Publications

Hu, CM; Qi, L; Hu, LB; Cui, TW; Xing, QG; He, MX; Wang, N; Xiao, YF; Sun, DY; Lu, YC; Yuan, C; Wu, MQ; Wang, CY; Chen, YL; Xu, HP; Sun, L; Guo, MH; Wang, MH (2023). Mapping Ulva prolifera green tides from space: A revisit on algorithm design and data products. INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION, 116, 103173.

Abstract
Since the first report in 2008, macroalgal blooms of Ulva prolifera (often called green tides) in the Yellow Sea have occurred every year, with their origins, transport pathways, temporal changes, as well as causes and consequences studied extensively. Of these studies, satellite remote sensing has been used widely to detect the bloom presence and quantify the bloom size (i.e., U. prolifera coverage in km2 or biomass in kilotons). However, substantial variability has been found in the refereed literature in the remote sensing methodology, results, and interpretation of the U. prolifera coverage, especially in the attempts to study inter-annual changes or long-term trends. There are often inconsistent or contradicting results even from the same satellite sensor. Such in-consistencies or contradictions create difficulty not only within the remote sensing community when presenting new methodology or results, but also to researchers when attempting to use the remote sensing results to make predictions or perform impact assessments. Here, we review the literature on the remote sensing methodology to detect and quantify U. prolifera blooms, and make recommendations based on physical principles. Specifically, we propose the following conceptual guidelines: 1) a reliable index or algorithm should be relatively tolerant to perturbations by non-optimal observing conditions (thick aerosols, thin clouds, moderate sun glint, cloud-adjacent straylight, which can all be found frequently in the study region) for presence/absence detection, as well as to small errors in the selected thresholds to quantify U. prolifera; 2) a reliable index or algorithm should also make it relatively easy to account for variability in subpixel coverage of U. prolifera (i.e., through pixel unmixing) in order to obtain an accurate estimate of total U. prolifera coverage from an image; 3) a reliable data product (i.e., U. prolifera maps) should be able to account for the variable clouds when interpreting spatial patterns or temporal changes, with uncertainty estimates provided whenever possible; and 4) both the algorithm and the data product should minimize manual work in order to make them more objective and repeatable by other researchers. Finally, we show different types of time series of U. prolifera amounts in the Yellow Sea using

DOI:
10.1016/j.jag.2022.103173

ISSN:
1872-826X