KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh #dsh-plugin #developer-tools #ai-agent #claude-code #codex #typescript #ai-agents #deepseek-harness #mcp #cli #rust

Bathymetrix-AI

★ 21 Open GitHub ↗

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

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

</td>
<td>

**Bathymetrix-AI: Advanced SDB Toolkit**

</td>
</tr>
</table>

> **Note:** Bathymetrix-AI supports ANY satellite imagery provided that the image is Atmospherically Corrected (Surface Reflectance) and the raster values are of type `Float`.

**Bathymetrix-AI** is a professional QGIS research toolkit designed for high-precision Satellite-Derived Bathymetry (SDB). It integrates Satellite Imagery multispectral data with ICESat-2 (ATL24) LiDAR data using a modular and adaptive Machine Learning pipeline.
The tool overcomes traditional bathymetry challenges like sun-glint and deep-water noise through a systematic 5-phase workflow, alongside powerful standalone analysis modules.

🔬 **Scientific Methodology (Core 5-Phase Workflow)**
The toolkit follows a modular workflow where each phase is designed to improve the accuracy of depth retrieval and understand coastal changes over time.

**Phase 01: Advanced Pre-processing**
This phase prepares the satellite imagery by isolating the aquatic domain and correcting radiometric noise.  
**Sun-Glint Removal:** Removes surface reflections to reveal the seabed signal (Hedley et al., 2005). The algorithm safely handles infinite/NaN values to ensure stability.
**Water Segmentation:** Advanced Water Masking using 3-Indices (NDWI, MNDWI, NWI) alongside adaptive thresholding to accurately isolate the aquatic domain.
**Deep Water OSW Filtering:** Deep Water Filter customized for ML algorithms. It features automatic statistical thresholding (NIR), manual polygon masking, and a dynamic **Elbow Point Detection** algorithm to intelligently remove scattered deep-water noise components.
**OSW Boundary Extraction:** Automatically extracts the Optically Shallow Water footprint as a GPKG vector polygon, inheriting the CRS of the source imagery to eliminate GDAL cutline projection warnings during map clipping.
**Log-Ratio Features & Indices:** Transforms spectral bands into depth-sensitive features based on light attenuation laws, computing physics-based Log-Ratio features (e.g., Blue/Green) as well as newly added indices like NDWI, Log(Green)/Log(NIR), and Log(Red)/Log(NIR).

**Phase 02: Robust Filtering**
To ensure high-quality training data, the tool filters altimetry data to remove outliers and environmental noise.
**Noise Removal:** Iteratively identifies high-confidence "inlier" depth points using Linear RANSAC, LS Variance Fit, or Huber Variance Fit (Zhang et al., 2021).
**Dynamic Plotting:** Outlier-robust, percentile-based zooming for diagnostic plots ensures clear, highly readable variance and trend charts by dynamically ignoring extreme noise.

**Phase 03: Global Auto-ML & Feature Analysis**
Instead of using one algorithm, the tool implements an automated machine learning workflow.
**Feature Analysis:** Implements advanced feature selection by dropping weak or redundant bands using statistical correlations (Pearson, Spearman) or automated dynamic thresholds (Automatic-RANSAC, Automatic-Random Forest).
**Algorithm Benchmarking:** Evaluates 15+ different Machine Learning models (e.g., Random Forest, Gradient Boosting, XGBoost, CatBoost, SVR, MLP) to find the best fit for specific coastal areas.
**Ensemble Blending:** Features an advanced **Uncertainty-Weighted Pixel Fusion** ensemble blending mode, improving the robustness of final maps using inverse variance weighting based on residual uncertainties, alongside Standard Average, Median, and Stacking.
**Memory-Efficient Prediction:** Uses chunk-based processing (batches of 500,000 pixels) to generate final maps, preventing memory crashes with extremely large images.
**Hyperparameter Optimization:** Automatically tunes model settings via Random Search, Grid Search, or Bayesian Optimization (Bergstra & Bengio, 2012).
**Spatial Cross-Validation:** Independent spatial block cross-validation control to evaluate base models.
**Customization:** Fully customizable hyperparameters for precise fine-tuning.

