modest increase in the mean temperature could, assuming no change in variability, imply that the number of hot days will increase, and that there will be fewer frosty nights. However, shifts in largescale weather systems, such as the mid-latitude depressions and anticyclones and the intensification of monsoons, are likely to affect the variability of weather at particular locations as well as mean values. At present, tropical cyclones are known to develop only over oceans that are warmer than about 26 °C. Numerical models predict that the area of sea with temperatures above this critical value will be enlarged. However, Houghton, Jenkins and Ephraums (1990: xxv) note that 'the critical temperature itself may increase in a warmer world ... [while] climate models give no consistent indication whether tropical storms will increase or decrease in frequency or intensity as climate changes; neither is there any evidence that this has occurred over the past few decades'.
It is interesting to compare the predicted temperature changes with the observed temperature rises over the 1980s. Figure 6.6(c) shows the 1980 9 surface air temperatures for the South-East Asian region plotted as anomalies from the 1951-80 average temperatures. The data are composites of land and ocean observations of surface air and sea surface temperatures, as described in Folland, Karl and Vinnikov (1990). Global distributions of 1980s anomalies show temperature increases greater than 0.75 °C over some areas of the northern hemisphere continents. The South-East Asian region, however, shows, on average, only about 0.25 °C temperature increase. This is in keeping with the numerical model predictions that the northern high latitudes will warm more than the tropical regions. The observed temperature variations (Figure 6.6(c)), at least in the 1980s in the region, seem to be opposite in distribution to the predictions (Figures 6.6(a) and 6.6(b)), with the greatest observed temperature anomalies occurring in the central part of the region, while smaller or no temperature increases are observed in the north.
This comparison is, of course, for one decade only and serves mainly to underline the fact that temperatures throughout the 1980s were high around the globe; the South-East Asian region was no exception. These observations of the 'warmest decade on record' have been seen by many as the first detection of humanity's additional greenhouse warming.
