A Comparative Analysis of the Rangeland Analysis Platform and Drone Imagery for Woody Cover in Texas, USA

Geospatial technologies have become essential to assess ecological processes, quantify changes in land use, and monitor vegetation dynamics in rangelands. The integration of remote sensing and field-based data collection has led to the development of online platforms that use remote sensing to estimate vegetation cover, forage production, and biomass. One of the most popular platforms in the United States is the Rangeland Analysis Platform (RAP). While RAP provides biomass and cover estimates across the United States, spatial resolution may limit the accuracy of fractional cover. Drones have been successfully used to estimate biomass and vegetation cover at fine scales. As concerns about woody plant encroachment grow, these technologies offer new opportunities to quantify both the amount and spatial distribution of vegetation cover. Therefore, the goal of this study was to evaluate the correspondence between RAP estimates and drone imagery data for woody cover. We used data from the state of Texas due to the diversity in vegetation communities and geographies. We used drone-based imagery at multiple sites with the image resolution at less than 2cm and flight area ranging from around 30 to 120 ha. We conducted the classification, built the confusion matrix, and performed the accuracy assessment in ArcGIS Pro, and we performed the comparison of drone imagery to RAP woody cover estimates using multiple focal window sizes (3 x3, 5x5, 7x7, and 9 x9). In our initial results, the accuracy of our woody cover classifications from drone imagery ranged from 87 to 90 percent. We hypothesize that estimates from RAP and drone imagery will differ in woody cover correspondence depending on woody cover spatial configuration and focal window size.