Abstract: Landscape patterns, dynamics, and processes are at the core of landscape ecology. The scientific analysis of dynamic changes in coastal landscape patterns plays an important role in regional development planning. This paper used land-use data from remote sensing products for the Guangxi Gulf of Tonkin Economic Zone (GGTEZ) from 2000, 2010, and 2020 as data sources. Then landscape pattern index and coefficient of variation were used to quantify the grain size effect of the landscape index at 30-1 000 m at the landscape level and class level. Information entropy models were elaborated to determine the optimal granularity. Further, the landscape index and the transfer matrix were used to analyze the dynamic changes in landscape structure in the GGTEZ between 2000, 2010, and 2020. The results show that:(1) At the landscape level and class level, landscape indices exhibited variable degrees of sensitivity to different grain sizes. The grain size effect of different landscape types was slightly different, and the optimum grain size for landscape pattern analysis was 60 m. (2) Forest was the main landscape type in the study area, and construction land was the fastest growing landscape type. Forest, agricultural land, and construction land areas changed dramatically due to human activities. (3) In the GGTEZ, the patch density increased, the landscape became fragmented and more complex, and the spatial heterogeneity increased. In the context of rapid social, economic, and tourist industry development, the landscape patterns of the GGTEZ changed at an accelerated pace. Therefore, when planning territorial space it is necessary to analyze the landscape structure from a regional perspective to adjust landscape structure, to strengthen the protection of natural resources, so as to achieve sustainable economic and ecological development.

Key words: granularity effect, landscape pattern, landscape index, transfer matrix, dynamic change, Guangxi Gulf of Tonkin Economic Zone