First seminar of the Geolearning chair

The Geolearning Chair held its first working seminar from March 31 to April 3, 2025, at Villa Clythia in Fréjus. The seminar brought together researchers, doctoral students, and postdoctoral fellows involved in the chair to discuss ongoing scientific projects. Invited academics from Sorbonne University, AgroParisTech, Ca' Foscari University, Venice (Italy), and the University of Lausanne (Switzerland) also actively participated in this conference. The scientific presentations can be downloaded by following the links below.

Spatio-temporal

Mike PEIRERA Higher time regularity of SPDE-based Gaussian processes using fractional Brownian motion

Denis ALLARD Modeling and simulating spatio-temporal multivariate and non-stationary Gaussian Random Fields: a Gaussian mixtures perspective

Xavier FREULON Spectral simulation of a spatio-temporal random field on the three-dimensional sphere

Charlie SIRE Statistical modeling of spatio-temporal data distributed over surfaces

Thomas ROMARY Deep kernel learning for geostatistics

Edith GABRIEL Integrating Spatial Modeling and Machine Learning for Plant Health Surveillance

Extremes and climate

Olivier WINTENGERGER Self-normalization of sums of dependent random variables

Thomas OPITZ Input-to-output propagation of tail behavior, with application to a dynamic space-time model for air pollution

Nicolas ECKERT A few statistical challenges in glaciology

Rita MAATOUK Graphical models for extreme events: Joint modeling of precipitation and extreme river flow in the Garonne watershed

Antoine HERENVAL Analyzing the dynamics of extreme events with marked point processes

Carlo GAETAN Multivariate modeling of low, moderate and large positive values ​​without threshold selection steps

Gregory JACQUEMIN Return period of non-concurrent climate compound events: a nonparametric bivariate Generalized Pareto approach

Gloria BURITICA Extreme rainfall temporal modeling and estimation of extreme concomitant events

Stochastic Weather Generators

Antoine DOIZE A stochastic precipitation generator with heavy rainfall and long periods of drought

Lionel BENOIT Improving rainfall gradients modeling by conditioning daily rainfall maps to monthly totals

Denis ALLARD  MSTWeathergen – and the SWG in the Geolearning chair

Generative approaches and PINNs

Sylvain LE CORFF VAE for state space models: theoretical guarantees, practical implementation and online learning

Garbriel VICTORINO CARDOSO Generative proxies of spatial non-homogeneous Gaussian processes: Opportunities and limitations

Ferdinand BHASVAR Deep Generative Models for Spatial and Spatiotemporal Simulation of Natural Phenomena

Gregory MARIETHOZ Generation of synthetic remote sensing images with ultrasimple but ultrafast approaches

Lucia CLAROTTO Parameter and density estimation in SDEs via PINNs and Normalizing Flows: applications to environmental sciences

Gstlearn

Nicolas DESASSIS   Gstelearn Roadmap

Pierre GUILLOU and Fabien ORS   Gstlearn Engineering