黑河流域上游植被及其环境要素空间插值与动态模拟分析

Upper reaches of Heihe River Basin has a complex landform. There exist both complex homogeneity and heterogeneity among the ecosystems so that most of vegetation and its environment elements are spatial non-stationarity. This project will classify data of the vegetation and its environment elements, which have been collected by ground monitoring, satellite observation and transect survey data in upper reaches of Heihe River Basin, into two sub-databases with spatial stationarity and spatial non-stationarity respectively. For the spatial stationarity elements, trend surfaces will be created by Ordinary Least Squares regression (OLS); for the spatial non-stationarity elements, Geographically Weighted Regression (GWR) will be used to produce trend surfaces. Two algorithm-bases are to be established. One algorithm-base is set up by combination of the OLS trend surfaces with a method for high accuracy surface modelling (HASM) that is employed to deal with their residues; another one is to be established by a combination of GWR trend surfaces with HASM treating their residues under a BOX-COX transformation of the spatial non-stationarity data. A high-speed numerical solver and a GPU parallel solver will be developed for personal computers and a MPI parallel solver is to be designed for high performance computer groups to considerably speed up the computation of HASM. A system is to be developed for spatial interpolation and dynamic simulation. This system, using LiDAR point-cloud data, is employed to extract wood multi-layer vertical structure and tree species classification. This system is also used to interpolate transect survey data and simulate vegetation types, vegetation structure, vegetation coverage and vegetation canopy height, by developing relationships between vegetation elements and their environment elements. We are to create spatial data sets of vegetation elements and their environment elements.

黑河流域上游地形复杂，大多数植被要素与其环境要素的关联存在着空间非平稳性。本项目拟对观测、监测和样带调查等手段获得的黑河流域上游植被及其环境要素数据进行空间非平稳性分析，实现空间非平稳和空间平稳要素的分类处理，据此采用不同的趋势面模拟策略，建立相应的算法库。根据空间插值的精度要求，运用高精度曲面建模（HASM）方法，对各个植被要素与其因变量的空间平稳或空间非平稳趋势面残差进行智能化处理；发展面向单机的快速数值求解器和GPU并行加速求解器及面向高性能计算机群的MPI并行加速求解器，构建面向大规模数据的空间插值和动态模拟分析系统。运用小流域LiDAR点云数据，提取树林多层垂直结构和树种分类等森林群落结构参数。运用样带调查数据，建立植被和相关环境因子的空间分布规律，实现植被类型、植被分层结构、植被盖度和植被冠层高度等植被要素的空间插值与动态模拟，完成植被和相关环境因子的空间分布图件及数据库。

本项目运用观测、检测和样带调查等手段获得了黑河流域植被及其环境要素数据，并对其进行空间平稳性分析，实现环境要素的空间非平稳和空间平稳的分类处理。并在此基础上，采用不同的趋势面模拟策略，建立了相应的算法库。根据空间插值的精度要求，运用HASM方法对各个植被要素及环境变量的空间平稳或空间非平稳趋势面残差进行智能化处理；发展了面向单机的快速数值求解器和GPU并行加速求解器及面向高性能计算机群的MPI并行加速求解器。运用小流域LiDAR点云数据，提取了树林多层垂直结构和树种分类等森林群落结构参数。运用样带调查数据，建立了植被和相关环境因子的空间分布规律，实现了植被类型、植被分层结构、植被高度和植被冠层高度等植被要素的空间插值与动态模拟，完成了植被和相关环境隐私的空间分布图件18件及数据库。