MassBalanceMachine
Getting started
Installation instructions
Requirements
Installation
Environment setup
CPU only
GPU
Install dependencies
Known Installation Issues
Additional Installation for Windows Users
Project structure
Tutorials
Data Preprocessing - convert to WGMS format
Foreword
Purpose
Process
Transform your Dataset to the WGMS Format
Reshaping the dataset to WGMS-format
Reproject Coordinates to WGS84 Coordinate Reference System
Data preparation
1. Load your Target Surface Mass Balance Dataset and Retrieve RGI ID per Stake
1.1 Match the Stake Measurements with a RGI ID
2. Get Topographical Features per Stake Measurement
3. Get Meteorological Features per Stake Measurement
3.1. Manually download ERA5 data
3.2. Automatically download ERA5 data using the CDSAPI
4. Transform Data to Monthly Time Resolution
Data Exploration
1. Plot the Point Surface Mass Balance for all Available Stakes
2. Plot the Cumulative Region-Wide Surface Mass Balance
XGBoost Model Training
1. Create the train and test datasets and the data splits for Cross Validation
2. Create a
CustomXGBoostRegressor
model
3. Training
3.1 Save the trained model
3.2 Show the predictions
3.3 Show the predictions per glacier
Neural Network Model Training
1. Create the train and test datasets
2. Create a
CustomNeuralNetRegressor
model
Create datasets
3. Training
3.1. With hyper-parameters tuning
3.2. Standard training without hyper-parameters tuning
3.1. Show the predictions
3.2 Show the predictions per glacier
3.3. Make cumulative predictions
Project information
Support and help
How to get help
What NOT to do
Project Roadmap
Contributing
How to contribute?
Developing locally
Formatting the code
API
API
massbalancemachine
MassBalanceMachine
API
massbalancemachine
Edit on GitHub
massbalancemachine