Assessing Global Landsat and Sentinel-2 Derived Land Cover Datasets Using GLOBE Observer Land Cover Photo Classification Protocols

Keywords: citizen science, land cover, sampling protocol, multiview, remote sensing
Abstract Type: Paper Abstract

Abstract

An ever-increasing number of geo-tagged field photographs are being collected and shared with the public, creating an exciting opportunity for human and environmental studies at multiple scales. GLOBE Observer (GO) is an international citizen science project with numerous subprojects covering various types and scales of earth observations. GO Land Cover has the most participants and the broadest scope of any citizen science project. Users collect one set of land cover observations by taking six directional photographs (north, south, east, south, upward, and downward). This sampling protocol accurately reflects the local land cover types and is unique. In addition, the GO land cover has significantly increased the spatial extent and sampling density of the measurements, which would require a great deal of time for an individual to replicate the environment. The land cover maps derived from remotely sensed satellite imagery may contain a significant amount of error. GO data can be used as a validation instrument for evaluating and controlling remote sensing-derived land cover measurements. In this study, we will create an evaluating network based on the GO mapping protocol and incorporate the networks into remotely sensed datasets (e.g., Landsat and Sentinel-2). Consequently, we will be able to evaluate remotely sensed datasets on a global scale. This study will begin to answer the question of how citizen science derived data can be used for research purposes and how those photos should be used to reestablish a robust citizen science field photo processing pipeline that can be applied to other use cases.