Supplementary code for "Emulation of high-resolution land surface models using sparse Gaussian processes with application to JULES"
This folder contains 2 scripts:
Emulator.py will fit an emualtor for a specified pft, given the existence of some training data "Training_Data_standardised.csv".
It then saves the results in a preexisting folder called "place_to_save_results"

Joint_Validation.py will then validate all 5 pft emulators (note this requires Emulator.py to be ran 5 times, once per pft), given the existence of some testing data "Testing_Data_standardised.csv"

gpflow 1.5.1 needs to be installed (which itself requires tensorflow 1 (last is 1.15), which then also requires 3.5<python<3.8)

Data used can be obtained from https://doi.org/10.5285/789bea37-0450-4822-9857-3dc848feb937
Note that csv files of this size can take a long time to load in. Originally I was using .pkl (pickle) files instead, but the EIDC required csv for online storage. 
If you are going to use run this multiple times, it may be worth converting them to pkl files in python for faster re-running.