Nebraska Invasive Species Mapping

Mapping is an integral part of management for any land manager. Mapping allows for a means to record a species temporally and spacially. While its one of many land management tools, its important to know that while a particular plant species can be mapped to particular area, they may not be consistently found for that transect. Continous monitoring aids mapping efforts and give valuable information to the land manager.

Links

The Nebraska Weed Control Association's state-wide mapping project is nearing its final stages in 2007. This unique system will allow for land managers and county agencies to map, monitor, and control invasive plants withing the State of Nebraska and keep everything standardized for all partners in Nebraska concerned with invasive plants .

NORTH PLATTE RIVER REMOTE SENSING PROJECT

Research conducted by Sunil Narumalani and Deepak Mishra (CALMIT) and Panhandle Research and Extension Center - Robert Wilson, Ann Koehle, and Patrick Reece

Objective:
The objective of our research was to quantify and map four dominant invasive plant species spread along the floodplain of the North Platte River, Nebraska using airborne hyperspectral remote sensing data and geographic information technology. Emphasis was given to four of the most common invasive species including Saltcedar (Tamarix sp.), Russian Olive (Elaeagnus angustifolia), Canada Thistle (Cirsium arvense), and Musk Thistle (Carduus nutans)along the flood plain of North Platte river corridor within 1 mile buffer. This research was aimed to protect the natural resources of Nebraska from the impacts of non-indigenous species, and help practitioners manage and mitigate the effects of established and potential non-indigenous species.

In September 26-27, 2005 an in situ investigation using a Trimble Pro-XRS (Trimble Navigation, Inc., Sunnyvale, CA) differentially corrected Global Positioning Systems (DGPS) and video/digital photography was performed. Several areas with dense invasive species coverage were accessed, and few patches occupied by homogeneous or relatively pure saltcedar and Russian olive species were delineated for classifier training (Figure 2). Musk Thistle and Canadian Thistle we found intermixed with each other and sometime with Reed Canary Grass along the wetland borders. Several GPS points of Cottonwood, Ash, Willow, and Water Cress plants were also collected during the field survey. However, the results revealed that most of these patches were smaller than 900 m2 and the LANDSAT data would not yield an accurate analysis because its spatial resolution is 30 meters (Figure 2). Consequently, alternative airborne data were sought to enable the mapping of the invasive species distribution along the river corridor. After the processing and classification of the airborne imagery, another field survey was performed on 1-3 March 2006 using DGPS equipment and video equipment to obtain accurate locational data for accuracy assessment of the classified maps. 223 GPS points were recorded for validation of the classified image resulting from spectral angle mapping.

An aerial remote sensing platform for hyperspectral data collection was used for this investigation. The instrument array included an AISA Eagle hyperspectral imager from Visible to Near Infrared (VNIR), a system which can provide high spatial and spectral resolution. AISA Eagle data used for the present study were acquired on 13 April 2005. AISA imagery for the entire study area was captured in 22 flight lines (strips) (Figure 3). Ground data indicated low wind (~ 3 m s-1), high visibility (30 km), and clear skies. The sensor altitude was (2.073 km), and the image was acquired at nadir at a spatial resolution of 1.5 m and spectral resolution of 62 bands. The spectral bands ranged from 392.39 to 981.68 nm with a 12 bit radiometric output. The image data were georectified and converted to at-platform radiance by applying the calibration coefficients provided by AISA Eagle processing software ‘Caligeo’ for subsequent processing.

The data processing technique utilized Spectral Angle Mapping (SAM) on the minimum noise fraction (MNF) transform of the 62 band AISA Eagle image. MNF determines the inherent dimensionality, identify and segregate noise in the image, and reduce the computational requirements for subsequent processing (Boardman and Kruse, 1994). It is essentially two cascaded principal components transformations. The first transformation, based on an estimated noise covariance matrix, decorrelates and rescales the noise in the data. This results in transformed data where the noise has unit variance and no band-to-band correlations. The second transform is a standard principal component transformation of the noise-whitened data. For further spectral processing, the inherent dimensionality of the data is determined by examining the final eigenvalues and the associated images. Based on the MNF output graph of eigenvalues and visually inspecting the output components, the first ten MNF bands were selected for SAM. SAM is a physically-based spectral classification that uses n-dimensional (n = number of bands) angles to match pixels to reference spectra (Kurse et al. 1993). The algorithm determines the similarity between the two spectra by calculating the angle and treating them as vectors in n-dimensional space. Smaller angles represent closer matches to the reference spectrum. Conversely, pixels that fall further away than the specified maximum angle threshold are not classified. The maximum acceptable angle between a endmember spectrum vector and a pixel vector (in number-of-bands dimensional space) was kept as 0.1 radians. Endmember spectra used by SAM were derived by overlapping the ground data on the MNF image. The selected fourteen endmembers were Saltcedar, Russian Olive, Canadian Thistle, Musk Thistle, Reed Canary Grass, Cottonwood, Willow, Ash, Water Cress, Wetland, Water, Sand, Cropland, and Roads.

Analysis is being conducted farther down the Platte River from Kingsley Dam to the city of North Platte. Stay tuned for an update on this stage of the project.

NOTE: Flight Line images below are in high resolution format and may take a little time to load on your screen.