Land Use Abstracts - MASTER
Transcrição
Land Use Abstracts - MASTER
Land Use and Land Cover Change 177 Effects of land-use on soil properties, macronutrients and carbon stocks in Eastern Acre, Western Brazilian Amazon Eufran Ferreira do Amaral 1 Irving Foster Brown 2 Divonzil Gonçalves Cordeiro 3 Antonio Willian Flores de Melo 4 The fate of soil nutrients and carbon stocks as a function of land-use is an essential information for understanding and planning development patterns in Eastern Acre. Three pastures and secondary forests of varying ages were analyzed regarding inter-time rates of change in apparent density, macronutrient content (0-20, 20-40, 40-60 cm), root biomass (0-20, 20-40 cm) in geo-referenced sampling localities with four replications within a 50 m radius. Current land use causes reduction of carbon stocks from 17±2 t C ha-1 in recently burned sites to 6±1 and 13±5 t C ha-1 in young (1-3 years) pastures and old(>6 years) secondary forests, respectively. Potassium increased from 65±42 kg ha-1 in young pastures to 160±58 kg ha-1 and 77±31 kg ha-1 in old and mature(3-6 years) pastures, respectively. Calcium was an efficient indicator in secondary pastures and dropped from 2500± 699 to 680± 154 kg ha-1 between young and old secondary forests. Apparent density varied from 1,3 ± 0,1 to 1,6 ± 0,1 g cm-3 in the recently burned area and old fallow, with an increase in pastures and a reduction in the young fallow areas. The fine roots had lower biomass in the areas of young pastures (4,6 ± 0,9 t ha-1). The land use system in Acre lead to inadequate soil management, since the cycles of slash and burn, in soils with low agricultural potential, without replacement of nutrients result in a gradual reduction in the total stocks, reflecting in decreasing crop productivity. 1 Centro de Pesquisa Agroflorestal do Acre, Embrapa, Rodovia BR-364, km 14, Caixa Postal 392, CEP 69908-980, Rio Branco, Acre, Brasil, Email: [email protected], [email protected], Tel.: 68 224 3931/3932/3933, Fax: 68 224 4035; 2 The Woods Hole Research Center, UFF, UFAC, Email: [email protected]; 3 Fundação BIOMA, WHRC, LBA, PZ/UFAC, Email: [email protected], 4 Centro de Pesquisa Agroflorestal do Acre - Embrapa, Email: [email protected] Land Use and Land Cover Change 178 A evolução do uso da terra na Reserva Extrativista Chico Mendes, Acre - Brasil Hiromi Suzana Yamasaki Sassagawa 1 Yosio Edemir Shimabukuro 2 Irving Foster Brown 3 Na Amazônia, as Reservas Extrativistas (Resex) são uma alternativa para o uso da terra que podem manter a cobertura florestal e gerar renda aos seus moradores. A Resex Chico Mendes (RECM) é a maior do país, com quase um milhão de hectares, e está localizada na região leste do Estado do Acre. Ela foi decretada em 1990 e está sob administração do IBAMA e associações dos moradores. Está dividida em 45 seringais que variam em área de 4.000 a 70.000 hectares. Neste trabalho, procuramos responder a pergunta: onde e como procede a ocupação da RECM? Usamos imagens Landsat-5 TM de 1986, 1992 e 1998 para determinar áreas desflorestadas associadas com colocações, isto é, unidades familiares. Em 1986, o desflorestamento foi de 0,7% da RECM, em 1992, 1,5% e em 1998 foi de 2,9%. Verificamos que esse desflorestamento distribuiu-se desigualmente dentro da RECM. Primeiro, poucas colocações ocupam áreas dominadas por bambu (aproximadamente 40% da RECM). Segundo, a ocupação na região leste e sudeste, perto das cidades de Xapuri e Brasiléia, foi maior do que na região oeste da RECM. Esta concentração de ocupação em seringais específicos pode gerar problemas administrativos, pois o limite de desflorestamento dentro da RECM é de 10%. Se continuarem com a taxa de desflorestamento de 1992 a 1998, oito seringais (Paraguassu, Carmem, Humaitá, Porongaba, Filipinas, Floresta, Independência e São João do Iracema) ultrapassariam o patamar de 10% dentro de uma década. Estes seringais representam 12% da área da RECM. 1 SETEM- Parque Zoobotânico,Universidade Federal do Acre, Rio Branco, 69.915, Email: [email protected], Tel.: 68 229 1642, Fax: 68 229 1246; 2 INPE; 3 WHRC/UFF/SETEM-PZ-UFAC, Email: [email protected] Land Use and Land Cover Change 179 Smoke across the borders in southwestern Amazonia: Fires in Bolivia affect air quality in Acre, Brazil Irving Foster Brown 1 Elaine Prins 2 Alberto Setzer 3 Carlos Nobre 4 C.E. Deus 5 Changing land use has resulted in fires and accompanying smoke that affect Amazonia as a regional entity. In mid-August 1999, a combination of meteorological conditions, large fires, and the need to predict airport closings provided an opportunity to identify smoke sources. By 15 August, GOES-8 satellite imagery identified major burning near the Santa Cruz region in Bolivia and the National Institute for Space Research (INPE) received personal accounts of the burning by email. A major cold front penetrated from southern Brazil and reached Rio Branch, Acre, Brazil (9° 57' S 67° 49' W), dropping temperatures from the 30°C to 12°C by the morning of 16 August. Concomitantly, limited visibility caused by smoke closed the airport in Rio Branco. GOES-8 imagery for 15 August showed a smoke plume over 1,000 km long, extending NW from the Santa Cruz region in Bolivia to Acre. Few hot pixels occurred in the region of Acre. A meteorological transport model showed low-level wind speeds of 10-15 m s-1 (ca. 1,000 km dia-1) during 14 and 15 August, heading from Santa Cruz to Rio Branco. Analysis of smoke distribution indicated that the smoke plume continued westward and reached western Peruvian Amazonia by 19 August. The above event underscores the international aspects of fire and smoke impact in southwestern Amazonia. 1 Woods Hole Research Center - UFF, UFAC, Parque Zoobotânico, Rio Branco, 69.915, Email: [email protected], Tel.: 68 229 1642, Fax: 68 229 1246; 2 NOAA/NESDIS/ORA, Email: [email protected]; 3 LTID/INPE, Email: [email protected]; 4 CPTEC/INPE, Email: 5 [email protected]; IMAC-Acre, Email: [email protected] Land Use and Land Cover Change 180 Impacto de Queimadas Acidentais no Estado do Acre, Brasil Elsa Renee H. Mendoza 1 Irving Foster Brown 2 Daniel Nepstad 3 Diogo Selhorst 4 O Acre representa uma expansão da fronteira agrícola e faz parte do arco de desmatamento, onde acontece a maior ação antrópica na Amazônia Brasileira. O principal processo para a conversão da floresta em pastagem e agricultura é a queima, assim como para a limpeza e manutenção das áreas convertidas. Este estudo vem sendo realizado próximo do município de Rio Branco. Para levantamentos de dados, foram realizadas checagem de campo e entrevistas junto a pequenos produtores (<100 ha), médios (101 a 1000 ha), grandes (1001 a 5000 ha) e muito grandes (>5000 ha). Resultados de 59 entrevistas mostraram que o fogo acidental era um problema sério para 60% dos entrevistados em 1994 e 1995, mas somente 30% confirmaram a presença de fogo entrando na mata. O alto custo da mão-de-obra fez com que 80% de pequenos e médios agricultores não realizassem aceiros. Nas 20 entrevistas realizadas no mês de janeiro de 2000, os dados indicam que este panorama mudou. Aproximadamente 85% dos entrevistados relatou que o fogo acidental foi um problema sério em 1998 e 1999, e 70% teve perdas de pastagem, cercas e de seus cultivos agrícolas implantados de dois a três anos. Quase 50% afirmaram que os aceiros não estavam conseguindo impedir o fogo, 50% concordaram que os ventos fortes e as árvores mortas são uns dos principais propagadores do fogo. Cerca de 80% afirmaram que o fogo rasteiro entrou na floresta. Estes dados preliminares indicam que as queimadas têm cada vez mais, impactos socioeconômicos e, que as florestas primárias no Acre estão deixando de ser uma barreira natural no controle do fogo. 1 Instituto de Pesquisa Ambiental da Amazônia, Parque Zoobotânico, UFAC, Rio Branco, 69 915, Email: [email protected], Tel.: 68 229 1642, Fax: 68 229 1246; 2 WHRC-UFF-UFAC, Email: [email protected]; 3 WHRC-IPAM, Email: [email protected]; 4 BIOMA, Email: [email protected] Land Use and Land Cover Change 181 Modeling Deforestation at the Landscape Level from Analysis at the Property-Farm Level: Findings and Forecast from the Altamira Case Bruce Boucek 1 Stephen D. McCracken 2 Eduardo Brondizio 3 Analysis of remotely sensed data at the level of individual farm properties provides many additional insights into the nature and pattern of frontier expansion at the landscape level. Trajectories of land-use are developed from the analysis of 402 household-farm surveys combined with information from remotely sensed analyses at the farm level. Both environmental and socio-demographic factors shape land-use decisions and the pattern of deforestation. Variables found to be important in patterns of land-use and deforestation are distances to highway, topography, and availability of water (environmental/contextual variables), initial capital, credit, and type of agricultural emphasis (e.g., cattle/pasture, perennials, and annual cash crops). Results from the multivariate analysis at the individual farm level are used as inputs for forecasting patterns of deforestation, using a Property grid, for the landscape under different scenarios. 1 Indiana University, Building 331, IN 47405, USA, Email: [email protected], Tel.: 812 855 6181, Fax: 812 855 3000; 2 Indiana University, Email: [email protected]; 3 Indiana University, Email: [email protected] Land Use and Land Cover Change 182 Scale and deforestation analysis in an Amazonian frontier: Understanding the variability across farm lots, farm cohorts, and the colonization landscape Eduardo S. Brondizio 1 The goal of this poster is to present a disaggregated analysis of land cover trajectories and deforestation at the scale of the farm, farm cohort, and the colonization landscape and to discuss variability in rates and extent of deforestation across these units of analysis. The role of time of settlement, farm development cycles, and period effects (credit, migration flows, inflation, etc) is discussed. Time series remote sensing data allow us the required multi-temporal synoptic view of this process that captures the arrival of colonization cohorts and the simultaneous process of frontier consolidation and expansion. Derived from Landsat MSS and TM images and from aerial photography, this poster examines land use and deforestation in the Transamazon highway (from Km 18 to KM 140 west of Altamira, Pará State) during the period of 1970 to 1996. Image classifications and maps representing 1970, 1973, 1975, 1976, 1978, 1979, 1985, 1988, 1991, and 1996 are geo-referenced to a property grid covering approximately 4,000 farms (100 hectares each) settled by the Brazilian government during the past 28 years. Land cover data encompasses classes of forest and deforested areas (Landsat MSS data and aerial photos), pasture, crop sites and three stages of secondary succession (Landsat TM data). This data set covering the entire period of colonization and linking fine-scale property level data to landscapebased maps will provide a baseline to examine deforestation across the scale of the farm, cohort of farms, and the colonization landscape. 