• Bayesian Spatial Modelling with geoML
• Modern geospatial data science with GeoStats.jl - From geodesics to multivariate simulation of random fields and point patterns
• Drone-based surveys of geological outcrops: best practices for acquisition, processing and interpretation (Tentative title)

Bayesian Spatial Modelling with geoML

Instructor:

Dr. Italo Gonçalves (Federal University of Pampa - UNIPAMPA)

Description

Conventional geostatistics treats kriging as a black box: fit a variogram, invert a matrix, get a map. This course opens the box. In one day, participants rebuild kriging from first principles as a Gaussian process (GP) — a Bayesian model — and then see how relaxing its assumptions one at a time (scalability, non‑Gaussian data, multiple correlated variables, non‑stationarity, compositional constraints) leads directly to the variational Gaussian process (VGP) network models implemented in geoML, an open‑source Python package for spatial and geoscientific modelling (Gonçalves, github.com/italo-goncalves/geoML). The day combines short lectures with hands‑on notebooks, using the package's own published "Learning materials" series as the practical spine of the course.

Audience:

Geoscientists, geostatisticians and ML practitioners with a working knowledge of linear algebra, basic probability and some Python. No prior Bayesian-statistics or GP background is assumed.

Format:

In‑person, one day (≈7 h contact time), lecture + live‑coded Jupyter/Colab notebooks. Participants should bring a laptop; no local install is required (the notebooks run on Google Colab).

Modern geospatial data science with GeoStats.jl - From geodesics to multivariate simulation of random fields and point patterns

Instructor:

Dr. Julio Hoffimann (Federal University of Rio de Janeiro (UFRJ) / Arpeggeo Tech)

Description

COMING SOON

Audience:

Geoscientists and geostatisticians with professional or academic interests. Basic programming knowledge is recommended; no prior background in Julia is required.

Format:

In‑person, two-day (≈ 14 h contact time), lecture + live‑coded notebooks. Participants should bring a laptop.

Drone-based surveys of geological outcrops: best practices for acquisition, processing and interpretation (Tentative title)

Instructor:

Dr. Leonardo Campos Inocencio (State University of Rio de Janeiro - UERJ)

Description

COMING SOON

Audience:

Geoscientists interested in digital outcrop models

Format:

In‑person, two-day (≈ 14 h contact time), including hands-on data acquisition. Participants should bring a laptop with sufficient RAM and a dedicated graphics card. Minimum system requirements will be provided soon. Basic knowledge of cartography required.