Contents - Previous - Next

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

4 Relations between várzea, and terra firme groups

Two scenarios have been proposed for interaction between the groups exploiting the várzea, of white-water rivers and those occupying the terra firme drained by black- and clear-water rivers. One envisions the várzea, as a region of constant population increase, creating demographic pressure that was relieved by expansion up the principal tributaries (Lathrap, 1970: 74-7). The other views the várzea, as prime agricultural land coveted by occupants of the terra firme.

Examining the affiliations of the phases identified thus far along the Madeira, Tocantins, and Xingu reveals a sharp boundary coincident with the first rapid on each river. Phases below this point are affiliated with the Polychrome Tradition and those above it belong to traditions restricted to the terra firme. This segregation is particularly clear on the Tocantins (fig. 4.4) and on the lower Xingu, where sites above and below the barrier are in close proximity. Expansions of the Polychrome Tradition up the Negro and the Solimões are late and seem attributable to the advent of the Incised and Punctate Tradition rather than to local population increase.

These distributions are compatible with environmental evidence that the várzea and terra firme habitats are distinct and require different specialized knowledge and procedures for effective utilization, such that groups adapted to either are ill equipped to exploit the other. A case can be made that the terra firme was a more stable environment than the várzea, which was subject to drastic fluctuations in subsistence productivity because of unpredictable variations in the rate, timing, and extent of annual inundation (Irion, 1984; Meggers, 1971: 146).

5 The impact of climatic fluctuation

It is generally accepted that Amazonia experienced episodes of climatic fluctuation during and since the Pleistocene, but until recently evidence for their impacts on humans was restricted to the disjunct distributions of languages and cultural traits (Meggers, 1987). The existence of detailed relative chronologies and numerous carbon-14 dates throughout the lowlands now makes it possible to correlate local cultural discontinuities with environmental oscillations.

When humans entered South America toward the end of the Pleistocene, some 12,000-14,000 years ago, the lowlands were less densely forested than at present. Although details are unclear, it seems likely that the savannas of Roraima and northern Pará are relicts of a corridor of relatively open vegetation that extended from Venezuela to eastern Brazil (Barbosa, 1992). The availability of similar kinds of resources across the basin would have facilitated movement by hunter-gatherers and carbon-14 dates indicate they had reached the south-eastern margin of the lowlands by 11,000 BP (Schmitz, 1987, table II). The only evidence thus far for their presence in the central Amazon takes the form of rare stone projectile points (Simões, 1976) and early carbon-14 dates from the lower levels of ceramic sites, which may represent campfires of pre-ceramic occupants of the same locations. The oldest result is 7320 ± 100 BP (SI-4277) from a tributary of the middle Madeira.

The vegetation assumed its current composition and extent during subsequent millennia, but the process was disrupted by several widespread episodes of climatic fluctuation. Pollen records from marginal locations indicate that prior to the Christian era savanna replaced forest during several centuries and briefer episodes occurred about 1,500, 1,200, 700, and 400 years ago (Absy, 1982; Van der Hammen, 1982). These episodes coincide with cultural replacements in welldated archaeological sequences throughout the lowlands.

Oscillations reflected in a pollen profile from Lago Ararí on eastern Marajó correlate with successive archaeological phases in the surrounding region (fig. 4.11). The Ananatuba Phase, the earliest pottery-making group on the island, arrived when forest vegetation was dominant. By 2590 i 100 BP (Beta-2289), forest pollen had declined from 65 per cent to 30 per cent, implying significant alteration in the climate and biota (Absy, 1985, fig. 4.9). The terminal date for the Mangueiras Phase probably marks the point during the transition at which declining subsistence resources could no longer sustain sedentary communities. The inception of the Formiga Phase coincides with re-expansion of the forest about 2000 BP. The arrival of the Marajoara Phase follows the shorter period of aridity about 1500 BP and its termination coincides with the 700 BP episode.

Similar cultural discontinuities are evident in all the regions where local sequences are sufficiently complete and well dated to minimize the probability of sampling error. With rare exceptions, the initial ceramic complexes throughout the lowlands postdate 2000 BP (fig. 4.12). On the Llanos de Moxos in north-eastern Bolivia, most phases begin or end about 1500, 1000, and 700 BP. In the Silves-Uatumã region on the left bank of the middle Amazon, replacements take place about 1,500, 1,100, 800, and 400 BP. On the lower Xingu, transitions occur about 1,500, 1,100, 800, and 400 BP. The only significant disagreement between the timing of climatic changes inferred from pollen profiles and the cultural replacements is the clustering of the latter about 1000 BP rather than 1200 BP. This may reflect more precise dating for the archaeological sequences. Droughts too brief to leave a pollen record are more frequent (Meggers, 1994; Sternberg, 1987: 206; Stockton, 1990) and failure to incorporate their effects places unwise schemes for "development" in greater jeopardy.

