The Cerrado, South America’s 2 million km^2 savanna, is recognized for both its overwhelming biodiversity and rapid Land Cover and Land Use Change (LCLUC). It supports 5% of the world’s biodiversity, including ~13,000 plant and vertebrate species, of which over 40% are endemic. The Cerrado lacks the same degree of international attention and legal protection as its carbon-rich neighboring tropical forests, and land clearing in recent decades has led to rapid changes; over 50% has been cleared in recent decades for crops and pasture.

This proposal directly addresses the first component of the NASA LCLUC announcement: identifying hotpot areas using multi-source data approaches to quantify land-cover and land-use changes (LCLUCs) including the expansion, intensification, and abandonment of croplands and rangelands. We will leverage machine learning algorithms to develop novel data sets of LCLUC and Land Use Transitions (LUTs) with our ultimate goal being to integrate socioeconomic and environmental datasets to better understand farmers’ decision making under a changing climate in a critically endangered global breadbasket. Below we detail the key objectives, methods, and significance of the work. 

>>Obj. 1: Develop LCLUC detection methods and data sets: How has Cerrado land use evolved? Our LCLUC detection will include specific crop types and rotations, irrigated agriculture, and pastoral and agricultural abandonment. We will detect LCLUC using a multi-scalar, spatio-temporal remote sensing approach. We will distinguish LCLUCs using green-leaf phenology from MODIS EVI time series with training and vicarious validation from higher resolution imagery (Landsat, Planet) and field sites. The results Obj. 1 will include new Cerrado-wide crop-specific LCLUC data set, filling an information gap with direct societal relevance for understanding both ecological impacts (e.g., nutrient cycling, surface temperature) and socioeconomic impacts (e.g., crop production). 

>>Obj.2: Characterize major land use transitions: Where and when do hot spots and hot moments of major LUTs occur? We will develop best practices for delineating LUT typologies as a new effort for LCLUC science. We expect significant spatial hotspots and hot moments from punctuating events or conditions like changes in deforestation laws (mid-2000s, early 2010s), economic (recession) and weather conditions (drought), and public health (COVID-19). We expect the results of this work to facilitate our ultimate objective, Obj. 3, below. 

>>Obj. 3: Build a transferrable framework for understanding drivers of LUTs: What causes one area to follow a particular LUT pattern? What are the local and global factors that influence LUTs? Here we will explore the causative socioeconomic (e.g., land tenure, frontier history, exchange rates) and environmental (e.g., soil, change in climate) factors influencing LUTs. Using quantitative and qualitative datasets, we will develop an LUT model tested via Bayesian Regression. The results of this work will provide information on enabling or suppressing factors for a LUTs, and thus have direct policy and climate adaptation applications to foster or discourage certain LUTs under a changing economic and global climate.

The project personnel, which include social and natural scientists, have a proven history of producing deeply integrated work on environment and socioeconomics. Galford, Fisher, Spera,  Macedo and Rattis have a combined experience of > 4 decades working on Brazilian land-use development, including generating novel LCLUC datasets; applying statistical, empirical and process-based models; analyzing socioeconomic data and surveys; and collaborating with researchers and students.