1 Indiana University, Student Building 130, Bloomington, IN, USA, 47405, Email: [email protected], Tel.: 812 855 6181, Fax: 812 855 3000 Land Use and Land Cover Change 183 Land use systems and landscape footprints in Indigenous, Caboclos, and frontier farmers settlements in Amazonia Eduardo S. Brondizio 1 Fabio de Castro 2 In this poster, we present an analysis of the relationship between land use system and landscape structure (e.g., pattern of forest fragmentation) by taking into account three sets of variables: -socioeconomic and cultural (e.g., land use composition, land tenure, settlement pattern), -biophysical (e.g., topography, drainage, soil distribution), and infrastructural (e.g., road network). These are key variables underlying a farmer's decision on where and what to deforest, what is planted, and for how long, therefore, impacting the spatial and temporal configuration of a region's land cover. In this poster we compare 7 research sites representing different types of land use systems distinguished according to two main criteria. First, land use systems are organized by length of fallow cycle (from long- to short-fallow shifting cultivation). Second, these systems are organized by degrees of co-existence of land use systems (from swidden manioc garden to a gradual combination of annual, pasture, and agroforestry systems). The sites representing the variation in land use systems are derived from long term anthropological and remote sensing research in areas of 2 Indigenous (Vaupes region), 3 Caboclo (Marajo island), and 2 immigrant farmers (Bragantina and Tome-Acu) communities. The role of each group of variable in driving a site's land cover spatial pattern is assessed through 3 groups of spatial metrics: composition (type and number of land cover classes and distribution), configuration (number, size, neighborhood, shape of patches), and correlation (types of trajectories, ratios between classes). A comparative analysis disaggregated at the level of patch, class, and landscape is presented. 1 Indiana University, Student Building 130, Bloomington, IN, USA, 47405, E-mail: [email protected], Tel.: 812 855 6181, Fax: 812 855 3000; 2 Indiana University, Email: [email protected] Land Use and Land Cover Change 184 Manaus Rainforest in the LBA: Physical, Ecological and Land Use Characterization of the site Alessandro Carioca de Araujo 1 A.O. Marques 2 F. Luizão 3 A.D. Nobre 4 R. Dallarosa 5 R.C. Luizão 6 In the big East-West LBA transect Manaus is at present the western-most site, with a stern climatic contrast to the nearest Santarém site, in terms of dry season length, and without the ocean influence of the Caixiuana site. Manaus also has much less deforestation disturbance in the vicinity of the site, if compared with Ji-Parana in Rondonia, and the prevailing wind blows over extensive expanses of undisturbed rainforest before it hits the two towers in the Cuieiras reservation. Since the late seventies this reservation has been extensively studied for hidrometeorology (bacia modelo) biomass and growth of trees (selective logging project; Jacaranda project) nutrients, ecology and disturbances associated to natural and artificial gaps in the forest (Bionte project), ecophysiology and soil-plant-atmosphere relationships (ABRACOS project) and a variety of Masters and Ph.D. thesis ranging in subjects from seed dispersal to botany and litterfall studies. We characterize the basics on climate, micrometeorology, geomorphology, soils, vegetation, land use, hydrology, biomass and ecology (litter and nutrient studies). We also critically analyze the site’s prospects for the next years. 1 INPA, Av. André Araújo, 2936, Petrópolis, Manaus, AM, Brasil, 69083 000, E-mail: [email protected], Tel.: 92 643 3238, Fax: 92 6433155; 2 INPA, Email: [email protected]; 3 INPA, Email: [email protected]; 4 INPA, Email: [email protected]; 5 INPA; 6 INPA Land Use and Land Cover Change 185 What Difference Does Soil Quality Make? Inter- and Intra-site Comparisons on How Amazonian Soils Affect Land Cover Trajectories, Crop Choice and Land Tenure Emilio Moran 1 Eduardo Brondizio 2 The presentation shows that the better soils of the Amazonian terra firme forests (TR, or terra roxa estruturada eutrofica) are associated with much faster rates of forest regrowth following deforestation and land abandonment and that this faster regrowth accelerates through time. We also show that the structure of the secondary forest is quite different in areas characterized by TR, with much taller canopies and higher basal area in the overstory than areas dominated by latosolos e podsolos. We have found that as the proportion of terra roxa increases, there is a larger propensity to diversify land use, and to reduce the proportion in pastures, whereas as properties with no terra roxa at all more than 80 percent of the area cleared is in pasture after 20 years of occupation. We have found that most farmers with lots with substantial areas in terra roxa are more likely to still be on their properties after 25 years, as compared with lots with poor soils which have suffered high rates of attrition. While we do not argue that soils determine the outcome of land cover in Amazonia after deforestation, we suggest that much greater attention needs to be given to soil quality assessment and mapping before road construction and colonization. Doing so will reduce the high environmental and social costs resulting from uncritical clearing of forest on poor soils. 1 Indiana University, ACT, [email protected] , Tel.: [email protected] 701, E. Kirkwood, SB 331, Bloomington, IN 47405, USA, Email: 812 855 6181, Fax: 812 8553000; 2 Indiana University, ACT, Email: Land Use and Land Cover Change 186 Designing Vegetation parameter Models for Selected LBA Study Sites in the Amazon, Brazil Hui Li 1 Emilio Moran 2 Genong Yu 3 Paul Mausel 4 Eduardo Brondizio 5 Statistical analysis and modeling interfaces were integrated into the master database to begin analysis of problems of core LBA issues. Integration of statistical methods and modeling algorithms has been initiated and through visual basic has been made available for use through user interfaces with all the field and spectral data in the Oracle data base. Through SQL, data selected for statistical analysis or modeling purposes can be extracted, displayed and analyzed to address specific LBA research needs. Currently, two test applications have been completed. One application focuses on factor, regression, and multiple regression analysis using soil and vegetation parameters to provide insight into soil changes associated with succession changes. The second application focuses on biomass modeling. The interface developed for biomass modeling permits instantaneous estimates of biomass derived from 13 allometric equations deemed to be most suitable for the study areas. Each biomass model directly extracts relevant data specified by an analyst from the database and calculates the data inherent in the model. In the current biomass model, data such as: 13 biomass estimates, tree/sapling biomass ratio, total basal area tree/sampling basal area, tree density, sapling density tree volume, and average stem height can be calculated at different geographic levels of interest for mature forests and various succession forest classes. The models and interfaces developed in this research have provided and proven the methodology for analysis and modeling using the integrated Oracle database. Proof of concept for analysis and modeling field-measured parameters with field-measured parameters and spectral parameters with field measured parameters have been accomplished. 1 Indiana State University, Email: [email protected]; 2 Indiana University, Email: [email protected]; 3 Indiana State University, Email: [email protected]; 4 Indiana State University, Email: [email protected]; 5 Indiana University, Email: [email protected] Land Use and Land Cover Change 187 Interface Development using SQL and Visual Basic for an Amazonian Soil Vegetation, and Spectral Database Supported by Oracle Ping Jiang 1 Emilio Moran 2 Paul Mausel 3 Eduardo Brondizio 4 Data from seven Amazonian study sites include detailed field measurements of vegetation parameters (primarily in mature forest and various succession classes) including trees, shrubs and economic species, soils including carbon at five levels, landuse history, and landscape metrics. A spectral data base with more than 30 Landsat TM scenes supplemented by JERS radar currently exists with additional data to be added. These data can be associated at four geographical levels which include region, site, plot, and subplot. A vital element in using this vast data resource to address LBA research requires that it be organized in a manner which permits retrieval of all data in any combination in both tabular, image, and graphic form. Organization, query, display, and integration of the various databases are being conducted with Oracle using SQL and Visual Basic programming to create user interfaces. This system permits tabular displays of all fieldacquired data in any combination designated. In addition, an image viewer function can display raw remote sensing images, location of sample sites on remotely sensed images, and graphs of multispectral and/or multitemporal spectral data associated with sample sites. The tabular field data can be displayed simultaneously with graphics and image displays. This has provided a comprehensive integration and organization of project data which permits sophisticated displays relevant to specific project needs. It is organized to accept additional interfaces designed to utilize data existing in the Oracle system which facilitates statistical analyses (i.e. factor, principal component, regression, multiple regression, correlation) and modeling needs of project research. 1 Indiana State University; 2 Indiana University, Email: [email protected], 3 Indiana State University, Email: [email protected]; 4 Indiana University, Email: [email protected] Land Use and Land Cover Change 188 The Role of Ecological Patchiness in Forest Fragmentation Patterns: The Upper Negro River Case Fábio de Castro 1 Eduardo Brondizio 2 We focus on the role of intra-regional ecological patchiness affects the pattern of forest fragmentation caused by land use activities. We analyze the fragmentation patterns in two indigenous communities located 10 miles apart on the Upper Rio Negro, Colombian Amazon. Land use in the region is driven mainly by subsistence agriculture based on traditional techniques. Market influences are limited by the limited accessibility to urban centers. Land tenure has not been an issue mainly due to the high degree of isolation and recent governmental policy for indigenous rights to land. Despite similar economic and social features, the two communities present a different landscape configuration - dendritic and concentric. Both communities are located in a region of low fertility soils, the two dominant vegetation types - moist forest and savanna - are related to soil types with different level of fertility - ultisols and spodosols - respectively. We describe the landscape structure and configuration based on ecological differences in landscape patchiness that affect both land access (distance from settlement and pathways) and land value (soil productivity) in both communities. The analysis is based on satellite image classification of land cover (TM Landsat, 1991) followed by fragmentation analysis (Fragstat). The results are based on patch, class and landscape metrics, and the discussion includes the importance of ecological patchiness in defining patterns of forest fragmentation, and the role of scaling of variables defining land access and land value. 