Hyperspectral Signal Processing Applications in Environmental Monitoring Identification and Mapping of the Invasive Plant Species Purple Loosestrife

The invasive plant Purple Loosestrife (Lythrum salicaria) is considered a threat to numerous ecosystems across the United States. It was introduced from Europe to North America in the 1800s. Now known as the "purple plague", it has spread across 48 of 50 states in the US and all provinces in Canada. The U.S. Fish and Wildlife Service declared Purple Loosestrife "Public Enemy #1 on Federal Lands", and The Nature Conservancy listed it as the 2nd most troublesome weed in wetlands. Monitoring and mapping of Purple Loosestrife is a challenge due to the difficulty of access to the vast, remote and most often privately owned marsh lands of New York State. This paper presents the results of a recent study of hyperspectral signal processing for the identification, classifying and mapping of Purple Loosestrife. A set of combined methods of signal processing were used including: geographic calibration, atmospheric correction, minimum noise fraction, pixel purity index, regions of interesting locating, hyperspectral library creation and spectral angle mapper classification. The classification accuracy is significantly increased (91.7%) due to the application of the self-built hyperspectral library STIHL