Forest Change Detection — LandTrendr

Time-series analysis of satellite imagery (Landsat / Sentinel-2 / HLS) for forest disturbance detection. Methodology: Kennedy et al. 2010, 2018.

Analysis parameters

Area of interest (AOI)

Draw a polygon or rectangle on the map. Maximum area: 500,000 ha.

Zoom to level ≥9 to start drawing
Satellite system and index
HLS recommended for the most complete coverage (from 2013)
NBR is most sensitive to forest disturbances
Time range
Minimum 5 years. Landsat: from 1984. Sentinel-2: from 2017. HLS: from 2013.
Advanced parameters
Pixels below the threshold are excluded (Hansen GFW 2000)
Higher value → fewer false detections, lower sensitivity

Running LandTrendr analysis on GEE…
Typically 30–120 s depending on area and time range

AOI:

Analysis results

Detailed statistics by year

Year Area (ha) 95% CI (ha) Pixel count Confidence (%)

Note: the confidence interval is estimated as a function of mean LandTrendr confidence. A higher CI corresponds to lower segmentation confidence.