State-of-art on Geostatistical Methodology

郭怀成,周丰,刀谞

DOI: https://doi.org/10.3321/j.issn:1000-0585.2008.05.022

2008-01-01

Geographical Research

Abstract:Geostatistical methodology becomes an important tool for spatial prediction and uncertainty analysis.In terms of scientometric analysis and methodology development,advanced research for geostatistical methodology from 1967 to 2005 has been proposed in this study.First,its developing trend,application and patterns were identified;its development mode,suitability and choosing principles were then summarized.Development mode could be summarized as five transformations from stationary to non-stationary,single-variable to multivariate,parametric to non-parametric,linear to non-linear and spatial static to spatiotemporal dynamic geostatistics.Finally,the future researches were preliminarily discussed in this field,which is of great significance to geostatistical methodology and applications in the future.It includes:1)developing new methods for variogram estimation to reduce the analytical complexity,2)refining uncertain geostatistics to reflect hybrid uncertainties(input/output dataset,model structure,model parameters),3)exploiting spatiotemporal/multi-point geostatistics and its algorithm/software to simulate complex realizations,4)developing hybrid approach of process-mechanism and geostatistics to comprehensively uncover potential process-mechanism,in spite of the difficulties within the process simulations,and 5)expanding geostatistical-based decision-making under uncertainty,which supports the risk-based decision-making in social work,environmental pollution,agricultural production,public health,etc.