CHAPTER 4 - LULC DYNAMICS: THE COLONIZATION IMPACT
'A época da rapadura passou. Agora é a do chimarrão.
Mas o chimarrão é amargo, a rapadura é doce.
Só que a rapadura é doce, mas não é mole…'
4.1. Geotechnologies and LULC dynamics in Amazônia: potentials and pitfalls
The enhanced capabilities in terms of data production and methods of analysis for Earth surface feature information have led to new approaches and to a more integrative vision about LULC change within and across research sites (Burrough and Frank 1995). Using multi-temporal satellite data allows a better understanding about the dynamics of deforestation, land abandonment, pasture conversion, agriculture, and secondary succession within rural landscapes in transformation (Lambin 1997). In Amazônia, the study of LULC change and its human dimensions through the use of geotechnologies may contribute to a more sustainable development of communities under investigation.
Remote sensing and geoprocessing techniques have allowed the integration of spatial data at several scales (Quattrochi and Pelletier 1991, Fotheringham and Rogerson 1995). The generation of global- and local-scale data, the reworking of historical data sets, and some programmatic aspects of land database development have become fundamental themes to the research community (Goodchild et al. 1992, Justice et al. 1995). This has highlighted the mutual benefits of closer links between Geographic Information Systems (GIS) and methods of spatial data analysis. The evolution of these tools has consolidated important elements to solve environmental problems and help decision-making tasks (Coulson et al. 1991). The possibility of testing spatial models through the use of georeferenced databases and algorithms to measure spatial heterogeneity has opened new pathways to research issues such as LULC dynamics in Amazônia.
The increasing interest in ecosystems spatial dynamics have led to the need for new quantitative methods capable of analyzing patterns, determining the importance of spatial processes, and developing models about landscapes (Gardner and Turner 1991, Fortin 1999). Thus, ecological studies have described landscape features in terms of number, diversity, distribution, complexity, and dispersion of their spatial elements (Jurdant et al. 1977, Domon et al. 1989, Robbins and Bell 1994).
Advanced airborne and satellite technologies, image processing and analysis, and extensive capabilities to analyze spatial data through GIS and associated software has catalyzed the development and test of new quantitative methods of spatial assessment (Goodchild et al. 1993, Sample 1994, Burrough and McDonell 1998). The variety of aerial and orbital data in distinct spatial, temporal, and spectral resolutions have required the generation of digital image processing techniques in applications related to the characterization and management of natural resources (Johannsen and Sanders 1982, Szekielda 1988, Richards 1993, Jensen 2000, Lillesand and Kiefer 2000). By the same token, GIS packages have integrated an always increasing and diverse amount of spatial information (Maguire et al. 1991, Dangermond 1992, ESRI 1997, DeMers 2000).
In the Amazon, several applications have been implemented at regional and local scales. The first systematic survey about natural resources for the entire region used side-looking airborne radar (Radambrasil 1972). Experts interpreted an impressive number of images to map geology, geomorphology, soils, vegetation, and other themes. This multi-task survey took more than ten years to accomplish the manual compilation of data in several layers for 335 sheets at 1:250,000 scale. Since that initiative, the technological advances mentioned above have taken place, together with the pressing need to understand and monitor recent ecological changes occurring in the region in the wake of development efforts.
Many Amazonian issues have attracted a great deal of attention during the last thirty years. Deforestation is among the most important of them. The Brazilian Amazon has been deforested at an average rate of approximately 0.5% per year (Skole and Tucker 1993, INPE 2000) as a result of several factors, including road building, colonization programs, land speculation, and demographic and geopolitical reasons. Besides being a very large area, the process is complex due to environmental heterogeneity, distinct socioeconomic factors affecting LULC outcomes, and strategic interests at several decision-making levels.
GIS and associated software provide a data structure to efficiently store and manage ecosystems data for large areas; enable aggregation and disaggregation of data between multiple scales; locate study plots and/or environmentally sensitive areas; support spatial statistical analysis of ecological distributions; improve remote sensing information-extraction capabilities; and provide input data and parameters for ecosystem modeling (Haines-Young et al. 1996). Aware of the potential of integration between remote sensing, GIS, and associated software, and based on an international concern about possible broad-scale environmental effects due to the conversion of tropical rain forests, research teams have carried out distinct initiatives. These initiatives include the study of climatic and meteorological mechanisms of interaction between the rain forests and the atmosphere; the rates; extension and possible consequences of deforestation processes; the biogeochemical cycles in the region; the 'greenhouse effect' of trace gases; the human dimensions of LULC change; and landscape structure, function, and dynamics (Hall et al. 1996).
Several questions have arisen: When, where, and why is deforestation taking place in Amazônia and what are its consequences? What happens after deforestation? Is the amount of carbon uptake from secondary successional vegetation comparable to the amount of carbon released by deforestation? When and where is deforestation and secondary succession likely to be detected by using remotely sensed digital data? Which sensors and techniques will be necessary to achieve accurate results for these research questions?
