What question did this study set out to answer?

The aim is to develop a framework for detecting anomalies in vegetation recovery after wildfires using remote sensing data.

April 19, 2026Open Access

TerraHeal: Post-Wildfire Vegetation Recovery Anomaly Detection Using Sentinel-2 Time-Series and Isolation Forest: A Google Earth Engine Framework Applied to California Wildfires

Key Points

The aim is to develop a framework for detecting anomalies in vegetation recovery after wildfires using remote sensing data.
Utilizes Sentinel-2 imagery for multi-temporal NDVI analysis.
Employs Isolation Forest for anomaly detection based on remote sensing data.
Implements a persistence verification step to differentiate between transient delays and sustained failures.
Framework executed in Google Earth Engine, focusing on major wildfire events.
Successfully identified localized areas of delayed vegetation recovery following wildfires.
Demonstrated effective detection of recovery cold spots using the proposed methods.

Abstract

This paper presents TerraHeal, an automated framework for detecting persistent post-wildfire vegetation recovery anomalies ("recovery cold spots") using Sentinel-2 imagery and machine learning. The framework combines multi-temporal NDVI analysis with heuristic thresholding and Isolation Forest-based anomaly detection. A persistence verification step distinguishes transient recovery delays from sustained failures. The system is implemented using Google Earth Engine and applied to major wildfire events, demonstrating the ability to identify localized areas of delayed vegetation recovery. This is a preliminary preprint version. Results are based on remote sensing analysis and require further validation with field data.

Read Full Paperexternally

Mark Helpful

Bookmark

Relay

View Full Paper