LunarIce-360

LunarIce-360 is an end-to-end mission planning and scientific processing pipeline for lunar water ice detection, volumetric estimation, landing site selection, and multi-objective rover traversal optimization. The pipeline is designed to work with dual-frequency (L-band and S-band) polarimetric Synthetic Aperture Radar (SAR) data, such as those from Chandrayaan-2's Dual Frequency SAR (DFSAR).

---

Architecture & Pipeline Flow

The processing flow from raw SAR inputs to optimized rover traversal and volumetric estimation:

graph TD
    %% Inputs
    Sub1[L-Band Stokes Parameters S1-S4] --> Pre[Preprocess & Speckle Filter]
    Sub2[S-Band Stokes Parameters S1-S4] --> Pre
    DEM[Digital Elevation Model] --> Terr[Terrain Analysis]

    %% Feature Extraction
    Pre --> Feat[Polarimetric Feature Extraction: CPR, DOP, Entropy, Alpha]
    
    %% Ice Detection
    Feat --> Det[Ice Detection Fusion <br> GMM, Anomaly detection, Sinha et al. Zone Classifier]
    
    %% Landing and Traverse
    Terr --> Land[Landing Site Safety Scoring]
    Det --> Land
    
    Land --> Trav[NSGA-II Rover Path Planning <br> Distance, Hazard, Solar Illumination]
    
    %% Volume Estimation
    Det --> Vol[Bayesian MCMC Volumetric Inversion <br> Dielectric & Thickness Modeling]
    
    %% Outputs
    Trav --> Vis[Visualization & Interactive UI]
    Vol --> Vis

Loading

---

Key Modules & Pipeline Steps

1. Preprocessing (preprocessing.py)

• Filters speckle noise using a modified Lee Speckle Filter on Stokes parameters ($S_1, S_2, S_3, S_4$).
• Computes spatial averages for coherent matrix calculations.

2. Polarimetric Feature Extraction (polarimetry.py)

• Computes Circular Polarization Ratio (CPR) and Degree of Polarization (DOP).
• Builds the $2 \times 2$ wave coherency matrix $[J]$.
• Performs Cloude-Pottier Eigenvalue Decomposition to extract scattering Entropy ($H$), Anisotropy ($A$), and average scattering angle Alpha ($\alpha$) for scattering mechanism identification.

3. Ice Detection Fusion (ice_detection.py)

• Fuses multiple heuristic and machine learning methods:
  • Thresholding: Traditional criteria (CPR > 1.0, DOP < 0.13 indicating volume scattering).
  • Cloude-Pottier H-Alpha Zones: Sinha et al. (2026) classification.
  • Gaussian Mixture Models (GMM): Clustering polarimetric signatures.
  • Isolation Forest: Detecting anomalies relative to typical dry lunar regolith.

4. Terrain & Shadow Analysis (terrain.py & landing_site.py)

• Calculates slopes and RMS surface roughness from the Digital Elevation Model (DEM).
• Estimates solar illumination fractions and shadow patterns.
• Scores candidate landing sites using a multi-attribute utility function weighting safety, illumination, flatness, and proximity to detected ice.

5. Multi-Objective Rover Traverse Path Planning (traverse.py)

• Optimizes rover traversals using the NSGA-II genetic algorithm.
• Optimizes paths across three concurrent objectives:
  1. Minimize path length.
  2. Minimize slope and hazard exposure.
  3. Maximize solar illumination for battery charging.
• Integrates physical constraints including rover speed, power generation, and communication limits.

6. Volumetric Ice Inversion (volume_estimation.py)

• Uses Bayesian MCMC (Markov Chain Monte Carlo) to invert physical parameters from SAR backscatter.
• Estimates ice fraction, layer thickness, density, and roughness.
• Produces confidence bounds and corner plots showing parameter correlations.

---

Installation & Setup

1. Clone the repository:

   git clone https://github.com/yashs00/BAHood.git
   cd BAHood

2. Set up a virtual environment (optional but recommended):

   python -m venv .venv
   # Windows:
   .venv\Scripts\activate
   # macOS/Linux:
   source .venv/bin/activate

3. Install dependencies: Ensure you have the required scientific libraries installed (NumPy, SciPy, Matplotlib, scikit-learn, emcee, corner).

---

Running the Pipeline

Option 1: Run the Synthetic Demo Pipeline

Run the fully self-contained simulation which generates a synthetic crater terrain, injects ice signatures, runs the entire detection, site selection, NSGA-II traversal optimization, and MCMC volume estimation:

python -m lunarice360.demo_synthetic

All generated heatmaps, path maps, MCMC corner plots, and logs are saved in the outputs/ folder.

Option 2: Run the Web UI App

Run the local HTTP server to launch the pipeline from your web browser:

python -m lunarice360.ui_app

Once started, navigate to http://localhost:8000 in your web browser.

---

Project Structure

BAHood/
│
├── lunarice360/
│   ├── __init__.py
│   ├── config.py              # Configuration & Physical Constants
│   ├── data_loader.py         # GeoTIFF and PDS dataset loaders
│   ├── preprocessing.py       # Speckle filtering (Lee filter)
│   ├── polarimetry.py         # Stokes & Cloude-Pottier features
│   ├── ice_detection.py       # Heuristics & Machine Learning Fusion
│   ├── terrain.py             # Slope & Roughness analysis
│   ├── landing_site.py        # Safety & Proximity scoring
│   ├── traverse.py            # NSGA-II rover path planning
│   ├── volume_estimation.py   # MCMC thickness & volume estimation
│   ├── visualization.py       # Grid plotting & Corner plots
│   ├── ui_app.py              # Local Web UI launcher
│   └── main.py                # Pipeline orchestrator
│
├── tests/                     # Smoke tests
├── .gitignore
└── README.md