1999. Valentina Camaran

There is an ongoing controversy about the use of satellite data for studying deforestation in the Amazon Basin. It would seem that technological advances in remote sensing, especially in the form of earth observing satellites, has made it easier to the scientific community to analyze the impact and extent of human impact on the environment, as well as naturally occuring environmental changes. Nonetheless, scientists have apparently grown weary of using binary data alone to interpret and analyze deforestation issues. Some investigations and reports have been made in this fashion, but today researchers make a more detailed approach to the analysis of the causes of deforestation when combining field research and remotely sensed data.

In 1984 Salati and Vose warned about the consequences of deforestation in regional climatic patterns, concluding that the present state of equilibrium could be broken. Thus augmenting the risk of natural forest fires in some areas, and increasing the frequency and degree of flooding in others, especially in the lower Amazon. They recommended against large-scale clearance for pastures or crops that would precipitate these regional climatic changes, and offered alternatives such as sustainable development of forest resources. Through managed forestry and agroforestry techniques, plantations of rubber, jacaranda, palm, cocoa and coffee under shade, would not diminish the canopy (Salati and Vose 1984).

The Legal Amazon of Brazil refers to the area covered by the states of Acre, Amapa, Para, Amazonas, Rondonia, and Roraima, plus part of Maranhao, Tocantins and Mato Grosso. It covers approximately 5,000,000 km2 of which 4,090,000 km2 are forests, 850,000 km2 cerrado and 90,000 km2 of water (Skole and Tucker 1993). Other studies calculate the original area forested at 3,964,000 km2 (Moran 1993), 3,562,000 km2 (FAO-UNEP in Skole and Tucker 1993), 4,195,660 km2 (Fearnside et al., in S. & T. 1993), and 3,793,664 km2 (INPE in S. & T. 1993). These differences do not allow an accurate comparison between rates of deforestation (S. & T. 1993).

Deforestation in the Amazon began in the late 50s with the construction the Belem-Brasilia Highway (Moran 1993). But it was in the 80s that deforestation reached the highest rates. The assertion by J. Terborgh (in Place 1993) that population growth was the chief cause of deforestation, has been deemed inaccurate in repeated occasions. Moran (1993) cites a plan from 1966 that gave priority to large-scale operators in the occupation of the Amazon, as the main responsible for deforestation in the Southeastern Amazon. This initiative made possible for investors not to pay taxes to the Federal Government, to receive a three dollar rebate for each dollar invested, and to keep all four dollars and capital gains tax-free (Moran 1993). In a posterior paper, Moran and others propose a study to quantify the effects of policy shifts in deforestation, and to determine what forms of credit are associated with more and less destructive policies (Moran et al., 1994). Scholars agreement that the main cause of deforestation is forest to pasture conversion, as a direct result of land speculation. Policy that terms clearing of land as an "improvement" and gives land titles to short-time settlers has caused land speculation to be the first cause of deforestation (Moran 1993). In 1975 the Brazilian government announced and implemented a Program of National Integration that would exacerbate the clearing of forest for conversion to cattle pastures. By 1978 cattle had increased from near zero to 5 million, and 2 million people had settled along the Belem-Brasilia Highway. In the 1970s the rate of conversion was approximately 8,000 – 10,000 km2/yr and it averaged 35,000 km2/yr in the 80s. In 1987, 8 million hectares of previously uncut forest were burned (Booth in Moran et al., 1994), and by 1994 cattle ranches covered 8.4 million hectares (Moran et al., 1994).

Although mining in itself is not responsible for a great amount of land cleared, charcoal smelting for the Great Carajas pig-iron project accounted for the destruction of 610,000 hectares of forest per year (Treece in Moran 1993).

Logging activities account for more than 25% of the industrial output in the Legal Amazon, and for 60% in Rondonia and Roraima. In 1996, more than 278 millions of cubic meters of roundwood were extracted from the Brazilian Amazon. Selective logging does not clear entire patches of forest, thus making difficult to portray damage from satellite images alone, however, studies reflect that 10-40% of living biomass of forests is killed through the harvest process. These percentage amounts to 10,000 – 15,000 km2/yr of damaged forest that are not included in current deforestation estimates (Nature 1999).

