Bathymetrix-AI

Plugin ID: 4513

An advanced Machine Learning pipeline for Satellite-Derived Bathymetry (SDB). Features ICESat-2 integration, Linear RANSAC filtering, and Spatial Residual Correction.

(1) votes

Download latest

About
Details
Versions

<h3>Bathymetrix-AI: Advanced SDB Modeling</h3>
Bathymetrix-AI is a specialized QGIS research toolkit designed to derive high-precision bathymetry from corrected multispectral satellite imagery (e.g., Sentinel-2 L2A). It integrates physics-based filtering with machine learning to overcome traditional SDB limitations.

Core Workflow (The 5-Module System):
<ul>
<li>1. Pre-processing: Automated sun-glint correction (Hedley), robust water masking (Otsu), and feature engineering (Log-Ratios).</li>
<li>2. Data Filtering: Physics-constrained Linear RANSAC to remove outliers from ICESat-2 photons and calculate uncertainty-based weights.</li>
<li>3. Global Modeling: Competitive training of 11 ML algorithms (e.g., ExtraTrees, RF, XGBoost) with hyperparameter optimization.</li>
<li>4. Spatial Refinement: Application of Universal Spatial Residual Correction (Regression-Kriging) to minimize local biases and merge global trends with local accuracy.</li>
<li>5. Rigorous Validation: Automated generation of stratified statistical reports and wMAPE analysis.</li>
</ul>

Key References: 
- Stumpf et al. (2003): Log-Ratio Algorithm for SDB. 
- Hedley et al. (2005): Sunglint Correction using NIR. 
- Fischler & Bolles (1981): RANSAC Algorithm for robust estimation. 
- Otsu (1979): A threshold selection method from gray-level histograms. 
- Breiman (2001): Random Forests machine learning. 
- Parrish et al. (2019): Satellite-derived bathymetry using ICESat-2.

Developed for scientific research and hydrographic applications.