An automated algorithm for thresholding hot pixels in AVHRR data is pr
esented. The algorithm, applied to cloud-free sub-images of channel 3
minus channel 4 brightness temperature, compares each target pixel wit
h its immediate background and then compares this difference with the
natural variation in the surrounding region. Application to images of
Mount Etna identified thermally anomalous pixels containing active lav
a or vents.