The integration of statistical numeric and georeferenced spatial data is an important topic in geographic information science and has been one of the bases of broad-scale research initiatives in the Amazon. Procedures take into account the relationships of predefined regions in space with attributes derived from numeric variables (Bastedo and Theberge 1983, Goodchild and Brusegard 1989, Kareiva 1994). This is not a trivial question when working with the Amazon, mainly due to the diversity of data, inconsistency between different sources, heterogeneity of data collection methods, and differences in presentation of data (scale, legend, material, format, and quality) (Davis et al. 1990, Wickhman and Norton 1994, Burrough and Frank 1995).
Recent development of remote sensors, GIS tools, and associated software (e.g., database management, programming languages, and spatial analysis) has facilitated the achievement of results with significant precision and accuracy, based on objective procedures (Burrough and McDonnel 1998, Lillesand and Kiefer 2000). These techniques have allowed assessment of the spatial organization of Amazonian agroecosystems and natural resources, defining potentials for conservation and development (Coulson et al. 1991).
From this point of view, land zoning is one of the most important recent applications of GIS and associated software, yielding a better management of natural resources in Amazônia. Some states (e.g., Rondônia, Tocantins, and Maranhão) already have an extensive database generated in a GIS and based on 1:250,000 maps. They include several layers of information and synthetic maps defining areas with distinct potentialities for conservation, management, agriculture development, and urbanization (Bognola and Miranda 1999, Miranda 1999, Olmos et al. 1999, Rondônia 2000). All these initiatives used Landsat TM images to produce LULC maps.
Other examples of GIS applications include the spatial analysis done by Alves (1999) and Alves et al. (1999). Making use of maps produced through visual interpretation of Landsat TM images at 1:250,000 scale (INPE 2000), the authors analyzed geographical patterns of deforestation for states, municipalities, and road buffers. Twenty-five percent of the total deforestation was found in less than 4% of the cells and 50% of the total deforestation in less than 10% of the cells. Forty percent of the cells amassed 95% of the deforestation, and 86% of the total deforestation in Amazônia was found within 25 km from areas already deforested in 1978. This is the typical case where geotechnologies were central for the achievement of results at regional scale, showing the concentration of processes of LULC change along roads.
Skole and Tucker (1993), also using Landsat TM images and GIS integration, mapped LULC change for the entire Brazilian Amazon. Deforestation, fragmented forest - defined as areas smaller than 100 km2 surrounded by deforestation -, and edge effects - within 1 km into forest from adjacent areas of deforestation - were measured for 1978 and 1988. The results showed that deforestation increased from 78,000 km2 in 1978 to 230,000 km2 in 1988, while tropical forest habitat, severely affected with respect to biological diversity, increased from 208,000 km2 to 588,000 km2. Besides reaffirming the concentration of deforestation processes, the findings supported analyses on the human dimensions of colonization within the region (Skole et al. 1994).
Miranda et al. (1994) used a grid approach to monitor Amazonian fires. They have followed the spatial-temporal magnitude of the phenomenon for almost ten years, counting burning spots indicated by the thermal bands of (National Oceanographic and Aerospace Administration (NOAA) satellites. Although technical problems are always associated with these estimates (Setzer and Pereira 1991, Setzer and Verstraete 1994), the results achieved so far have defined the seasonality, timing, and interannual variations of fires in the region. Recent initiatives are using a multiple sensors for estimating biomass burning at a regional scale (Eva and Lambin 1998).
Another recent application for the entire region is the link of satellite, census, and survey data done by Wood and Skole (1998). The project looks for general trends in deforestation, using regression analysis on socioeconomic variables and GIS database links to census data.
Several other research initiatives using remote sensing and GIS techniques have taken place at distinct sites and more detailed scales in the Amazon. Studies of physical, biological, and social processes have helped understand how human decisions affect local and regional land use (Mausel et al. 1993, Moran et al. 1994, Skole et al. 1994, Brondizio et al. 1996). Nepstad et al. (Flames 1999) show the role of logging and fire on the impoverishment of Amazonian forests. A myriad of articles discuss the use of new sensors and techniques to monitor LULC within the region (Adams et al. 1995, Foody et al. 1996, Steininger 1996, Rignot et al. 1997, Saatchi et al. 1997, Yanasse et al. 1997, Lucas et al. 1998).
Following this literature, my research about colonization impacts in Rondônia is a contribution to the use of geotechnologies in spatial-temporal assessments of LULC. Producing multi-temporal information about two distinct settlement designs (i.e., Machadinho and Anari) may improve the capability to include detailed LULC information in regional- and global-scale simulation models through the use of accurate landscape- and property-based data. My goal is to use this research to stimulate a rethinking of settlement designs in the Amazon.