DAS (Distributed Acoustic Sensing) Analysis for Store Glacier, Greenland

This repository contains analysis code for processing and interpreting DAS data from Store Glacier in Greenland. The project combines Python and Julia for various seismological tasks including event detection, ambient noise correlation, and glacier structure monitoring.

This codebase supports three main analysis categories:

• Event Monitoring: Detection and phase picking of icequakes using STA/LTA algorithms, template matching, and machine learning models
• Data Exploration: Spectrogram computation, noise characterization, and data quality assessment
• Structure Monitoring: Ambient noise correlation, velocity change (dv/v) analysis, and glacier structure imaging

greenlanDAS/
├── event-monitoring/          # Icequake detection and phase picking
│   ├── detection/            # Detection algorithms (STA/LTA, template matching)
│   ├── icequake_search.ipynb
│   └── icequake_phase_picking.ipynb
├── data-exploration/          # Data quality and characterization
│   ├── spectrograms.ipynb
│   ├── compute_spectrogram.py
│   ├── characterize_noise.ipynb
│   ├── run_rms.jl
│   └── file_wrangling.ipynb
├── structure-monitoring/      # Correlation analysis
│   ├── correlation/          # Main correlation workflow (Julia)
│   └── Olinger_etal_2026_scripts/  # Publication analysis scripts
├── deprecated/                # Archived older versions
├── issues/                    # Troubleshooting and debugging
├── obspy_local/              # Custom ObsPy modifications (if needed)
├── requirements.txt          # Python dependencies
├── Project.toml              # Julia dependencies
└── README.md                 # This file

• Python: 3.8, 3.10, or 3.11
• Julia: 1.8.5 or compatible version
• GPU (optional but recommended): CUDA-capable GPU for accelerated correlation and detection
• CUDA Toolkit: 11.8 or 12.1 (if using GPU)

# Create a virtual environment
python -m venv greenlandas_env
source greenlandas_env/bin/activate  # On Windows: greenlandas_env\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Install ELEP (custom ensemble learning package)
git clone https://github.com/Denolle-Lab/ELEP.git
cd ELEP
pip install -e .
cd ..

# Create conda environment
conda create -n greenlandas python=3.11
conda activate greenlandas

# Install dependencies
pip install -r requirements.txt

# Install ELEP
git clone https://github.com/Denolle-Lab/ELEP.git
cd ELEP
pip install -e .
cd ..

If you have a CUDA-capable GPU, install PyTorch with CUDA support:

pip install torch==2.0.1 --index-url https://download.pytorch.org/whl/cu118

1. Install Julia 1.8.5 from julialang.org

2. Activate the project environment:

cd /path/to/greenlanDAS
julia --project=.

3. Install dependencies (in Julia REPL):

using Pkg
Pkg.instantiate()

This will install all packages specified in Project.toml with compatible versions.

4. Set up GPU support (optional, for CUDA acceleration):

using Pkg
Pkg.add("CUDA")
using CUDA
CUDA.functional()  # Should return true if GPU is available

Navigate to event-monitoring/ for icequake detection and phase picking workflows.

Quick start: Detect icequakes

cd event-monitoring
jupyter notebook icequake_search.ipynb

This notebook demonstrates:

• Loading DAS SEGY data
• Applying STA/LTA detection algorithm
• Template matching for event identification
• Saving detection catalogs

Phase picking with machine learning

jupyter notebook icequake_phase_picking.ipynb

Uses seisbench and ELEP ensemble models for automated P- and S-wave phase picking.

See event-monitoring/README.md for detailed workflow descriptions.

Navigate to data-exploration/ for data quality assessment and characterization.

Compute spectrograms

python compute_spectrogram.py
# Or use the interactive notebook:
jupyter notebook spectrograms.ipynb

Characterize noise levels

# Python notebook for noise analysis
jupyter notebook characterize_noise.ipynb

# Or Julia script for RMS computation
julia run_rms.jl

See data-exploration/README.md for detailed usage instructions.

Navigate to structure-monitoring/ for ambient noise correlation and velocity change analysis.

Run correlation workflow (Julia, GPU-accelerated):

cd structure-monitoring/correlation
julia --project=../.. run_correlation.jl

Batch processing for different wave types:

cd structure-monitoring/correlation/batch
julia --project=../../.. run_correlation_p.jl    # P-wave correlations
julia --project=../../.. run_correlation_s.jl    # S-wave correlations
julia --project=../../.. run_surface_correlation.jl  # Surface wave correlations

Post-processing and visualization:

cd structure-monitoring/correlation
jupyter notebook postprocessing.ipynb
jupyter notebook autocorrelation_visualization.ipynb

See structure-monitoring/README.md for detailed correlation workflows and parameter descriptions.

The structure-monitoring/Olinger_etal_2026_scripts/ folder contains finalized analysis code for the forthcoming publication:

• autocorrelation_dvv.ipynb: Velocity change analysis from autocorrelations
• surface_cross_correlation.ipynb: Surface wave cross-correlation analysis
• plot_study_site.ipynb: Study area maps and figures
• ice_flow.ipynb: Ice flow stress/strain modeling

This codebase primarily works with:

• Input: SEGY format DAS data (1 kHz and 4 kHz sampling rates)
• Output: HDF5 (.h5) and JLD2 (.jld2) for processed results
• Metadata: Station XML files for instrument response

Many workflows support GPU acceleration for faster processing:

• Python: PyTorch-based detection and phase picking
• Julia: CUDA.jl-accelerated correlation computations

GPU usage is automatic if a CUDA-capable device is detected. CPU fallback is available.

Load SEGY → Filter/Preprocess → STA/LTA Detection → Template Matching → Phase Picking → Catalog

Load SEGY → Resample → Detrend/Taper → Bandpass → FK Filter → Whitening → 
1-bit Normalization → Cross-Correlate → Stack → Save JLD2

Load Autocorrelations → Time Stacking → Frequency Filtering → 
Stretching Method → Compute dv/v → Visualization

• SEGY reading issues: Check issues/ folder for known problems and solutions
• GPU errors: Verify CUDA installation with nvidia-smi (Linux/Windows) or system settings (macOS)
• Julia package conflicts: Try Pkg.resolve() or Pkg.update() in Julia REPL
• Python import errors: Ensure all packages from requirements.txt are installed

This is research code for the Denolle Lab. For questions or contributions, please contact:

• Stephanie Olinger
• Marine Denolle ([email protected])

If you use this code, please cite:

Olinger, S. D., et al. (2026). [Title]. [Journal]. [In preparation]

MIT License - See LICENSE file for details.

This work was supported by [funding sources]. DAS data collection was conducted at Store Glacier, Greenland.