1,2,2,4,1,- 1Department of Public Health Science, Graduate School of Public Health, Seoul National University, Seoul, Republic of Korea
- 2IBS Center for Climate Physics, Busan, Republic of Korea
- 3Pusan National University, Busan, Republic of Korea
- 4Institute of Health and Environment, Graduate School of Public Health, Seoul National University, Seoul, Republic of Korea
- 5School of Biomedical Convergence Engineering, College of Information and Biomedical Engineering, Pusan National University, Yangsan, Republic of Korea
Background
Accurate estimation of temperature-related mortality under climate change may be influenced by the spatial resolution of climate data. Recently, the km-scale global warming simulations provide improved representation of regional climate processes. However, it remains unclear how differences in spatial resolution influence the quantification of health impacts. This study quantifies temperature-related mortality using climate simulations with different spatial resolutions and evaluates the sensitivity of mortality estimates to climate model resolution.
Methods
We used two simulations from the same fully coupled climate model (AWI-CM3) that differ only in atmospheric resolution: a medium-resolution setup (TCo319, ~31–38 km) and a high-resolution setup (TCo1279, ~9–10 km). Daily temperatures were statistically bias-corrected using the ISIMIP trend-preserving approach. The mortality data were obtained from the Multi-Country Multi-City Collaborative Research Network and were linked to climate data by matching each of the 761 cities worldwide to the nearest model grid cell.
Temperature–mortality associations were estimated through a two-stage time-series approach. In the first stage, distributed lag non-linear models with lag periods up to 21 days were fitted for each city to capture non-linear and delayed temperature effects on mortality. Relative risks were estimated using the minimum mortality temperature as the reference, distinguishing heat-related and cold-related risks. In the second stage, city-specific estimates were pooled using multivariate meta-regression to derive Best Linear Unbiased Predictions at the regional level.
Baseline temperature-attributable mortality for 2002–2012 was estimated using 1,000 Monte Carlo simulations. Future changes in attributable mortality were projected and compared between the TCo319 and TCo1279 simulations to assess the impact of spatial resolution.
Results
Despite sharing the same model structure and bias-correction method, the two simulations produced different estimates of temperature-attributable mortality. The TCo1279 simulation captured finer-scale temperature variability and extremes, leading to larger and more spatially heterogeneous estimates of heat-related mortality, particularly in future periods. These differences were most pronounced in regions with complex topography or strong climate variability, including Europe and the Americas. Cold-related mortality was generally less sensitive to spatial resolution, although regional differences remained.
Conclusions
Spatial resolution in km-scale global warming simulations plays a critical role in quantifying temperature-related mortality. High-resolution climate data improve the detection of heat-related mortality burden, especially for extreme temperature events, and provide more detailed regional patterns. Reliance on coarser-resolution data may underestimate both the magnitude and spatial heterogeneity of future health impacts. Incorporating fine-resolution climate projections is therefore essential for robust and policy-relevant assessments of climate change–related mortality.
How to cite: Oh, J., Seo, Y., Lee, J.-Y., Moon, J.-Y., Min, J., Kang, C., Kim, H., and Lee, W.: Quantifying temperature-related mortality from km-scale global warming simulation data with different spatial resolutions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9433, https://doi.org/10.5194/egusphere-egu26-9433, 2026.
