反射地震学中的一些数据处理问题可以先化为普通反演问题,而后在已知观测数据和模型先验概率分布的情况下,再川求模型后验概率极大值的方法予以求解。贝叶斯解法和应用计算技术一般都可应川,且不需要作局部线性或高斯统计特怀的假设。唯一的限制是模型参数的条件概率呈现为局部相关。或者更具体地说,这种模型参数定义了一种叫做马尔柯夫随机域的随机过程。这就等于把其联合先验概率分布说成是统计物理学中的吉普斯分布(或叫Canonical分布)。 Some data processing problems in reflection seismology can be transformed into ordinary inversion problems first. Then, the known posterior probability distribution of the observed data and models can be used to solve the model posterior probability maximum. Bayesian solutions and applied computing techniques are generally applicable and do not require the assumption of local linear or Gaussian statistics. The only limitation is that the conditional probabilities of the model parameters appear to be locally related. Or more specifically, this model parameter defines a stochastic process called Markov random fields. This is equivalent to describing its joint prior probability distribution as a Gibbs distribution (or Canonical distribution) in statistical physics.