Research
Research lines, funded projects, datasets, and open-source tools in remote sensing and agricultural monitoring.
My work develops satellite-based methods to map and monitor agriculture — from smallholder fields in Mexico and South Asia to large-scale commodity crops in Brazil. It is organized around two main research lines, supported by open code and data.
🌱 Smallholder Agricultural Monitoring
University of Michigan · NASA LCLUC program
In Dr. Meha Jain’s lab at SEAS, I develop remote sensing products that estimate crop sowing times, areas, and dynamics in smallholder farming systems, where small field sizes and data scarcity make conventional approaches fail. The core study region is Mexico’s maize systems, funded by the NASA LCLUC program, where we use multi-decadal Sentinel and Landsat time series in Google Earth Engine to track how farmers adapt sowing decisions to a changing climate.
Highlights
- Sowing-date (Start of Season) detection from optical imagery time series, validated with ground-truth fieldwork — published in Remote Sensing Applications: Society and Environment (2025).
- Mapping maize transitions across Mexico from 2000 to the present to understand policy, market, and climate drivers.
- Ongoing work: automating agricultural ground-truth generation by fusing Google Street View, satellite time series, and generative AI (Google Cloud Research Credits, 2026–2027).
🛰️ Brazilian Agriculture & Satellite Data Cubes
INPE · UNESP · CNPq · FAPESP · Fundação Araucária collaborations
With collaborators in Brazil, I work on large-scale crop monitoring that combines optical and SAR time series, machine learning, and Analysis-Ready Data cubes.
Active projects
- Large-Scale Crop Monitoring Integrating Medium Spatial Resolution Satellite Data Cubes (FAPESP, 2026–2029) — within-season classification of soybean, corn, and cotton in the Cerrado, differentiating crops from native vegetation using harmonized Sentinel-2/Landsat data cubes. PI: Dr. Michel E. D. Chaves (UNESP); my role: Associate Researcher.
- SinergIA (CNPq, 2025–2027) — remote sensing and AI for mapping energy crops in the Brazilian Semi-Arid Region, fusing Sentinel-1 SAR and Sentinel-2 optical time series with deep learning. My role: International Technical Lead.
- NAPI AGROGENÔMICA (Fundação Araucária, 2025–2030) — Support Center for Research and Innovation studying agricultural soil microbial communities in Paraná, Brazil, and their interactions with agroenvironmental, chemical, physical, and geographic variables. PI: Dr. E. Mercante; my role: International Collaborator.
- GEEadas — a Google Earth Engine tool for automatic detection of adverse-frost stress in crops, published in RSASE (2025) and featured by Agência FAPESP. Explore the interactive GEE app.
- Brazil-Crop dataset (in development) — an open benchmark of ~9,400 field observations and 2,200 geotagged photos across Paraná and Mato Grosso for agricultural remote sensing applications.
- SafraWatch (proof of concept) — an interactive hub that reconstructs a field’s multi-year crop history from Sentinel-2/Sentinel-1 and automatically detects sowing, SOS/POS/EOS and cropping intensity per agricultural year, extending my Start-of-Season method (RSASE, 2025). Study region: Medianeira, PR. Try the live demo · read more. Source code to be released once the demo is validated.
💻 Code & Tools
Open-source repositories, including small Python utilities with step-by-step tutorials in their READMEs.