Figure 4.11 Correlation between the inceptions of archaeological phases on Marajó and local fluctuations in arboreal vegetation. A carbon-14 date of 2590 * 100 BP for a decline in tree pollen falls within the hiatus between the end of the Mangueiras Phase c. 2800 BP and the inception of the Formiga Phase c. 2000 BP. (After Meggers and Danon, 1988.)

Figure 4.12 Carbon-14 durations of successive phases and traditions in four widely separated parts of Amazonia, compared with episodes of climatic fluctuation inferred from pollen cores. Cultural discontinuities in all four regions coincide with the 1500 BP episode and also tend to be associated with the episodes c. 700 and 400 BP. Disagreement between the concentration of cultural discontinuities c. 1000 BP and vegetational changes c. 1200 B.P. probably reflects the greater number of dates for this episode from archaeological sites.

The contemporaneity of climatic fluctuations and cultural discontinuities implies a cause and effect relationship. Evidence that many plants respond to variations in rainfall by failing to flower and fruit indicates that the cause was deterioration of local subsistence resources below the requirements of small semi-sedentary communities (Leigh et al., 1982). The effect may have been emigration, decimation, adoption of a wandering way of life, or a combination of solutions. Although emigration cannot yet be traced archaeologically, it is implied by the disjunct distributions of the principal Amazonian languages as well as by the different ceramic affiliations of successive phases in the archaeological sequences.

6 Overcoming environmental constraints

From the perspective of temperate-zone observers, the failure of indigenous inhabitants of Amazonia to equal the population density and sociopolitical complexity achieved in the adjacent Andean region signifies cultural stagnation. From the perspective of the tropical environment, however, their way of life represents highly successful exploitation of unpromising and unpredictable resources. Wild plants and animals are solitary and dispersed, agricultural intensification is precluded, and food cannot be stored for future consumption (Bergman, 1980: 109-10; Good, 1989: 78). The number, variety, and ingenuity of the cultural practices that have developed for manipulating environmental constraints, inhibiting overexploitation, and optimizing the productivity of this complex ecosystem are no less remarkable than the intricate interactions among the climate, soils, and biota.

Comprehensive knowledge of the flora provides alternatives when primary resources fail (Berlin, 1984: 32; Boom, 1989: 82-3; Carneiro, 1978; Cavalcante and Frikel, 1973: 5). The Yanomami are reported to experiment continuously with new plants they encounter (Fuentes, 1980: 23). The abundance of useful species is enhanced by selective cutting and weeding and by transplanting (Irvine, 1989). Multiple varieties of the principal cultigens with differing tolerances for disease, moisture, soil, and other variables are usually planted to minimize loss (Baster, 1987: 412; Johnson, 1983: 44-5). Knowledge of the fauna is equally detailed and includes many species not ordinarily consumed (Berlin and Berlin, 1983: 320-2).

Prenatal and postnatal practices offsetting population increase are numerous and varied (Meggers, 1971: 103-10), and include prolonged lactation, contraception, abortion, infanticide, abstinence from intercourse, blood revenge, and warfare. Their effectiveness is implicit in calculations of the consequences of uncontrolled reproduction. A four per cent rate of increase would have created a population of 4-5 trillion in 5,570 years, half of the time since human colonization of the lowlands (Cowgirl, 1975: 510; cf. Frank, 1987: 114). Both ecological studies (Clark and Uhl, 1987; Fearnside, 1990) and ethnographic evidence for environmental degradation following forced sedentarism (Gross, 1983: 438) also indicate that the densities of surviving unacculturated groups represent sustainable carrying capacity.

Although formal and informal trade for commodities not locally available has been characteristic of human groups since the Upper Palaeolithic, the extensive networks of the neo-tropical lowlands are peculiar in several respects. Items exchanged are often necessities that could be manufactured locally, trading partners must accept what is offered whether they desire it or not, and hostile relations among the groups involved make participation perilous (Chagnon, 1968; Coppens, 1971; Jackson, 1983: 99; Mansutti R., 1986: 13-15; Meggers, 1971: 65; Oberg, 1953). Several ethnographers have pointed out the role of these networks in creating and perpetuating regional and ethnic interdependence (Bodley, 1973: 595; Colson, 1985: 104; Coppens, 1971: 40). They also serve as channels for transmitting all kinds of potentially useful information among groups with different linguistic and tribal affiliations and occupying different environments.

