贝叶斯最大熵地统计方法研究与应用进展

Review on Bayesian Maximum Entropy Geostatistics Method

以克里格估算为基础的插值和随机模拟为代表的经典地统计方法是目前研究地理属性空间分布的主要方法,但仍存在精度不高及不能有效利用其他有价值信息的缺陷。近年来贝叶斯最大熵地统计方法在国外逐渐流行,该方法能够在有效利用多源数据的基础上,提高空间分布研究精度,是一种新的非线性方法。本文详细阐述了贝叶斯最大熵方法的数据内容、实施步骤、一般算法及计算结果,并介绍了该方法的应用情况,最后对该方法的优点和不足作出了评价。

The classical geostatistics methods, including kinds of Kringing and stochastic simulation methods, are the main approaches to research spatial distribution of geographical attribute. However, these methods have some shortcomings, including low quality and disable of making use of other valuable information effectively. In recent years, Bayesian Maximum Entropy is becoming widely used in various studies on evaluation of natural resources. This method is a new nonlinear method with more rigorous theoretical foundation than Kriging for integrating uncertain information into space mapping. It provides new and powerful means from incorporating various forms of physical knowledge （include hard and soft data） into space mapping process, and produces the complete probability distribution at each estimation point, thus allowing the calculation of elaborate statistics. This paper introduced a Bayesian Maximum Entropy approach with its data content, process, algorithm, result and sample of application. At last, advantages and disadvantages of the approach were analyzed.