Fine-scale Earths monitoring using a novel integration of multi-scale remote sensing observations

Abstract

<p>Our Earth has experienced rapid environmental changes tightly tied to anthropogenic activities. Satellite remote sensing offers a quantitative means to monitor such changes but is often limited to coarser spatial or temporal resolutions. Only very recently, the arrival of PlanetScope CubeSats, which provides daily-to-weekly observations at a 3-m spatial resolution and global coverage, offers an unprecedented opportunity for fine-scale Earths monitoring. However, several outstanding challenges remain with the CubeSat observations that further hinder its broad applications: 1) CubeSat observations source from over 170 satellite sensors with varying solar geometry, causing the data inconsistency issue across different sensors, 2) frequent clouds and cloud shadows often contaminate the satellite signal, and 3) the accurate biophysical interpretation of satellite signal remains lacking. To address the first challenge, we developed a rigorous method to cross-calibrate the CubeSat observations into the same level using a more stable single-sensor satelliteModerate Resolution Imaging Spectroradiometer (MODIS) as reference. To address the second issue, we developed a novel cloud/shadow screening method (STI-ACSS) which leveraged the spatial and temporal information of reflectance bands to enable automatic and accurate cloud and shadow screening. To address the third issue, we developed a spectral unmixing approach that effectively classified the forest canopy into leafy vs. leafless statuses, from which it would enable fine-scale accurate phenology monitoring in the tropical forests. Similarly, by integrating the proximate drone-survey data (with crown-scale tree segmentation) with CubeSat observations, we demonstrated the feasibility to monitor plant phenology at the individual tree-crown scale. Such an integration of multi-scale remote sensing observations shows the potential to facilitate the monitoring of Earths environmental changes, especially for those rapid and fine-scale changes.<br></p>