LAUSR

ABSTRACT Due to the weak information about cultural targets in the complex marine environment, an omission problem exists in large-scale mariculture extraction using single-view and single-source images. To overcome this problem, we developed a mariculture extraction method that combines dense time-series Sentinel-2 and Sentinel-1 data. A high-precision Chinese mariculture distribution map for 2020 was produced with an overall accuracy of 94.00% and a kappa coefficient of 0.91. The results show that (1) the total area of mariculture was 1173249.22 ha on the national scale, which was significantly larger than the previous studies (459595.70 and 205920.28 ha, respectively), with Shandong Province (39.09%) having the largest proportion; (2) China’s mariculture presented a spatial distribution characteristic of ‘Denser North and Sparser South’, and mariculture was centralized in the coastal zones of the northern provinces (60.76%) rather than the southern provinces; (3) the official production statistics and remote sensing-derived mariculture area revealed a highly corresponding trend at the provincial level, with an R2 reaching 0.78, which is much higher than the 0.07 and 0.41 values of the comparison data. The results directly provide data reference for mariculture production estimation and site selection or ideas for mariculture extraction in other regions and globally.