Extraction and spatio-temporal distribution of raft aquaculture areas based on U-Net deep learning model

Effective monitoring of the spatial and temporal distribution changes of mariculture areas is of great significance for the dynamic marine supervision, ecological environment protection and disaster prevention. In this paper, the remote sensing image datas of ALOS and GF-6 are selected, and the regional network adjustment method is adopted to effectively solve the migration problem of offshore aquaculture area. The distribution information of Rongcheng raft aquaculture area is obtained through U-Net deep learning model. Moreover, the spatial and temporal change characteristics of aquaculture area and the spatial expansion mode are deeply analyzed based on the center of gravity transfer model and convex shell principle. The results show that from 2009 to 2023, the raft aquaculture area increased by 10,224.58 hm2 , and the annual average change rate of the area was 3.84%; 75% of the raft culture in the research area shifted in the south-west by 11,743.22 meters. Based on the spatial distribution of the raft aquaculture areas in2023, there occupied some port shipping areas, industrial and urban use areas, marine protection areas, special utilization areas and reserve areas.