1 Indiana University, ACT, Email: [email protected]; 2 Indiana University, Email: [email protected] Land Use and Land Cover Change 189 Assessment of Atmospheric Correction Models for Landsat TM Data Applied to Amazon Basin Research Dengsheng Lu 1 Emilio Moran 2 Paul Mausel 3 Eduardo Brondizio 4 More than 30 Landsat TM scenes in seven Amazon study areas are being used in the research to develop land use/land cover (LULC) inventories, LULC change with a focus on succession, and selected biophysical modeling such as biomass. These research foci absolutely require that the remotely sensed data used has been atmospherically corrected or calibrated. Four dates of TM data from a study area in the region of Altamira, Para, were used to identify the most appropriate method to calibrate remotely sensed data in this research. Physically-based models, such as 6S, are usually considered the most accurate, but require detailed atmospheric data concurrent with remotely sensed data collection. Most of the TM images used in this research do not provide the data needed for a physically-based model. The Improved DOS Model proved to correct/adjust the data from one date to another almost perfectly. After calibration, using this DOS method, each one of the four TM scenes had reflectance values that were comparable for each feature of interest. The conclusions from this research are: 1) the Improved DOS Model will be very satisfactory to calibrate all the remotely sensed data currently used and 2) the quality of the DOS-based correction is almost as good as a good physically-based model, but requires much less effort and cost. 1 Indiana State University, Email: [email protected]; 2 Indiana University, Email: [email protected], Tel.: 812 855 6181, Fax: 812 855 3000; 3 Indiana State University, Email: [email protected]; 4 Indiana University, ACT, Email: [email protected] Land Use and Land Cover Change 190 Landscape Fragmentation and Flammability in the Interface Between Land Ernesto Alvarado 1 David V. Sandberg 2 Joao Carvalho 3 Renata Alves 4 Increased deforestation and change in land use patterns throughout the Amazons Arc of Deforestation is leaving a highly fragmented forest landscape. Additionally, fire is the most common management tool used in the region for land clearing, and maintenance of pastures and other agricultural crops. During the fire season, many fires escape from the target area into the forest, where fires will burn and may smolder, and increasing risk of burning non-target pastures or agricultural crops. Additionally, forest health on that interface is deteriorated and the value of forest remnants is threatened. This study monitors the interface between newly created land clearings of primary forest until the slash is burned at the end of the dry season. We are monitoring fuel moisture, microclimate, fuel loading, and canopy disturbance along transects from the land clearing into the forest throughout the entire dry season. The study includes simulation of escaped fires from the land clearing burn into the forest, measurement of fire spread, and assessment of tree mortality one year after the fire. The study has been implemented for three seasons (97, 98, 99) in Alta Floresta, Mato Grosso. It is a collaboration between the Fire and Environmental Research Applications team of the USDA Forest Service, INPE, the Universidade Estadual do Estado do Sao Paulo, and IBAMA It is conducted under the collaborative program "Fire and Environmental Change in Tropical Ecosystems" between the USDA Forest Service and IBAMA. 1 University of Washington, 4043, Roosevelt Way NE, Seattle, Washington, 98105, USA, Email: [email protected], Tel.: 206 732 7842, Fax: 206 732 7801; 2 USDA Forest Service Research, Fire and Environmental Research Applications Team, Email: [email protected]; 3 INPE, Email: [email protected]; 4 Fundação de Desenvolvimiento Científico e Tecnológico, NESP, Email: [email protected] Land Use and Land Cover Change 191 Landscape fragmentation analysis in two settlement designs in Rondonia, Brazilian Amazon Mateus Batistella 1 Eduardo S. Brondizio 2 Emilio F. Moran 3 Deforestation processes in the Brazilian Amazon have attracted a great amount of attention. Colonization incentive policies induced the implementation of rural settlements, without considering the constraints of each region and consequent environmental impacts. This paper analyzes the influence of different architectures of colonization upon landscape fragmentation. Two settlements were selected in Rondônia: Vale do Anari, where the orthogonal road network is often referred to as a 'fishbone' pattern, and Machadinho d’Oeste, a settlement designed to conform to local topography. Landscape structure indices were calculated, using a LANDSAT TM image from July 1998. The results show distinct fragmentation patterns in Machadinho and Anari. The method, based on landscape ecology and remote sensing techniques, may be useful to evaluate potential environmental impacts resulted from the establishment of colonization projects in the Amazon. 1 Indiana University, ACT, Student Building, 331, Bloomington IN 47405, Email: [email protected], Tel.: 812 855 6181, Fax: 812 855 3000; 2 Indiana University, ACT, Email: [email protected]; 3 Indiana University, ACT, Email: [email protected] Land Use and Land Cover Change 192 Detecção da Degradação Florestal pela Exploração Madeireira e Queimadas na Amazônia Oriental Utilizando Imagens de Satélite Carlos Moreira de Souza Jr. 1 Jeffrey J. Gerwing 2 Luciano S. Moreira 3 As florestas da Amazônia Oriental têm sofrido mudanças rápidas impostas pela exploração madeireira e por queimadas. Imagens de satélite Landsat TM e SPOT podem detectar “cicatrizes” e assinatura espectral de exploração madeireira e queimadas, respectivamente. Após 2-3 anos, as “cicatrizes” e assinaturas espectrais desaparecerão com a regeneração, dificultando a detecção nas imagens. Neste trabalho, propomos uma metodologia para corrigir este problema, e mapear a evolução da degradação florestal. Utilizamos imagens georretificadas (Landsat TM de 1984, 1988, 1991 e 1996 e SPOT de 1999) cobrindo uma área de 3.600 km2 do Município de Paragominas. Inicialmente, as imagens foram classificadas individualmente em: i) floresta; ii) floresta explorada; iii) floresta queimada; iv) não floresta (áreas desmatadas e floresta secundária) e v) água. Para esse propósito, aplicamos classificação não supervisionada, agrupamento de “clusters” para compor as classes de interesse, finalizando com interpretação visual para remoção de ambiguidades espectrais. As florestas exploradas foram mapeadas utilizando-se imagens de abundância de solos, obtidas com modelos de mistura de pixel. Por último, aplicou-se um algoritmo de detecção de mudança para corrigir a classificação inicial, o que permitiu incluir novas classes no mapa final: vi) floresta explorada antiga; vii) floresta queimada antiga; viii) floresta degradada; e ix) floresta superdegradada. O método mostrou uma boa correlação com os dados de campo. Os erros de classificação estão associados com a resolução temporal utilizada, que não deve ser maior que três anos. Por último, o método mostrou-se eficaz para mapear a distribuição de biomassa nos ambientes de florestas degradadas da região. 1 IMAZON, Caixa Postal 1015, Belém, Pará, 66 017 000, Brazil, Email: [email protected], Tel.: 91 235 4214, Fax: 91 235 4214; 2 Interdepartmental Program in Ecology, Pennsylvania State University; 3 IMAZON, Email: [email protected] Land Use and Land Cover Change 193 Ecological aspects of forest degradation by logging and fire in eastern Amazonia Jeffrey J. Gerwing 1 Carlos Souza, Jr. 2 In the eastern Brazilian Amazon, widespread unplanned logging and fire are degrading forest composition and structure in ways that are not captured by the traditional forest nonforest classification schemes used to determine deforestation rates. This forest degradation often consists of a stepwise process commencing with one or more unplanned logging episodes making way for one or repeated fires of varying intensities. The purpose of this study was to determine the relative impacts of logging and fire on forest biomass, structure, and species composition. Field inventories in forests ranging from intact to logged and heavily burned were conducted in 15, 10 x 500-m plots located on 10 properties in the region of Paragominas, Pará. Live aboveground biomass of intact forest was estimated at 309 t/ha. This value declined 20% following moderate intensity logging and 48% following repeated or high intensity logging. Logged and moderately burned forest also showed a 48% reduction in live biomass whereas this parameter was reduced by 83% in logged and heavily burned forest. Increased amounts of coarse woody debris and reduced canopy cover, particularly in forests subjected to repeat logging, have greatly increased fire risk in these degraded forests. Furthermore recovery of pre-degradation biomass and forest structure is likely to be impeded by abundant lianas and high incidence of crown or stem damage among the remaining trees. As timber industries in senescing logging frontiers, such as Paragominas, begin to close, it is likely that repeat logging of previously logged stands will become more common thus exacerbating the problem of forest degradation. 1 IMAZON, Brazil, Interdepartmental Program in Ecology, Pennsylvania State University, University Park, PA 16802, IMAZON, Caixa Postal 1015, Belém, Pará, 66 017 000, Tel.: 235 4214, Fax: 235 4214; 2 IMAZON Land Use and Land Cover Change 194 Análise espacial dos focos de calor: Um confronto com a vegetação Amazônica Flávio Barbosa Justino 1 Paulo Cesar Pires de Menezes 2 Este trabalho visa analisar a distribuição espacial dos focos de calor (FC) e sua localização sobre a vegetação na região Amazônica. A análise foi feita com base em dois mapas de vegetação, o proveniente do Radam-Brasil (1973-1983) e o gerado pelo macro diagnóstico da Amazônia do IBAMA-MMA (1993). Os FC foram determinados com base nas imagens termais AVHRR do satélite NOAA-12, passagem noturna. A comparação entre os dois mapas mostrou uma evolução no desmatamento na região sul do Pará e norte do Mato Grasso, onde a ombrofila aberta deu origem a savanas. A distribuição da vegetação pioneira apresentada pelo Radam-Brasil em Roraima, compreende boa parte do sudoeste do Estado, o que não é mais observado no mapa do IBAMA. A observação dos FC mostra que estas regiões de maior desmatamento encontram-se onde se tem a maior taxa de incidência de queimadas. No leste do Pará, na área de transição entre as savanas e a floresta ombrofila, foram observados em setembro de 1999, cerca de 50 focos de calor, numa área de 27,2 x 27,5 km, valor similar ao encontrado no norte do Mato Grosso. É importante observar que os FC não foram observados apenas nas áreas de savanas, os Estados do Maranhão e Roraima também mostram alta incidência na campinarana e floresta ombrofila. 