Skip to content

CMCC-Hybrid-EBM collects all the experiments releated to the estimation of salt-wedge intrusion length and salinity concentration using hybrid and machine learning based approaches.

License

Notifications You must be signed in to change notification settings

CMCC-Foundation/CMCC-Hybrid-EstuaryBoxModel

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CMCC-Hybrid-EstuaryBoxModel DOI

Table of Contents:

Description [to ToC]

The CMCC-Hybrid-EstuaryBoxModel (CMCC-Hybrid-EBM) is an hybrid model for the estimation of the salt-wedge intrusion length (Lx) and the salinity concentration in the estuaries. The Hybrid-EBM has been implemented by combining the ML-based model with the fully-physics EBM model. In particular, the first and the second component of this new model has been obtained replacing the two equations of the fully-physics model by ML algorithms like Random Forest and LSBoost. The Po River (Po-Goro-Branch) has been selected as test-case.

Project Structure [to ToC]

The project structure is organized as follows:

  • data folder contains two subfolders:
    • raw folder contains three subfolders with the raw data:
      • Component-1-Lx: Contains Excel files with the raw dataset related to the Component-1 of Hybrid-EBM.
      • Component-2-Ck: Contains three subfolders:
        • Ck-Obs-LSBoost: Contains Excel files with the raw dataset related to the Component-2 of Hybrid-EBM, generated using the Component-1-LSBoost.
        • Ck-Obs-RF: Contains Excel files with the raw dataset related to the Component-2 of Hybrid-EBM, generated using the Component-1-RF.
        • Input-Features-For-Synthetic-Ck-Obs-Generation: Contains Excel files with the raw dataset required to generate the synthetic Ck observations.
      • Component-4-Sul: Contains Excel files with the raw dataset related to the component-4 of hybrid-model.
    • processed folder contains two subfolders with the precessed training and test dataset:
      • Component-1-Lx: Contains Excel files with the processed dataset related to the Component-1 of Hybrid-EBM.
      • Component-2-Ck: Contains two subfolders:
        • Ck-Obs-LSBoost: Contains Excel files with the processed dataset related to the Component-2 of Hybrid-EBM, generated using the Component-1-LSBoost.
        • Ck-Obs-RF: Contains Excel files with the processed dataset related to the Component-2 of Hybrid-EBM, generated using the Component-1-RF.
  • src folder contains the source code files and subfolders:
    • models folder contains four subfolders, each of one contains the main scripts for running the modeling and analysis :
      • Component-1-Lx folder includes the following files:
        • train_model_component_1_lx.m: The script to trains ML models for Component-1 of Hybrid-EBM.
      • Component-2-Ck folder includes the following folder and file:
        • train_model_component_2_ck.m: The script to trains ML models for Component-2 of Hybrid-EBM.
        • Component-2-1-Generate-Syntethic-Ck-Observations folder with the files:
          • run_equation_synthetic_ck_observations.m: The script to run creation of new synthetic observations for ck values.
          • generate_synthetic_ck.m: The function (equation) to generate the ck observations.
      • Component-3-Qul folder includes the following files:
        • compute_qul.m: The function (equation) to compute the Component-3 of Hybrid-EBM.
      • Component-4-Sul folder includes the following files:
        • run_experiment_component_4_sul.m: The script to run the Component-4 of Hybrid-EBM.
        • compute_sul.m: The function (equation) to compute the Component-4 of Hybrid-EBM.
    • lib folder contains libraries for analysis, machine learning, and utility functions.
  • models folder contains the trained models, models predictions and model summaries for each component of hybrid-model.
  • reports folder contains a brief reports of the analysis with graphics and figures.

Requirements [to ToC]

Setup [to ToC]

To set up the project, follow these steps:

  1. Clone the repository:
    git clone --single-branch --branch master https://github.com/CMCC-Foundation/CMCC-Hybrid-EstuaryBoxModel.git
    
  2. Navigate to the project directory:
    cd CMCC-Hybrid-EstuaryBoxModel
    

Usage [to ToC]

To run the experiment follow these steps:

  1. Run the script to train machine learning models for the Component-1 of Hybrid-EBM (Lx):
 \src\models\Component-1-Lx\train_model_component_1_lx.m
  1. Run the script to generate synthetic observations for the Ck coefficient:
\src\models\Component-2-Ck\Component-2-1-Generate-Syntethic-Ck-Observations\run_equation_synthetic_ck_observations.m
  1. Run the script to train machine learning models for the Component-2 of Hybrid-EBM (Ck):
\src\models\Component-2-Ck\train_model_component_2_ck.m
  1. Run the script to compute Component-4 of Hybrid-EBM (Sul):
\src\models\Component-4-Sul\run_experiment_component_4_sul.m

Contact [to ToC]

For any questions or inquiries, please contact Rosalia Maglietta, Leonardo Saccotelli or Giorgia Verri.

License [to ToC]

This project is licensed under the Apache License 2.0.

About

CMCC-Hybrid-EBM collects all the experiments releated to the estimation of salt-wedge intrusion length and salinity concentration using hybrid and machine learning based approaches.

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages