Soundscapes to Landscapes (S2L): Monitoring Animal Biodiversity from Space Using Citizen Scientists

Soundscapes to Landscapes (S2L) is a citizen science project with the broad goal of advancing scientific understanding of biodiversity and conservation using a combination of new and existing spaceborne sensors. The prototype phase of this project is focused on modeling bird diversity at the watershed scale in Sonoma County, California. The main objectives are to: 1) involve citizen scientists in the collection of in situ field information on bird diversity; 2) assess the accuracy and precision of bioacoustics for the detection and monitoring of individual species and richness; 3) test the predictive strength of spaceborne imaging spectroscopy, lidar, synthetic aperture radar (SAR) sensors for spatial modeling of species occupancy and overall species richness; and 4) use occupancy/richness maps to better understand factors related to conserving animal diversity. In the prototype field campaign, citizen scientists (birders, undergraduate students) deployed portable sound recorders for 3-4 days in various habitats in a local watershed. Over 50,200 minutes (866 hours) of "soundscape" recordings were collected over 3 months. Through a series of "bird blitzes", citizen scientists used the ARBIMON II web-based system to listen to spectrograms (time vs. frequency) of 1-minute recordings, tag bird calls for identifiable species, validate presence/absence of bird species, draw training-set boxes around well-formed calls, and help evaluate Random Forests machine-learning model performance. Bird-call models were applied to all soundscapes to identify presence/absence of 10 indicator species. Another phase of this project involves species distribution modeling in conjunction with C- and L-band SAR imagery, simulated Hyperspectral Infrared Imager (HyspIRI) and Global Ecosystem Dynamics Investigation (GEDI) large-footprint lidar data. Metrics derived from these data provide unique, wall-to-wall information on vegetation chemistry (HyspIRI) and three-dimensional structure (GEDI, SAR) that are linked with soundscape-derived bird data to model species occupancy and overall richness at the landscape scale.