The legacy of the Tasseled Cap Transform: A development of a more robust Kauth -Thomas Transform derivation.

摘要

As remote sensing sensors and data systems become more complex, the need for a simple, meaningful index becomes essential in applications. The Tasseled Cap Transform (TCT) is a practical vegetative index and spectral enhancement and has been adapted to other multispectral sensors; most notability, the TM, ETM+ and ASTER. Unfortunately, there have been modifications to the Tasseled Cap derivation presented in literature that have created confusion regarding the transform and the proper orientation of the defined feature space.;For this research, the Tasseled Cap-like Transform coefficients have been derived for DigitalGlobe s QuickBird 2 11-bit, DRA-off product; Terra Advanced Spaceborne Thermal Emission and Reflection Radiometer Level 1 B raw counts product as well as the at-sensor radiance and at-sensor reflectance (exoatmospheric) products. During the derivation of these coefficients, different techniques were used and compared. The Gram-Schmidt Orthogonalizal Process was determined to be the most sensitive in preserving the shape of the Tasseled Cap space as well as intuitive in application.;The resulting Tasseled Cap-space is defined by specific directions of the coordinate axes. These directions and the overall shape of the space are typically the result of bandwidth selection of the sensor capturing the reflected electromagnetic energy. For most multispectral the "land-bands" are most common and each sensor in general, has bandwidths comparable to those contained in Landsat-class sensors. The results are similarly shaped Tasseled Cap-space and the lower-order coordinate axes having similar directions to specific phenomena (i.e. soil brightness and healthy vegetation).;Conclusions are presented that highlight the need for a better understanding of the Kauth-Thomas Transform by the user community. Detail must be given the sensor and the data model (units) before the Tasseled Cap coefficients are properly applied.;The future of the Kauth-Thomas Transform lay with continued development of the Tasseled Cap-like derivation for new sensors and incorporation of hyperspectral datasets. Both will require methods to eliminate spectral bands and use only those bands similar to Landsat-class sensors. Likewise, the Kauth-Thomas Transform may be used to discover new coordinate axes that do not contain the directions of the Tasseled Cap-like space.