In this study, a historical scene-based setting method is proposed for the parameter Weight of Neighborhood of the FLUS model by simulating the land use scenario of urban agglomeration of the Golden Triangle of Southern Fujian in 2030. Firstly, based on the land use data in 2015, the occurrence probability and spatial distribution of each land use type were calculated with the artificial neural network algorithm, which integrates 12 natural, social and economic driving factors. According to the analysis of historical scenarios, the parameters of the FLUS model were confirmed via the Markov Chain and the landscape pattern indexes analysis method. Finally the Self-Adaptive Inertia and Competition mechanism Cellular Automata were used to simulate the land use scenario of urban agglomeration of the Golden Triangle of Southern Fujian in 2030. According to the analysis, the variation of TA (Total Area) at the same time scale could better reflect each land use type's expansion intensity and were ranked follows:construction land, water area and tidal flat, other land, grassland, woodland, and cultivated land. The dimensionless value of TA variation was in good agreement with the Weight of Neighborhood of the FLUS model in terms of data meaning and data structure. Combined the relationship between the TA variation and the expansion intensity of each land use type, the cultivated land would be most affected by the expansion of construction land until 2030. A large amount of land will be converted constantly from farmland, woodland, grassland and other land into construction land or water area and tidal flat. The construction land will continue to expand, the spatial integration pattern of the urban agglomeration of the Golden Triangle of Southern Fujian will basically form, other land use types will be further separated, and the patches of same land use types would progress independently. Taking the parameter setting process and simulation results into account, the dimensionless value of TA variation can provide an objective and feasible method for setting the Weight of Neighborhood parameters of FLUS model.