Achieving Conservation of Energy in Neural Network Emulators for Climate Modeling
This addresses a critical issue for climate scientists by enabling more reliable neural network emulators in climate models, though it is incremental as it builds on existing methods to enforce physical constraints.
The authors tackled the problem of neural network emulators in climate modeling not conserving energy and mass, which hinders long-term predictions, by proposing two methods to enforce linear conservation laws, showing that architecture constraints achieve satisfactory numerical precision and improve generalization to conditions like global warming.
Artificial neural-networks have the potential to emulate cloud processes with higher accuracy than the semi-empirical emulators currently used in climate models. However, neural-network models do not intrinsically conserve energy and mass, which is an obstacle to using them for long-term climate predictions. Here, we propose two methods to enforce linear conservation laws in neural-network emulators of physical models: Constraining (1) the loss function or (2) the architecture of the network itself. Applied to the emulation of explicitly-resolved cloud processes in a prototype multi-scale climate model, we show that architecture constraints can enforce conservation laws to satisfactory numerical precision, while all constraints help the neural-network better generalize to conditions outside of its training set, such as global warming.