Monitoring Narrow Mangrove Stands in Baja California Sur, Mexico Using Linear Spectral Unmixing

摘要

Small stands of mangrove trees are difficult to detect and monitor using satellite remote sensing because the width of the narrow strips of vegetation are typically much smaller than the spatial resolution of the imagery. Every mangrove pixel also contains water and bare soil reflectance. Linear spectral unmixing, which estimates the fractional presence of specific land cover types per pixel, was performed on Landsat 8 imagery to detect mangroves on the eastern shoreline of the Bay of La Paz on the Baja California Peninsula of Mexico. Low-altitude aerial imagery collected from a DJI Mavic Pro drone was used as ground-reference data in the accuracy assessment. Continuous fractional presence of mangroves was detected with 80% accuracy and 85% of mangrove area was found. Future work will use linear spectral unmixing to systematically monitor mangrove extent and health in the region relative to expected growth in tourism development.[GRAPHICS].