1 INPE/CPTEC, Rod. Pres Dutra, km 39, Cachoeira Paulista, SP, Email: [email protected], Tel.: 12 560 8530, Fax: 12 561 2835; 2 INPE/CPTEC, Email: [email protected] Land Use and Land Cover Change 195 Insights into tropical forest regeneration using time-series of remotely sensed data Richard Lucas 1 Christina Potts 2 The use of time-series of remotely sensed data for understanding the successional and carbon dynamics of tropical regenerating forests in response to different land uses prior to abandonment is outlined for a site near Manaus, Brazil. Spatial datasets of deforestation extent, clearance and fire history, periods of active land use, vegetation reclearance frequency, and forest age were generated from a time-series comparison of Landsat sensor data (1972-93). Spectral reflectances were extracted from each Landsat Thematic Mapper image within the time-series (1985-95) for regenerating forests of varying age, for which species composition, structure and biomass data were available. Community analysis using TWINSPAN identified several regeneration pathways, each of which was associated with different land use intensities, as determined from the historical land use surfaces. Species composition/diversity differed between pathways. Biomass accumulation rates were similar within all pathways during the first 10 years but diverged thereafter, with the data suggesting that subsequent rates of biomass recovery be least for forests regenerating on more intensively used sites. Distinct near and mid infrared reflectance trajectories was observed within different pathways, which were attributed largely to changes in upper canopy structure associated with species turnover. Two cycles of tropical forest succession (approximately 0-20 years and 20-35 years) were identified. Evidence is presented that these temporal changes in reflectance are typical for many regenerating forests across Amazonia. The study concludes that time-series reflectance and land use/cover data can be used to assist the understanding and modeling of forest ecosystem dynamics on abandoned clearances. 1 School of Geographical Sciences, University of Bristol, University Road, Bristol, BS8 1SS, Email: [email protected], Tel.: 00 11 44 117 954 6857, Fax: 00 11 44 117 928 7878; 2 The University of New South Wales Land Use and Land Cover Change 196 Logging Activity in Acre State, Brazil. How much is occurring? Silvia Helena C. Brilhante 1 Irving Foster Brown 2 Official data for logging activity in Amazonia are often contradictory. This pattern occurs in Acre State, located at the western end of the Brazilian ‘arc of fire’ and soon to be linked to the Pacific with all-weather highways through Bolivia and Peru. In 1996, IBGE estimated 400,000 m3 of timber were logged, while IBAMA licensed only 21,000 m3 for the same period, a value 20 times smaller. The quality and limited availability of data in subsequent years make it difficult to analyze the trends in this activity. In a survey of 17 sawmills, 1 plywood factory, 43 wood processors, and 349 woodworking shops, the Executive Secretary of Forests and Extractivism (SEFE) of Acre State estimated that approximately 210,000 m3 of timber was consumed, producing 69,000 m3 of processed wood in 1999. The municipality of Rio Branco was responsible for about half of the total processing. Eleven timber genera are the most commercialized: Amburana acreana; Apuleia molares; Calycophyllum spruceanum; Cedrela odorata; Ceiba pentandra; Copaifera spp; Dipteryx odorata ferrea; Hymenaea courbaril; Hymenolobium spp; Swietenia macrophylla; and Tabebuia spp. The use of common names and lack of voucher specimens makes identification to the species level problematic. Using the SEFE survey as a base for the later 1990s, official IBGE and IBAMA data would super-estimate logging activity by a factor of two or sub-estimate it by a factor of 10. The data available suggest that about 90% of logging activity in Acre was illegal in the late 1990s. 1 Secretaria Executiva de Florestas e Extrativismo do Estado do Acre e UFAC, Edifício Banacre DEF-SEFE, 4° andar, E mail: [email protected], Tel.: 68 223 5354; 2 WHRC-UFF-UFAC, Email: [email protected] Land Use and Land Cover Change 197 A New Regional-Scale Amazonia Land Use Data Set Developed from Satellite and Ground-Based Data Jeffrey A. Cardille 1 Jonathan A. Foley 2 Marcos Heil Costa 3 Recent global satellite-based land cover classifications disagree on the location and extent of croplands in Amazonia, and have not included identification of pastures, an important human land use in this region. As part of the Large Scale Biosphere-Atmosphere (LBA) Experiment, we are seeking to identify crop and pasture extent for the mid-1990s in the Amazon and Tocantins River drainage basins at the 5 minute (~10 km) spatial resolution. To this end, we have produced a GIS-based county-level map of these major land use activities using mid-1990s agricultural census data for the countries of the basins. Results of this effort are quite consistent both with existing maps of agricultural activity and with observed patterns of fire frequency in the region. However, these county-level agriculture maps do not cover the entire region due to missing data for some countries in this period. Furthermore, we would like a map at finer resolution to better represent the spatial variation in land use within counties throughout the basins. Here we present the results of merging finer-scale satellite-based land cover classifications with these ground-based maps of human land use to estimate, at the 5 minute scale, the extent and intensity of cropland and pasture throughout the basins. Merged maps coincide well with existing limited-extent classifications within the study region, while preserving important aspects of the spatial structure of both the census-based and satellite-based data. These maps of Cultivated Area, Planted Pasture, and Natural Pasture should be suitable for input to ecosystem models, and are calculated and presented for both the Legal Amazon and the entire Amazon and Tocantins Basins. 1 Climate, People and Environment Program (CPEP), University of Wisconsin, Madison, WI 53706 USA, Email: [email protected], Tel.: 1 608 262 4775, Fax: 1 608 263 4190; 2 Climate, People and Environment Program (CPEP), Email: [email protected], 3 Federal University of Viçosa, Email: [email protected] Land Use and Land Cover Change 198 Economic aspects of ground fires in Amazonia Larissa Steiner Chermont 1 Joshua Bishop 2 William Cavendish 3 Ground fires set by land users are an important factor in the destruction of Amazonian forests (Alencar et al. 1999). While much has been written about Amazonian deforestation, understanding of land-user decision-making with respect to fire use and fire prevention remains rudimentary. An economic model of land-user behaviour can be used to predict both the incidence of fires, and the level of investment in fire prevention and control. We hypothesize that land users’ propensity to set fires decreases, while their investment in fire prevention increases, as agriculture becomes more intensive and permanent. In other words, we predict that investments in fire-vulnerable improvements to the land, such as fencing, agroforestry systems, tree plantations and other perennial crop production systems, pasture renewal and buildings, act as a disincentive for landholders to burn, and as an incentive for landholders to invest in fire prevention and control. Other factors, such as distance to markets, land tenure security, and absenteeism, may also be significant determinants. We will test these relationships using data from property-level interviews and satellite images, linking the use of fire and the level of fire prevention effort in rural properties to the history of fire occurrence on that property, and other property characteristics. Empirical data will be used to analyze: 1. the economic determinants of fire use and prevention; and 2. the costs and benefits of fire prevention activities. Finally, we will explore potential policy and development options to encourage fire prevention by land users. 1 IPAM - UFPA - Imperial College, 48, Prince's Gardens, London - SW7 2PE - Reino Unido, Email: [email protected], Tel.: 44 171 594 9291, Fax: 44 171 581 0245; 2 Int. Institute for Environment and Development - IIED, Email: [email protected]; 3 University of London, Email: [email protected] Land Use and Land Cover Change 199 Development of a Coupled Ecological and Hydrologic model of the Amazon basin: II – Vegetation dynamics and disturbances Aurélie Botta 1 Jonathan A. Foley 2 Kirsten Thonicke 3 We are developing a coupled ecological and hydrologic model of the Amazon basin. The main frame of the project is described in a companion talk. In order to apply the coupled model, IBIS-HYDRA, to questions related to human activity and climatic fluctuations, we are making a number of improvements to the models. We are adding managed ecosystems to IBIS (e.g., pastures and croplands). Second, to account for the effects of natural or human disturbance on ecosystems and their response over time, we are adding explicit disturbance and age-dependence algorithms. This study presents the first step of the ecological part of this project, and focuses on impacts of interannual climate variability and natural disturbances such as fire on vegetation dynamics. Introducing climate variability increased the risk of drought in the drier areas of the basin, which changed the vegetation distribution modeled by IBIS in two ways. First, growth of the lower canopy was favored, increasing the area of savanna by 37%. Second, deciduous forest area expanded into evergreen forest area. Using the more realistic variable climate, we performed different model simulations applying a disturbance rate fixed for the whole basin ranging from 0% to 2%. With increasing rates of disturbance, IBIS simulations showed a subsequent decrease in total biomass and soil carbon. When a disturbance rate higher than 0.5% was used, IBIS simulated the expansion of savanna into tropical forest areas. The disturbance rate required to maintain the current potential patterns of savanna is around 0.5%. Finally, to simulate fire, we introduced a climate-dependent disturbance. This last simulation showed similar results to the 0.5% fixed disturbance simulation, but allowed full development of the tropical forest in wet areas and subsequently, higher carbon storage. 1 Climate, People, and Environment Program (CPEP), Institute for Environmental Studies, University of Wisconsin-Madison, USA, Email: [email protected], Tel.: 1 608 262 4775, Fax: 1 608 263 4190; 2 CPEP, Email: [email protected]; 3 Postdam Institute for Climate Impact Research (PIK), Email: [email protected] Land Use and Land Cover Change 200 Regional Amazon Basin Analyses with MODIS Vegetation Indices: Early Results and Comparisons Alfredo Huete 1 Kamel Didan 2 Yosio Shimabokuro 3 Laerte Ferreira 4 Edson Sano 5 Vegetation indices have emerged as important tools in the monitoring, mapping, and resource management of the Earth's terrestrial vegetation. They are radiometric measures of the amount, structure, and condition of vegetation and serve as useful indicators of seasonal and inter-annual variations in vegetation. In this study, early images from the Moderate Resolution Imaging Spectroradiometer (MODIS) are used to evaluate the normalized difference and enhanced vegetation indices (NDVI, EVI) for regional vegetation monitoring of the Amazon Basin. We analyzed the MODIS vegetation indices over the primary and secondary forests as well as Cerrado biomes and investigated the use of the MODIS vegetation indices in discriminating different types of land use conversions in both biomes. Comparisons of the MODIS products were made with Landsat, SeaWiFS, and AVHRR images. The histograms of the individual bands and the vegetation indices were analyzed and compared with the different sensor NDVI products. Chlorophyll saturation over the dense vegetated areas are investigated and the aerosol resistance properties of the EVI and its extended sensitivity over dense vegetation are evaluated. We found the MODIS vegetation indices to significantly improve upon the detection and discrimination of land conversions within densely vegetated environments. 