Short course 2

R-INLA in geoscience

Summary

The short course aims at providing a basic understanding of Point Processes in the context of landslide predictive modeling. As a result, the attendees will go through a quick theoretical introduction on traditional landslide susceptibility models and subsequently move to landslide intensity. The latter corresponds to the outcome of a Point Process under a different paradigm compared to what has been traditionally done in landslide science.
The theoretical introduction will be linked to a practical component during which the attendees will have their hands on a small dataset and a code to perform all the analyses. The code is based on the R programming languages and will heavily rely on the package INLA which allows for advanced spatial Bayesian models and much more. However, knowledge of R (or INLA) is not required. In fact, the concepts previously introduced will be put into practice following line by line, with associated explanation, the example R-code.
Overall, the practical will cover some pre-processing concepts related to the representation of space for the study area as well as the optimal use of the covariate set. From there, the attendees will dive into the modeling and performance assessment phases. Ultimately, the interpretation of the results will be executed as a shared task.

Instructors

 

The course will be held by Dr. Luigi Lombardo, assistant Professor in soil sciences at ITC (University of Twente), and Dr. Thomas Optiz, researcher at the Biostatistics and Spatial Processes lab of INRAE (Avignon, France).
Luigi primary research interest gravitates around mass-wasting processes approached via geostatistics; more specifically, he has moved the bulk of his research towards space-time Bayesian application, entirely implemented in R-INLA. His work has shifted the paradigm from the traditional susceptibility concept to a more statistically rigorous intensity framework via Log-Gaussian Cox Processes.
In his work, Thomas develops new methodology for spatio-temporal statistical modeling of climatic, environmental, ecological and epidemiological phenomena, with a focus on extreme and rare events that produce a strong impact in our world of environmental and ecological transitions. He has become an enthusiastic contributor to INLA-based methods over the past years.