Research

Research areas and related publications

Research by Daniela Castro-Camilo on extreme value theory, spatial and spatio-temporal statistics, Bayesian inference, causal inference and environmental risk.

Click a research area to jump to a short description and related publications.

Causality for extremes

Spatial extremes

INLA/SPDE

Landslide hazard

Ecology & environment

Energy risk

Data fusion

Tail modelling

Risk assessment

Causality for extremes

This area develops statistical tools for causal questions where the object of interest is not the mean response, but the tail of a distribution: high quantiles, rare events, severe impacts and extreme-event attribution. Current work includes causal inference in the tails, treatment effects for extremes, structural discovery and climate/weather attribution.

Related publications

  • Li, M., & Castro-Camilo, D. (2026). On the importance of tail assumptions in climate extreme event attribution. arXiv preprint arXiv:2507.14019.
  • Li, M., & Castro-Camilo, D. (2026). Tail-Calibrated Estimation of Extreme Quantile Treatment Effects. arXiv preprint arXiv:2603.23309.
  • Li, M., & Castro-Camilo, D. (2026). Causal Discovery in Multivariate Extremes via Tail Asymmetry. arXiv preprint arXiv:2604.21620.

Spatial and spatio-temporal extremes

I develop models for extremes observed over space and time, with applications to precipitation, temperature, wildfires, environmental pollution and other spatial hazards. The goal is to understand where and when rare events occur, how extremes cluster, and how dependence changes across environmental conditions.

Related publications

  • Castro-Camilo, D. and Huser, R. (2020). Local likelihood estimation of complex tail dependence structures, applied to U.S. precipitation extremes. Journal of the American Statistical Association, 115(531), 1037–1054.
  • Castro-Camilo, D., Mhalla, L. and Opitz, T. (2021). Bayesian space-time gap filling for inference on extreme hot-spots: an application to Red Sea surface temperatures. Extremes, 24(1), 105–128.
  • Castro-Camilo, D., Huser, R. and Rue, H. (2022). Practical strategies for GEV-based regression models for extremes. Environmetrics, 33(6), e2742.
  • Cuba, M. D., Wilkie, C., Scott, M., & Castro-Camilo, D. (2025). Spatio-temporal data fusion of censored threshold exceedances. arXiv preprint arXiv:2504.20268.
  • Hu, C., Bispo, R. B., 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, 37(3), e70089.

INLA and SPDE models

A substantial part of my work uses Bayesian latent Gaussian models, INLA and SPDE-based representations of spatial and spatio-temporal random fields. I am interested both in methodological development and in making these tools usable for environmental and hazard applications.

Related publications and projects

  • Krainski, E. T., Gómez-Rubio, V., Bakka, H., Lenzi, A., Castro-Camilo, D., Simpson, D., Lindgren, F. and Rue, H. (2018). Advanced Spatial Modeling With Stochastic Partial Differential Equations Using R and INLA. Chapman & Hall/CRC.
  • Hu, C., Bispo, R. B., 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, 37(3), e70089.
  • Bryce, E., Castro-Camilo, D., Dashwood, C., Tanyas, H., Ciurean, R., Novellino, A. and Lombardo, L. (2025). An updated landslide susceptibility model and a log-Gaussian Cox process extension for Scotland. Landslides, 22(2), 517–535.
  • GEOBEx: Geostatistical Binary Models for Extremes, EPSRC-funded project.

Landslide hazard modelling

This work develops statistical approaches for landslide susceptibility, hazard, intensity and density. A key aim is to produce models that are statistically principled but still useful for natural-hazard assessment and decision-making.