The effects of deforestation on habitat degradation remained largely undocumented until 1993, when D. Skole and C. Tucker published a pioneering research on the effects of deforestation in biodiversity. This research, funded by the NASA, was based almost solely on interpretation of data obtained by the Landsat TM sensor. The study affirms that massive extinction of species is the primary adverse effect of deforestation and classifies the consequences in biodiversity in three main points: destruction of habitat, isolation of forest fragments and edge effect. The investigators set up to document the extent and rate of deforestation through elaboration of a base map. They parted from 210 black and white, band 5 photos taken by the TM, at a large scale of 1:500,000, primarily from 1988. In these images, deforested areas were delineated using visual deforestation interpretation, including areas of secondary growth and pastures. With the help of a GIS, a seamless data set was created for the entire Brazilian Amazon. This made a fundamental difference with previous studies that based estimates in extrapolation of information obtained from Landsat and meteorological satellites, but that were limited to study areas. The use of satellite data in a GIS allows for habitat fragmentation and edge effect calculation through spatial analysis of the geometry of deforestation. The conclusion indicates that the total affected habitat for 1988 was 588,000 km2, an increase from the 208,000 km2 in 1978. That means that habitat destruction, isolation and edge effect affected approximately 15% of forest. The results for deforestation rates and total deforestation were not very different from previous studies, although they were smaller. For 1988 a total of 230,000 km2 and an average rate of 15,000 km2/yr were deforested. Most dramatic is the rate of habitat degradation of 38,000 km2/yr, with disastrous implications for biodiversity (Skole and Tucker 1993).

In 1994, a study by E. Moran, E. Brondizio, P. Mausel and Y. Wu, criticized the previous research as limited to monitoring rates of deforestation and falling short of addressing land-use and policy alternatives. The propose a multidisciplinary approach that encompasses biological, social and physical scientific technique to ideate policies based on locally used strategies currently in use, that are of lesser environmental consequences.

This is a very complex study that aims to analyze the process of land cover change following deforestation, along a fertility gradient from eutrophic to oligotrophic conditions. That is, from poor soils to less poor soils. Its long-term objective is to discover areas of rapid regrowth and analyze through field studies and multitemporal satellite imagery, the different impact of land use, soil type and size of area cleared on secondary succession. The results could be used to make environmental preservation policies that draw on local strategies, rather than documenting the extent and effects of deforestation.

For the purposes of the research, areas of relatively rich soils were chosen because they enhance regrowth and permit successional processes to occur in a short period. They rely on two images, 1985 and 1991, from Landsat TM, with a high resolution of 30 m that allows the identification of 1-hectare fields. The two midinfrared spectral bands were used to distinguish successional forest classes. After land cover classes of interest were identified, they selected training sites for each class. A complicated method of classification and clustering of areas based on classes was designed to develop a conceptual spectral model used to define 9 land cover classes considering reflection and absorption characteristics, using 6 different bands (1-5+7). The study classifies secondary succession in three stages:

-Initial secondary succession: The proverbial "degraded" pastures start with woody invasion two years after clearing.

-Intermediate secondary succession: Following a period of abandonment of 6-10 years, show woody growth of more than 8 m.

-Advanced secondary succession: After approx. 15 years. Reflectance is close to mature forest and are considered reforested or almost reforested areas.

The time range refers only to eutrophic soils. In poorer oligotrophic soils the stage of initial secondary succession can last for 15 years. Using the spectral signature of the field studies, the paper concludes that areas of abandoned pastures in the intermediate secondary stage are the most common culturally induced feature, and consequence of a decline in subsidies and changes in credit policies. The rate of regrowth in these areas, called "degraded" by ranchers, depends on many variables including, but not limited to soil nutrients and seed bank, how thoroughly the area was burned and the grade of slope. The research also mentions that the uniqueness and importance of satellite data is that it permits a regionally broad analysis. But that linking it with in-depth analysis of fieldwork would "address the relation between global, regional and local environmental change." (Moran et al., 1994).

David Skole, with colleagues, published a paper later that year in the magazine Bioscience. Its title "Physical and Human Dimensions of Deforestation" is a clear signal that this time a more inclusive approach was undertaken. The article starts stating that the study of deforestation unequivocally implies study of its causes. Their main intent is to create a set of deforestation data by comparing and overlapping information from remote sensing satellite binary data and governmental land census data from 3973 towns in Brazil. They later analyze the possible causes of deforestation and its temporal correlation with population growth and both national and international economical factors. They introduce a new scheme to measure deforestation: utilization of data from the Brazilian Census of Agriculture which reports the amount of land that is used for diverse kinds of agriculture and pasture. While this information is not indicative of the total amount of land that has been deforested, it gives other details such as land use patterns.