1 University of Arizona, 1200 E. South Campus Drive, 38, Tucson, AZ, USA, Email: [email protected], Tel.: 520 621 3228, Fax: 520 621 1791; 2 University of Arizona, Email: [email protected]; 3 INPE; 4 University of Arizona; 5 EMBRAPA/CPAC Land Use and Land Cover Change 201 Multi-scale analysis of inundation and wetland vegetation with microwave and optical remote sensing in the Amazon Basin John Melack 1 Evlyn Novo 2 Leal Mertes 3 Laura Hess 4 Doug Alsdorf 5 Claudio Barbosa 6 Collaborative activities conducted at UCSB and INPE focused on analysis of basin-scale optical and microwave data to characterize inundation and wetland vegetation, development of new interferometric and polarimetric synthetic aperture radar (SAR) techniques for examination of floodplains, and coordination and execution of a basin-wide airborne campaign to acquire videographic and laser altimetry data over wetlands and other LBA sites. Using the methods and algorithms that we have developed for determining flooded area with the 37 GHZ polarization difference observed by the Scanning Multichannel Microwave Radiometer, we have completed a monthly time series for seven years of inundation area for the central Amazon and Llanos de Mojos. The high and low water acquisitions of JERS SAR coverage of almost the whole Amazon Basin offers us a rich dataset; we have applied INPE's segmentation algorithms to conduct of polygon-based classification of wetlands using the JERS data. Further, we have developed a method for classifying and mapping floodplain habitats by linking multi-date SAR imagery with river stage records, incorporating information on both vegetation structure and inundation periodicity, and applied the method to sites in the central Amazon. In addition, we have demonstrated that interferometric SAR is capable of measuring subtle changes in water level in regions with inundated vegetation. Using SIR-C data from the central Amazon, we were able to determine 2 to 9 cm/day changes in water level in hundreds of floodplain lakes. Recent analyses of the mainstem Amazon floodplain based on Landsat Thematic Mapper images indicate considerable landscape heterogeneity. Upstream reaches are dominated by forests, which transition through the middle reaches to lower reaches dominated by macrophytes. By combining field measurements of methane emissions with our analyses of inundation and vegetation, we calculate a regional estimate of methane emission from the central Amazon of about 1.8 Tg C per year. 1 University of California, ICESS, Email: [email protected], Tel.: 805 893 3879, Fax: 805 893 4724; INPE, Email: [email protected]; 3 UCSB, Email: [email protected]; 4 UCSB, Email: [email protected]; 5 UCSB, Email: [email protected]; 6 INPE and UCSB, Email: [email protected] 2 Land Use and Land Cover Change 202 The Effects of Amazonian Deforestation on Regional and Global Climate David Werth 1 Roni Avissar 2 Using the GISS Model II Global Climate Model, we simulate an extreme Amazonian deforestation to see what effects such a change will have on the climates of Amazonia and North America. We run the model twice: a control run with the vegetation set to its 1950 level (a mixture of several vegetation types), and a deforestation run in which the rainforest in Amazonia is converted to a mixture of shrubs and grassland. Each is run out to 12 years, and two complete realizations are done. The effect of deforestation on the Amazonian precipitation is substantial, with the largest reductions in precipitation and evaporation occurring in summer (the ITCZ-dominated rainy season. The precipitation-evaporation (P-E) value normally undergoes an annual cycle, whereby the Amazon experiences a net moisture convergence in summer and net divergence in winter. The effect of deforestation is to reduce the amplitude of the cycle, with a corresponding change in the soil moisture cycle (wetter winter/drier summer). The reason for the change in summer is related to the reduction of convection and its feedback to the large-scale flow. The effects of Amazonian deforestation in North America were strongest in the North Central region of the United States, an area that includes the Dakotas, Nebraska, Minnesota, Iowa, and parts of Montana and Wyoming. We see the largest effects in the warmer, rainier months, with the largest reductions of precipitation and evaporation in June. The North American changes are related to changes in Atlantic moisture transport and in the large-scale flow aloft. 1 Center for Environmental Prediction-Rutgers University, 14 College Farm Road, New Brunswick, New Jersey 08901 8551, Email: [email protected], Tel.: 732 932 9083, Fax: 732 932 8644; 2 Center for Environmental Prediction-Rutgers University, Email: [email protected] Land Use and Land Cover Change 203 A Model For Mapping Amazon Forests that are Susceptible to Fire RisQue Paul A.Lefebvre 1 Daniel C. Nepstad 2 Peter Schlesinger 3 Urbano L. Silva Jr. 4 Carlos Nobre 5 Paulo Moutinho 6 During severe droughts, Amazon forests shed a portion of their leaves and become vulnerable to fire. We have developed a model, "RisQue", that identifies those forest areas that may undergo drought-induced fire susceptibility. We present results for the 1997/98 El Niño event. The model tracks soil water beginning on May 1, 1997 through the end of 1998 rainy season. Amazon forests can tap the water stored in deep soil layers to maintain evapotranspiration during periods of low rainfall. We assume that forests become flammable only when soil moisture is depleted to ten meters depth, based on field studies of soil moisture, leaf shedding, fine fuel moisture, and the propagation of experimental fires. The maximum amount of plant-available water that can be stored in the soils was calculated for the forested areas of Brazilian Amazonia using soil texture from 1500 soil profiles. Rainfall data through December 30, 1998, were obtained from 30 to 60 automated weather stations scattered across Brazilian Amazonia (INPE, unpublished data), which also provided air temperature data that we employed to estimate evapotranspiration (ET) using a corrected Thornthwaite equation. Because of the severe drought of 1997 and 1998, we calculate that approximately 1,500,000 km2 of Amazonian forest had completely depleted plant-available water stored in the upper five meters of soil by the end of the 1998 dry season. 1 Woods Hole Research Center, 13 Church Street Woods Hole - MA USA, Email: [email protected], Tel.: 1 508 540-9900, Fax: 1 508 540 9700; 2 Woods Hole Research Center, Email: [email protected], 3 Woods Hole Research Center, [email protected]; 4 Instituto de Pesquisa Ambiental da Amazônia, Email: [email protected]; 5 INPE/CPTEC, Email: [email protected]; 6 Instituto de Pesquisa Ambiental da Amazônia, Email: [email protected] Land Use and Land Cover Change 204 Processes and Conditions Governing Amazon Seasonally Dry Forests Flammability Gustavo Hees de Negreiros 1 David Sandberg 2 Ernesto Alvarado 3 Thomas Hinckley 4 João Andrade Carvalho Jr. 5 José Carlos dos Santos 6 Ecological processes underlining current efforts to model flammability of Amazon forests have not adequately been validated in the field and need to be revised. Field experiments conducted from 1996 to 1999 in Santana do Araguaia PA and Alta Floresta MT show several significant things. The variation of leaf area index (LAI) during the dry season does not significantly affect the light intensity inside the forest floor, thus the LAI variation does not appear to affect the heating of materials near the forest floor or the fuel moisture content (FMC) of fine fuels. LAI variation may change the stand resistance to air movement within the forest for both vertical (convergence) and horizontal (wind), resulting in faster decrease in humidity within the stand. With the absence of a duff layer restricting the air circulation between the surface soil and the litter, the surface soil humidity was found to significantly influence the litter supplying between 20 and 30% of its FMC. The FMC of the forest floor is not the only factor defining the flammability of the stand; atmospheric conditions, fuel availability, and other factors may individually or in combination promote or suppress flammable conditions. Finally, some methods currently being used to access FMC of the forest floor in the region, like the 10h standard field sticks, are not effective monitoring devices for Amazon forests. The research presented is part of the first author Ph.D. Dissertation at University of Washington with support from USDA Forest Service FERA Research Team, LCP/INPE, and NASA. 1 University of Washington (FERA), 4043 Roosevelt Way NE Seattle WA 98105, USA, Email: [email protected], Tel.: 206 732 7842, Fax: 206 553 7709; 2 USDA Forest Service FERA, Email: [email protected]; 3 University of Washington, Email: [email protected]; 4 University of Washington, Email: [email protected]; 5 LCP/INPE, Email: [email protected]; 6 LCP/INPE, Email: [email protected] Land Use and Land Cover Change 205 Development of a remote sensing data compatible tropical forest secondary succession model Dalton de Morisson Valeriano 1 Flávio Jorge Ponzoni 2 Secondary forest stands are conspicuous feature of Eastern Amazon landscape as result of land abandonment and of a land use system that requires a long fallow period after a few crop rotations. These stands are important for regional carbon balance as sites of intense carbon uptake. Regional assessment of the dynamic of these stands require the understanding of the process of tropical forest secondary succession and the identification of discriminate features of the seral stages in remote sensing data. The development of a forest succession model aiming at the identification of seral stages that can be identified in remote sensing data is proposed based on floristic surveys of secondary forest stands and Landsat and JERS-1 data. Forest stands will be classified on the basis of their floristic similarity indexes. Merged Landsat and JERS-1 data will be classified by image segmentation and field classification. Non-parametric correlation between image derived classes and vegetation classes will be evaluated. Class aggregation will be applied in order to attain a significant correlation between the two classifications. Test sites for the development of the methodology will be Altamira, Tome-Acu, Marajo Island and Bragantina area. 1 INPE, Av. dos Astronautas, 1758, 12227 010, São José dos Campos, SP, Brazil, Email: [email protected], Tel.: +55 12 345 6436, Fax: +55 12 345 6450; 2 INPE, Email: [email protected] Land Use and Land Cover Change 206 Predicting fire susceptibility as a function of microclimatic conditions and forest structure in Amazonia José Benito Guerrero 1 Daniel Curtis Nepstad 2 Paulo Roberto de Souza Moutinho 3 Ane A. C. Alencar 4 Each year accidental ground fires burn forest area in Amazonia, similar in size to the area that is slashed and burned. Fires have the potential to reduce forest evapotranspiration, affecting local rainfall and provoking additional burning of forests. This study aims to understand relationships among microclimatic and forest structure conditions on fire susceptibility and to develop a prediction model of forest flammability in Amazonia. The field campaigns were carried out in two subsequent years (1998 and 1999) during the dry season, in fragmented primary and transitional primary forests. Canopy cover, microclimatic conditions, fuel and litter biomass, plant available water and flammability parameters were measured. We related those variables with results from experimental fires to determine forest fire susceptibility. Over 300 experimental fires were conducted, generating data about ignition, combustion and fire intensity and propagation conditions. The preliminary results showed significant (P<0.001) and positive relationship between fire propagation and canopy openness, air temperature, litter and fuel depth, significant (P<0.005) and negative relationship for litter moisture and relative humidity. The type, height and moisture content of litter are important parameters related to the ignition of accidental fires. However, fire development depends on the canopy openness and favorable microclimatic conditions, such as high air temperature, low relative humidity, wind speed and days without rain, and on the type of plant and its physiological conditions resulting from soil available water, logging and past fires. For prediction model purposes, further data from other Amazonian regions will be included in future studies and analysis. 