The modification of the Earth’s surface constitutes the most impactful way in which humans affect their surrounding environment, with broad and lasting consequences. Changes in land cover accelerate biodiversity loss, contribute to climate change, and affect the provisioning of ecosystem services. Such negative environmental impacts can have important effects on human health and livelihoods. The South American continent, in particular, has undergone significant transformations over the past decade, due in large part to the conversion of natural land to more economically productive land uses, such as crops, pastures, and tree plantations. The agricultural commodities produced in South America are traded and consumed globally, and land will likely continue to be converted if demand for these commodities continues to rise. Despite the environmental and commercial importance of land cover and land use change dynamics in South America, the extent and rates of land change have not yet been thoroughly characterized and quantified. This dissertation aims to advance scientific knowledge on the extent and rates of change of important land covers and land uses, especially as they relate to the production of agricultural commodities, by leveraging the 34-year Landsat archive of Earth observation data. The general approach employed throughout follows a two-step process of mapping and sampling, in order to provide spatially explicit information on the patterns of land cover/land use change, as well as associated unbiased area estimates. This approach is first employed for the use-case of Brazilian cropland expansion from 2000 to 2014, and results show a near doubling of cropland area, the majority of which (80%) came about through the conversion of existing pastures. The methodology is then repeated at broader thematic, temporal, and geographic scales, resulting in area estimates of changes in cropland, pasture, plantation, natural tree regrowth, semi-natural land, tree cover and degraded tree cover from 1985 to 2018. Altogether, these changes amount to a 60% increase in human impact on natural land over the study period. Finally, an analysis and evaluation of the methodology employed for mapping and sampling when there is a multitude of target classes instead of a single one is provided as an assessment of methodological approaches.