基于Dempster-Shafer证据理论的流域不透水表面扩张特性研究

[目的]揭示驱动盘龙江流域不透水表面扩张的影响因子,以及影响因子各属性值对不透水表面扩张的影响程度,并在分析驱动机制的基础上,模拟预测盘龙江流域的扩张趋势,为流域生态建设合理规划提供依据。[方法]采用Dempster—Shafer(D—S)证据理论来描述和融合多种空间数据,在已有的不透水表面(impervious surfaces,IS)信息与多种空间数据的量关系的基础上,采用数据驱动方法分配基本概率函数(basic probability assignment,BPA)。经过定义多种空间数据的BPA函数,然后采用D—S证据理论的融合规则融合多个BPA函数以获取研究区域IS的信任函数、不信任函数、不确定函数、似真函数。[结果]距道路距离,距居民点距离,距水系距离,人口密度,GDP,IS邻域单元数,坡度,高程驱动因子对盘龙江流域不透水表面的扩张影响比较大,而坡向对不透水表面扩张的影响程度变化不明显。不透水表面扩张模拟的精度达到78.04%。[结论]采用D—S证据理论方法来描述空间数据和融合多种空间数据具有比传统逻辑回归模型更好的分析和预测功能。

A Study of Expansion Features of River Basin's Impervious Surface Based on Dempster-Shafer Evidence Theory

[Objective] This paper aimed to reveal the driving factors of impervious surface(IS) expansion in Panlong River basin, and to quantize the expansive responses of IS to the values of these factors. Upon these, a simulation of the expansion in the Panlong River basin was conducted to provide some bases for ecological improvement scheme. [Methods] Dempster-Shafer(D-S) evidence theory was used to described the spatial data and to integrate it into the basic probability assignment function based on the quantitive relationship between the existing IS information and the multiple spatial data by data-driven way. With the basic probability assignment functions, the belief, disbelief, ignorance, plausibility functions were obtained by merging multiple spatial data. [Results] Panlongjiang River basin was taken as an example to verify the validity with D-S evidence theory for the research on IS expansion properties, and the simulation accuracy reached 78.04%. [Conclusion] In predicting IS expansion, the D-S evidence theory to describe and integrate multiple spatial data was better than the traditional logistic regression model.