Object-based mapping of native vegetation and para grass (Urochloa mutica) on a monsoonal wetland of Kakadu NP using a Landsat 5 TM Dry-season time series

A multi-temporal Landsat 5 TM time-series analysis significantly enhances the accuracy of mapping native wetland vegetation and para grass (Urochloa mutica) in Kakadu National Park, achieving an overall accuracy of 82% (kappa = 0.80) compared to single-date classifications. This approach improves segmentation quality and spectral separability, offering a reliable tool for strategic conservation efforts. The study demonstrates that integrating object-based classification with multi-temporal satellite data can effectively distinguish vegetation classes in monsoonal wetlands.

Boyden, J., Joyce, K. E., Boggs, G., & Wurm, P. (2013). Object-based mapping of native vegetation and para grass (Urochloa mutica) on a monsoonal wetland of Kakadu NP using a Landsat 5 TM Dry-season time series. Journal of Spatial Science, 58(1), 53–77. https://doi.org/10.1080/14498596.2012.759086