地理研究 ›› 2016, Vol. 35 ›› Issue (3): 534-550.doi: 10.11821/dlyj201603012

• 研究论文 • 上一篇    下一篇

城市近郊区景观格局变化特征,潜力与模拟----以成都市龙泉驿区为例

欧定华(), 夏建国()   

  1. 四川农业大学资源学院,成都 611130
  • 收稿日期:2015-10-09 修回日期:2016-01-04 出版日期:2016-03-20 发布日期:2016-03-28
  • 作者简介:

    作者简介:欧定华(1984- ),男,四川宜宾人,博士研究生,主要从事景观生态规划与设计,土地利用规划与管理,"3S"技术应用研究.E-mail: 357881550@qq.com

  • 基金资助:
    国家自然科学基金项目(31270498);四川省学术和技术带头人培养经费(2014);四川农业大学双支计划项目(2015)

Characteristics, potential and simulation of landscape pattern change in peri-urban areas: A case of Longquanyi district, Chengdu city

Dinghua OU(), Jianguo XIA()   

  1. College of Resources, Sichuan Agricultural University, Chengdu 611130, China
  • Received:2015-10-09 Revised:2016-01-04 Online:2016-03-20 Published:2016-03-28

摘要:

以成都市龙泉驿区为研究区,利用TM/OLI影像,ASTER GDEM和景观变化驱动因子数据,在ArcGIS和IDRISI Selva软件支持下,建立Ann-Markov-CA复合模型,分析了1992-2014年景观格局时空演变特征和变化潜力,对2021年,2028年景观变化趋势进行了模拟.结果表明:近22年,交通运输,果园,城乡人居及工矿景观增加显著,分别增加329%,184%,125%,农田,森林,水体景观减少明显,分别减少67.85%,59.94%,41.00%;主要景观均发生频繁转入转出,其中农田向果园,森林向果园,农田向城乡人居及工矿的转化最明显.参与景观变化潜力预测驱动因子越多其预测准确率不一定越高,需根据预测准确率选择恰当驱动因子组合进行变化潜力模拟.未来14年,大部分景观保持原有变化趋势,但变化剧烈程度逐渐减弱.过去14年(2000-2014年)和未来14年(2014-2028年),农田,森林景观总体上呈减少趋势,成为其他景观增加的稳定补给源.因此,遏制农田,森林景观无节制缩减,对维持区域生态平衡,实现地方生态建设与经济发展互动双赢具有重要意义.

关键词: 景观格局, CA-Markov模型, 多层感知人工神经网络模型, 变化模拟, 城市近郊区

Abstract:

With the support of ArcGIS and IDRISI Selva platform, based on remote sensing images of Landsat TM/OLI, the ASTER GDEM and data of landscape change driving factors, Ann-Markov-CA composite model was established to analyze characteristics and potential of landscape pattern change of Longquanyi district, Chengdu city in 1992-2014, and to simulate landscape pattern change trend of this district in 2021 and 2028. The results showed that, landscape area of transportation, orchard, urban-rural residential and industrial-mining increased by 329%, 184%, and 125%, respectively, and landscape area of farmland, forest and water body decreased by 59.94%, 67.85%, and 41% during the past 22 years. Landscape pattern experienced a great conversion process. The transformation of farmland to orchard landscape, forest to orchard landscape and farmland to urban-rural residential and industrial-mining landscape were significantly apparent. The forecast accuracy rate of landscape pattern change potential was not necessarily increased with the increase of the number of driving factors. According to the forecast accuracy rate, appropriate combination of driving factors were adopted to simulate landscape pattern change potential. For the next 14 years, most of the landscapes will maintain the changing trend in the past years, but the intensity of landscape change gradually weakened. The simulation showed a decreasing trend of farmland and forest in the past 14 years (2000-2014) and the future 14 years (2014-2028). Farmland and forest were taken as the stable supply source for other landscape. It is suggested that a guiding plan is required to protect farmland and forest landscape. It is most significant to maintain regional ecological balance and promote the interaction between ecological construction and economic development.

Key words: landscape pattern, CA-Markov model, multi-layer perceptron artificial neural network model, change simulation, peri-urban area