1 Instituto de Pesquisa Ambiental da Amazônia, Tv. Enéas Pinheiro, 1424, Belém,PA, 66087 430, Brasil, Email: [email protected], Tel.: 91 276 2403, Fax: 91 276 3576; 4 IPAM, Email: [email protected] Land Use and Land Cover Change 207 Efeitos do fogo florestal na biomassa vegetal e nas populações de grandes mamíferos na Amazônia Oriental Oswaldo de Carvalho Júnior 1 Daniel Nepstad 2 A exploração madeireira e o fogo florestal, além de aumentarem a vulnerabilidade de florestas para a ocorrência de novos incêndios, liberam carbono da floresta para a atmosfera e tem efeito ainda pouco conhecido sobre a fauna da região. Com o objetivo de medir os efeitos da ocorrência de fogo florestal na biomassa vegetal e nos mamíferos, estamos avaliando fragmentos florestais (n=6) com os seguintes históricos: 3 com cobertura vegetal primária (grupo controle); 3 de floresta com exploração madeireira, sendo que aproximadamente 50% da área é queimada. O estudo está sendo desenvolvido na região de Paragominas, nordeste do Pará. Na quantificação da biomassa vegetal estamos medindo a liteira e diversas classes de diâmetro (0-2; 2-5; 5-10, 10-30, e >30cm) da vegetação, localizadas em parcelas distribuídas em 3 transectos de 1000m. Para a quantificação dos mamíferos, adotamos o “Line Transect Method”, onde percorremos 3 km de trilhas em cada área, uma vez por semana. Os resultados preliminares mostram que aparentemente não existe uma grande diferença na diversidade nem na abundância de espécies entre os ambientes. Porém, em locais onde o fogo parece ter sido mais intenso, poucas são as observações de indivíduos/grupos utilizando este local. 1 IPAM - Trav. Enéas Pinheiro, 1424, Marco, Belém-PA, 66087 430, Email: [email protected], Tel.: 91 276 3576, Fax: 91 276 3576; 2 IPAM - WHRC, Email: [email protected] Land Use and Land Cover Change 208 Forest Fragmentation and Probability of Forest Fires in Eastern Amazon Ane A. C. Alencar 1 Luis A. Solórzano 2 In this study, the probability of fire occurrence is assessed for the forest areas located within a 30km radius around Paragominas City, an old agriculture and logging frontier located on northeast of Pará. The fire risk for the forest fragments was retrieved from a logistic regression in which ten years of forest ground fire history, based on satellite images and interviews, was used to indicate the probability of a burn event. Nine variables, including the flammability of the forest fragments, proximity to ignition sources, and the size and shape of forest fragments were used as predictors in this probability model. Stepwise forward regression was used to determine the most important variables for explaining forest fires. The percentage of the fragment degraded and the proximity to sources of ignition such as charcoal pits were the two most significant variables for determining a forest fire risk, followed by proximity to roads, and neighboring land use. The results indicate that in this old landscape, roads and fragment size and shape are not the most important variables to explain the occurrence of a forest ground fire. The flammability status of the vegetation and the sources of ignition play the most important roles in predicting those fires. A single model using only flammability and ignition information showed that 85% of the area burned was predicted to burn again. It is believed that in new frontier area variables such as road and, fragment size and shape will contribute more in predicting the risk of forest fires. 1 2 IPAM - Tv. Enéas Pinheiro, 1424, Email: [email protected], Tel.: 91 2763576, Fax: 91 2763576, WHRC, Email: [email protected] Land Use and Land Cover Change 209 High-Resolution Remote Sensing Imagery (IKONOS) for Eddy-Flux Tower Sites George Hurtt 1 X. Xiao 2 M. Palace 3 M. Cardoso 4 S. Frolking 5 B. Moore 6 C. Nobre 7 M. Keller 8 Scaling issues are of central importance to understanding the climatological, ecological, biogeochemical, and hydrological functioning of the Amazon. In addressing the issues of scaling, two key science questions arise: How important is fine-scale heterogeneity to large-scale questions? How can coarse-resolution data be interpreted in light of known fine-scale heterogeneity? In order to address these and other questions, high-resolution (14m) optical remote-sensing images from the IKONOS satellite are being obtained for 15 operational eddy-flux tower sites in Brazil. These are important actively-studied research sites that provide direct measurements of carbon and water fluxes at key locations across gradients in climate, ecosystem types and land use, and are in many cases are the focus of supplemental ecological field research. IKONOS images for these sites will be archived and made available to all LBA scientists. It is expected that these data will be used in a variety of LBA projects, particularly for addressing scale and validation issues in remote sensing analysis (e.g., Landsat 7 ETM+, ASTER, MODIS, Radarsat, ERS-2). 1 Complex Systems Research Center, Morse Hall, University of New Hampshire, Durham, NH 03824, USA, Email: [email protected], Tel.: 603 862 4185, Fax: 603 862 0188; 2 University of New Hampshire, Email: [email protected]; 3 University of New Hampshire, Email: [email protected]; 4 University of New Hampshire, Email: [email protected]; 5 University of New Hampshire, Email: [email protected]; 6 University of New Hampshire, Email: [email protected]; 7 INPE/CPTEC, Email: [email protected]; 8 University of New Hampshire, International Institute of Tropical Forestry, Email: [email protected] Land Use and Land Cover Change 210 The effects of rainfall exclusion on an Amazon forest: Simulating El Niño Daniel C. Nepstad 1 Paulo Moutinho 2 Moacyr Bernadino Dias Filho 3 Eric Davidson 4 Gina Cardinot 5 Noemi Viana 6 Leonel Sternberg 7 Marcelo Moreira 8 Daniel Markewitz 9 Amazonian forests are exposed to increasingly severe and frequent droughts associated with El Niño episodes. The consequences of these droughts include potentially large effects on forest metabolism, reproductive biology, tree mortality, biogeochemistry, and forest flammability. We are examining the effects of severe drought on an Amazon forest (Floresta Nacional do Tapajós) using a rainfall exclusion experiment. The rainfall exclusion began in January of 2000, following a one-year pre-treatment intercalibration period. We measure canopy photosynthesis, leaf water potential, and phenology from wooden canopy towers and catwalks. Tree growth, tree mortality, sapflow, litterfall, LAI, forest floor decomposition, soil respiration, trace gas emissions, water and nutrient fluxes, and forest floor flammability are measured in the forest understory. Volumetric soil water and fine root dynamics are monitored in 12-m-deep soil shafts. Soil water movement is studied using deuterium pulses. Models of water flux, ecosystem productivity, biogeochemical processes, and forest fire prediction are being parameterized through this experiment. 1 The Woods Hole Research Center, P.O. Box 296, Woods Hole, MA 02543-0296, Email: [email protected], Tel.: 508 540 9900, Fax: 508 540 9700; 2 INPA, Email: [email protected]; 3 EMBRAPA, Email: [email protected]; 4 WHRC, Email: [email protected]; 5 INPA, Email: [email protected]; 6 EMBRAPA, Email: [email protected]; 7 University of Miami, Email: [email protected]; 8 USP/CENA, Email: [email protected]; 9 University of Georgia, Email: [email protected] Land Use and Land Cover Change 211 Influence of Land Use and Land Cover on Biophysical Feedbacks to Atmosphere E. Shevliakova 1 P. Moorcroft 2 G. Hurtt 3 S. Fan 4 J. Casperson 5 M. Gloor 6 S. Pacala 7 A key LBA research topic is to describe the potential consequences of changes in land cover and land use on climate in Amazonia. To address this topic, we have developed a coupled biosphere-atmosphere model. The two major components of the coupled model are 1) the Ecosystem Demography Model (ED) and 2) the Penn State/NCAR Mesoscale atmospheric modeling system (MM5). ED predicts ecosystem structure and biophysical parameters such as height of vegetation, leaf area index, and plant stomatal conductance. ED also predicts fluxes of CO2 and water from subgrid patches that represent the effects of natural and human disturbances. MM5 simulates the state of the atmosphere: temperature, wind, water vapor mixing ratio, radiation fluxes, precipitation, pressure, etc. We use the coupled model to explore a range of perturbations to the land-atmosphere system that represents scenarios of land use and vegetation distributions in the Amazonian region. Our analysis includes an explicit discussion of the key uncertainties and assumptions including: 1) heterogeneity in surface characteristics (e.g., LAI, surface roughness, and albedo), 2) atmospheric turbulent fluxes and 3) distribution of CO2 sources and sinks. Such analysis is an important pre-requisite to the development of a new generation of interactive biosphere-climate models. 1 Carbon Modeling Center, Department of Ecology and Evolutionary Biology, Princeton University Guyot, hall 27, Department of EEb, Princeton University, Princeton, NJ 08544, Email: [email protected], Tel.: 609 258 2594, Fax: 609 258 6818; 2 Princeton University, Email: [email protected]; 3 University of New Hampshire, Email: [email protected]; 4 Princeton University, Email: [email protected]; 5 Princeton University, Email: [email protected]; 6 Max-Planxk Institut fuer Biogeochemie, Email: [email protected]; 7 Princeton University, Email: [email protected] Land Use and Land Cover Change 212 Diagnosis and Potential Socioeconomic and Environmental Impacts of Pasture Death in the Western Brazilian Amazon Judson Ferreira Valentim 1 Eufran Ferreira do Amaral 2 Maria de Jesus B. Cavalcante 3 Murilo Fazolin 4 Segundo S. Urquiaga Caballero 5 Robert M. Boddey 6 Ravi D. Sharma 7 Antônio Willian F. de Melo 8 Cattle ranching accounts for approximately 75% of the 1,420,300 hectares of deforested areas in Acre (9.3%), being responsible for the biggest changes in the natural ecosystems in the State. As a result, ranching has been the focus of a heated debate regarding the environmental and socioeconomic impacts of the conversion of vast areas of forest of high biodiversity to a grass monoculture, mainly of Brachiaria brizantha cv. Marandu. In 1994, farmers observed tha Marandu pastures were dying in Acre. Since 1998, the areas affected by the problem in Acre expanded rapidly, causing total pasture degradation in some farms. As the cost of recovering or reforming degraded pastures is two to three times that of clearing new areas of forest and forming new pastures, this problem is creating strong pressure from the landowners to clear more forest, and an economically and environmentally viable solution to this problem is urgently required. The hypothesis proposed to explain the death of pastures so far are: 1) the establishment of this grass on Ultisols, of high clay content, subject to waterlogging during they rainy season, which causes stress on the plants, owing to the accumulation of toxic elements and favours the proliferation of parasitic micro-organisms that turn pathogenic; 2) pasture degradation due to overgrazing, which causes compaction, erosion and loss of soil fertility; and; 3) the proliferation of spittle-bug species, heterefore not registered as pests in the region. 