The similarities between prehistoric and recent settlement behaviour (territoriality, impermanent residence, centripetal village movement, site reoccupation, small homesteads) imply that associated practices maximizing sustainable exploitation of essential resources had evolved by the beginning of the Christian era (when the adoption of pottery permits their recognition) and probably earlier. The archaeological data provide no support for the existence during preColumbian times of urban centres, highly stratified sociopolitical or ganization, or expansive states. Rather, they suggest a dispersed pattern of settlement by small communities that were politically autonomous but socially and economically interdependent.

7 Conclusion

During tens of thousands of years, the changing courses of rivers and the fluctuations of climate divided and redivided the landscape, segregating and reuniting populations of plants and animals. Drift and natural selection enhanced divergence and guided interactions, creating increasingly intricate configurations that not only conserved the limited nutrients but also moderated the range of variation in heat and humidity. Arriving at the end of the Pleistocene, humans were the last of a series of mammalian immigrants that entered via the Central American passage and melted into the ecosystem. Like the rest of the fauna, they returned in services as much as they took in sustenance, sometimes knowingly, sometimes unaware.

Contemporary humans in the northern temperate zone came to terms with a different set of edaphic, climatic, topographic, and biotic conditions. Initially, their cultural development followed a similar course. They too conserved resources and enhanced their productivity. The motivation was the same: degradation meant extinction. Slowly at first and then with increasing success, some groups expanded their sustaining areas and their capacities to transport commodities. By neutralising the immediate impact of overexploitation, they were able to increase consumption while degrading resources both locally (since deficits were compensated by imports) and at a distance (since decimation in one location could be compensated by moving to another).

During the past decade, the human sustaining area has become synonymous with the surface of the planet. The scale of our activities is now sufficient to alter the global climate, an achievement equalled only once before in the history of the earth. The blue-green algae that added oxygen to the atmosphere eons ago and established the conditions for terrestrial life did so unaware. Although we are conscious of our impact and its potential consequences, we appear as helpless as the algae to alter our behaviour. They survive inconspicuously today. Whether humans will be so resilient is questionable.

Amazonia will play a critical role in the future of the biosphere because of its influence on global climate. During the past several millennia, the vegetation has suffered the vicissitudes of repeated cli matic fluctuations and recovered. Whether it will survive the impacts of accelerating human-induced deforestation, erosion, and pollution seems less likely. The pursuit of inappropoate policies will persist as long as incentives and perceptions of alien origin remain dominant (Ledec and Goodland, 1989: 448-51). Their continuing strength in spite of negative economic, social, and environmental results bodes ill for the future of the tropical forest and, if the climatologists are correct, for the future of the biosphere as well.


The archaeological investigations undertaken under the Programa Nacional de Pesquisas Arqueológicas na Bacia Amazônica have been funded principally by the Neotropical Lowland Research Program of the National Museum of Natural History, Smithsonian Institution. I am grateful to the following colleagues for use of their data: Bernardo Dougherty, Museo de La Plata, Argentina; Ondemar F. Dias, Instituto de Arqueologia Brasileira, Rio de Janeiro; Eurico Th. Miller, Eletronorte, Brasilia, and Celso Perota, Universidade Federal do Espírito Santo, Vitôria. The sequences on the lower Negro and Tocantins are based on the work of Mario F. Simões, Museu Paraense Emílio Goeldi, Belém.


Absy, Maria Lucia. 1982. "Quaternary palynological studies in the Amazon basin." In G.T. Prance (ed.). Biological diversification in the tropics (Columbia University Press, New York), 67-73.

. 1985. "Palynology of Amazonia: The history of the forests as revealed by the palynological record." In G.T. Prance and T.E. Lovejoy (eds.), Amazonia (Pergamon Press, Oxford), 72-82.

Balée, William. 1987. "A etnobotânica quantitative dos índigos Tembé (Rio Gurupí, Pará)." Boletim do Museu Paraense Emílio Goeldi 3: 29-50.

Barbosa, Altair Sales. 1992. "A tradição Itaparica: Uma compreensão ecológic e cultural do povoamento inicial do planalto central brasileiro." In B.J. Meggers (ed.), Prehistoria sudamericana: Nuevas perspectivas (Taraxacum, Washington, DC), 145-60.

Bergman, Roland. 1980. Amazonian economics: The simplicity of Shipibo Indian wealth. Dellplain Latin American Studies no. 6. Syracuse, NY.

Berlin, Brent. 1984. "Contributions of native American collectors to the ethnobotany of the neotropics." Advances in Economic Botany 1: 24-33. New York Botanical Garden, New York.

Berlin, Brent, and Elois A. Berlin.1983. "Adaptation and ethnozoological classification: Theoretical implications of animal resources and diet of Aguaruna and Huambisa." In Hames and Vickers, 1983), 301-25.

