基于遥感反演参数与双约束空间聚类算法的湖泊营养物生态分区  

Lake Nutrient Concentration Oriented Ecological Division Based on Remote Sensing Inversion Parameters and Dual-constraint Spatial Clustering Algorithm

作　　者：柯新利[1,2] 刘曼[2] 邓祥征[1,3] 

机构地区：[1]中国科学院地理科学与资源研究所,北京100101 [2]咸宁学院资源与环境科学学院,湖北咸宁437100 [3]中国科学院农业政策研究中心,北京100101

出　　处：《地理科学进展》2012年第3期315-323,共9页Progress in Geography

基　　金：国家水体污染控制与治理科技重大专项(2009ZX07106-001-03;2009ZX07106-001-07);国家科技支撑项目(2008BAC43B01)

摘　　要：中国湖泊营养物水平和富营养化效应存在显著区域分异特征,开展湖泊营养物生态分区研究,是实现湖泊分区控制的前提和基础。本文以云贵高原亚热带湿润区为案例区,通过遥感反演得到研究区域湖泊水体的叶绿素a(Chl-a)、总氮(TP)和总磷(TN)含量,从自然地理要素、生态系统和人类活动3方面建立指标体系;根据研究区域DEM数据划分小流域,将各指标因子投影到小流域;在小流域尺度上,采用数理统计方法定量分析各指标分别对Chl-a/P与Chl-a/N的作用,计算两种湖泊营养物生态分区评价分值;采用双约束空间聚类算法对小流域尺度上两种湖泊营养物生态分区评价分值共同进行聚类,形成湖泊营养物生态分区结果。研究发现:①通过遥感反演模型可以将少数呈点状分布的监测数据推演到面上,这在一定程度上弥补了有限的野外监测数据量难以支撑湖泊营养物生态分区的不足;②通过构建各指标因子分别与湖泊水体中Chl-a/P、Chl-a/N的定量关系,实现了湖泊营养物生态分区中湖泊水体微观指标与湖泊流域宏观指标的有机结合;③采用双约束空间聚类进行湖泊营养物生态分区,保证了分区结果在空间上的连续性和评价分值上的接近性;④根据自然地理条件、土地利用和人类活动强度的不同,将云贵高原亚热带湿润区分为4个区:北云南高原湖区、滇南河谷山原湖区、桂西-滇东-黔南岩溶湖区、贵州高原东北部湖区。本文提出的基于双约束空间聚类的湖泊营养物生态分区方法也可以作为其它地区乃至国家尺度湖泊营养物生态分区的借鉴。There exist remarkable regional differences on the lake nutrient and its eutrophication effect in China.It is the premise to explore the features of lake nutrient and its determinacy on the ecological division to carry out effective measures on the breaking-out of lake eutrophication.In the case study of Yunnan-Guizhou Plateau,the contents of Chlorophyll-a（Chl-a）,total nitrogen（TN） and total phosphorus（TP） of lake water are inverted via the remote sensing inversion algorithm.An index system is also established by including the information from geophysical,geographical,ecosystem and human dimensions to evaluate the levels of land nutrients and all the information is mapped onto the small watersheds delimitated based on DEM data covering the entire study region.Meanwhile,the effect of each index on the ratio of Chl-a over P（Chl-a/P） and the ratio of chl-a over N（Chl-a/N） is quantitatively analyzed by applying mathematical statistics method on the small watershed scale so as to calculate the evaluation scores of two lake nutrient ecological zones who will then be clustered according to the small watershed scale via the dual-constraint space clustering algorithm.After a couple rounds of handlings,ecological zones reflecting the pattern of the lake nutrient concentrations are finally formed.The paper concludes a couple of key findings.A small number of sites based monitoring data can be used to deduce the surface of lake nutrient level through the remote sensing inversion model,which makes up the deficiency that the few sites-based monitoring data are always hard to supply enough information for the lake nutrient ecological division.By exploring the relationship between each index factor and Chl-a/P and Chl-a/N on the lake water,organic combination of microscopic indexes of the lake water and macroscopic indexes of the watersheds on the lake nutrient ecological division is thus realized.By applying the dual-constraint spatial clustering algorithm to carry on the lakes nutrient ecological zoning to

关 键 词：湖泊营养物 生态分区 遥感反演 双约束空间聚类 云贵高原亚热带湿润区 

分 类 号：X524[环境科学与工程—环境工程] X87