物探与化探
物探與化探
물탐여화탐
GEOPHYSICAL AND GEOCHEMICAL EXPLORATION
2015年
3期
562-566,571
,共6页
大地电磁%有限差分%一维正演%连续介质%网格剖分方法
大地電磁%有限差分%一維正縯%連續介質%網格剖分方法
대지전자%유한차분%일유정연%련속개질%망격부분방법
finite difference%MT%1D inversion%continuous medium%new mesh generation method
介绍了有限差分算法在大地电磁测深法中的应用,推导了一维层状介质和连续介质模型下的有限差分算法,提出了一种新的网格剖分方法,并通过两层和三层介质模型的正演计算,验证了算法的正确性以及网格剖分方法的合理性。设计了一种连续介质模型,与两种不同程度近似的多层层状介质模型进行了正演结果的对比,指出了研究连续介质模型正演的必要性。对于连续介质模型,计算中剖分得越详细,其结果越接近真实值,而随着误差的逐渐减小,精确度的提高不再明显,但资源与时间的消耗将大幅增加。因此,无限制地减小网格步长是不必要的。同时,为了兼顾误差大小与网格步长的关系,采用了一种将一次网格剖分进行二次差分计算的方法得到新的结果,使计算结果明显得到了改善。
介紹瞭有限差分算法在大地電磁測深法中的應用,推導瞭一維層狀介質和連續介質模型下的有限差分算法,提齣瞭一種新的網格剖分方法,併通過兩層和三層介質模型的正縯計算,驗證瞭算法的正確性以及網格剖分方法的閤理性。設計瞭一種連續介質模型,與兩種不同程度近似的多層層狀介質模型進行瞭正縯結果的對比,指齣瞭研究連續介質模型正縯的必要性。對于連續介質模型,計算中剖分得越詳細,其結果越接近真實值,而隨著誤差的逐漸減小,精確度的提高不再明顯,但資源與時間的消耗將大幅增加。因此,無限製地減小網格步長是不必要的。同時,為瞭兼顧誤差大小與網格步長的關繫,採用瞭一種將一次網格剖分進行二次差分計算的方法得到新的結果,使計算結果明顯得到瞭改善。
개소료유한차분산법재대지전자측심법중적응용,추도료일유층상개질화련속개질모형하적유한차분산법,제출료일충신적망격부분방법,병통과량층화삼층개질모형적정연계산,험증료산법적정학성이급망격부분방법적합이성。설계료일충련속개질모형,여량충불동정도근사적다층층상개질모형진행료정연결과적대비,지출료연구련속개질모형정연적필요성。대우련속개질모형,계산중부분득월상세,기결과월접근진실치,이수착오차적축점감소,정학도적제고불재명현,단자원여시간적소모장대폭증가。인차,무한제지감소망격보장시불필요적。동시,위료겸고오차대소여망격보장적관계,채용료일충장일차망격부분진행이차차분계산적방법득도신적결과,사계산결과명현득도료개선。
This paper describes the application of finite difference algorithms in MT method, deduces the algorithms of 1D layered and continuous medium, and proposes a new mesh generation method. Meanwhile, with two-layer or three-layer medium model inversion calculation, the paper demonstrates the correctness of the algorithms and rationality of the grid subdivision method. The authors also de?signed a kind of continuous medium and made a comparison of the inversion result between two kinds of multiple layered medium mod?els which have different layer approximations, and indicated the necessity of continuous medium model inversion research. For continu?ous medium model, the fine the grid, the better the approach result. Nevertheless, with the gradual decreasing of the error, the im?provement of accuracy becomes not so distinct. However, resource and time consumption will dramatically increase. So decreasing grid spacing without restriction is not necessary. What's more, taking both the error and grid spacing into consideration, the authors conduc?ted second difference calculation in first mesh generation and achieved a new conclusion, which can distinctly improve the calculation result.