Using Deep Learning <scp>AI</scp> to Detect Riparian Vegetation Recovery Following a Major Flood in the Brisbane River

ABSTRACT Post‐flood riparian vegetation recovery demands significant attention; however, the complexity of traditional remote sensing methods often hinders environmental managers from implementing rapid vegetation monitoring. This paper developed a model using a Deep Learning Model within ArcGIS to classify and detect recovery of vegetation with high resolution aerial imagery following the 2022 major flood of the Brisbane River at 4 sites: Colleges Crossing, Corinda, Kholo and Priors Pocket. The model was 91% accurate across all sites and was most accurate in post‐flood imagery and in the water and low vegetation classes at 100% for both. Low vegetation was found to recover at all sites except Priors Pocket whilst tall vegetation and bare ground did not. By leveraging this streamlined and efficient solution for monitoring post‐flood riparian vegetation recovery, damaged vegetation can be more effectively identified, allowing more precise and cost‐effective rehabilitation efforts. Future studies should further develop the model to increase accuracy and applicability to other catchments.