Coastal land cover mapping using UAV imaging on the southeast coast of Brazil

Author(s)

Bispo dos Santos, Gabriel Almeida ; Conti, Luis Americo

• Document type: Journal article
• Source: JOURNAL OF COASTAL CONSERVATION; v. 26, n. 5, p. 10-pg., 2022-10-01.

Abstract

The use of Object Based Analysis (OBIA) from drone images (Unmanned Aerial vehicles - UAVs) has been increasingly employed in land use mapping, specifically in coastal areas, regions of high dynamics and complexity. Such approaches can be very useful as they allow the elaboration of high spatial resolution maps and multitemporal analysis. However, few studies have been conducted in order to evaluate different classification methods and establish best practices for the use of these applications with a standardized and comparable way. In this work four supervised classification methods were analysed (Logistic Regression, Decision Forest, Decision Jungle and Neural Networks); in addition, we have analysed the role of different parameters (such as texture and shape indices) in the classification. The results may significantly contribute to the dissemination of the use of UAVs for coastal areas mapping, specifically for monitoring and management as well as in the detection of areas vulnerable to erosion and in the delimitation of regions sensitive to environmental changes. (AU)

• FAPESP's process: 19/22028-0 - Multihab - multispectral unmanned aereal vehicle imaging for coastal habitat characterization and monitoring
• Grantee: Luis Americo Conti
• Support Opportunities: Research Program on Global Climate Change - Regular Grants