Related publications and reports

  • Castro-Camilo, D., Lombardo, L., Mai, P. M., Dou, J. and Huser, R. (2017). Handling high predictor dimensionality in slope-unit-based landslide susceptibility models through LASSO-penalized Generalized Linear Model. Environmental Modelling and Software, 97, 145–156.
  • Amato, G., Eisank, C., Castro-Camilo, D. and Lombardo, L. (2019). Accounting for covariate distributions in slope-unit-based landslide susceptibility models. A case study in the alpine environment. Engineering Geology, 260, 105237.
  • Lombardo, L., Tanyas, H., Huser, R., Guzzetti, F. and Castro-Camilo, D. (2021). Landslide size matters: a new data-driven, spatial prototype. Engineering Geology, 293, 106288.
  • Bryce, E., Lombardo, L., van Westen, C., Tanyas, H. and Castro-Camilo, D. (2022). Unified landslide hazard assessment using hurdle models: a case study in the Island of Dominica. Stochastic Environmental Research and Risk Assessment, 36(8), 2071–2084.
  • Di Napoli, M. et al. (2023). On the estimation of landslide intensity, hazard and density via data-driven models. Natural Hazards, 119(3), 1513–1530.
  • Novellino, A. et al. (2023). Mitigating Landslides Impact in Scotland - MLIS. Summary Report. National Centre for Resilience.
  • Bryce, E. et al. (2025). An updated landslide susceptibility model and a log-Gaussian Cox process extension for Scotland. Landslides, 22(2), 517–535.

Extremes in ecology and the environment

This area covers environmental and ecological applications of extremes, including marine temperatures, wildfires, precipitation, weather hazards and environmental risk. The statistical challenge is to model rare events in settings where data are spatially structured, noisy, incomplete or physically constrained.

Related publications

  • Castro-Camilo, D., Mhalla, L. and Opitz, T. (2021). Bayesian space-time gap filling for inference on extreme hot-spots: an application to Red Sea surface temperatures. Extremes, 24(1), 105–128.
  • Castro-Camilo, D. and Huser, R. (2020). Local likelihood estimation of complex tail dependence structures, applied to U.S. precipitation extremes. JASA, 115(531), 1037–1054.
  • Hu, C. et al. (2026). XGBoost meets INLA: a two-stage spatio-temporal forecasting of wildfires in Portugal. Environmetrics, 37(3), e70089.
  • Li, M. et al. (2025). A wee exploration of techniques for risk assessments of extreme events. Extremes, 28(1), 101–121.
  • Cuba, M. D., Wilkie, C., Scott, M., & Castro-Camilo, D. (2025). Spatio-temporal data fusion of censored threshold exceedances. arXiv preprint arXiv:2504.20268.
  • Li, M., & Castro-Camilo, D. (2026). On the importance of tail assumptions in climate extreme event attribution. arXiv preprint arXiv:2507.14019.

Energy and weather-driven risk

This research focuses on probabilistic forecasting and risk assessment for energy systems affected by weather. The emphasis is on modelling both routine behaviour and rare damaging events, especially where extremes matter for infrastructure resilience.

Related publications and projects

  • Castro-Camilo, D., Huser, R. and Rue, H. (2019). A spliced Gamma-Generalized Pareto model for short-term extreme wind speed probabilistic forecasting. Journal of Agricultural, Biological and Environmental Statistics, 24(3), 517–534.
  • Maia, M., Castro-Camilo, D. and Browell, J. (2026). Probabilistic forecasting of weather-driven faults in electricity networks: a flexible approach for extreme and non-extreme events. Technometrics. Early online publication.
  • Shen, T., Browell, J., & Castro-Camilo, D. (2026). Adaptive Bayesian Very Short-Term Wind Power Forecasting Based on the Generalised Logit Transformation. arXiv preprint arXiv:2505.06310.
  • Predict4Resilience beta phase, funded by Ofgem, in collaboration with Scottish Power, SIA Partners and the Met Office.

Data fusion for environmental extremes

I am interested in combining multiple environmental data sources while preserving the behaviour of extremes. This includes blending in-situ observations, reanalysis products, model outputs and other imperfect sources while representing uncertainty and disagreement.

Related publications and projects

  • Hu, C. et al. (2026). XGBoost meets INLA: a two-stage spatio-temporal forecasting of wildfires in Portugal. Environmetrics, 37(3), e70089.
  • Cuba, M. D., Wilkie, C., Scott, M., & Castro-Camilo, D. (2025). Spatio-temporal data fusion of censored threshold exceedances. arXiv preprint arXiv:2504.20268.
  • Current PhD project: Bayesian Spatio-Temporal Modelling of Extremes: Forecasting, Scoring, and Data Fusion.