To use the information in a comparative manner, they constructed a digital map parting from base maps and converted the paper data in a geographic information system. The resulting pattern of deforestation for agriculture is reported as "similar" to that of satellite obtained maps of deforestation. They conclude that land census data alone can be used to analyze deforestation for the years before remotely sensed data was available. The paper addresses the dynamic of secondary growth areas, stating that usually these are included as areas of deforestation and not considered as producers and storers of carbon for carbon release studies. The cycle of conversion for secondary growth areas is estimated as a 5 year turnover time for areas of primary forest that have been cleared for agriculture, then abandoned for a while and cleared a second time of secondary growth. Maintenance of this dynamic is deemed essential for the mode of production of this area.

The article explains the relation between international fall of oil prices and deforestation in the following manner: The transfer of wealth from industrial nations to OPEC nations created a need for international banks to find borrowers in order to pay interest to national accounts that had deposited "petrodollars". Brazil’s response to this international phenomenon was twofold: to reduce the importation of oil and fund projects of domestic energy sources, such as construction of dams, with money borrowed from foreign lenders. On the other hand, Brazil modernized its agricultural sector in order to export products that would generate income in dollars, to service the debt. This created a key cause for the clearing of new land for the cultivation of exportable crops, which are mainly mechanized crops such as soybeans and wheat which are effective only on large sized farms. By 1977, half of the value of total crops was export crops.

Finally, the authors conclude that in order to create a comprehensive set of measures to fight against deforestation, an interdisciplinary approach must be taken:

-Use remote sensing data to measure and analyze deforestation trends.

-Compare the data with field investigations and case studies analysis with information from governmental sources, such as the Brazilian land census, and

-Consider international political, institutional and economic forces that can be considered as external factors for Amazonian deforestation.

The most recent article published regarding estimated effects and causes of deforestation, is an article from Nature magazine. It argues that while satellite data is the fastest and cheapest way to account for deforested land, it neglects on factors that do impoverish forest conditions without completely clearing land, concentrating primarily in logging activities and surface fires.

The article explains that selective logging does not clear entire areas of land, but that during the harvest process companies kill or damage 10% - 40% of living biomass. And that by reducing forest canopy, sun is allowed to penetrate and dry the floor, thus creating vulnerability to fire. They interview logging companies, obtain harvest records in number of tree trunks and harvest rates as m3/hectare of forest, to calculate the area of disturbed forest necessary to supply each company. They compare the accuracy of the companies’ reports by comparing with previous studies. They distinguish three different categories of intensity:

-Low, or approx. 19 m3/hectare

-Moderate, or approx. 28 m3/hectare

-High, or approx. 40 m3/hectare

They show in a table that while Para and Mato Grosso are the main producers of timber, followed by Rondonia, in Maranhao 100% of logging is done at high intensity. To prove the lack of detail of satellite data in regards to forest fires, they interview landholders in some states and had them draw deforested areas and areas burned by surface fires on satellite images. The researchers compared the results with satellite obtained images and reported similarities but underestimation by the landholders. The article reports that areas attacked by surface fires are 1.5 times greater than areas reported as deforested for the years studied (1994-1995). It also addresses the potential graveness of fires during the severe drought conditions, specially ought to El Nino Southern Oscillation. They assume that forest becomes potentially flammable when moisture from soil is depleted five meters deep.

They accuse that impoverishment by logging and fires, is not accounted for in estimates of Amazon carbon balance. Superimposing their newly obtained figures and maps, they conclude that for the study area of Paragominas, where 62% of land is officially deemed as forested, on reality only 10% supports undisturbed forest. They conclude that while satellite image is essential in determining human impact, there is a need to further document consequences and point to the effects of efforts to sustain migration to the area as contrary to deforestation prevention efforts.

The Landsat project properly started in 1972 when the first-of-a-series satellite entered orbit. Its mission was the same of today’s Landsat 7, to assess landcover dynamics at regional and global levels, to determine its causes, and to generate a fresh, cloud cover-free global surface inventory of images. The Landsat 7 satellite was built by Lockheed Martin Missiles and Space, and the instruments by Hughes Santa Barbara Remote Sensing, contracted by the NASA Goddard Space and Flight Center. It uses solar energy and nickel-hydrogen batteries to move its 4632 lbs. in a near polar circular orbit 14 times each day, using thrusters and other apparatus. It has a high spatial resolution of 30 m, a swath width of 185 km. and a repeat cycle of 16 days. The Eros Data Center works together with the Goddard Space and Flight Center in flying and operating Landsat 7’s instruments. (1)

