Professional Documents
Culture Documents
Spatial and Temporal Patterns of Amazon Rainfall
Spatial and Temporal Patterns of Amazon Rainfall
BioOne Complete (complete.BioOne.org) is a full-text database of 200 subscribed and open-access titles
in the biological, ecological, and environmental sciences published by nonprofit societies, associations,
museums, institutions, and presses.
Your use of this PDF, the BioOne Complete website, and all posted and associated content indicates your
acceptance of BioOne’s Terms of Use, available at www.bioone.org/terms-of-use.
Usage of BioOne Complete content is strictly limited to personal, educational, and non - commercial use.
Commercial inquiries or rights and permissions requests should be directed to the individual publisher as
copyright holder.
BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit
publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to
critical research.
poorly known. Existing rainfall maps are based on the data on hydrographic, vegetational and le- 1400–1600
from full-scale, long-term meteorological stations, operated gal considerations. It concerns the ar- 1600–1800
1800–2000
by national organizations linked to the World Meteor- eas of the Amazon river basin down- 2000–2200
ological Organisation, such as INMET in Brazil. Stations stream of the Andean Cordillera, the 2000–2400
with 30 or more years of uninterrupted and reliable re- Araguaia-Tocantins basin, the minor 2400–2800
cordings are very few, considering the size of the region, catchments directly bordering the At- 2800–3200
3200–3600
and most of them are located along the major rivers. It has lantic ocean, as well as the river sys- 3200–4000
been suggested that rainfall conditions away from these tems of the Guyanas and southern 3600–4000
rivers are substantially different. An analysis has been Venezuela. The vegetational criterion 4000–6400
made of the records of a network of simple pluviometric is the Amazon humid tropical forest < 6400
sites in the Brazilian part of the region as maintained by biome, including its internal edaphic July, October:
the National Agency for Electric Energy (ANEEL) since savannahs. Legally, it concerns the 9
Central month of dry season
1970. The latter data sets were used to draw more detailed countries of the Amazon Cooperation
maps on annual rainfall, and on the strength of the dry Treaty, and the 9 states of Brazilian
Meteostation
season in particular; average number of consecutive Amazonia Legal. The total area is
Full meteostation, included
in 30 year normals of WMO/FAO
months with less than 100 mm, 50 mm, and 10 mm, about 6 x 10 km2, which is equiva- Full meteostation, not included
in WMO/FAO database
respectively. Also, some data were obtained on the spatial
lent to the size of Europe, excluding Pluviometric station of ANEEL
expression of El Niño events within the region. Sub-
Russia. Pluviometric station of ANEEL
regional differences are large, and it is argued that they with short and/or less reliable data
are important for the success or failure of agricultural The digitized topographic base
settlements; for the hazard of large-scale fire damage of map of Amazonia Legal, scale
Other
the still existing primary forest vegetation; for the 1:3␣ 000␣ 000, of the Brazilian Geo-
Outside study area
functioning of this land cover as stock and sink of CO2, and graphic and Statistics Institute (2),
Ocean and inland water
for the likelihood that secondary forests on abandoned was used as cartographic base. The Altitude above 500 m
agricultural lands will have less biomass. The effects of western and northern edges were
past El Niño rainfall anomalies on the biodiversity of the taken from a map of the World Me- River
natural savannahs within the forest region are discussed. teorological Organisation (WMO) International boundary
State boundary
(3). Areas above 500 m altitude are Lowland boundary
indicated on the maps with a shading, Road
fore a valuable source of additional information, especially on by the National Agency of Electric
the absence or strength of a dry season, which is decisive for Energy (ANEEL, formerly DNAEE)
planning of agricultural settlement, avoidance of forest fires, as- on the basis of field recordings car-
sessment of biomass and biodiversity, and ecological-economic ried out by institutions such as CPRM
zoning in general. (Geological Service) and IDESP
388 © Royal Swedish Academy of Sciences 2001 Ambio Vol. 30 No. 7, Nov. 2001
http://www.ambio.kva.se
Ambio Vol. 30 No. 7, Nov. 2001 © Royal Swedish Academy of Sciences 2001 389
http://www.ambio.kva.se
(Socio-economic Development Institute of Pará State). The Hy- reliability as described above. To a degree, the hydrographic pat-
drologic Resources Department of ANEEL kindly provided all tern was taken into account, once it became apparent that away
computerized pluviometric recording data, on a monthly basis, from the main rivers and their broad floodplain the annual rain-
for all of the Brazilian Amazon. Such pluviometric information fall is higher and the dry season less pronounced than near these
networks also exist for the Amazon parts of neighboring coun- rivers, in agreement with a hypothesis of Molion and Dallarosa
tries, but they are of much lower density, except for the Guy- (5). This “Manaus rule” or brisa fluvial is defined as the effect
ana’s, and their data were not available to the author. of local river breeze circulation away from the waterbodies, due
There are 2 types of fully equipped meteorological stations. to early-daytime preferential ascending cloud formation over the
Those qualifying for the 30-year averaged “normals” criterion forest-covered land area. It apparently holds true for the stretches
of WMO (full periods 1930–1960 and 1960–1990) are indicated Solimões – Meio Amazonas – Baixo Amazonas, but is replaced
with large black dots within circles, while those that do not have by a contrasting rule for the very wide water-bodies at the mouth
recordings of sufficient length, gauge, or reliability to be used of the river system, where the tropical Atlantic ocean causes high
statistically are indicated with large black dots without circles. annual rainfall but a pronounced dry season. This is called
The recordings of the simple pluviometric sites are of varying “Marajó rule” or brisa do mar (6). In the southeastern part of
quality. Some are quite long-term, having started around 1965 the region the parts of higher altitude (above 500 m) appeared
and continued uninterruptedly till 1998. Others started recently to have somewhat higher annual rainfall and lower dry-season
or have been abandoned 10 to 15 years ago. Those sites with strength. This phenomenon was subsequently used in the fine-
less than 10 years of recording, and/or low reliability—gaps in tuning of interpolation lines.
