What does this research mean for the field?

The Bayesian information criterion (BIC) outperforms other criteria for selecting generalized extreme value (GEV) regression models to predict changes in extreme temperature return values, and applying this method reveals significant projected increases in extreme temperatures across global desert regions under strong forcing scenarios. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To identify optimal model selection criteria in non-stationary generalized extreme value (GEV) regression for estimating changes in 100-year return values of extreme desert temperatures under CMIP6 projections.

August 29, 2026Stochastic Environmental Research and Risk AssessmentOpen Access

Model selection in extreme value regression for estimation of changes in return values over time applied to extreme annual regional CMIP6 desert temperatures

View Full Paper

Ask AI

Bookmark

Authors

CLCallum LeachKnowledge Unlatched (Germany)KEKevin EwansOcean Energy (Norway)PJPhilip JonathanLancaster University

Discussion

Member takes

Implication

Modeling study reveals escalating 100-year extreme temperatures across global desert regions, indicating faster warming of extreme maxima in hot deserts but extreme minima in Antarctica.

Key Points

To identify optimal model selection criteria in non-stationary generalized extreme value (GEV) regression for estimating changes in 100-year return values of extreme desert temperatures under CMIP6 projections.
Formulated non-stationary GEV regression models with time-varying parameters estimated via Bayesian inference using CMIP6 annual maximum and minimum near-surface temperatures (2015–2125) across global desert regions.
Conducted a simulation study using ground-truth models to assess information criteria (AIC, BIC, DIC, and WAIC) and stacked Bayesian model averaging for predicting changes in 100-year return values (ΔQ).
The Bayesian information criterion (BIC) minimized prediction error for ΔQ, outperforming divergence and widely-applicable information criteria.
Aggregated posterior estimates showed significant increases in 100-year return values for annual maxima across all desert regions under stronger climate forcing scenarios, with weaker trends for annual minima.
Extreme minimum temperatures warmed more rapidly than extreme maxima in Antarctica, whereas hot desert regions exhibited the opposite pattern.

Cite This Study

Leach et al. (2026) studied this question.

synapsesocial.com/papers/6a92994d8e5d7d1fc0c112eb https://doi.org/10.1007/s00477-026-03351-7

View Full Paper

Ask AI

Bookmark

Implication

Modeling study reveals escalating 100-year extreme temperatures across global desert regions, indicating faster warming of extreme maxima in hot deserts but extreme minima in Antarctica.

Model selection in extreme value regression for estimation of changes in return values over time applied to extreme annual regional CMIP6 desert temperatures

Discussion

Member takes

Implication

Key Points

Cite This Study

Model selection in extreme value regression for estimation of changes in return values over time applied to extreme annual regional CMIP6 desert temperatures

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider