Berkeley Indices Trajectory Extractor (BITE)

BITE derives disturbance maps from satellite image stacks

Berkeley Indices Trajectory Extractor (BITE) is an algorithm to derive disturbance maps from multi-temporal remote sensing image stacks. Particularly, BITE can distinguish persistent forest, slow-onset disturbances and rapid-onset disturbances. Slow-onset disturbances are usually caused by diseases and insects, which result in forest loss over a long period. Rapid-onset disturbances are abrupt loss of canopies that are usually caused by clearcut, logging, prescribed fires, wildfires and other natural disasters. BITE features a distinctive processing flow that requires almost no parameters for tuning but a training dataset to fit statistical learning models. The models are used to predict the features extracted from the trajectories, which are derived from segmenting time-series for multiple spectral indices. For each spectral index, a disturbance map can be derived, and by integrating these maps a final integrated disturbance map is produced via a plurality voting, thus is more accurate than any disturbance map of a single spectral index. BITE algorithm was tested to be resistant to data gaps (clouds/shadow/snow) and noises (haze, temporal fluctuation, minor misregistrations).
The algorithm is introduced in,
Chen, Y., Liang, L., Hawbaker, T.J., Gong, P., Biging, G.S., Zhu, Z., BITE: An algorithm for mapping slow-onset forest disturbances caused by mountain pine beetles with Landsat image stacks. Remote Sensing of Environment, submitted.
Read BITE_UserGuideV1.1.pdf for instructions.