Environmental
What is Environmental Modelling?
Environmental Modelling refers to the use of statistical methods and computational techniques to understand natural systems and their interactions with human activities. It aims to predict environmental changes, assess impacts, and guide sustainable management of resources. Models integrate data from various sources, including climate, land use and interactions with natural emergencies (e.g. wildfires, avalanches), and infrastructure impacts (e.g. wildlife mortality on wind farms and power lines) to represent interactions between environmental processes, ecosystem dynamics and human activities. By testing scenarios, environmental modelling provides insights into complex environmental problems, supports decision-making and helps designing policy responses.
What has CREEM contributed?
CREEM’s contributions are mostly related to the development and application of statistical and computational methods for environmental modelling, particularly in areas related to wildlife mortality estimation, spatio-temporal modelling, and environmental risk assessment.
CREEM has advanced the use of spatial-temporal models to understand and predict environmental processes such as wildfire occurrences and avalanche risk, integrating machine learning with Bayesian hierarchical frameworks to improve forecasting of these hazards. Recent work aims to translate spatiotemporal wildfire forecasting research into a practical tool to support proactive fire prediction and strategic planning. We have also been improving methods to estimate wildlife mortality at wind farms and power lines, addressing challenges such as imperfect detection, carcass persistence, and spatial heterogeneity.
What species are these methods used for?
Most of the work on wildlife mortality relates to species of birds and bats and work on wildfires is also relevant to many taxa dependent on the landscape.
Who in CREEM works on these methods?
PhD Students: Gordon Hannah
A few relevant and recent publications by CREEM staff
Bispo, R., Vieira, F. G., Yokochi, C., Marques, F.J., Espadinha-Cruz, P., Penha, A., & Grilo, A. (2025). Using spatial point process models, clustering and space partitioning to reconfigure fire stations layout. International Journal of Data Science and Analytics, 20, 687–697.
Hu, C., Bispo, R., Rue, H., DaCamara, C.C., Swallow, B. and Castro-Camilo, D., (2026) ‘XGBoost meets INLA: a two-stage spatio-temporal forecasting of wildfires in Portugal’ Environmetrics
Biscaia, E., Bernardino, J., & Bispo, R. (2025). An Approach for Predicting Spatially Indexed Carcass Persistence Probability. New Frontiers in Statistics and Data Science: SPE2023, Guimarães, Portugal, October 11-14, 469, 169.
Hu, C., Swallow, B. and Castro-Camilo, D., (2024) ‘A Bayesian multivariate extreme value mixture model’ arXiv:2401.15703
Gomes, M., Lopes, V.M., Mai, M.G., Paula, J.R., Bispo, R., Batista, H., … & Pimentel, M.S. (2023). Impacts of acute hypoxia on the short-snouted seahorse metabolism and behaviour. Science of the Total Environment, 904, 166893.
Villejo, S.J., Illian, J. B., & Swallow, B. (2023). Data fusion in a two-stage spatio-temporal model using the INLA-SPDE approach. Spatial Statistics, 54, 100744.
Yokochi, C., Bispo, R., Ricardo, F., & Calado, R. (2023). Regularization Methods for High-Dimensional Data as a Tool for Seafood Traceability. Journal of Statistical Theory and Practice, 17(3), 44.