In this tutorial notebook, we will run through the process of fitting XGBoost models to meteorological data from the SHEBA campaign, in order to predict surface turbulent fluxes of sensible heat over sea ice in the Arctic. This application to polar turbulent fluxes, inspired by the work of Cummins et al. (2023) and Cummins et al. (2024), is an example of a parametrization problem that is hard to solve with traditional physics, and illustrates how modern boosting methods allow us to easily obtain performant models.
Readers are assumed to have basic familiarity with the programming language Python, and to have already completed the tutorial notebook on Random Forests from the Leeds Institute for Fluid Dynamics (LIFD).
If you're already familiar with Git, Anaconda and virtual environments, the environment you need to create is found in XGB.yml and the code below will install, activate and launch the notebook. The .yml file has been tested on the Windows 11 operating system.
git clone [email protected]:cemac/XGB-notebook.git
cd XGB-notebook
conda env create -f XGB.yml
conda activate XGB
jupyter-notebook
XGBoost-notebook by CEMAC is licensed under a Creative Commons Attribution 4.0 International License.