Whitney Broussard JESCO, Inc.

14F

Unmanned Aircraft Systems (UAS) are capturing our imagination through high resolution aerial photography, producing imagery that resolves the finest details. Workflow development, however, from data capture to actionable information is proving difficult in both upland and coastal environments. Coastal wetlands, for example, are difficult to map as a result of elevation errors produced by Structure from Motion algorithms over water areas. Furthermore, these high-resolution datasets are prone to high spectral variability, which complicates pixel-based classification techniques. Here, we present two case studies to demonstrate the ability of UAS technology to collect hyperspatial (2-3 cm), multispectral (4 band) aerial images and to produce 2-dimensional orthomosaics and 2.5-dimensional digital surface models in two coastal wetland environments in southern Louisiana. The first is a 107-acre mitigation bank developed by the Louisiana Department of Wildlife and Fisheries. The second is an 818-acre marsh creation site funded by the Coastal Wetlands Planning, Protection and Restoration Act. We demonstrate workflows that utilize Object-Based Image Analysis (OBIA) techniques to first classify water features then use this value-added information to improve the intermittent point clouds and final elevation models of the site. We then use OBIA techniques and machine-learning algorithms to analyze the UAS-derived data stacks and create highly accurate, hyperspatial information products. We classified and mapped the dominant species, mapped the land-water interface, quantified the average plant height, produced a digital surface model of dredge spoil material over 1,000 acres, and calculated a Normalized Difference Vegetation Index (NDVI) surface of the project area. Model results were validated with on-the-ground surveys and compared with conventional monitoring products produced by the U.S. Geological Survey. When compared with conventional land-water interface models, we quantified overestimation of land and, as such, improved our understanding of the landscape and limitations of previous workflows. We suggest that these OBIA methods could be readily applied in multiple urban and coastal settings and could support other project operations and monitoring needs such as land use land cover mapping, flood mapping, disaster response, and long-term monitoring efforts for both wetland mitigation banks and coastal restoration projects.

14:15 Improving Elevation Models and Wetlands Mapping with OBIA and UAS Imagery in Coastal Environments, Whitney Broussard