基于过程模型的长白山自然保护区森林植被净第一性生产力空间尺度转换方法

净第一性生产力(Net Primary Productivity,NPP)是反映林分和生态系统结构特征的最重要的参数之一,其空间分布可为自然资源管理和可持续发展提供高度综合的信息,是全球变化和碳循环研究中的一个非常重要的组成部分。为了得到NPP的空间尺度转换算法,利用改进的北部生态系统生产力模拟模型(Improved Boreal Ecosystem Productivity Simulator,IBEPS),分别采用高分辨率(30m)的ETM 和低分辨率(1km)的MODIS遥感数据,模拟了长白山自然保护区不同分辨率森林植被的净第一性生产力。对模拟得到的30m分辨率的NPP结果,采用尺度上推(upscaling)的方法将其转换为1km分辨率的NPP结果(分布式的NPP,Distributed NPP),并将转换后的NPP与直接利用1km分辨率的遥感数据计算得到的NPP结果(聚合式的NPP,Lumped NPP)进行比较,以分布式模式计算的结果为基准,对聚合式的计算结果进行尺度效应校正。在对NPP进行尺度效应校正时,根据不同分辨率下土地覆盖类型(Land cover)的亚像元面积比信息,采用基于结构的分析方法对模拟的NPP结果进行校正,结果表明经过尺度效应校正后NPP的模拟精度要比原始的NPP计算精度有所提高,相关系数从校正前的0.898提高到0.960,标准偏差由校正前的49.84gC/m2减小到41.02gC/m2。采用这种方法可以很好的解决大尺度遥感应用的需要,为定量遥感分析和应用提供理论基础。

Spatial scaling of net primary productivity based on process model in Changbai Mountain Natural Reserve

Being one of the most important parameters to reflect the structural characteristics of the forestland ecosystem,Net Primary Productivity(NPP)becomes a very important area in the research of global change and carbon cycling as its distribution can provide the integral information for a sustainable development in terms of managing the natural resources.This paper presented an algorithm for the spatial scaling of NPP.Based on the Improved Boreal Ecosystem Productivity Simulator(IBEPS),the algorithm produced the NPP results at different resolutions in the area of Changbai Mountain Natural Reserve in Jilin Province,Northeast China,by using the remotely sensed images with high spatial resolution,Enhanced Thematic Mapper plus(ETM ),and moderate spatial resolution,Moderate Resolution Imaging Spectroradiometer(MODIS)respectively.Field experiments were carried out in order to get some ancillary data and also to validate the simulated NPP values.The NPP at the spatial resolution of 30 meters was scaled up to the scale of 1 kilometers spatial resolution and the distributed NPP(NPPd)was obtained.Based on the distributed NPP,the lumped NPP(NPPl,simulated NPP with 1 kilometer resolution parameters)was corrected(as NPPl_corrected)by using the contextual approach of spatial scaling,where the area fractions were used to derive the surface parameters at different resolutions.The corrected result indicates that the precision after scaling gets improved in comparison with the lumped NPP.The correlation coefficient is increased from 0.898 to 0.960,and the standard difference gets decreased from 49.84 gC/m2 to 41.02 gC/m2.This method can better meet the requirements of remote sensing applications of large scale.