Examining current bias and future projection consistency of globally downscaled climate projections commonly used in climate impact studies

• Published in: Climatic change Vol. 176; no. 12; p. 169
• Main Authors: Berio Fortini, Lucas, Kaiser, Lauren R., Frazier, Abby G., Giambelluca, Thomas W.
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01-12-2023; Springer Nature B.V
• Subjects: Annual precipitation; Annual temperatures; Atmospheric Sciences; Bias; Climate change; Climate Change/Climate Change Impacts; Climate studies; Datasets; Distribution; Earth and Environmental Science; Earth Sciences; Environmental impact; Future climates; Geographical distribution; Impact analysis; Local climates; Regional climates; Standardization; Uncertainty; GCMs; Downscaling; Regional projections; Future scenarios; Climate shifts; Bioclimatic variables
• ISSN: 0165-0009; 1573-1480
• DOI: 10.1007/s10584-023-03623-z

The associated uncertainties of future climate projections are one of the biggest obstacles to overcome in studies exploring the potential regional impacts of future climate shifts. In remote and climatically complex regions, the limited number of available downscaled projections may not provide an accurate representation of the underlying uncertainty in future climate or the possible range of potential scenarios. Consequently, global downscaled projections are now some of the most widely used climate datasets in the world. However, they are rarely examined for representativeness of local climate or the plausibility of their projected changes. Here we explore the utility of two such global datasets (CHELSA and WorldClim2) in providing plausible future climate scenarios for regional climate change impact studies. Our analysis was based on three steps: (1) standardizing a baseline period to compare available global downscaled projections with regional observation-based datasets and regional downscaled datasets; (2) bias correcting projections using a single observation-based baseline; and (3) having controlled differences in baselines between datasets, exploring the patterns and magnitude of projected climate shifts from these datasets to determine their plausibility as future climate scenarios, using Hawaiʻi as an example region. Focusing on mean annual temperature and precipitation, we show projected climate shifts from these commonly used global datasets not only may vary significantly from one another but may also fall well outside the range of future scenarios derived from regional downscaling efforts. As species distribution models are commonly created from these datasets, we further illustrate how a substantial portion of variability in future species distribution shifts can arise from the choice of global dataset used. Hence, projected shifts between baseline and future scenarios from these global downscaled projections warrant careful evaluation before use in climate impact studies, something rarely done in the existing literature.