
Spatial Data Science
I have worked in urban problems, including growth models, spatial segregation, social exclusion, and urban regionalization. To extract contextual information from images, I proposed segmentation using spatial autocorrelation and landscape pattern detection. These methods were consolidated in the GeoDMA software.
My current work focus on pattern matching and machine learning for satellite image time series analysis (see dtwSat and sits R packages).

Land Use Change
My research deals with intra-regional and spatio-temporal analysis of changes in Amazonia and impacts of public and private sector policies. I also worked on software for building scenarios of land-use change.
I have helped to project future land use in Brazil and to argue for better supply chain agreements.
Our team provided science-based policy guidance for Brazil’s NDCs to the Paris Agreement.

Data Semantics
My early work on object-oriented models for GIS led to an OO map algebra. Recently, I proposed an algebra that links observations to events, a model of geo-fields for big data and a calculus of land-use trajectories.
I worked with ontologies for GIS and then considered the semantic content of satellite images. My ongoing work deals with semantics of machine learning and time series analysis of big Earth observation data.