Estimating past and future impacts of multiple dimensions of climate change

Abstract:

A robust empirical statistical literature has emerged documenting the relationship between rising temperatures and a range of negative human outcomes across physical and spatial scales. These observational exposure-response relationships have been linked to projected temperature changes to estimate the anticipated impacts of future warming scenarios. But global environmental changes are expected to happen in numerous dimensions. should we think about (a) correctly estimating statistical causal relationships when climate is changing in multiple dimensions that are not independent from one another, and (b) projecting those estimated relationships into the future, when the future structure of climate may differ from the past? Here, we demonstrate these issues with simple examples of aerosols, temperature, and precipitation, and propose methodological solutions to recover and project unbiased causal estimates in these more complicated coupled systems.