High spatial resolution monthly temperature and precipitation data modelling for Tunisia from 1950 to 2023

Taheri, Mehdi; Mol, Lisa; Leslie, Alick

Abstract

High spatial and temporal resolutions climate data play a fundamental role in enhancing our understanding of climate change and addressing its environmental and social consequences. However, challenges such as the sparse distribution of meteorological stations and the limited duration of statistical data collection have led to climate data being predominantly available at lower temporal and spatial resolutions, in many countries worldwide, including Tunisia. This limitation hampers accurate climate impact assessments and the development of effective adaptation strategies. One approach to address this limitation is to utilize global or regional gridded datasets and apply downscaling techniques to achieve the desired resolution. There-fore, this study aimed to produce high-resolution (~1 km) monthly air temperature and precipitation data for Tunisia, spanning from 1950 to 2023, by downscaling the state-of-the-art ERA5-Land reanalysis dataset through utilizing two main methods: the delta downscaling framework and statistical downscaling with the KrigR R package. While applying the delta downscaling method, three interpolation methods were employed, including bilinear, nearest neigh-bour, and kriging with elevation as a covariate. Following a comprehensive evaluation of the results against observa-tions from meteorological stations, the KrigR R package method demonstrated the highest accuracy and was subse-quently selected as the final product. The assessment indi-cated acceptable accuracy for the downscaled data pro-duced using this method. Therefore, the KrigR R package offers a promising alternative to traditional delta downscal-ing methods for generating more accurate, high-resolution climate data. Importantly, downscaled data enables the analysis of highly accurate and spatially detailed trends in temperature and precipitation changes across Tunisia. Analysis of the downscaled data reveals a robust and con-sistent warming trend in Tunisia, with an average increase of 0.0265°C per year. Conversely, there is a general ten-dency for precipitation to decrease, with a mean annual re-duction of 0.363 mm/year. However, statistically signifi-cant trends in precipitation are mostly confined to specific areas. Finally, the detailed spatial variability revealed in the downscaled data for Tunisian temperature and precipi-tation trends presents valuable opportunities for further in-depth studies on regional climate change. This methodolo-gy for generation of high-res climate data improves impact and adaptation assessments in data-scarce regions.

• Presentation Conference Type: Presentation / Talk
• Conference Name: Mediterranean Geosciences Union Annual Meeting (MedGU-24)
• Start Date: Nov 25, 2024
• End Date: Nov 28, 2024
• Acceptance Date: Jul 20, 2024
• Deposit Date: Dec 10, 2024
• Peer Reviewed: Peer Reviewed
• Keywords: Downscaling, ERA5-Land, air temperature, precipitation, Tunisia
• Public URL: https://uwe-repository.worktribe.com/output/13521092