Abstract Climate change impacts include contributions from both anthropogenic forcing and internal (natural) climate variability. Large ensembles of Earth System Model (ESM) simulations have been used to quantify the influence of natural variability in climate change impact projections. However, such ensembles have high computational costs. Future climate scenarios can be simulated at much lower cost using statistical climate emulators that mimic the output of more complex ESMs. Most climate emulators do not include internal variability, limiting their utility for climate impacts assessment. We present a new method for emulating spatially resolved monthly natural temperature variability using Vector Auto‐Regression models and empirical orthogonal function analysis of temperature data from existing ESM simulations. The emulator—Internal Variability Emulator for Regional Temperature—preserves spatial, and temporal characteristics of temperature variability in ESMs and can be quickly trained on any target data set.
Saenger et al. (Sun,) studied this question.