Title: Constraining Climate Model Projections of 21st-Century Global and Regional Warming
Abstract: Projections of future climate change by climate models come with uncertainty due to the use of different emissions scenarios, model imperfections, and natural internal climate variability. One approach to reducing this uncertainty is to constrain model projections using observations based on emergent relationships between historical metrics and future projected climate across multi-model ensembles. This presentation provides an overview of the application of observational constraints to global and regional warming. Firstly, we used past global mean surface air temperature (GSAT) trend and climatology of low-cloud as two metrics to constrain the projected temperature change. We found that the use of climatology of low-cloud as a constraint produced better results than using the past GSAT trend in cross-validation. Our evidence suggests the lower bound of the projected warming range to be higher when projection is constrained by climatology of low-cloud than that by the past GSAT trend. Secondly, the use of GSAT trend with the Pacific natural internal variability removed as constraint metric produced projections consistent with the constrained projections based on climatology low-cloud metric. This implies that accounting for the influence of internal variability in the constraint metrics can lead to more robust constrained projections. Lastly, we found that using global-scale low-cloud metrics alone could produce considerably narrower uncertainty ranges of 21st-century warming over sub-continental land regions in the Northern Hemisphere. The narrower constrained uncertainty ranges produced by our observationally constrained framework are relevant to climate change policy and adaptation decisions.
