A Remote Sensing Perspective: Classifying the Human Footprint in the Zambezi Region Using Random Forest and Object-Based Classification Methods

Authors: Ariel Weaver*, University of Louisville
Topics: Remote Sensing, Africa, Coupled Human and Natural Systems
Keywords: Random Forest, OBIA, Object-Based Image Analysis, Africa, Human Footprint, Remote Sensing, Zambezi Region, LULCC, Conservation, Development, Namibia, Settlements, Cropland, Agriculture, Zambezi Region, KAZA, Kavango Zambezi Transfrontier Conservation Area
Session Type: Paper
Day: 4/3/2019
Start / End Time: 12:40 PM / 2:20 PM
Room: Washington 2, Marriott, Exhibition Level
Presentation File: No File Uploaded

Land use and land cover change is an important arena of research in remote sensing, particularly in terms of understanding the impacts of human-induced environmental change (Turner et al. 2011). An analysis of the “human footprint” (or the extent of land areas impacted by human-use) can be used as to estimate human activities across the landscape and to assess conservation metrics, such as habitat quality, connectivity, etc. (Leitoa & Ahern 2002). Computer vision and machine learning techniques have made great strides over the last several decades, and gives hope for processing larger data-sets with greater ease and improved accuracy. Object Based Image Analysis (OBIA) (Blaschke 2010) and Random Forest (RF) (Breiman 2011) have become more widely known for their ability to improve classification accuracies of remotely sensed imagery (Breiman 2011, Pal 2005, Liaw & Wiener 2002). This study attempts to map the human footprint for the eastern portion of the Zambezi Region of Namibia (Zambezi East), which is centrally located in a major southern African savanna wildlife corridor. The land covers in the Zambezi Region are characteristically spectrally difficult to separate, due to overlapping spectral signatures across the highly heterogeneous savanna landscape (Wingate et al. 2017). Specifically, this study investigates the efficacy of utilizing OBIA, RF, and a hybrid Object-Based Random Forest (RF-OBIA) approach to classify built and fallow agricultural land during the dry season of 2017 as a snapshot of the rural human footprint in the Zambezi Region.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login