Using a Simple Climate Model as an Earth System Model Emulator
Earth System Models (ESMs) provide highly detailed predictions of the future state of the Earth, but take a long time and require extensive computing power to run. Reduced form climate models (RCMs), such as Hector, provide lower resolution climate information but are comparatively computationally inexpensive and fast. Many applications do not require the detailed data provided by ESMs, but need similarly accurate climate information. To accomplish this, RCMs must be trained to reproduce the final results of an ESM. Tuning the Hector parameters enables it to emulate the long-term behavior of ESMs.
Demonstrating Hector’s emulating capabilities enables its use in a wide range of applications. These parametrizations can be used in model coupling exercises to quickly and inexpensively explore a range of climate impacts. This also acts as a foundation to guide future calibration activities that will inevitably be necessary as both Hector and ESMs undergo further development.
Researchers calibrated an RCM (Hector) to efficiently emulate the output of more complex climate models. Using RCMs as emulators is a common practice in the climate science and climate-economic research communities when data on long-term changes in climate are needed for energy, water, land, or other sectors and when fine-scale climate information is not needed. In order to use Hector as an emulator that satisfactorily reproduces an ESM’s modeled climate, researchers found that it must be constrained with multiple output variables. Otherwise, the parameters can trade-off with arbitrarily large values; for example, an unrealistically high climate sensitivity balanced by unrealistically large aerosol forcing can produce temperature change comparable to those produced by an ESM.