# Spatial trend analysis of gridded temperature data at varying spatial scales

## TL;DR

This paper introduces a hierarchical Bayesian model to analyze spatial temperature trends, accounting for spatial correlations and enabling joint significance assessments, applied to 65 years of European temperature data.

## Contribution

It develops a novel space-time Bayesian approach that models trend coefficients as a Gaussian random field for improved spatial trend analysis.

## Key findings

- Spatial smoothing increases regions with significant trends
- Not all sub-regions show increased significance after smoothing
- Joint significance assessment provides more accurate trend detection

## Abstract

Classical assessments of trends in gridded temperature data perform independent evaluations across the grid, thus, ignoring spatial correlations in the trend estimates. In particular, this affects assessments of trend significance as evaluation of the collective significance of individual tests is commonly neglected. In this article we build a space-time hierarchical Bayesian model for temperature anomalies where the trend coefficient is modeled by a latent Gaussian random field. This enables us to calculate simultaneous credible regions for joint significance assessments. In a case study, we assess summer season trends in 65 years of gridded temperature data over Europe. We find that while spatial smoothing generally results in larger regions where the null hypothesis of no trend is rejected, this is not the case for all sub-regions.