The bulk of the information generated by Landsat 7, both images and scientific data, are processed and transformed from raw data into Hierarchical Data Format (HDF) files, by The Goddard Space Flight Center. The Landsat 7 captures images with its Enhanced Thematic Mapper plus (ETM+) sensor. The ETM+ sensor works with 8 different spectral bandwidths: two infrared; one multispectral or panchromatic; one each for red, green and blue. This sensor is the next generation of the one that was lost to the Landsat 6 in 1993, and its predecessor is in Landsat 4 and 5. For the first time since the start of the Landsat project, the basic data product of this satellite is available in level 1R, a radiometric correction by the Landsat Processing System. Data is also geometrically corrected as a second step, for level 1G, which makes the product able to support 7 different map projections, Mercator, polyconic and others. (1)

The Eros Data Center in Sioux Falls, South Dakota receives approximately 200 scenes everyday from the satellite, and 50 or 60 more scenes per day in stored form from the two other ground stations at Alaska and Norway. The first two stations collect and distribute scenes from the contiguous United States and Alaska, and the station in Norway collects images of foreign land masses. There are also International Ground Stations (IGS), a network of centers in several countries around the world, that capture scenes from the Landsat satellites, within their acquisition circle. Thus, the US Ground Stations downlinks and stores images from foreign land masses, but it is a more comprehensive effort at each IGS. The IGS in Brazil is the Instituto do Pesquisas Espaciais. Mr. Paulo Martins Serra is the head of the project. (2)

Landsat7 has an onboard recorder with a capacity of 380 Gb of data, or 100 scenes. The Eros Data Center is capable of capturing and processing 250 scenes a day, or roughly 1.2 Terabytes of information. Users from all around the world will have access to 100 new images each day, at the cost of fulfilling their request. The Landsat 7 is scheduled to last 6 years (1).

Figures 1 and 2, are pictures of the Bay of San Francisco, taken by the Landsat 5 and 7. Despite that no information as to what bands are represented in the Landsat 5 image, it is clear the difference in resolution introduced by Landsat 7. Landsats 1-5 carried versions of a sensor called the multispectral scanner (MSS), which collected data simultaneously from four broad bands of the electromagnetic spectrum, from visible green through near-infrared wavelengths (6). Figure 2, taken by Landsat 7 is a typical full scene. This composit overlaps bands 2, 3, and 4, with a resolution of 30 m, and the panchromatic black and white band, with a resolution of 15 m. It is originally a high definition digitally engineered image, converted to a small file size for web presentation purposes (4)(5).

Since 1991 there is a Burning Monitoring Project in Brazil. AVHRR data obtained from a NOAA satellite is processed in Cachoeira Paulista by the INPE and sent to NMA and ECOFORCA, who are the teams that advanced this project. The Agencia Estado later publishes this data as maps (7). See figure 3. AVHRR is a sensor in NOAA satellites that are used for meteorological purposes. This sensor has a resolution of roughly 1km. (6)

CBERS is a joint initiative between the governments of Brazil and China to monitor Earth resources and human impact. Currently it consists of two satellites, however, there are plans for a third and fourth launching in the next 6 years. CBERS-1 was launched as recently as October 1999, and CBERS-2 is projected to start orbiting the Earth two years after. Each new generation of this satellite will have an improved version of the sensors aboard the previous. With this venture, Brazil hopes to cut its dependence on foreign satellites for meteorology and communication and become the leader of the Third World in remote sensing (10). It has three onboard sensors. The Wide Field Imager (WFI) has two bands, one green and one near infrared, a ground swath of 890 km, and resolution of 260 m, or almost 9 times less than the average resolution of Landsat 7. There is a CCD camera with a high resolution of 20 m, which may be used to zoom in any phenomenon, and to get stereoscopic view of a terrain. The third and last sensor is an Infrared Multispectral Scan, IR-MSS has a resolution of 80 m and covers the infrared thermal range in four bands (9). Figure 4 shows a section of the Jupura river (on the bottom of the image) and the Valpes river (on the top), southwest of Amazonas State. The image was obtained on October 21, 1999 (8).

Terra is the first bird of Earth Observing System, an international effort to study environmental and climatic changes globally. Formerly known as EOS-1, it was successfully launched on December 1999 and the first images are expected to arrive in early 2000. Its payload is drastically different to Landsat 7, it produces much coarser data that is easier to handle. (12) and pro HAVE TO COMPLETE INFO

 SPOT is the French satellite: http://edcftp.cr.usgs.gov/glis/hyper/guide/spot

ENVISAT-1 is the first of a new generation of observation satellites. It is scheduled to be launched in mid 2001 by the European Space Agency. Its payload will include: two radar instruments, three spectrometers of different types and measurement characteristics, two different radiometers (broad and narrow band), the first high-resolution spaceborne interferometer for long-term observation, and two instruments for range measurements (11). See figure 5.