1 Embrapa Acre, km 14, BR-364, C. Postal 392, CEP 69901 180, Rio Branco, Email: [email protected] , Telefone: 68 224 3933, Fax: 68 224 4035; 2 Embrapa Acre; 3 Embrapa Acre; 4 Embrapa Acre; 5 Embrapa Agrobiologia; 6 Embrapa Agrobiologia; 7 Embrapa Cerrados; 8 Fudação Bioma/LBA Land Use and Land Cover Change 213 Landscape dynamics in areas of slash and burn agriculture in eastern Amazonia based on multi-temporal analysis of Landsat data Thomas Stone 1 Ima Célia Guimarães Vieira 2 Arlete Silva de Almeida 3 We have used a combination of Landsat data from 1984 (MSS), 1996 (TM5)and 1999 (TM7) and field work to examine the land use, land cover, and land cover change dynamics in the municipio or county of Sao Francisco do Para located in the Bragantina region, the oldest agriculture frontier in Amazonia. Covering some 480 sq. km, this municipio is unusual in this part of Amazonia in that there is a range of ages of capoeira from newly formed up to more than 70 years old. Igapo forest exists here as well as the largest remnant terra firme forest in the Bragantina region and a rubber plantation. Floristic inventories have been made in successional forests of 3, 6, 10, 20, 40 and 70 years old and in terra firme and igapo forest. Species composition, diversity and structure (basal area and biomass) were determined for each type of forest. Analysis of the August 6, 1996 TM data using a supervised classification showed that about 40% of the municipio was in some form of successional forest, 1% was terra firme forest, 9% was igapo, 20% was in agriculture, 17 % was in pasture, and almost 3% was bare soil. The largest class was intermediate (10 to 20 yrs old) successional forest. We are now developing comparisons with both the earlier and the more recent Landsat data. 1 Woods Hole Research Center, PO Box 296, Woods Hole, MA, 02543, USA, Email: [email protected], Tel.: 15085409900, Fax: 15085409700; 2 Museu Paraense Emilio Goeldi, Email: [email protected]; 3 Faculdade de Ciências Agrárias do Pará, Belém, Email: [email protected] Land Use and Land Cover Change 214 A remote sensing educational CDROM developed to support LBA educational needs Dar A. Roberts 1 Getulio Batista 2 During the 1999 LBA Remote Sensing Workshop, the need for improved remote sensing education was recognized. Here we propose the development of a remote sensing educational CD ROM tailored to LBA. We recognize that excellent educational products exist, and thus one objective is to design modules that complement existing material, yet provide a unique LBA focus. Our ultimate goal is to develop a full system, including introductory material, remote sensing fundamentals, analysis techniques, laboratory exercises and applications. We propose a design that differs from most systems. Unlike systems that focus on remote sensing fundamentals, followed by applications, we propose an application driven system where examples are derived from Amazonia. Land-cover mapping and change, fire detection/risk, habitat loss, water quality, natural disturbance, and vegetation response to climatic extremes are examples of topics. Applications modules would be complemented by a general introduction to remote sensing. Fundamental remote sensing concepts and analysis techniques would be addressed by linking the applications modules to background material. For example, an instructor might choose the topic landcover mapping. If interested in background information on sensors, mapping techniques or land-cover reflectance properties, these details could be accessed through links in the module. We have begun assembling a team of LBA researchers who could contribute to the module. To facilitate its development, we propose treating the CD ROM as an edited volume, where team members contribute to a topic and rely on existing material. We envision the CD ROM could be used in intensive short courses, a lecture setting or individually on a computer. 1 U.C. Santa Barbara, Dept. of Geography, EH3611, University of California, Santa Barbara, CA 93106, Email: [email protected], Tel.: 805 893 2276, Fax: 805 893 3146; 2 INPE, Email: [email protected] Land Use and Land Cover Change 215 Large Scale Remote Observations of Disturbance in the Amazon S.C. Hagen 1 B.H. Braswell 2 W.A. Salas 3 S. Frolking 4 Studies of the impact of human activities in the tropics on terrestrial ecosystems and the carbon cycle are limited by a lack of spatially and temporally comprehensive data on land use change. We report initial results using 8 km AVHRR data to evaluate historical disturbance patterns (1981- 1994) for the entire Amazon region. AVHRR has been used extensively for monitoring vegetation globally, including many studies of tropical deforestation. Early work focused on deforestation detection using the 1km resolution local area coverage (LAC) data. While these LAC data have limited availability temporally, a recent reprocessing of the AVHRR global area coverage data (NOAA/NASA Pathfinder GAC) has excited the possibility of using AVHRR as a long ecological time series. These optical measurements almost certainly contain artifacts that mask subtle year-to- year changes in reflectance associated with the removal of some small fraction of forest from a in 8km grid cell. Despite this, we have developed indices based on NDVI and channel 3 (3.8 micron) brightness temperature which appear to capture many details of the spatial and temporal patterns of disturbance. In this preliminary analysis, we compare the results obtained here with time series of disturbance patterns from high resolution TM data on a site basis and with tabular data on a state- by-state basis for the Amazon Basin. Though tropical rainforest monitoring is in many ways a remote sensing worst-case scenario because of the region's high cloud frequency and the dominance of other atmospheric constituents in the signal, this study suggests that AVHRR data contain information about disturbance patterns and rates. A calibrated product based on this information would have the advantage of being Basin-wide and nearly temporally continuous. 1 Complex Systems Research Center, EOS, University of New Hamphire, CSRC, Morse Hall, University of New Hampshire, Durham, NH 03824, Email: [email protected], Tel.: 603 862 0272; 2 3 4 Land Use and Land Cover Change 216 Land-cover dynamics in Rondonia determined from multi-temporal Spectral Mixture Analysis of Landsat data Dar A. Roberts 1 Getúlio Batista 2 Izaya Numata 3 Oliver Chadwick 4 Karen Holmes 5 Thelma Krug 6 We report on remote sensing analysis of multi-temporal Landsat data acquired over the state of Rondonia. A key objective is to establish rates of land-cover change and determine the age and duration of pasture and second-growth forest. This will enable us to select field sites that are undergoing differing rates of change, analyze their soil-landscape relationships, select pasture chronosequences, and determine the differing responses to land-use change. Our research focuses on a mosaic consisting of between 6 and 9 scenes ranging from PortoVelho in the west to Vilhena in the east. We sample the time period between 1986 and 1999, with a two year time interval from 1986 to 1996 followed by a one year time interval to 1999. We apply techniques developed by Roberts et al., 1998, that employ spectral mixture analysis and reference endmembers to derive standardized fractions for Green Vegetation, Non-photosynthetic Vegetation and Shade for each scene. Standardized fractions and training sites are used to develop a binary decision tree classifier that can be applied to any data set within the mosaic. Land-cover classes mapped included pasture, second-growth/crop, upland forest, water, cloud obscured and urban/soil. Maps showing changes in spectral fractions and the dynamics of land-cover classes are produced. Additional products include georeferenced, inter-calibrated Landsat data that can be used by collaborators within LBA and INPE. In this paper we report on scenes from Ariquemes (P232R67), Jiparana (P231R67) and Luiza (P231R68). 1 U.C. Santa Barbara, Dept. of Geography, EH3611, University of California, Santa Barbara, CA 93106, Email: [email protected], Tel.: 805 893 2276, Fax: 805 893 3146; 2 INPE, Email: [email protected]; 3 U.C.S.B. Dept. of Geography, Email: [email protected]; 4 U.C.S.B. Dept. of Geography, Email: [email protected]; 5 U.C.S.B. Dept. of Geography, Email: [email protected]; 6 INPE, Email: [email protected] Land Use and Land Cover Change 217 Fire Risk Assessment in Low Impact Harvesting at the Floresta Nacional de Tapajos, Para Ernesto Alvarado 1 David V. Sandberg 2 Niro Higuchi 3 Ligia Toledo 4 Johan Zweede 5 Ordinarily, undisturbed tropical rain forest is considered a fireproof ecosystem. Severe fire can only occur during sustained drought when it is dry enough to sustain smoldering combustion, or when the canopy is heavily disturbed by intensive selective logging. Application of low-impact harvesting techniques have been suggested to reduce fire hazard to levels similar to those of undisturbed forest within a short time after harvesting. This presentation includes results of two years of monitoring fuel moisture, microclimate, regional weather, fuel loading, canopy disturbance, and fire behavior at the FLONA Tapajos in the state of Para. Monitoring sites were located in undisturbed forest and plots where low impact harvesting techniques were implemented in 1997 and 1999. The objective is to identify the conditions that sustain fire spread and evaluate the reduction of fire hazard in forest stands under the LIH regime for several years after the harvesting. Results from this study will improve the ability to assess changes in flammability that result from partial harvesting or canopy removal in other parts of Amazonia. The study is conducted in collaboration between the Fire and Environmental Research Applications team of the USDA Forest Service, INPA, INPE, and IBAMA It is conducted under the collaborative program "Fire and Environmental Change in Tropical Ecosystems" between the USDA Forest Service and IBAMA. 1 University of Washington, 4043 Roosevelt Way NE, Seattle, Washington 98105-6497, Email: [email protected], Tel.: +1 206 732 7842, Fax: +1 206 732 7801; 2 USDA/FSR, Email: [email protected]; 3 INPA, Email: [email protected]; 4 INPA, Email: [email protected]; 5 TFF, Email: [email protected] Land Use and Land Cover Change 218 Fire Dynamics in Amazonia Manoel F. Cardoso 1 George Hurtt 2 Berrien Moore 3 Carlos Nobre 4 Elaine Prins 5 Fire is an important phenomena in Amazonian ecosystems. It is involved in the depletion of the vegetation, deforestation, changes in nutrients and carbon cycling, and release of reactive gases and aerossols into the atmosphere. Thus fire should be included as an important aspect in models of Amazonian ecosystems. In order to develop and test relevant fire models, data on the patterns and dynamics of fire must be studied and synthesized. To this end, we analyzed the remote-sensing based dataset from the UW-Madison CIMSS Biomass Burning Monitoring Program for 1995 and 1997. These data, based on the geostationary satellite GOES-8, provide fire occurence information at four daily detection times during the fire season, from June to October. Results from this analysis summarize important features of the fire dynamics in Amazonia, providing information on frequency, temporal length, and spatial distribution of fires. This information can be related to information on climate, vegetation, and humans in order to construct and test hypotheses for the observed patterns. 