A classical two-stage method to stochastic inverse problems in groundwater and petroleum engineering starts from the generation of a series of independent seed flelds and then calibrates those flelds to inverse-condition on nonlinearly dependent state data from difierent sources, which is known as m