In this chapter we will explore some key decisions we need to consider when deciding which model representation we are going to employ. Below is a whirlwind tour of what's to come.
Dynamical Systems. The make-up of a dynamical system is the basis of most processes in geosciences. Here, we give an overview of a.
Known, Unknown, Hybrid. Our knowledge is always limited. However, we can always make some assumptions about which assumptions we would like to assume based on the architectures
Continuous Fields. Admittedly, many continuous systems are limited to solutions which have an analytical form. However, we also introduce neural fields as a neural network representation of a continuous field that can be learned from observations.
Discrete Fields. Continuous representations are expensive so most times we end up with discrete representations of our systems.
Mesh Fields (TODO).