基于时空地统计学的生态环境监测网布局优化方法研究

Optimal selection of observation locations is an essential task in designing an effective dynamic ecological environment monitoring network, which provides information on target variables by capturing their spatial variations and distribution. Currently, the most widely used design is based on the geostatistical model-based sampling methods. However, these traditional methods are simply based on the assumption that the surface process is a static stationary process, which neglect to consider the dynamics and heterogeneities in surface. To overcome these weakness, we proposed two new monitoring network optimization methods to optimize sampling location for monitoring a surface process with high spatial heterogeneities. For the case that target variable could be regarded as a discrete spatial random field, a multi-cokriging was proposed to model the spatial cross-variability between times. The monitoring locations were optimised by defined the objective function that the mean cokriging error variance is minimum. Furthermore, to be more generality, a space-time kriging model-based sampling design was proposed to solve the problem that the variable vary from both space and time. For real world application, the proposed methods will be used to design a wireless sensor network for monitoring the characteristics of the eco-hydrological process in the Heihe river basin and the air quality monitoring network in Beijing, China. The aim of the optimized monitoring network is to efficiently capture the spatial and temporal variations of the ecological environment variables and improving the monitoring efficency and save construction investment.

分布式的地面观测对于定量刻画生态环境过程的时空动态特征与场分布具有重要意义。在监测网设计过程中，观测点的空间布局将直接影响到监测网的整体观测水平。目前的监测网布局设计是基于传统地统计学的空间采样优化方法。而这些方法往往简单假设观测对象是一个静态的平稳过程，缺乏对地表特征动态性和异质性的考虑，使得设计的监测网时空代表性较差。鉴于此，本项目将对空间采样优化方法进行时间维扩展，发展面向复杂生态环境过程监测网布局优化方法。针对生态环境过程，提出了多时间协同优化策略和发展了时空克里格优化方法，分别用于解决目标对象在时间上离散变化或连续变化情况下监测网优化问题。突破监测网布局优化局限在平稳过程的传统思路，发展异质过程监测网布局优化新方法。最后将提出的方法应用到黑河流域生态水文无线传感器网络和北京市空气质量监测网评估和优化设计中，从而提高生态环境监测网设计的科学性、观测精度和效率，节省投入。

分布式的地面观测对于定量刻画生态环境过程的时空动态特征与场分布具有重要意义。在监测网设计过程中，观测点的空间布局将直接影响到监测网的整体观测水平。目前的监测网布局设计是基于传统地统计学的空间采样优化方法。而这些方法往往简单假设观测对象是一个静态的平稳过程，缺乏对地表特征动态性和异质性的考虑，使得设计的监测网时空代表性较差。鉴于此，本项目将对空间采样优化方法进行时间维扩展，发展面向复杂生态环境过程监测网布局优化方法。针对生态环境过程，提出了多时间协同优化策略和发展了时空克里格优化方法，分别用于解决目标对象在时间上离散变化或连续变化情况下监测网优化问题。突破监测网布局优化局限在平稳过程的传统思路，发展异质过程监测网布局优化新方法。最后将提出的方法应用到生态水文无线传感器网络、空气质量监测网评估和优化设计以及空气质量时空分布估计中，提高生态环境监测网设计的科学性、观测精度和效率。在本项目的支持下累计发表论文4篇，其中第一和第二标注论文各两篇，受邀在国际计量土壤学大会上做主题报告一次。