A study of Data Generation in Saturated Hydraulic conductivity with Spatial Distribution

碩士 === 國立雲林科技大學 === 環境與安全工程系碩士班 === 89 === The primary goal of this research is to apply Monte Carlo method and Spectrum theory into the data generation simulating formula in order to produce a featured heterogenous saturated hydraulic conductivity in spatial distribution, and then further statistic...

Full description

Bibliographic Details
Main Authors: Hsing-Hsiang Chang, 張湘杰
Other Authors: Jet-Chau Wen
Format: Others
Language:zh-TW
Published: 2001
Online Access:http://ndltd.ncl.edu.tw/handle/26256862965407205316
Description
Summary:碩士 === 國立雲林科技大學 === 環境與安全工程系碩士班 === 89 === The primary goal of this research is to apply Monte Carlo method and Spectrum theory into the data generation simulating formula in order to produce a featured heterogenous saturated hydraulic conductivity in spatial distribution, and then further statistically analyzes its results to verify the accuracy of the formula in this research. In the 2 and 3-dimensional experiments, we discovered that the correlation function, which we got from the random variable of, becomes gradually closer to the exponential decay correlation function once the loss of spectrum declines clearly. And the two factors which lead to the loss of are truncation error and discretization error. Moreover, when the number of grids increases, the chance which the results of random variable of shows larger than the results between 95﹪erroneous acceptable range will become larger. In other words, the larger the no-dimension relative distance of becomes, the better results the random variable of we are going to have. Plus, we also assured in our 2-dimensional experiment that the limitation of random variance must be less than 0.5 according to Bakr et al., creators and proposers of incorporating the minor disturbed approach into sequential procedure in 1978. That is, when the variance of is 0.3, the subsequent generating correlation function of random disturbed variance will close to the correlation function of exponential decay. On the contrary, the difference between the correlation function of random disturbed variance and exponential decay becomes larger and larger while the variance of moves upward. We can learn that the integral of angular frequency turns out to be identical to the disturbed variance of according to Wiener-Khintchine relation [Yahlom, 1987; Bakr, 1979]. In addition, this formula also reveals the same results which conducted in 2 and 3 dimensional research, proving a feasible program if we want to simulate the exponential modeling of saturated hydraulic conductivity.