In the context of global biodiversity loss and the ongoing degradation of ecosystems, it is indisputable that people must take urgent actions on forest management and conservation. As the most biodiverse ecosystem on earth and the largest above-ground carbon reservoir, tropical forests play a vital role in adjusting the global climate and atmosphere, which makes understanding, monitoring, and predicting the spatiotemporal dynamics across this biome and exploring tropical ecosystems composition, structure, and function a high priority for mitigating and halting biodiversity loss. The main goal of this study is to track key plant functional traits (including chemical, morphological, and photosynthetic traits) across the tropics by combining with the high-resolution Sentinel-1 synthetic aperture radar (SAR) imagery, in-situ plot vegetation census data collected in six countries covered the four tropical continents, and other ancillary data (e.g. climatic and soil data), on the free-to-use Google Earth Engine cloud platform. Specifically, three vegetation indices and texture features were calculated and derived from SAR imagery. Then the machine learning algorithm Random Forest was applied to map and predict plant functional traits distributions across the tropics. Eventually, we conducted variable importance computation and accuracy assessment to analyse the potentialities of Sentinel-1 SAR data for large-scale biodiversity monitoring and ecosystem conservation. Our study aims to investigate the application of Sentinel-1 SAR imagery for mapping and predicting plant functional traits distributions across the tropical forest biome and to evaluate the potentialities of Sentinel-1 data for biodiversity monitoring and ecosystem conservation. To the best of our knowledge, we are the first to propose the method for mapping and predicting the distributions of plant functional traits using Sentinel-1 SAR imagery and assess the robustness of Sentinel-1 data in ecological applications.
Plant functional traits, Radar, Tropics, Google Earth Engine, Random forest