**Phase 04: Adaptive Refinement**
This phase corrects local errors that global models might miss by analyzing the "residuals" (differences) between predicted and observed depths.  
**Zero-Mean Centered Spatial Error Mapping:** Zero-mean centered residual bias correction to eliminate global offset drift, followed by spatially localized corrections and residual analysis (Alevizos, 2020) to fix local geographic biases.
**LOO Robust Huber & Smoothed IDW Weighting:** Uses Leave-One-Out (LOO) robust Huber weighting and distance-decay smoothing ($1 / (d + 1.0)$) to prevent divide-by-zero spikes and outlier distortion in spatial error maps.
**Empirical Residual Uncertainty Modeling:** Fits a RandomForest Residual Regressor to generate spatial prediction uncertainty maps at 95% confidence (`Final_Uncertainty_95.tif`).
**Spatial Cross-Validation:** Independent spatial block cross-validation control to evaluate residual modeling.
**Adaptive Re-training & Pixel Fusion:** Combines spectral data with error maps to produce a refined, high-accuracy final depth map. Incorporates **Uncertainty-Weighted Pixel Fusion** for robust model stacking.

**Phase 05: Validation & Reporting**
**Independent Accuracy Assessment:** Validates the finalized models on unseen test data to ensure robust accuracy reporting, IHO Order 1a/2 TVU compliance evaluation, and scientific validity.
**Interactive Validation Dashboard:** Generates a premium HTML dashboard containing comprehensive AutoML leaderboards, validation metrics, and diagnostic plots with a high-performance responsive layout.

🚀 **Standalone Modules**

**Coastal Dynamics (Temporal Intelligence)**
An independent, standalone Temporal Analysis module to monitor how the seabed changes across multiple years.
**Volumetric Sand Tracking & Target ROI Analytics:** Utilizes robust **Linear Regression** across multiple years to track sediment volume (m³). Features a dedicated **Target ROI Vector Polygon** input for tracking accretion, erosion, and net balance ($m^3$) in targeted sub-regions alongside the full study area. This is **highly valuable for Model Quality Control**, allowing users to validate the model's accuracy by checking if the calculated dredging (Erosion) and shoaling (Accretion) volumes in a known ROI match real-world physical expectations. It incorporates a Statistical Level of Detection (StatCD) (Wheaton et al., 2010) and Minimum Mapping Unit (MMU) spatial coherence filters (Lane & Chandler, 2003) to flawlessly categorize dredging and shoaling while ignoring noise.
**Temporal Uncertainty Modeling (Quantile Regression vs Classical):** Supports both Classical Z-score (1.96σ) and **Quantile Regression (Adaptive σ)** uncertainty modes at customizable confidence levels (0.50–0.99) to account for depth-varying noise.
**Morphological Stability Index (MSI):** Computes the volatility and stability of the seabed (0 to 1 scale) based on temporal depth variance, vital for infrastructure planning.
**Shoreline Migration:** Extracts land-water boundaries for each year, creating explicit Erosion and Accretion polygons.
**Full Process File Logging:** Automatically streams all execution logs to `Full_Process_Log.txt` in real time.
**Automated HTML Reporting:** Generates a premium, dynamic HTML report that beautifully formats all statistics and highlights engineering metrics.

**ICESat-2 Downloader**
A specialized standalone tool to query, filter, and download ICESat-2 (ATL24) LiDAR bathymetry data directly from NSIDC for integration into the MasterFlow pipeline.

📊 **Performance Metrics**

The tool evaluates results using three main standards:  
**R2** (Coefficient of Determination): Measures how well the model fits the data.  
**RMSE** (Root Mean Square Error): Measures the average vertical error in meters.  
**wMAPE** (Weighted Mean Absolute Percentage Error): Measures the relative error across different depth ranges.

🛠️ **Installation & Dependencies**  
Open **OSGeo4W Shell** (as Administrator) and run the following command to install all required libraries. This version is optimized for **QGIS 4.0 (Qt6)** and **NumPy 2.0** support:

```bash
pip install numpy pandas rasterio matplotlib seaborn scikit-learn>=1.5.0 scipy joblib scikit-optimize sliderule icepyx geopandas parquet netCDF4 xgboost lightgbm catboost optuna
```

📧 ***Contact & Citations***

**Author:** Mohamed Aly Nasef  
**Email:** Eng.m.nasef2017@gmail.com, Nasefm.aly@alexu.edu.eg

Nasef M.Aly. (2026). Nasef2017/Bathymetrix-AI: Bathymetrix_AI V6.5 (Version v6.5) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.21828682

🤖 **AI Acknowledgment**  
The development of the Bathymetrix-AI code, its logical structure, and the technical documentation were significantly enhanced and optimized using Google Gemini. The AI assisted in debugging complex workflows and ensuring the implementation follows best practices in data science.

Bathymetrix-AI

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README