在遥感反演中对先验知识的表达和应用方法是多阶段目标反演中急需解决的关键问题。论述一种在遥感反演中处理不确定性信息的一种数学方法 ,引入未确知有理数和盲数的概念和运算方法 ,定量计算反演参数在可能取值区间上反演前后的可信度及其改变量。为遥感反演中先验知识的积累、更新和应用提供定量的方法依据。

Remote sensing inversion algorithms exist the problems of land surface parameters retrieval by means of remote sensing physical models, which is also a key problem in remote sensing image interpretation. We developed a priori knowledge based remote sensing inversion strategy. All the available information on parameters gained from the former inversion stages are taken as the prior knowledge in the next stage's inversion. To do the inversion objectively, we need to describe the information content that the parameters get in each inversion stage, and the reliability of the accumulated knowledge as well. The concepts and theory of unascertained number and blind number are developed from the requirements of describing the uncertain information in application. They have been used in some techniques and producing fields to solve the practical problems, such as that used in the engineering theory. Our new approach is to describe the reliability of the inverted parameter and accumulated knowledge by the definition of the unascertained number. For the inverted parameter, during the inversion stage, the changing of its reliability then can be calculated by using math method of the unascertained mathematics, which is used to judge the inversion result and to modify the prior knowledge about the parameter. This is expected to be a quantitative expression of the accumulation of a priori knowledge. We take the land surface temperatures inversion as an example to show how the new method works.