DID YOU KNOW? DTIC has over 3.5 million final reports on DoD funded research, development, test, and evaluation activities available to our registered users. Click HERE
to register or log in.
Optimal LANDSAT Transforms for Forest Applications,
JET PROPULSION LAB PASADENA CA
In agricultural applications of remote sensing, linear transforms of Landsat data, such as those of Kauth and Thomas, are known to be highly effective both for data compression and enhancement of crop identification accuracies. Typically, such transforms are based on the time-trajectory of crop pixels through measurement space as the crop increasingly obscures the soil, matures, scenesces, and is harvested. In natural vegetation applications, temporal variations are less important-- life-form differences among vegetation types lead to distinctive signatures for natural vegetation types that are more or less distinctive, independent of season. However, the signatures of natural vegetation types are greatly influenced by their topographic position on the landscape, due to factors of differential illumination and complex bidirectional reflectance distribution functions. Thus, the question arises whether there are one or more transforms of Landsat data, beyond those already explored, that can accentuate the separability of natural vegetation classes in areas of diverse topographic relief. To answer this question, we investigated eleven transforms of four Landsat MSS channels. These transforms were evaluated for their ability to distinguish among thirteen classes of natural vegetation in a small area of the Klamath Mountains in northern California, USA.
This article is from 'Papers Selected for Presentation at the International Symposium on Remote Sensing of Environment (16th) Held at Buenos Aires, Argentina on 2-9 June 1982. Volume 1,' AD-A134 719. p455-468.