Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Clustering Methods for Identifying and Modelling Areas with Similar Temperature Variations

Edoardo Otranto

TL;DR

The paper addresses how to improve modelling of global temperature dynamics by identifying groups of countries with similar temporal behavior. It introduces three clustering schemes based on annual warming rates, annual temperature variations, and the persistence of the sign of changes, using Euclidean distances for the first two and a Hamming distance for signs. It then constructs six distance- or cluster-based spatial weight matrices for Space-Time AutoRegressive (STAR) models and compares them against a contiguity-based baseline, finding that distance-based weights yield superior in-sample and out-of-sample forecasts, with the Hamming-based weights delivering the strongest in-sample fit and the Euclidean distance on full distances (dA) offering the best out-of-sample performance. The results showcase that statistical similarity among countries can outperform geographic proximity in capturing temperature dynamics and suggest broader applicability to other environmental and socioeconomic datasets.

Abstract

This paper proposes a novel data-driven approach for identifying and modelling areas with similar temperature variations throufigureh clustering and Space-Time AutoRegressive (STAR) models. Using annual temperature data from 168 countries (1901-2022), we apply three clustering methods based on (i) warming rates, (ii) annual temperature variations, and (iii) persistence of variation signs, using Euclidean and Hamming distances. These clusters are then employed to construct alternative spatial weight matrices for STAR models. Empirical results show that distance-based STAR models outperform classical contiguity-based ones, both in-sample and out-of-sample, with the Hamming distance-based STAR model achieving the best predictive accuracy. The study demonstrates that using statistical similarity rather than geographical proximity improves the modelling of global temperature dynamics, suggesting broader applicability to other environmental and socioeconomic datasets.

Clustering Methods for Identifying and Modelling Areas with Similar Temperature Variations

TL;DR

The paper addresses how to improve modelling of global temperature dynamics by identifying groups of countries with similar temporal behavior. It introduces three clustering schemes based on annual warming rates, annual temperature variations, and the persistence of the sign of changes, using Euclidean distances for the first two and a Hamming distance for signs. It then constructs six distance- or cluster-based spatial weight matrices for Space-Time AutoRegressive (STAR) models and compares them against a contiguity-based baseline, finding that distance-based weights yield superior in-sample and out-of-sample forecasts, with the Hamming-based weights delivering the strongest in-sample fit and the Euclidean distance on full distances (dA) offering the best out-of-sample performance. The results showcase that statistical similarity among countries can outperform geographic proximity in capturing temperature dynamics and suggest broader applicability to other environmental and socioeconomic datasets.

Abstract

This paper proposes a novel data-driven approach for identifying and modelling areas with similar temperature variations throufigureh clustering and Space-Time AutoRegressive (STAR) models. Using annual temperature data from 168 countries (1901-2022), we apply three clustering methods based on (i) warming rates, (ii) annual temperature variations, and (iii) persistence of variation signs, using Euclidean and Hamming distances. These clusters are then employed to construct alternative spatial weight matrices for STAR models. Empirical results show that distance-based STAR models outperform classical contiguity-based ones, both in-sample and out-of-sample, with the Hamming distance-based STAR model achieving the best predictive accuracy. The study demonstrates that using statistical similarity rather than geographical proximity improves the modelling of global temperature dynamics, suggesting broader applicability to other environmental and socioeconomic datasets.
Paper Structure (9 sections, 9 equations, 3 figures, 5 tables)

This paper contains 9 sections, 9 equations, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. The Dataset
  3. Clusters of Countries with Similar Temperature Changes
  4. Clustering A: Similar Annual Warming Rates
  5. Clustering B: Similar Temperature Variations
  6. Clustering C: Persistence of the Sign of Variations
  7. STAR Models with Distance-Based Weights
  8. Final Remarks
  9. Countries and Clusters

Figures (3)

  • Figure 1: Temperature time series (solid gray lines) of six countries and their corresponding linear trends (dotted black lines).
  • Figure 2: Boxplots of the estimated slope coefficients for each cluster. The cross indicates the mean slope within the corresponding cluster.
  • Figure 3: Distribution of warming-rate clusters across geographical zones.