geospatial-machine-learning

A curated list of resources focused on Machine Learning in Geospatial Data Science.

Utilizing machine learning to examine deforestation rates in the undeveloped region of Paraguay's Chaco

Google Earth Engine code for forest cover change mapping based on the LandTrendr algorithm

The public GitHub repository of Ruben Ahrens' Master's Thesis, where the importance of SO2 and HCHO data in ship plume detection was studied

Reproducible evaluation audit of GeoFM label proxies, baselines, and frozen versus fine-tuned decoders for field-extent segmentation.

contains our team submission for Hackathon Bank Indonesia. Problem that we want to solve is the unfairness of fish price. In summary, we propose to (1) using geospatial ML when to decide the next location of cold storage (2) build app to help fish auction decide the fair price and monitor the recommended storage location based on stock management

Visit the website below to view the html files in this portfolio

Geospatial ML pipeline for OSM-derived building floor and residential prediction

Machine Learning for GEOspatial applications

Linear-probe land-cover classifier on precomputed Tessera (Sentinel-1+2 foundation model) embeddings — one tile, one day, CPU-only.

Applied ML system predicting urban accessibility across Barcelona using geospatial features, SMOTE and Random Forest. Built for inclusive mobility.

A precipitation prediction model using the ERA5 hourly data to predict hourly rainfall over the continental U.S. This project was made for Statistics 421 at VCU.

Reproducible geospatial machine-learning workflow for urban heat exposure, temperature forecasting, and heat vulnerability assessment in Bologna, Italy.

🏙️ Transform urban data into insights with the Barcelona Accessibility Intelligence System, enhancing mobility and inclusive city planning through machine learning.

NeurIPS 2026 (Climate Change AI workshop). Leakage-controlled, support-matched harness showing that how counties are grouped into regions flips the winner of cross-region crop yield benchmarks: 7 methods, 3 spatial partitions, 8,691 US county-years.