the monthly data or obvious anomalies—are indicated on the Interpolation lines for the non-Brazilian parts of the region are
maps with small open dots, the others with small black dots. The very tentative. They are based on the WMO (3) map on annual
poor accessibility of higher-altitude and broken landscape situ- rainfall, while for the seasonal characteristics use was made of
ations of the Guyana and Brazilian Shields, and the occurrence the Length-of-Growing-Period (LGP) approach of FAO (7) and
of Indian Reserves—which often coincide with high and inac- its database on monthly rainfalls. Also a French study (8) on the
cessible terrain—have resulted in several subareas of poor or in- geography of the rainfall in the Guyana’s and southern Ven-
existent rainfall recording sites. Examples are the whole fron- ezuela was taken into account.
tier zone of the Guyana’s with northern Pará, and the interfluve Computer-driven kriging of the interpolation lines was not at-
zone between the Araguaia and the Middle Xingu rivers. tempted, because both the reliability categories of the recording
Interpolation lines of annual and monthly rainfall character- stations and sites, and the influence of large waterbodies or
istics were drawn manually, taking into account the variation in orographic features are as yet too vague to be put in formal
390 © Royal Swedish Academy of Sciences 2001 Ambio Vol. 30 No. 7, Nov. 2001
http://www.ambio.kva.se
weight classes. However, gridding of the pattern that results from A recent NGO study on the risks of forest fires (9) recognizes
the present manual interpolation is probably a relatively simple the importance of the dry season, but does not show it in map
exercise for ecosystem modelling specialists. Translation of the form, though it refers to an attempt on gridded input in a NASA-
map pattern into the Köppen or Thornthwaite climatic schemes CASA ecosystem production model (10). Characterizing a dry
is feasible as well. month as having less than 45 mm rain, it used only the normal-
Four different digital maps were prepared (Maps 1–4): ized 1961–1990 records of the recognized WMO meteorologi-
Map #1: annual rainfall pattern, with isohyets of the WMO sys- cal stations for the whole of Brazil. An early attempt was made
tem: 4000, 3600, 3200, 2800, 2400, 2000, 1600, 1400, 1200 and by the present author (11), applying the 100 mm month–1 and
1000 mm, and 2 extra isohyets in the lower range: 2200 and 1800 50 mm month–1 criteria to the recordings of the 25 meteostations
mm. available in 1959 (12).
Map #2: the number of consecutive dry months, averaged over Attention was given to the occurrence of years with excep-
the years of observation, with 100 mm monthly rainfall as cri- tionally low rainfall, which would give indications of local in-
terion. fluence of the El Niño phenomenon. However, the relatively
Map #3: number of consecutive dry months, averaged over the short length of many of the pluviometric records did not allow
years of observation, with 50 mm monthly rainfall as criterion. for delineation of spatial drought patterns with any degree of ac-
Map #4: number of consecutive dry months, averaged over the curacy on an additional map.
years of observation, with 10 mm as criterion, also implying a
low atmospheric humidity during the dry months.
In all 4 cases the central month of the dry season is indicated, RESULTS
but without line boundaries. The use of the computerized pluviometric database of ANEEL
The set of 3 maps (Maps 2–4) on the characteristics of the yields a more detailed pattern of annual rainfall (Map #1) for
dry season is rather uncommon. Many studies on the Amazon the Brazilian Amazon, when compared to published maps (3, 13,
climate provide maplets with isohyets of selected dry months, 14) and a number of page-size maps in review articles (15–18).