Extreme value regression and tail modelling

This work develops flexible regression models for extremes, including generalized extreme value, generalized Pareto, blended and spliced models, and approaches for tail dependence. The aim is to make tail models more robust, interpretable and useful in real applications.

Related publications

  • Castro-Camilo, D. and de Carvalho, M. (2017). Spectral density regression for bivariate extremes. Stochastic Environmental Research and Risk Assessment, 31(7), 1603–1613.
  • Castro-Camilo, D., de Carvalho, M. and Wadsworth, J. (2018). Time-varying extreme value dependence with application to leading European stock markets. Annals of Applied Statistics, 12(1), 283–309.
  • Castro-Camilo, D., Huser, R. and Rue, H. (2019). A spliced Gamma-Generalized Pareto model for short-term extreme wind speed probabilistic forecasting. JABES, 24(3), 517–534.
  • Castro-Camilo, D., Huser, R. and Rue, H. (2022). Practical strategies for GEV-based regression models for extremes. Environmetrics, 33(6), e2742.
  • Vandeskog, S. M., Martino, S. and Castro-Camilo, D. (2021). Modelling Block Maxima With the Blended Generalised Extreme Value Distribution. 22nd European Young Statisticians Meeting.
  • Hu, C., Swallow, B., & Castro-Camilo, D. (2024). A Bayesian multivariate extreme value mixture model. arXiv preprint arXiv:2401.15703.
  • Hu, C., & Castro-Camilo, D. (2025). GPDFlow: Generative multivariate threshold exceedance modeling via normalizing flows. arXiv preprint arXiv:2503.11822.
  • Cuba, M. D., Wilkie, C., Scott, M., & Castro-Camilo, D. (2025). Spatio-temporal data fusion of censored threshold exceedances. arXiv preprint arXiv:2504.20268.
  • Li, M., & Castro-Camilo, D. (2026). On the importance of tail assumptions in climate extreme event attribution. arXiv preprint arXiv:2507.14019.
  • Li, M., & Castro-Camilo, D. (2026). Tail-Calibrated Estimation of Extreme Quantile Treatment Effects. arXiv preprint arXiv:2603.23309.
  • Li, M., & Castro-Camilo, D. (2026). Causal Discovery in Multivariate Extremes via Tail Asymmetry. arXiv preprint arXiv:2604.21620.

Risk assessment of extreme events

This theme connects extreme value methodology with decision-relevant risk summaries. It includes quantifying rare-event probabilities, high quantiles, return levels, joint risks and measures that are interpretable for stakeholders.

Related publications

  • Li, M., Cuba, D., Hu, C. and Castro-Camilo, D. (2025). A wee exploration of techniques for risk assessments of extreme events. Extremes, 28(1), 101–121.
  • Castro-Camilo, D. and Huser, R. (2020). Local likelihood estimation of complex tail dependence structures, applied to U.S. precipitation extremes. JASA, 115(531), 1037–1054.
  • Castro-Camilo, D., de Carvalho, M. and Wadsworth, J. (2018). Time-varying extreme value dependence with application to leading European stock markets. Annals of Applied Statistics, 12(1), 283–309.
  • Hu, C., Swallow, B., & Castro-Camilo, D. (2024). A Bayesian multivariate extreme value mixture model. arXiv preprint arXiv:2401.15703.
  • Hu, C., & Castro-Camilo, D. (2025). GPDFlow: Generative multivariate threshold exceedance modeling via normalizing flows. arXiv preprint arXiv:2503.11822.
  • Li, M., & Castro-Camilo, D. (2026). On the importance of tail assumptions in climate extreme event attribution. arXiv preprint arXiv:2507.14019.
  • Li, M., & Castro-Camilo, D. (2026). Tail-Calibrated Estimation of Extreme Quantile Treatment Effects. arXiv preprint arXiv:2603.23309.
  • Li, M., & Castro-Camilo, D. (2026). Causal Discovery in Multivariate Extremes via Tail Asymmetry. arXiv preprint arXiv:2604.21620.
  • Shen, T., Browell, J., & Castro-Camilo, D. (2026). Adaptive Bayesian Very Short-Term Wind Power Forecasting Based on the Generalised Logit Transformation. arXiv preprint arXiv:2505.06310.