Uncertainties in MODIS-Derived <i>Ulva Prolifera</i> Amounts in the Yellow Sea: A Systematic Evaluation Using Sentinel-2/MSI Observations
Lin Qi, Menghua Wang, Chuanmin Hu
Abstract
Uncertainties are an integral part of remote sensing data products in order to quantify changes, yet due to patchiness and spatial heterogeneity, it is difficult to use field measurements to estimate uncertainties in the satellite-derived Ulva prolifera (U. prolifera, also called green tides) amounts in the Yellow Sea. This is perhaps why such estimates are missing in nearly all remote sensing literature on U. prolifera mapping. Here, by comparing all available data collected by the Moderate Resolution Imaging Spectroradiometer (MODIS) on the Terra/Aqua satellites and the MultiSpectral Instrument (MSI) on the Sentinel-2A/ 2B satellites for the period of 2015–2022, we evaluate uncertainties in the MODIS-derived U. prolifera amounts. The relative uncertainties are found to decrease with increasing Ulva amounts in individual images, ranging from 58.8% for Ulva areal coverage (after pixel unmixing) of < 50 km2 to 8.7% for Ulva areal coverage of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$>$ </tex-math></inline-formula> 200 km2. Such uncertainties decrease in the monthly composite data products because of the increased number of observations, reducing to 3% in the total Ulva amount during the peak months. Such uncertainty estimates, in relative terms, are expected to serve as a reference when interpreting temporal changes in long-term Ulva estimates derived from satellite data.