关键词: 神经网络, 土地利用, 模拟, 乌鲁木齐市

Abstract: Based on the land use data in 2000, 2010 and 2018, this paper revealed the evolution characteristics of land use pattern in Urumqi city; On this basis, the FLUS model was applied to simulate the development trend of land use types in the study area under three scenarios in 2035. The results showed that the land use area from 2000 to 2018 showed the characteristics of “three increases and three decreases”; Among them, the dynamic change in the growth of construction land area was the largest, 7.79%, and the dynamic change in the decrease of arable land area was the largest, 1.02%; During the 18 years period, the area of grassland shifted to unused land was the largest, 435.40 km², followed by the transfer of forest land to grassland and grassland to cropland; Under the natural development scenario, the area of grassland to cropland was the largest, 105.26 km², followed by the transfer of unused land to grassland, grassland to unused land and grassland to construction land. Under the basic conservation scenario, the largest area of grassland to construction land was 59.91 km², followed by conversion of grassland to cropland, conversion of grassland to unused land and conversion of forest land to grassland. Under the ecological protection scenario, the area of both unused land and forest land was shifted to grassland, with 36.83 and 25.34 km², respectively. The study provided a scientific basis for integrated decision making in land use planning, sustainable development and resource management.

Key words: neural network, land use, simulation, Urumqi city