wavelet-lab

Interactive web app for real-world signal exploration, wavelet analysis, event-based sampling, and SVM-based event detection — powered by Python and Flask.

Overview

Wavelet Lab is an experimental web application for exploring signal processing techniques using real-world data. It combines data retrieval, interactive visualization, and analytical tools in a shared Signal Processing Lab.

The project currently supports seismic waveforms from EarthScope and hydrological time series from the USGS Water Data API.

Live Demo

Wavelet Lab is deployed on Render and is available at:

https://wavelet-lab.onrender.com

The application is automatically deployed from the master branch, so each new release is published to the live environment after it is merged.

Features

Data Explorers

• EarthScope Explorer: Retrieve seismic waveforms using dynamic network, station, and channel selection.
• USGS Earthquake Catalog: Discover recent earthquakes with magnitude filters and configurable result limits.
• Water Data Explorer: Explore monitoring stations, retrieve historical streamflow observations, and visualize hydrographs.
• Transfer selected signals from either explorer directly to the Signal Processing Lab.

Signal Processing Lab

• Wavelet compression: Apply discrete wavelet transforms and compare the original and reconstructed signals, including retained coefficients and reconstruction RMSE.
• CWT analysis: Explore time-frequency representations using Morlet and Mexican Hat wavelets with configurable parameters.
• Event-based sampling: Compare Send-on-Delta, energy-domain sampling, predictive Send-on-Delta, and the integral criterion.
• Optional wavelet preprocessing: Investigate the effect of wavelet coefficient thresholding on event-based sampling.
• SVM event detection: Fit an RBF Support Vector Regression model and identify observations outside the epsilon-insensitive tube.
• Interactive Plotly visualizations for signals, reconstructions, sampling results, and detected events.

Screenshots

SVR model, epsilon-insensitive tube, and detected events highlighted in red.

Technologies

• Python
• Flask
• NumPy
• SciPy
• PyWavelets
• scikit-learn
• Plotly
• JavaScript

Project Status

Wavelet Lab is under active development. The current implementation provides an experimental environment for comparing signal processing and event detection methods.

Planned work includes improved reconstruction-error metrics, comparative evaluation of sampling algorithms, and SVM-based analysis of wavelet scalograms.

Changelog

See CHANGELOG.md for the release history.