轮回搜索-贝叶斯法及其在大地测量反演中的应用

Cyclic-Bayesian Searching Method and Its Application to Geodetic Inversion

分析了反演中常用的贝叶斯逼近法、轮回搜索法两种算法的优缺点 ,提出轮回搜索 贝叶斯联合算法 ,该算法可以很好地反演出先验信息不明的参数。利用喜马拉雅区域GPS速度场 ,通过位错模型结合轮回搜索 贝叶斯方法 。

The advantages and disadvantages about Bayesian method and Cyclic searching method are discussed detailedly,and a Cyclic-Bayesian searching method is put forward. The Bayesian-cyclic searching method has two obvious advantages:one is of wide searching space for unknown parameters with Cyclic searching method and the result can act as initial value for Bayasian method, and the other is of considering prior information about unknown parameters with Bayasian method. It is useful for inversing problems which are short of prior information with cyclic-Bayesian searching method. The surface contraction and uplift rates in the Himalaya zone based on fault dislocation models and GPS velocity field are discussed with the cyclic-Bayesian searching method. The results show that the surface contraction rates across the Himalaya zone about 13.22mm/a to 20.38mm/a, and the uplift rates in the Himalaya zone is about 8.25mm/a to 9.34mm/a.