Research and computational progress in integrated modeling of the co-evolution of human and Earth systems open the potential to more routinely represent climate impacts in a fully endogenous manner. To create a consistent chain of effects and feedbacks, we need to represent climate variables (many, jointly, at high resolution spatially and temporally) that drive impact models, whose outcomes influence economic activity and emissions and land-use in our energy models. The resulting forcing pathways in turn translate to a new set of climate outcomes. Many iterations may be required, and therefore emulators are critical to a computationally tractable, consistent modeling effort, also amenable to uncertainty exploration.

Here we present our solution to the climate emulation problem, STITCHES, as well as an application to the production of reference scenarios reflecting climate impacts in the Global Change Analysis Model, GCAM. STITCHES can create new climate scenarios by using existing CMIP6 output as building blocks, and delivers jointly coherent variables at the ESM original spatial resolution and time step. We show how it is being used to drive impact models connected to GCAM and thus test the sensitivity of emissions and land use to the effects of climate on agricultural production, energy demand and water availability. The goal is to provide scenarios of anthropogenic forcings that reflect the coevolution of climatic and socio-economic changes, to explore in more depth uncertainties and feasibility of future emission pathways.