Bodley, J.H. 1973. "Deferred exchange among the Campa Indians." Anthropos 68: 589-96.

Boom, Brian M. 1989. "Use of plant resources by the Chácobo." Advances in Economic Botany 7: 78-96. New York Botanical Garden.

Boster, James S. 1987. "Classification, cultivation, and selection of Aguaruna cultivars of Manihot esculenta (Euphorbiaceae)." Advances in Economic Botany 1: 34-47. New York Botanical Garden, New York.

Carneiro, Robert (1978). "The knowledge and use of rain forest trees by the Kuikuru Indians of central Brazil." In The nature and status of ethnobotany (University of Michigan Anthropological Papers 67), 201-16. Ann Arbor.

Cavalcante, Paulo B., and Protasio Frikel. 1973. "A farmacopeia Tiriyó." Museu Paraense Emílio Goeldi, Pubs. Avulsas 2. Belém.

Chagnon, Napoleon A. 1968. Yanomamo: The fierce people. Holt, Rinehart and Winston, New York.

Clark, Kathleen, and Christopher Uhl. 1987. "Farming, fishing, and fire in the history of the upper Rio Negro region of Venezuela." Human Ecology 15: 1-26.

Collins, Jane L. 1986. "Smallholder settlement of tropical South America: The social causes of ecological destruction." Human Organization 45: 1-10.

Colson, Audrey Butt 1983/4. "The spatial component in the political structure of the Carib speakers of the Guiana highlands: Kapon and Pemon." Antropológica 5962: 73-124.

. 1985. "Routes of knowledge: An aspect of regional integration in the circumRoraima circum-Roraimaarea of the Guiana Highlands." Antropológica 63-4: 103-149.

Coppens, Walter. 1971. "Las relaciones comerciales de los Yekuana del Caura-Paragua." Antropológica 30: 28-59.

Cowgill, George L. 1975. "On the causes and consequences of ancient and modern population changes." American Anthropologist 77: 505-25.

D'Ans, Andre-Marcel. 1982. L'amazonie péravienne indigene. Payot, Paris.

Dufour, Darna. 1983. "Nutrition in the Northwest Amazon." In Hames and Vickers, 1983, 329-55.

Fearnside, Philip M. 1990. "Estimation of human carrying capacity in rainforest areas." Trends in Ecology and Evolution 5:192-6.

Frank, Erwin. 1987. "Delimitaciones al aumento poblacional y desarrollo cultural en las culturas indigenes de la Amazonía antes de 1942." In G. Kohlhepp and A. Schrader (eds.), Hombre y naturaleza en la Amazonía (Tübinger Geographische Studien 95, Tübingen), 10923.

Fuentes, Emilio. 1980. "Los Yanomami y las plantas silvestres." Antropológica 54: 3-138.

Good, Kenneth. 1989. "Yanomami hunting patterns: Trekking and garden relocation as an adaptation to game availability in Amazonas, Venezuela." Ph.D dissertation, University of Florida.

Gross, Daniel R. 1983. "Village movement in relation to resources in Amazonia." In Hames and Vickers, 1983), 429-49.

Hames, Raymond B. 1983. "The settlement pattern of a Yanomamo population block: A behavioral ecological interpretation." In Hames and Vickers, 1983, 393427.

Hames, R.B., and W.T. Vickers (eds.). 1983. Adapfive responses of nafive Amazonians. Academic Press, New York.

Henley, Paul. 1982. The Panare: Tradition and change on the Amazonian frontier. Yale University Press, New Haven, CT.

Irion, G. 1984. "Sedimentation and sediments of Amazonian rivers and evolution of the Amazonian landscape since Pliocene times." In Harald Sioli (ed.), The Amazon (Dr W. Junk, Dordrecht), 201-14.

Inine, Dominique (1989). "Succession management and resource distribution in an Amazonian rain forest." Advances in Economic Botany 7: 223-37. New York Botanical Garden, New York.

Jackson, Jean E. 1983. The fish people. Cambridge University Press, Cambridge.

Johnson, Allen. 1983. "Machiguenga gardens." In Hames and Vickers, 1983, 29-63.

Lathrap, Donald W. 1970. The upper Amazon. Praeger, New York.

Ledec, George, and Robert Goodland. 1989. "Epilogue: An environmental perspective on tropical land settlement." In D.A. Schumann and W.L. Partridge (eds.), The human ecology of tropical land settlement in Latin America (Westview, Boulder.), 435-67.

Leigh Jr., Egbert G., A. Stanley Rand, and Donald M. Windsor (eds.). 1982. The ecology of a tropical forest. Smithsonian Institution Press, Washington, DC.

