ROSES Proposals

Both natural and man-made floating matters have been reported in the oceans and lakes, include floating macroalgae (Sargassum fluitans/natans, Sargassum horneri, Ulva prolifera), microalgae scums (Microcystis, Trichodesmium, Noctiluca), marine mucilage (a.k.a. sea snot), pollens, and marine debris. This project will combine the hyperspectral PACE/OCI and computer artificial intelligence (through feature-based deep learning) to detect and map these various floating matters, with three specific objectives: 1) to develop a unified hybrid approach/algorithm to map, classify, and quantify the various types of floating matters; 2) to generate new PACE/OCI data products for both floating matters, as well as a data quality flag; 3) to serve the new data products in near real-time to user groups for informed decision making. In addition, a biproduct of this effort is a new quality control flag to label image pixels containing floating matters. The combination of hyperspectral and machine learning capacity is expected to overcome the difficulty in discriminating floating matter types as well as in distinguishing floating matters from other confusing image features such as clouds, cloud shadow, sun glint, ship wakes, and optically shallow bottom.

Postdocs: Yingjun Zhang and Junnan Jiao, University of South Florida