Using crowd sourcing and geopositioned images to document near real time rangeland condition

Abstract

Monitoring rangeland vegetation is essential for sustainable land management, biodiversity conservation, and climate change mitigation. Traditional vegetation monitoring methods often require extensive fieldwork, which can be time-consuming and costly. Crowdsourcing, which leverages the collective power of a large number of volunteers, offers a promising alternative. This study examines the effectiveness of using crowdsourcing to collect geopositioned images for monitoring rangeland vegetation. By engaging herders and pastoralists with smartphones featuring built-in GPS capabilities, a substantial dataset of geotagged photographs from diverse rangelands was amassed. These images were transferred to a central repository when an internet connection was available, ensuring continuous data flow from even the most remote areas. Subsequently, the images were analysed using advanced image processing and machine learning techniques to assess vegetation and ground cover in near real time. Preliminary results indicate that our protocol can provide high spatial and temporal resolution imagery, which complements traditional monitoring methods by offering more immediate and detailed insights. These images also serve as ground truth for supervised classification of large-scale remote sensing satellite scenes. Additionally, this approach enables sampling of inaccessible remote areas while promoting community engagement and environmental awareness among pastoral communities. The necessary steps for implementation are discussed, along with examples from various locations. The findings highlight the potential of crowdsourcing as a cost-effective and scalable tool for rangeland monitoring and management, showcasing its ability to enhance both data quality and stakeholder participation.