Mansutti Rodriguez, Alexander. 1986. "Hierro, barro cocido, curare y cerbatanas: El comercio intra e interétnico entre los Uwotjuja." Antropológica 65: 3-75.

Meggers, Betty J. 1971. Amazonia: Man and culture in a counterfeit paradise. Harlan Davidson, Arlington Heights.

. 1987. "The early history of man in Amazonia." In T.C. Whitmore and G.T. Prance (eds.), Biogeography and quaternary history of South America (Clarendon Press, Oxford), 151-74.

. 1992. "Prehistoric population density in the Amazonian Basin." In J.W. Verano and D.H. Ubelaker (eds.), Disease and demography in the Americas (Smithsonian Institution Press, Washington, DC), 197-205.

. 1994. "Archeological evidence for the impact of mega-Niño events on Amazonia during the past two millennia." Climatic Change 28: 1-18.

Meggers, Betty J., and Jacques Danon. 1988. "Identification and implications of a hiatus in the archeological sequence on Marajó Island, Brazil." Journal of the Washington Academy of Sciences 78: 245-53.

Miller, Eurico Th., et al. 1992. Arqueologia nos empreendimentos hidrelétricos da Eletronorte; Resultados preliminares. Centrais Elétricas do Norte do Brasil S.A., Brasilia.

Oberg, Kalervo. 1953. Indian tribes of Mato Grosso, Brazil. Institute of Social Anthropology Publ. 15. Washington, DC.

Roosevelt, Anna C. 1989. "Resource management in Amazonia before the conquest: Beyond ethnographic projection." Advances in Economic Botany 7: 30-62. New York Botanical Garden, New York.

Ruddle, Kenneth. 1971. "Notes on the nomenclature and the distribution of the Yukpa-Yuko tribe." Antropológica 30: 18-27.

Schmitz, Pedro Ignacio. 1987. "Prehistoric hunters and gatherers of Brazil." Journal of World Prehistory 1: 53-126.

Schumann, Debra A., and William L. Partridge, eds. 1989. The human ecology of tropical land settlement in Latin America. Westview, Boulder.

Seeger, Anthony. 1981. Nature and society in central Brazil: The Suyá Indians of Mato Grosso. Harvard University Press, Cambridge, MA.

Simões, Mario F. 1976. "Note sobre duas pontas-de-projetil da Bacia do Tapajós (Pará)." Boletim do Museu Paraense Emílio Goeldi, Antropológia 62.

Smith, Nigel J.H. 1980. "Anthrosols and human carrying capacity in Amazonia." Annals of the Association of American Geographers 70: 553-66.

Spielman, R.S., E.C. Migliazza, and J.V. Neel. 1974. "Regional linguistic and genetic differences among Yanomama Indians." Science 184: 637-44.

Sternberg, Hilgard O'Reilly. 1987. "Aggravation of floods in the Amazon River as a consequence of deforestation?" Geografiska Annaler, Series A, 69A: 201-19.

Stockton, Charles E. 1990. "Climatic variability on the scale of decades to centuries." Climatic Change 16: 173-83.

Van der Hammen, Thomas. 1982. "Paleoecology of tropical South America." In G.T. Prance (ed.), Biological diversification in the tropics (Columbia University Press, New York), 606.

Verswijver, Gustaaf. 1978. "Separations et migrations des Mekragnoti: Groupe Kayapó du Brésil central." Boletin, Société Suisse des Américanistes 42: 47-59.

Vickers, William T. 1983. "The territorial dimensions of Siona-Secoya and Encabello adaptation." In Hames and Vickers, 1983, 451-78.

Distribution and interannual variability of rainfall in Brazil

1 Introduction
2 Data source and distribution of rainfall in South America
3 Interannual variability of rainfall in brazil
4 Relationship to southern oscillation Index
5 Seasonal variability of rainfall in Brazil
6 Comparison of the rainfall in northern Brazil to other tropical regions
7 Conclusion


Minoru Tanaka, Akio Tsuchiya, and Toshie Nishizawa

1 Introduction

Climatological studies on annual change in rainfall regimes in South America were conducted by Ratisbona (1976) and Caviedes (1981). Similar studies on the atmospheric circulation patterns were conducted by Virji (1981). The rainfall regime of the Brazilian Nordeste is divided into northern (north of 10°S) and southern Nordeste by Hastenrath and Helter (1977). Markham and McLain (1977), Moura and Shukla (1981), and Hastenrath (1984) suggest a link between rainfall in the northern Nordeste and sea surface temperature (hereafter called SST) anomalies in the Atlantic Ocean. Walker (1928), Ramos (1976), Kousky and Chu (1978), and Kousky (1979, 1985) studied the rainfall in the Nordeste and atmospheric circulation over Brazil and the Atlantic Ocean. The influence of the northern hemisphere circulation on the rainfall in the Nordeste was investigated by Namias (1972). Yamazaki and Rao (1977) studied the tropical cloudiness over the South Atlantic Ocean.

