Integrating Survey and Remote Sensing Data to Analyze Land Use at a Fine Scale: Insights from Agricultural Households in the Brazilian Amazon

Abstract

The ability to model factors influencing land use can be significantly improved by incorporating variables derived from geographic information systems with more detailed survey data. While remote sensing data have the advantage of providing land cover measurements for large areas, survey data collected from households provide a more detailed account of land use. We estimate land use decisions in the Brazilian Amazon with land cover data derived from satellite images merged with observations from a household panel. We focus on the rapidly expanding cattle industry or ``pecuarizaçao'' (cattleization) as well as a potentially competing land use strategy: the planting and harvesting of annual and perennial crops. We identify spatial error in our initial estimations and as a result use corrected models of land use with greater explanatory power and efficiency. Estimation results indicate that both pasture creation and agriculture are determined by similar household and spatial characteristics; however, the impacts of these determinants are in opposing directions for both land uses estimated. We conclude from our survey and satellite comparisons that while remote sensing data may overestimate the extent of pasture, the overestimation is minor. Finally, based on the current extent of pasture, we also conclude that policy that focuses on reducing deforestation within these long-established settlements is most likely inefficient. Rather, policy that addresses the land use practices employed by recent settlers may be more effective in reducing further pasture creation.

Long‐term, high‐spatial resolution carbon balance monitoring of the Amazonian frontier: Predisturbance and postdisturbance carbon emissions and uptake

“…Any remaining instances of the cloud class, which had not been reassigned by the multitemporal filters, were reassigned to the class from the preceding year. Our land cover classification scheme does not include a crop class due to the difficulties in separating secondary growth and crops and the limited extent in Rondônia [ Caviglia‐Harris and Harris , ]. An exception is the southeastern portion of the state where mechanized soy production dominates land use [ Brown et al ., ]; however, this land use differs little from pasture in terms of carbon dynamics [ Kauffman et al ., ].…”

Drivers of Households’ Land-Use Decisions: A Critical Review of Micro-Level Studies in Tropical Regions

“…Yet, very few studies make an attempt to address this problem. One exception is Caviglia-Harris and Harris [66] who use lagged variables of income-instead of current income-in their analysis of cattle ranching expansion in the Brazilian Amazon. They find a positive correlation between income and pasture but not for cropland.…”

“…A typical finding of micro-level studies with regard to labor is the correlation between deforestation and agricultural extension and the use of hired labor (for example, [66,[77][78][79]), particularly for commercial agriculture [78]. Unfortunately, these studies do not take into account that hired labor is endogenous to land-use change: Labor use, be it family or hired labor or a combination, is always determined by the production technology and labor market conditions, i.e., wages and the availability of labor for hire-rather than vice versa.…”

Linking Spatial Pattern and Ecological Responses in Human‐Modified Landscapes: The Effects of Deforestation and Forest Fragmentation on Biodiversity

“…Research needs to focus not only on the ecological effects of fragmentation but on understanding the drivers of landscape transformation and predicting future landscape change. In recent years, our understanding of the broad‐scale ecological impacts of deforestation and forest fragmentation and the drivers of ongoing landscape change has been deepened by coupling measures of forest pattern derived from satellite imagery with field‐based measures of species and community characteristics (Lees and Peres 2006) and household survey data (Caviglia‐Harris and Harris 2008). The results of such studies can be used to predict future landscape changes through the use of agent‐based models that combine a cellular landscape model with agent‐based representations of decision making, integrating the two components through specification of interdependencies and feedbacks between agents and their environment (Parker et al.…”