Quantitative analysis of the impact of realization selection on future climate change impact assessments using CMIP6 data

Future climate projection data are increasingly employed to evaluate the potential impacts of global warming across a wide range of domains, including meteorological variables (e.g., temperature and precipitation), hydrological processes, ecosystems, human health, and societal activities. The Coupled Model Intercomparison Project Phase 6 (CMIP6) provides an extensive dataset produced through international collaboration, incorporating multiple General Circulation Models (GCMs), diverse future scenarios, and numerous initial conditions. Despite the comprehensive nature of these datasets, most impact assessments rely on a limited subset of realizations, with no standardized methodology guiding their selection. This lack of consensus introduces potential biases into the outcomes of impact studies. This study quantitatively assesses the influence of realization selection on future climate impact assessments. Monthly precipitation and temperature data from CMIP6 were analyzed for both historical experimental periods and multiple Shared Socioeconomic Pathways (SSP) scenarios. Comparisons were conducted between outcomes obtained using all available realizations for each GCM and those derived from a single realization per GCM. Additionally, combinations of GCMs and realizations commonly used in prior studies were evaluated for their representativeness. The findings reveal that global average monthly precipitation is consistently higher when all realizations are utilized compared to scenarios based on a single realization. The inclusion of all realizations captures a broader range of variability, whereas subsets exhibit narrower variability and more localized trends. These results emphasize the significant impact of realization selection on future climate prediction outcomes. Moreover, an analysis of existing studies indicates that while selected datasets often reflect average trends, their overall representativeness requires further scrutiny. This research highlights the necessity of adopting uncertainty-aware methodologies in climate change studies. The findings offer valuable insights for improving the robustness and reliability of future climate impact assessments, paving the way for more informed decision-making in addressing climate change challenges.