A pilot study on annual change in tropospheric circulation and its relationship to the monthly mean rainfall in South America was conducted by Nishizawa and Tanaka (1983), as was a similar survey on interannual change also by Tanaka and Nishizawa (1985). Collec lively, these studies show a trough at the 150mb level over the Nordeste which can be linked to regional subsidence and to the relatively low amount of the rainfall in the Nordeste in the drought year of 1983.

Empirical orthogonal function (EOF) analyses of rainfall in Brazil on monthly, seasonal, and interannual time scales were carried out by Tsuchiya et al. (1988) and Tanaka et al. (1988). In addition, Aceituno (1988, 1989) analysed the rainfall and atmospheric circulation associated with the Southern Oscillation. These studies show that the inverse relationship between the northern Nordeste and southern Brazil (hereafter called the NS pattern), which appears in the First EOF, is best developed for an interannual time scale of more than one year. The rainfall difference between northern and southern Nordeste appears for all time scales in the Second EOF. For the present study, the relationship between the NS pattern and the Southern Oscillation Index (SOI) is based on the normalized sea level pressure difference between Tahiti and Darwin in the tropical Pacific Ocean and the SST in the tropical Atlantic Ocean.

2 Data source and distribution of rainfall in South America

The source of the data for rainfall distribution and 850mb height are from the Monthly Climatic Data for the World published by the US National Oceanic and Atmospheric Administration (NOAA) for the 10-year period from 1969 to 1978. The data for EOF analysis (36 stations, 1968 to 1985) come from the Superintendência do Desenvolvimento do Nordeste (SUDENE) for the Nordeste region and from Boletim Agroclimatológico published by the Agricultural Ministry of Brazil for the rest of Brazil. Because locations of the rainfall data for the EOF analysis are unevenly distributed, six of the locations shown are the average of closely spaced stations. This averaging was conducted because the EOF components can shift toward the area of high concentration of the stations. Annual cycles in rainfall are removed by subtracting the long-term mean (1968-85 average) for each month of the year. The interannual component of rainfall variability was extracted by filtering rainfall data by the 24-month triangular weighted running means (see Burroughs, 1978 for details). The triangular running means reduce short period oscillations much more efficiently, compared to the simple unweighted running means. However, about twice as many terms are required compared to the simple running means. Hence, filtering frequency below 12 months re quires about 24-month triangular means. This filter was used because the regional variations in the rainy season in Brazil are large. This means that the rainy season is observed in certain regions in Brazil during any months of the year. Hence it is difficult to define a cutoff month for obtaining the annual totals. The EOF analysis of the anomaly rainfall time series employed filtered data. Similar smoothing was applied to the 850mb height and SST over the Atlantic Ocean. Distribution of rainfall in South America and its relationship to the tropospheric circulation are shown in detail by Nishizawa and Tanaka (1983), as shown only by the examples for December (fig. 5.1) and April (fig. 5.2). In December, rainfall is especially heavy in the Amazon Basin and the interior of Brazil. Because the Intertropical Convergence Zone (ITCZ) over the Atlantic Ocean is located north of the equator, dry areas are observed in the Nordeste. In April, rainfall in the interior of Brazil begins to decrease, while a southward dis placement of the ITCZ over the Atlantic Ocean produces a brief rainy season in the Nordeste.

Figure 5.1 Monthly mean rainfall over South America in December (1969-78) (mm). Stations are shown by dots.

Figure 5.2 Monthly mean rainfall over South America in April (1969-78) (mm).

3 Interannual variability of rainfall in brazil

The interannual variability of rainfall in Brazil is analysed by empirical EOF analysis for an 18-year period from 1968 to 1985 by using filtered data, as discussed previously.

The primary reasons for conducting EOF analysis are its capability to reduce the dimensionality and to describe coherent variability in the rainfall data. Since 27 locations are shown in figure 5.3, there are 27 dimensions in the rainfall data. The EOF analysis reduces the dimensions to few major components while retaining the information contained in the original data. These components are calculated to maximize the variance explained in the original data. Hence, coherent variability in the original data can be described by the first few components which contain most of the variance. For these reasons, we believe that EOF analysis depicts large-scale variability of the rainfall much more clearly than the simple correlations between the 27 locations.

