Recently, a large
numbers of coast live oak (Quercus Agrifolia), Tanoak (Lithocarpus
densiflorus) and black oak (Q. kelloggii) trees have died in coastal
areas of California, USA. This
epidemic is characterized by a distinct set of symptoms, and has been termed;
“Sudden Oak Death”(SOD). In
addition to a suite of symptoms found on the trunks of infected trees, the
foliage of infected trees appears to die rapidly, changing color from dark
green to reddish-brown within a few weeks. Oak leaf samples were picked up from different positions of
crowns in the different seasons (later October, 2000, September, 2001, and
April and July, 2002) at China Camp State park, California. These samples were carried to CAMFER
lab, UC Berkeley to take spectometer measurements with an ASD full range
spectrometer (covering 350 – 2500 nm).
At same time, the relative water content (RWC) was also taken
immediately after spectral measurement for each sample. With those spectral and RWC data sets,
a study on relationship between spectrum and corresponding RWC, spectral
differentiation between two health levels: healthy and infected, and seasonal
change of the relation and spectral difference as well was conducted. A plan of working on SOD with multi/hyperspectral
image data is being made. The
major work analyzed with the image data sets will be related to determining and
mapping abundance of different health components: healthy and heavy infected /
newly dead oak trees or directly labeling a pixel as one health level, either
healthy or heavy/newly dead using some classifiers such as neural network
algorithm and penalized discriminant analysis etc.
Please go back my main homepage