The technology advances in remote sensing during the last 30 years have been extremely helpful in studying regional processes in the Amazon Basin. With the new Lansat 7 satellite and its fine resolution sensor, more conclusions can be drawn from each image. Other countries are starting to see results of their efforts to parallel their technology to the current paradigm, and achieve independence from the United States managed remote sensing apparatus.

The use of satellite imagery is fundamental to the analysis of regional trends. Given the size of the terrain, and the difficulty it poses to travel inside it, researchers have to be able to watch the entire area to achieve a complete understanding of the processes in the Amazon Basin. Extrapolation of field studies alone cannot portray accurately the phenomena that occur in the area and the land cover change dynamics. However, as the investigation effort conducted by Moran et al. showed, the details of site specific events can be accurately identified linking fieldwork and the corresponding area’s satellite data, and through identification of its spectral signature similar processes can be spotted in a regional wide map.

I definitively agree that research studies that aim at discovering currently used environmentally sound strategies, to try to encourage them through legislation and policy, is an initiative that can prove more successful than blind prohibitions of adverse activities without offering any possible alternatives in exchange.

Chomentowski, W., B. Salas, and D. L. Skole, "Landsat Pathfinder project advances deforestation mapping. GIS World 7.4 (1994): 34-38.

Laurance William F., Laurance Susan G., Ferreira Leandro V., Rankin-de Merona Judy M., Gascon Claude, and Lovejoy Thomas E. "Biomass Collapse in Amazonian Forest Fragments." Science 278 (1997): 1117-1118.

Place, Susan, ed. Tropical Rainforests: Latin America Nature and Society in Transition. Delaware: Scholarly Resources, 1993.

Moran, Emilio F. "Deforestation and Land Use in the Brazilian Amazon." Human Ecology 21.1 (1993): 1-21.

Moran E. F., Brondizio E., Mausel P., Wu Y. "Integrating Amazonian Vegetation, Land Use and Satellite Data." Bioscience 44 (1994): 329-338.

Nepstad, D., Verissimo A., Alencar A., Nobre C., Lima E., Lefebvre P., Schlesinger P., Potter C., Moutinho P., Mendoza E., Cochrane M. and Brooks V. "Large-scale impoverishment of Amazonian forests by logging and fire." Nature 398 (1999): 505-508.

Skole, D. L., W.H. Chomentowski, W. A. Salas and A. D. Nobre, 1994. "Physical and human dimensions of deforestation in Amazonia." BioScience 44.5 (1994): 314-322.

Skole, D. and Tucker, L. "Tropical Deforestation and Habitat Fragmentation in the Amazon: Satellite Data from 1978 to 1988." Science 260 (1993): 1905-1910.

 (1) Dr. Williams, D., Dr. Irons, J., Dr. Goward, S. "Landsat 7." http://ltpwww.gsfc.nasa.gov/LANDSAT/CAMPAIGN_DOCS/Guides/LANDSAT-7_dataset.html (8 Dec. 1999)

(2) "Landsat 7: Science Data Users Handbook." 15 October 1999 http://ltpwww.gsfc.nasa.gov/IAS/handbook/handbook_htmls/chapter13/chapter13.html#section13.1 (8 Dec. 1999)

(4) "More Details about Landsat 7 Browse Images" 28 April 1999. http://ltpwww.gsfc.nasa.gov/LANDSAT/CAMPAIGN_DOCS/DATA/Browse/Landsat7_BrowseInfo.html (8 Dec. 1999)

(5) "Landsat 7 Browse Image Path Row Gallery." 6 May 1999. http://ltpwww.gsfc.nasa.gov/LANDSAT/CAMPAIGN_DOCS/DATA/Browse/Full_Scenes/L7_PathRowGallery.html (8 Dec. 1999)

(6) "The U.S. Geological Survey and Remote Sensing." Aerial Photographs and Satellite Images. November 1999. http://mapping.usgs.gov/mac/isb/pubs/booklets/aerial/aerial.html (8 Dec. 1999)

(7) Agencia Estado. http://agest.ecof.org.br/ingles/projetos/burnings/ (8 Dec. 1999)

(8) INPE. http://www.dss.inpe.br/programas/cbers/english/imagens.html (8 Dec. 1999)

(9) INPE. http://www.dss.inpe.br/programas/cbers/english/sensor.html (8 Dec. 1999)

(10) INPE. http://www.dss.inpe.br/programas/cbers/english/historia.html (8 Dec. 1999)

(11) ESA. http://envisat.estec.esa.nl/m-s/space/payload.html (8 Dec. 1999)

(12) Terra. http://terra.nasa.gov/