Figure 5.3 Distribution of eigenvector of First EOF (24 M1) of rainfall EOF explaining 46.7% of variance, for 27 locations as shown by dots and squares for averages of closely spaced stations.

Figure 5.3 shows the distribution of the eigenvector of the First EOF (24 M1) explaining 46.7 per cent of variance. The pattern shows an inverse relationship in rainfall between northern and southern Brazil (NS pattern). This pattern is analysed by Aceituno (1988, 1989). The highest values are concentrated near São Luiz and Fortaleza, where rainfall variability is strongly influenced by the interannual variation in intensity and location of ITCZ located near the equator. Figure 5.4 shows the time coefficients for the First EOF. Positive values in 1974 and 1985 show wet years in the northern Nordeste. Negative values in 1972, 1976, and 1983 coincide with dry years in northern Brazil and severe flooding in 1983 in southern Brazil (see Tanaka and Nishizawa, 1985 for a detailed case study).

Figure 5.5 shows the correlation of the 850mb height over South America to the First EOF (24 M1). A region of high negative correlation near Recife indicates a decrease (increase) in the 850mb height in wet (dry) years in northern Brazil. This pattern shows intensification of the ITCZ in wet years in northern Brazil.

Figure 5.4 Time coefficients for First EOF (24 M1) of rainfall EOF.

Figure 5.5 Correlation of smoothed 850mb height to First EOF (24 M1) of rainfall EOF.

Figure 5.6 Correlation of smoothed SST (C) to First EOF (24 M1) of rainfall EOF.

Figure 5.6 shows the correlation of the smoothed Atlantic SST to the First EOF (24 M1). High positive correlation values over +0.8 are observed near 5-10°S, 15-10°W. The negative correlation values over -0.4 are observed near 12°N, 40°W. This SST dipole pattern was discovered by Hastenrath and Heller (1977) and confirmed by Hastenrath (1978), and by Moura and Shukla (1981). For 17 years of annual totals, the significant levels of the correlations are 0.49 for the 5% level and 0.61 for the 1% level. We believe that these values will be slightly higher for the smoothed data.

Figure 5.7 shows the smoothed Atlantic SST at 5-10°S, 15-10°W. The wet years in northern Brazil of 1974 and 1985 (see fig. 5.4) are the years with positive values of SST anomalies. The dry years of 1972, 1976, and 1983 have negative values of SST anomalies.

4 Relationship to southern oscillation Index

The relationship to SOI using normalised sea level pressure at Tahiti and Darwin was investigated. Figure 5.8 shows the smoothed SOI from 1967 to 1986. High positive values in 1974 and negative values in 1972 and 1983 coincide with extreme rainfall in the northern Nordeste (see fig. 5.4). This relationship is reversed for the periods from 1969 to 1971 and from 1976 to 1978. Figure 5.9 shows the correlation of smoothed rainfall values (not EOF) for 27 locations in Brazil compared to the smoothed SOI. The correlation pattern is similar to the First EOF (24 M1) (see fig. 5.3). However, the regions with moderately high correlation over 0.4 are restricted to northern Brazil and to central and southern Brazil. Figure 5.10 shows the correlation of the smoothed 850mb height to the smoothed SOI. This pattern is very similar to Hastenrath (1984). When this pattern is compared to the Atlantic SST (fig. 5.5), it shows a high correlation to the subtropical high near 20°S.

Figure 5.7 Time change of smoothed Atlantic SST (C) at 5-10°S, 15-10ºW.

5 Seasonal variability of rainfall in Brazil

EOF analyses of monthly and seasonal (defined here as a consecutive 4month period) variability of rainfall in Brazil show the shorter timescale interaction between SST and rainfall. The monthly EOF analysis of rainfall (Tanaka et al., 1988) reveals rainfall distribution patterns of three distinct seasons.

Table 5.1 shows the variance explained by the first three components of the seasonal and smoothed interannual rainfall (24 M). The First EOF in the December to March season is centred on interior Brazil and has a component which shows a low month-to-month persistence pattern. (Persistence is defined here as the persistence from one month to the next month.) The low persistence indicates that the monthly EOF pattern does not persist in the next month. The First EOF (fig. 5.11) in the April to July season is very similar to the First EOF (24 M1) of the interannual time scale, which has a high month-to-month persistence. The First EOF in August to November season is centred on southern Brazil and has a low month-to-month persistence.

Figure 5.8 Time change of smoothed Southern Oscillation Index (SOI) using normalized sea level pressure at Tahiti and Darwin (Tn-Dn).

Figure 5.9 Correlation of smoothed rainfall values to smoothed SOI.

