Mapping wattle rust induced defoliation within South Africa’s wattle commercial forest plantations over two seasons

Effective suppressive measures of pest and pathogen outbreaks require effective monitoring efforts. Therefore, this study used two Landsat 8 images to detect and map wattle rust induced defoliation within black wattle commercial forest plantations across the KwaZulu-Natal and Mpumalanga provinces of South Africa. To our knowledge, this was the first study to tested the utility of remote sensing to detect and map levels of wattle rust induced defoliation on South Africa’s wattle plantations over two seasons. Therefore, this study integrated the Gradient Boosting Machine approach with Landsat 8 data over the summer and spring seasons to detect and map wattle rust induced defoliation. The results indicated that wattle rust occurrence was effectively detected and mapped with an overall classification accuracy of 69% and 72% for March of 2015 and November of 2015, respectively. The study also found the blue, ultra-blue, SWIR2 and NIR spectral bands to be the most significant in model development. In addition, TWI, NDVI and SAVI were the most influential vegetation indices in model development. Overall, the study illustrated the effectiveness of remote sensing in detecting and mapping wattle rust induced defoliation, which can be used as an effective suppressive measure to curtail further impacts.