地质模型平滑后处理方法的改进

Improvement of Smoothing Method for Geological Models

摘要: 基于像元的建模方法建立的地质模型通常会含有少量不符合地质认识的“噪点”，这些噪点对后续的物性参数建模和油藏数值模拟都有影响。Clayton提出的后处理平滑方法MAPS在多数情况下能够取得较好的去噪效果。当模型中存在薄的地质体时，如薄的泥岩夹层，MAPS方法处理时会将单层网格泥岩夹层完全平滑掉，没有考虑夹层的延伸长度。针对这个问题，对MAPS方法平滑窗口的搜索范围和权值进行改进。改进后的方法能够处理掉含泥岩夹层模型中的噪点，并能够保留具有一定延伸长度的薄夹层。对比改进前后模型及原始模拟模型中泥岩夹层的长度分布，结果表明改进后的方法得到的泥岩延伸长度更接近于先验的泥岩长度，更加符合实际认识。

Abstract: Categorical variable models built by pixel-based modeling methods frequently present unrealistic small scale variations (noise) that may have an impact on subsequent petrophysical property modeling and flow simulation. The Maximum A-Posteriori Selection (MAPS) method proposed by Clayton can clean such models in most situations. When there are some thin geological bodies such as mud interlayer, the MAPS will clean those thin geological bodies with thickness of one cell, no matter how long they are. A new method is proposed to solve this problem. The key idea behind the proposed method is to design a rectangle moving Windows according to the character of geological bodies, and assign bigger weights to center layer cells. In the situation with thin geological bodies, the proposed method can clean the noise, and at the same time, the thin geological bodies which are longer than the given minim length can be kept.