Since the First EOF in April to July (4-7 M1) seasonal rainfall has high month-to-month persistence and is very similar to the NS pattern (24 M1) in interannual time-scale, correlation to Atlantic SST was computed (see fig. 5.12). This pattern is similar to the interannual time-scale (see fig. 5.6). An area of negative correlation is located near 5-10°N, 45-40°W. This resultant pattern is very similar to that of Moura and Shukla (1981), who employed data from Fortaleza and Quixeramobim. Our study shows that rainfall in northern Brazil is related to this SST pattern.

Figure 5.13 shows the time change of the SST at 0-5°S, 15-10°W and the First EOF (4-7 M1) of seasonal rainfall. The time changes of both parameters are very similar.

Figure 5.10 Correlation of smoothed 850mb height to smoothed SOI.

Table 5.1 Variance explained by first three EOFs of seasonal and smoothed interannual rainfall (24 M)

Months Variance explained %
M1 M2 M3
12-3 30.44 18.66 14.93
4-7 50.52 13.85 8.54
8-11 27.93 16.91 10.18
24M 46.66 12.46 10 .28

6 Comparison of the rainfall in northern Brazil to other tropical regions

The rainfall in the semi-arid region has high interannual variability. However, Nishizawa et al. (1986) have shown that part of the subhumid northern coastal region in north-eastern Brazil, with more than 1,000 mm of annual rainfall, has a high (30-50 per cent) coefficient of variability in rainfall. Figure 5.14 shows annual rainfall variability in Brazil based on the 27 locations used in this study. As shown in the
lower part of this figure, most locations in Brazil show a decrease in rainfall variability as the total rainfall increases. However, Fortaleza and São Luiz show both greater variability and annual rainfall, which indicates the unstable nature of the climate at these locations. An inspection of other regions of tropical climate shows that unstable rainfall of similar magnitude (approximately over 1,000 mm per year) in relatively humid locations is restricted to the regions directly influenced by El Niño-Southern Oscillation (ENSO) in the equatorial Pacific (e.g., Ocean Island and Tarawa) and coastal Ecuador (Guayaquil).

Figure 5.11 Distribution of eigenvector of the First EOF (4-7 M1) of seasonal rainfal EOF explaining 50.5% of variance.

Figure 5.12 Correlation of March to May SST (C) to First EOF of seasond seasonalrainfall EOF (4-7 M1). 5% level = 0.49;1% level = 0.61.

Figure 5.13 Time change of SST at 0-5°S, 15-10°W and First EOF of seasonal rainfall EOF (47 Ml).

Figure 5.14 Annual rainfall variability shown as function of annual rainfall in mm (horizontal scale) and coefficient of variability (vertical scale) for 27 locations in Brazil and three selected stations in tropical area.

An interesting study on this topic was published recently by Nicholls and Wang (1990), which showed that the annual rainfall variability is typically higher for a specific mean rainfall in areas affected by ENSO. In these areas of unstable climatic regimes, any human development must consider careful water and forest resource management in order to reduce the possible impact of drought or floods caused by unusually wide fluctuations in rainfall.

7 Conclusion

EOF analysis of interannual variability of rainfall in Brazil confirmed an inverse relationship in the rainfall in the First EOF between northern and southern Brazil (NS pattern), analysed by Aceituno (1988, 1989). This pattern is best developed in interannual and seasonal (April to July total) timescales. The relationship between the NS pattern and the atmospheric circulation pattern shows an intensification of the ITCZ in wet years in northern Brazil. The correlation analysis of SOI and the tropical Atlantic SST confirmed the earlier findings by Hastenrath and Heller (1977) and Moura and Shukla (1981), that the NS pattern is highly correlated (+0.80 to 0.86) to the SST near 0-10°S, 15-10°W on the interannual and seasonal timescales. Although detailed atmospheric circulation data over the South Atlantic ocean are not available, weakening of the trade winds in the South Atlantic is the probable cause of the rise in SST in the equatorial South Atlantic. The negative correlation in the SST near 5-10°N, 45-40°E suggests that southward migration in South America of the ITCZ coincides with the increase in rainfall in northern Brazil. The influence of the Southern Oscillation is found to be less than Atlantic SST. However, the extreme cold event in 1974 (wet year in northern Brazil) and the warm event in 1983 (dry year) coincided with rainfall extremes of the NS pattern.

The EOF analysis of the seasonal variability of rainfall in Brazil have shown that the First EOF in the December to March season is centred on interior Brazil and has a low month-to-month persistence pattern. The First EOF in August to November season is centred on southern Brazil and has a low month-to-month persistence.


The writers wish to express their sincere gratitude to the SUDENE office in Recife, Pernambuco, for providing part of rainfall data in this study.


Contents - Previous - Next