Multispectral imaging of land surface

An airborne spectroscopic image was flown over an intensively used agricultural area and over a location which had suffered from a forest fire. Many surfaces feature details could be indentified either from synthesized near true color images, from the normalized differential vegetation index (NDVI), from a principal component analysis or by treating the data with an unsupervised neural network. A comparison of the individual results demonstrates the strength of each method.