Forecasting climate risk and heat stress hazards in arid ecosystems: Machine learning and ensemble models for specific humidity prediction in Dammam, Saudi Arabia

Adel S. Aldosary (), Baqer Al-Ramadan (), Abdulla Al Kafy (), Hamad Ahmed Altuwaijri () and Zullyadini A. Rahaman ()
Additional contact information
Adel S. Aldosary: King Fahd University of Petroleum and Minerals (KFUPM)
Baqer Al-Ramadan: King Fahd University of Petroleum and Minerals (KFUPM)
Abdulla Al Kafy: The University of Texas at Austin
Hamad Ahmed Altuwaijri: King Saud University
Zullyadini A. Rahaman: Sultan Idris Education University

Natural Hazards: Journal of the International Society for the Prevention and Mitigation of Natural Hazards, 2025, vol. 121, issue 8, No 14, 9309 pages

Abstract: Abstract Climate change is intensifying weather-related hazards in arid regions, making precise predictive models crucial for effective adaptation strategies. This study employs machine learning (ML) and ensemble learning (EL) models to predict specific humidity (SH) levels in Dammam, Saudi Arabia, as a foundation for forecasting climate risk and heat stress hazards in arid ecosystems. Using daily climate data from 1982 to 2023, we applied three ML models (Support Vector Machine, M5-Pruned Tree, Reduced Error Pruning Tree) and six EL models (Random Forest, Additive Regression, Random Subspace, Light Gradient Boosting Machine, Adaptive Boosting Regression, eXtreme Gradient Boosting) to predict SH levels. Descriptive statistics showed significant seasonal variations, with the highest SH levels in August (15.87 g/kg) and the lowest in January (7.16 g/kg). The region's minimal annual precipitation (average 60.57 mm) and extreme summer temperatures (average July maximum of 44.06 °C) further underline Dammam’s vulnerability to climate-induced stress. A significant increasing trend in annual SH levels was confirmed through the Mann–Kendall Test and Innovative Trend Analysis, highlighting a rising trend with a Sen's Slope of 0.025 g/kg/year. The most substantial increases occurred in July and August, reaching 0.059 and 0.054 g/kg/year, respectively, indicating escalating heat stress risks during summer season. Among the tested models, LightGBM and XGBoost stood out for accuracy (R2 = 0.99897 and 0.99883 respectively), while Additive Regression achieved the best balance across all performance metrics (R2 = 0.9987). The performance of these models demonstrates strong potential for early warning systems, enabling proactive responses to heat-related hazards. By integrating ML and EL models, this study provides a robust framework for forecasting humidity trends, contributing to improved risk assessment, disaster preparedness, and climate adaptation strategies for arid regions like Dammam.

Keywords: Climate risk; Heat stress; Specific humidity; Machine learning; Arid ecosystems; Early warning systems (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s11069-025-07140-3 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:nathaz:v:121:y:2025:i:8:d:10.1007_s11069-025-07140-3

Ordering information: This journal article can be ordered from
http://www.springer.com/economics/journal/11069

DOI: 10.1007/s11069-025-07140-3

Access Statistics for this article

Natural Hazards: Journal of the International Society for the Prevention and Mitigation of Natural Hazards is currently edited by Thomas Glade, Tad S. Murty and Vladimír Schenk

More articles in Natural Hazards: Journal of the International Society for the Prevention and Mitigation of Natural Hazards from Springer, International Society for the Prevention and Mitigation of Natural Hazards
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().