The 2008 Joint Meeting of the Society for Range Management and the America Forage and Grassland Council. |
Monday, January 28, 2008 - 2:40 PM
A New Remote Sensing System for Assessing Seeded Pasture Condition in Western Canada
Duane McCartney, Research Branch, Agriculture and Agri Food Canada, 6000 C and E Trail, Lacombe, AB T4L1W1, Canada and John Polson, Polson Environmental, 2001 St Henry Street, Saskatoon, SK S7M0P4, Canada.
A new system has been developed to assess Aspen Parkland pasture condition using enhanced Landsat satellite data. Forage yield samples were obtained at various locations at the Pathlow Community pasture in north eastern Saskatchewan and the GPS co ordinates were located on the satellite image. A buffer strip was created around the sampling points of 1 pixel in radius and allowed for the inability to locate the exact yield location on the image. A series of digital enhancements were developed to find enhancements that allowed for the best separation of a range of biomass groups. The ability of various enhancements and band combinations to reflect the biomass were determined by plot clipping included a) a visual look at the image, b) classification of the image and c) an analysis of the reflectance values of the image data bounded by the plot point buffers. The original bands, and band 5, 4, 3, 2, resulting from the Taylor enhancement algorithm, and NDVI enhancements were used in various combinations. The more promising colour combinations were classified using an Unsupervised Classifier. Biomass point data was then used to determine which classes represented which biomass. This process will still need to be refined in future research efforts but this basic system provides a computer satellite method for land managers to monitor large pasture areas in western Canada.