1 UNH, Durham, NH 03824 USA, Email: [email protected] , Tel.: +1 603 862 0819, Fax: +1 603 862 0188; 2 CSRC/UNH, Email: [email protected]; 3 CSRC/UNH, Email: [email protected]; 4 CPTEC/INPE, Email: [email protected]; 5 NOAA/NESDIS/ORA/ASPT - Univ. of Wisconsin, Email: [email protected] Land Use and Land Cover Change 219 Measuring spatial and temporal dynamics of regrowing forest in Amazonia using Landsat William A. Salas 1 Walter Chomentowski 2 David Skole 3 Estimates of land cover change dynamics for three sites (Alto Paraiso, Rondonia; Uruara and Rurupolis, Para) in the Amazon were derived from multi-temporal Landsat TM data. The objectives of this analysis were to document the spatial and inter-annual variability in the rates of forest clearing, secondary growth formation, and re-clearing of secondary regrowth, and to estimate spatial patterns of land use from temporal patterns of land cover change. Annual, or nearly annual, Landsat TM data from 1986 to 1999 were classified into four broad land cover classes (forest, non-forest, secondary regrowth, and water) using unsupervised classification with knowledge based manual editing to facilitate high classification accuracies (e.g. 1998 analysis had overall classification accuracy of 98.2%). Pasture, perennial crops, rotational agriculture, and abandonment into long term fallow are the dominant land uses in these regions of Amazonia. The spatial-spectral-temporal sequences of the landscape were used to map land use based on an a priori definition of common land use practices in the region. Results indicate secondary growth was a large, rapidly changing pool (varied annually between 24 to 45% of the total deforested area). Variability in the observed annual rates of deforestation and secondary growth formation was high. The transition probabilities of both regrowth formation and clearing of regrowth decreased with age of the clearing and regrowth stands, respectively. Persistence of the secondary growth pool was also highly variable. The observed variability is likely due to a combination of the effects of land use practices and the seasonality of the Landsat data acquisition. 1 2 University of New Hampshire, 26 Mill Rd, Durham NH, Tel: (603) 868-3948, E-mail: [email protected] ; MSU, E-mail:[email protected]; 3 MSU, E-mail: [email protected]. Land Use and Land Cover Change 220 Fire and Environmental Change in Tropical Ecosystems: A synthesis of a Collaborative research program between the USDA Forest Service and IBAMA in the Amazon Forest David V. Sandberg 1 Ernesto Alvarado 2 Roger D. Ottmar 3 Heloisa Miranda 4 Joao Carvalho 5 Jan Engert 6 This presentation is a synthesis of the ongoing collaborative research program between the USDA Forest Service and IBAMA. The program "Fire and Environmental Change in Tropical Ecosystems" started in 1993 and was reauthorized in 1999 for five more years. It is supported by USAID, the USDA Forest Service, NSF. The objectives of the program include to conduct an integrated fire assessment, to develop a baseline for monitoring ecosystem productivity and health, support for monitoring forest resources, fire hazard and risk assessment, and flammability assessment on different ecosystems." The program is implemented by the Forest Service Fire and Environmental Research Applications team. The goal is to have replicated research results from Alaska to the Amazon providing a globally-consistent decision support system for fire management, air quality management, and global change response. Results to date in Brazil include development of a photo series inventory technique for cerrado ecosystems, assessment of comunity exposure to smoke in Amazonia, and validation of a biomass consumption mode. Brazilian collaborators include IBAMA, INPE, Universidade do Estado do Sao Paulo, INPA, Universidade de Brasilia, and Tropical Forest Foundation. Core studies include combustion and Carbon emissions from tropical biomass, predicting fire effects, flammability assessment in primary forest and adjacent to land clearings, risk assessment in low impact harvesting, and community smoke exposure. Study sites in the Amazon includes Alta Floresta in Mato Grosso, the FLONA Tapajos and Paragominas in the state of Para, and Theobroma in Rondonia. 1 USDA Forest Service, Fire and Environmental, Research Applications Team, 3200 SW Jefferson Way, Corvallis, Oregon 97331 USA, Tel: (541)750-7265, Fax: (541)758-7760, E-mail: [email protected]; 2 University of Washington, E-mail: [email protected]; 3 USDA Forest Service Research, Fire and Environmental Research Applications Team, E-mail: [email protected] ; 4 University of Brasilia, E-mail: [email protected] ; 5 INPE, E-mail: [email protected] ; 6 USDA Forest Service International Programs, E-mail: [email protected] . Land Use and Land Cover Change 221 Incêndios Amazônicos: uma abordagem em diferentes escalas de tempo Renato Campello Cordeiro 1 Bruno Turcq 2 Abdelfethah Sifeddine 3 Jeferson Aleixas Capitâneo 4 Arlei Oliveira da Silva 5 A reconstituição paleoambiental, baseada em sondagens de sedimentos lacustres, deposição de partículas carbonáceas e estudo da distribuição de carvões em solo procurou identificar a ocorrência, freqüência e dimensão de paleoincêndios recriando o cenário ambiental e as condições paleoclimáticas em que ocorreram eventos de queima da vegetação. Registro ambientais em diversas escalas temporais foram obtidos a fim de comparar processos climáticos que influenciaram a ocorrência de incêndios em ecossistemas amazônicos. Foram avaliados registros de ocorrência de incêndios nas seguintes áreas e escalas temporais: 1) na região de Manaus (AM), foi feito estudo da distribuição de carvões em solos associados a datações 14 C onde foram identificadas ocorrência de incêndios no Holoceno superior da região; 2) registros de deposição de partículas carbonáceas em um ciclo anual de 1998 foi obtido em Porto Velho (RO). Foram calculados os aporte relacionados aos ciclos de queimada durante as diversas fases climáticas sazonais; 3) em Alta Floresta (MT) o registro da deposição recente de partículas carbonáceas obtido, mostram indicadores dos fluxos de partículas carbonáceas e carbono grafítico associado a mudanças do uso da terra.; 4) registros geoquímicos associados a registros de deposição de partículas carbonáceas durante o Holoceno na região de Carajás (PA) mostraram mudanças ambientais importantes associados a ocorrências de incêndios entre 7000 e 4000 anos e no Holoceno superior. Os fluxos identificados em eventos de queimada nas fases secas do Holoceno foram similares aos dos fluxos antropogênicos na região de Alta Floresta; 5) registros dos últimos 45000 anos foram obtidos na úmida região de São Gabriel da Cachoeira (AM), mostrando importantes modificações ambientais durante o último glacial. Os fluxos de deposição de partículas carbonáceas detectados nesta região foram os de menores dimensões. 1 UFF, Niteroí –RJ; 2 E-mail: [email protected]. Land Use and Land Cover Change 222 A Systematic Model of Land Use/Land Cover Change in the Brazilian Amazon Yushuang Zhou 1 David Skole 2 The primary objective of the research is to understand the process of land use/land cover change (LUCC) and to project the future LUCC for the Brazilian Amazon. In most of the existing LUCC models, the human driving forces are depicted as simple explanatory variables treated as exogenous factors from outside of the land change process; and rarely is there any internal interactions or feedback explicitly included in the models. This paper provides an alternative systematic modeling method regarding the Brazilian Amazon with its population structure, economic structure and land use structure as a coherent system. It is the interactions between the socio-economic system and the population system as well as the land use system that lead to a dynamic LUCC process. The socio-economic system is comprised of two interacting parts: sources and causes. Sources are the major types of socio-economic activities that are directly responsible for the physical removal of trees, including agricultural land use, logging and firewood extraction, mining and hydrological construction, and fire. Causes are the factors that determine the volume of the sources, for example, road construction, market price etc. Demand expansion and inventory adjustment are the two main economic processes that interact to generate the equilibrium between agricultural supply and demand, so as to decide how much forestry land will be cleared as an input to agricultural production, and how much agricultural land will be abandoned and grow into secondary growth. The modeling results suggest that, if no major steps are taken, deforestation in the Brazilian Amazon will keep increasing at a high rate accompanied by agricultural land use expansion for the next half century. 1 Michigan State University, 218 Manly Miles, USA, Email: [email protected], Tel.: 517-432-3905; Fax: 517-432-3908; 2 Michigan State University, Email: [email protected] Land Use and Land Cover Change 223 Assessing the Use of the MODIS Vegetation Indices for Vegetation Mapping in the Brazilian Cerrado Laerte G. Ferreira1 Hiroki Yoshioka2 Alfredo Huete3 Edson Sano4 The Brazilian Cerrado, the second largest biome in South America, is marked by a distinct seasonal contrast and comprises an intricate mosaic of land cover types, vertically structured as grassland, shrubland, and woodland. This is a very complex biome to characterize due to the high degree of spatial heterogeneity and extensive land use alterations. We conducted a preliminary evaluation on the ability of the Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation indices (VI’s) to differentiate among the major Cerrado physiognomies. A field campaign was conducted from April 16th through May 5th, 2000, at the Brasilia National Park (BNP), a 30,000 ha area of preserved Cerrado near Brasilia, recently incorporated into the LBA science plan. Data collected at the BNP included distinct ground biophysical measurements (e.g., LAI and fAPAR) and radiometric measurements, acquired at both the ground and airborne levels. This was used to evaluate the response of the MODIS VI’s to the cerrado biophysical and structural parameters. A preliminary analysis of the airborne spectroradiometric data, convoluted to the MODIS and AVHRR bandpasses and converted to the normalized difference vegetation index (NDVI) and the enhanced vegetation index (EVI), indicates the ability of the MODIS EVI to discriminate the four major cerrado types encountered at the BNP (i.e., cerrado grassland, shrub cerrado, wooded cerrado, and cerrado woodland). On the other hand, both the MODIS NDVI as well as the AVHRR NDVI fail to differentiate the shrub cerrado from the cerrado grassland. Although not fully conclusive, our results demonstrate a higher performance of the MODIS sensor and the EVI algorithm for mapping and monitoring the vegetative cover in the Cerrado Region, Brazil. 1 University of Arizona, Department of Soil, Water, and Environmental Science, 1200 E. South Campus Drive, Room 429, Tucson, AZ 85721, USA, Tel.: (520) 621-1791, fax: (520) 621-1791, E-mail: [email protected]; 2 University of Arizona; 3 University of Arizona; 4 EMBRAPA/CPAC, E-mail: [email protected].