The paper Copula-based models for spatially dependent cylindrical data by Francesca Labanca, Prof. Anna Gottard (DiSIA, University of Florence), and Prof. Nadja Klein (Methods for Big Data, SCC) has received the EnviBayes Student Paper Competition Award 2026, awarded by the International Society for Bayesian Analysis (ISBA).
The paper introduces a novel statistical framework for modeling cylindrical data, combining a circular component (such as wind direction) and a linear component (such as wind speed) in the presence of complex spatial dependence. Typically, existing copula-based approaches for cylindrical data treat dependence parameters as constant and rely on restrictive marginal regression structures, often neglecting spatial autocorrelation. These limitations reduce flexibility and hinder the accurate representation of environmental processes.
To address these challenges, the authors propose a structured additive conditional copula regression model that jointly captures circular-linear dependence and spatial variability. The circular component is modeled using a wrapped Gaussian process, while the linear component follows a distributional regression framework. By exploiting the equivalence between Gaussian random fields and Gaussian Markov random fields, the approach reduces computational cost while allowing for flexible, non-stationary covariance structures. The framework is designed to handle real-world data complexity, including non-stationarity and covariate-driven effects, and is supported by a simulation study and an application to wind data over Germany.
The proposed approach enables the association between wind direction and speed to depend on covariates and to vary spatially across Germany (see figures).
We sincerely congratulate Francesca, Anna and Nadja.
Further information: Preprint of the paper on arXiv: https://arxiv.org/abs/2602.05778.
.