Average Ensemble Approach and Implications for Modeling Future Climate Projections of Flexible Pavement Subsurface Temperature

Increasing practitioner interest in quantifying the impacts of climate change in flexible pavement design has driven approaches to utilize spatially and temporally downscaled atmosphere-ocean general circulation model (AOGCM) future climate projections as inputs to mechanistic empirical design methodologies. Gaps in research include comparing subsurface temperature projections between different methodologies and how projected temperatures compare to baseline scenarios throughout pavement depths. The research question of this work was: are differences in temperature distributions sourced from statistical and temporal downscaling methods and baseline scenarios statistically significant throughout a pavement profile, even across various locations in the United States? The investigation was conducted using two established approaches for pavement applications, the Asynchronous Regional Regression Model (ARRM) and the Delta Method, to produce spatially and temporally downscaled Representative Concentration Pathway (RCP) 6.0 and RCP 8.5 AOGCMs with an average ensemble approach for 20 Seasonal Monitoring Program sites in the Long-Term Pavement Performance Program. Impact-relevant hourly ARRM and Delta Method future climate projections were obtained using the Modern-Era Retrospective Analysis for Research and Applications version 2 observed temperatures 2 meters above the pavement surface. One-sample t-testing accounting for uncertainty in the energy balance model was conducted between the base year and projected target year, 45 years in the future, for temperature distributions at the pavement surface, the penultimate, and bottom depths. In most cases, there were statistically significant differences between baseline temperature scenarios and future climate projections of temperature throughout pavement depths. Significant differences were also seen across the multiple downscaling techniques. Finally, differences in extreme temperatures were examined using 98% reliability performance classifications at the pavement surface, in which a myriad of changes in computed binder grades for future versus baseline scenarios were noted. These results were accompanied by an analysis of extreme temperature percentiles and trends in diurnal temperature range for air and pavement surface temperature.