Using Sentinel-2 to Detect and Predict Biodiversity Hotspots within a Vernal Pools and Grasslands Ecosystem

Abstract:

Sentinel-2 has emerged in recent years as a powerful and low-cost tool for studying temporal changes in plant cover and biodiversity. However, few studies have attempted to use satellite remote sensing to evaluate biodiversity in seasonal isolated wetlands, such as vernal pools, which change rapidly over short temporal scales. California’s vernal pools are hotspots of native and endemic diversity, and millions of dollars are spent annually to restore degraded grasslands across the state. Through integration of Sentinel-2 remote sensing products, microtopography from UAV-LiDAR, and hydrologic data, it may be possible to produce predictive models that can inform management actions that support conservation of biodiversity, such as grazing, fire and water management. The objectives of this study are 1) to test the feasibility of using a Sentinel-2 spectral species time series to monitor and detect biodiversity hotspots and changes within the University of California Merced Vernal Pools and Grasslands Reserve and 2) to incorporate a variety of derived Sentinel-2 products, UAV-LiDAR data and hydrologic data into a species distribution model to evaluate plant biodiversity within the Reserve. Sentinel-2 Level 2A imagery from December 2017 to July 2018 (51 images) and December 2018 to July 2019 (44 images) were analyzed for changes in plant biodiversity and ET through green-up to brown-down. UAV-LiDAR data (25 cm resolution) was acquired from November 2019 to March 2020, during low biomass season. Precipitation data was collected at an onsite meteorological station. Biodiversity products were validated against ecological data derived from 132 community structure plots in 2019 and 171 community structure plots in 2018. Results indicate there is moderate correlation between satellite-derived spectral diversity and plant biodiversity that is statistically significant (p < 0.05), demonstrating the potential for Sentinel-2 to monitor plant biodiversity in mixed grassland/wetland ecosystems. Inclusion of these products from Sentinel-2 into a species distribution model shows potential as a biodiversity predictor.