I am interested to learn about hybrid modeling (machine learning incorporated to conventional physics-based models) but unsure how to start. I have some background in numerical modeling, as in I have experience in preparing and running numerical model simulations. I would appreciate what would be a way to start getting into hybrid modeling. I have no experience in ML yet.
Thanks! 
Hey, you can check out the following colab notebook: Google Colab
This notebook is meant to serve as a code walkthrough for the following research paper: https://doi.org/10.1029/2022WR032404 and is for hybrid modeling for streamflow models. Not sure if you need something more specific to your goals, but this is a good start to understand the idea of integrating a Neural Net into your process-based model.
Thank you very much for your response. It seems the colab notebook is restricted. I requested for access.
Apologies for that. Here is an updated link with better access: Google Colab
As part of the M2LiNES project, we wrote a book called Learning Machine Learning with Lorenz-96.
The climate system is composed of many interacting components (e.g., ocean, atmosphere, ice) and described by complex nonlinear equations. To simulate, understand and predict climate, these equations are solved numerically under a number of simplifications, therefore leading to errors. The errors are the result of numerics used to solve the equations and the lack of appropriate representations of processes occurring below the resolution of the climate model grid (i.e., subgrid processes).
The goal of this book is to conceptualize the problems associated with climate models within a simple and computationally accessible framework. We will introduce the readers to climate modeling by using a simple tool, the [Lorenz, 1995] (L96) two-timescale model. We discuss the numerical aspects of the L96 model, the approximate representation of subgrid processes (known as parameterizations or closures), and simple data assimilation problems (a data-model fusion method). We will then use the L96 results to demonstrate how to learn subgrid parameterizations from data with machine learning, and then test the parameterizations offline (apriori) and online (aposteriori), with a focus on the interpretability of the results.
The book was created by and as part of M2LInES, an international collaboration supported by Schmidt Futures, to improve climate models with scientific machine learning. The goal for this book was for our team to work together and learn from each other; in particular, to get up to speed on the key scientific aspects of our collaboration (parameterizations, machine learning, data assimilation, uncertainty quantification) and to develop new ideas. Ultimately, we are happy to share these resources with the scientific community, to introduce our research ideas and foster the use of machine learning techniques for tackling climate science problems.
Thank you very much! This is very interesting.