| dc.description.abstract | AmeriCorps is a United States (U.S.) federal agency that works with governmental and non-governmental organizations to serve communities across the U.S. AmeriCorps members assigned to locally based Conservation Corps organizations create more resilient greenspaces through habitat restoration, enhancing biodiversity, and reducing invasive species coverage. While the application of remote sensing to evaluate species distribution is well-demonstrated, it has not yet been utilized to assess outcomes of Conservation Corps projects. This gap coincides with an opportunity to leverage high-resolution satellite imagery through Planet’s Education and Research (E&R) program, to produce fine-scale spatially comprehensive assessments of conservation outcomes. The goal of this study was to assess a remote sensing-based approach to evaluating work completed by Conservation Corps groups. Using PlanetScope imagery, our specific objectives were to: 1) assess the extent and distribution of species targeted for invasive species removal and fuel reduction programs at eight Conservation Corps project sites across the United States, 2) spatially and temporally evaluate the impact of Conservation Corps projects by analyzing changes in target species coverage following on-the-ground, site-specific removal treatments, and 3) develop standardized protocols for remote sensing–based evaluations of future Conservation Corps projects and disseminate the methodology and findings through a publicly accessible ArcGIS StoryMap. The project utilized field-based data collected by each Conservation Corps program using KoboToolbox and 3-m PlanetScope imagery from Planet’s E&R program. The field data consisted of geolocated observations (40 to 83 observations per project site) of seven target species designated for removal by Conservation Corps crews: Russian olive (Eleagnus angustifolia), buckthorn (Rhamnus cathartica), honeysuckle (Lonicera maackii), musk thistle (Carduus nutans), Douglas fir (Pseudotsuga menziesii), lodgepole pine (Pinus contorta), Rio Grande cottonwood (Populus deltoides wislizeni), and Ponderosa pine (Pinus ponderosa). A total of 30 predictors were extracted from PlanetScope imagery acquired between June 2024 and December 2025, and combined with field-based observations to train and evaluate random forest (RF) models. Trained, site-level RF models were used to generate 3 m resolution species distribution maps for pre- and post-treatment conditions and change detection analyses were carried out to determine the difference in invasive species coverage and fuel coverage following treatment. Overall accuracies ranged from 98.45% to 100% (kappa 0.97 to 1), with the best accuracy for detecting vegetation in New Mexico and musk thistle in Wyoming (OA: 100%, kappa: 1) and the lowest accuracies for detecting honeysuckle in Iowa (OA: 98:45%, kappa: 0.97). All programs showed a decrease in the distribution of their target species within their project areas, although the size of the change varied between sites. A protocol was developed to support reproducibility of this methodology, and the study was publicly communicated through an ArcGIS StoryMap. The StoryMap was designed to present an overview and findings from this work, including the background, methodology, results and implications, in an open and accessible format for current and future researchers as well as the general public. This evaluation builds evidence for how land management organizations can utilize satellite imagery to better understand the conservation outcomes of their work. It also demonstrates an effective approach for communicating these results through an ArcGIS StoryMap, illustrating how project outcomes can be presented in clear and visually engaging format to effectively support understanding for both technical and general audiences. | en_US |
