Study Background and Methodology
Thapa’s study concentrated on three ecologically important coastal areas: the Mobile Tensaw River Delta, the Bon Secour National Wildlife Refuge, and the Mississippi Sandhill Crane National Wildlife Refuge. These regions have witnessed the establishment of both Tallow tree and Chinese privet.
Thapa leveraged ENVI to conduct image classification using three distinct methods: ISODATA, Maximum Likelihood (ML), and Random Forest (RF), representing unsupervised, supervised, and machine learning techniques, respectively. The process began with an examination of a 1-meter National Agriculture Imagery Program (NAIP) orthoimage, further refined with vegetation structure and topography parameters derived from LiDAR data.
ENVI was pivotal in interpreting and synthesizing this multilayered data, transforming it into actionable insights for invasive species management. “ENVI has a user-friendly interface and a wide array of processing functions,” said Thapa. The software's capabilities include image classification, accuracy assessment, post-classification, user-defined filters, and the stacking of bands and variables derived from remote sensing data according to the user's interest. "In other words," she notes, "it has the potential to combine LiDAR, radar, optical, thermal, multispectral, and hyperspectral imagery."