for instance July and December, but examples of the integration The annual rainfall away from the main rivers is higher. The spa-
of all dry months providing quantitative information on the tial pattern is complex, not only in the water-divided areas where
length, strength and regularity of the whole dry season are very orographic unevenness may have an influence, but also in ho-
few. The Length-of-Growing-Period map of FAO (7) at scale mogeneously low altitude areas. An example is the configura-
1:5␣ 000␣ 000 shows, understandably, the wet season instead of the tion for Marajó island in the mouth of the Amazon river. A far
dry one. more intricate pattern was found than assumed thus far: less rain-
Ambio Vol. 30 No. 7, Nov. 2001 © Royal Swedish Academy of Sciences 2001 391
http://www.ambio.kva.se
fall in its western interior part without a dry season, contrasting –␣ weak droughts in: 1903 (Manaus), 1911 (Manaus), 1930
with high rainfall on its coasts, but partly with a pronounced dry (Parentins), 1933 (Manaus), 1939–40 (Barcelos), 1941–42
season. Areas with low annual rainfall (< 1800 mm) occur in (Taperinha), 1951–52 (Porto Velho, Taperinha), 1955 (Manaus),
Roraima, the upper Trombetas-Jarí area of northern Pará, south- 1963 (Porto Velho, Taperinha), 1980–81 (Tapauá, São Felix do
west of Altamira on the Trans-Amazonica, and the Açailândia Araguaia), 1987–88 (Açailandia), 1995–96 (Apui);
area on the frontier Pará-Maranhão. Together they constitute a –␣ moderate droughts in: 1909 (Sena Madureira, Manaus),
sort of dry belt or corridor. However, these are also the areas 1915–16 (Sena Madureira, Taperinha, Manaus), 1936 (Manicoré,
with relatively low-density field information, and many of their Manaus), 1957–58 (Itacoatiara, Parentins), 1967–68 ( Coarí,
recording sites have moreover become nonoperational. Humaitá, Manaus), 1971–72 (Boa Vista, Porto Velho, Tarauacá),
The bulge of dryness in the Paragominas-Açailândia area, with 1976 (Rio Branco), 1991–92 (Prainha, Humaitá, Coarí,
its high degree of deforestation since the early 1960s, is note- Barcelos);
worthy. During the combined forestry- and soil survey of FAO –␣ strong droughts in: 1906 (Manaus), 1925–27 (Fonte Boa,
in 1960/61, when the primary forest coverage was still complete, Erunepé, São Gabriel de Cachoeira), 1948 (Manaus), 1983 (eve-
both the high timber volumes per ha and the low base satura- rywhere), 1997–1998 (everywhere).
tion status of the soils in the Paragominas area suggested that
in those days the dry season was little pronounced. Only 200 The ANEEL data set provides also some information on the spa-
km south of Paragominas town the timber volumes dropped be- tial variation in the strength of the El Niño phenomenon within
low 100 m3 ha–1 and the base saturation of the soils reached levels the Amazon region. The central-northern zone apparently expe-
above 35% (19), indicating a strong seasonal drought. It is pos- riences stronger El Niño effects than the southern zone. In the
sible that the strength of the dry season and its duration have north, the dry season has a gradual start and an irregular length
become more evident because of the large-scale deforestation. (“wobbly” character); in some years there are one or more com-
However, records of the Paragominas meteostation—started only pletely dry months outside the long-term average period, which
in 1965 and out of operation since 1995 because of lack of mu- has its center in January. In the south, the dry season starts quite
nicipal interest—do not show such a negative trend. Any ten- regularly and abruptly, with the center in July–August (south-
dency to decreasing rainfall in the region after large-scale de- west) or September (southeast). The dry season in the eastern
forestation may be verified easier in northern Mato Grosso or part of the Brazilian Amazon, near the ocean coast, is transitional
central Rondônia where pluviometric records are of longer term. between the 2 systems, with its center in October.
An indirect indication of a negative effect of large-scale defor-
estation on the amount and regularity of subregional rainfall is
given by the 2-lobed extension of drier conditions in the CONSEQUENCES FOR ECOLOGIC-
Barcelos-Rio Branco area in the south of Roraima State. The ECONOMIC ZONING
abundance of edaphic white-sand savannahs there would imply The new set of rainfall maps can be used in various ways. In
reduced evapotranspiration, hence less cloud formation in the dry their present vectorized form they have applications at regional,
season. state and municipal levels. In a gridded and digital form, they
Screening of the 800 individual pluviometric records; analy- can be helpful in modelling biosphere processes for the region
sis of the data of the full Brazilian meteostations—the first ones as a whole, in different scenarios of human occupation and cli-
dating from 1910—as well as scanning of the oscillations in the mate change. Gridding with 30' x 30' latitude/longitude for glo-
discharge regime of the Rio Negro-Rio Solimões system at bal models, or at 5' x 5' pixels for Amazon-level models (10)
Manaus, measured since 1902 (20, 21) give the following indi- may want to combine the data of Maps #2, #3 and #4 through a
cations on El Niño occurrences since 1900: composite legend.
392 © Royal Swedish Academy of Sciences 2001 Ambio Vol. 30 No. 7, Nov. 2001
http://www.ambio.kva.se
Ambio Vol. 30 No. 7, Nov. 2001 © Royal Swedish Academy of Sciences 2001 393
http://www.ambio.kva.se
394 © Royal Swedish Academy of Sciences 2001 Ambio Vol. 30 No. 7, Nov. 2001
http://www.ambio.kva.se
Ambio Vol. 30 No. 7, Nov. 2001 © Royal Swedish Academy of Sciences 2001 395
http://www.ambio.kva.se
396 © Royal Swedish Academy of Sciences 2001 Ambio Vol. 30 No. 7, Nov. 2001
http://www.ambio.kva.se