An Analysis of the Marginal Value of Hyperspectral Features of the Mixed Pixel of Lotus Leaf and Water Body.

A linear regression and curve fitting method is used to study the mixed pixel of lotus leaf and water body under the experimental conditions. The marginal value of hyperspectral features of vegetation is discussed. Through an analysis of the regression result and fitted equation, a conclusion is reached as follows: Until the area ratio of the lotus leaf is 5.9 %, the vegetation reflection features cannot be observed even factually there are some lotus leaves in the mixed pixel; when it is 10.7 %, the reflection spectrum curve saturates and the water body reflection features cannot been detected. The result indicates that although the area ratio of the lotus changes the reflection features may not change. This research will help to understand the mixed pixel and the liner spectrum unmixing technique further. [ABSTRACT FROM AUTHOR]