High-Resolution Data for Capturing Wetland Vegetation Using Object-Based Classification Methods

Abstract

The purpose of this study is to use the Unmanned Aerial Systems (UAS) images for mapping wetland vegetation using object-based classification methods and to compare its performance with cropland data layers. The UAS imagery (~0.1-m resolution) and National Agriculture Imagery Program (~0.6-m) data were used to extract wetland vegetation using object-based classification methods in ArcGIS Pro. Spectral indices, such as green chromatic coordinate (GCC) and normalized difference vegetation index (NDVI) coupled with unsupervised classification have been used for vegetation classification. UAS imagery performed slightly better than NAIP for classification yielding 49% vegetation in 2019, while it was 45% in 2018 and 35% in 2016 for NAIP classification. According to cropland data classification, open water land cover class also covered a large portion of the study area. In conclusion, object-based classification using high-resolution imagery has good potential to integrate with ground survey to implement best management practices for restoring wetlands.