考虑相关长度误差的含水层渗透系数反演研究

Inversion of Hydraulic Conductivity of Aquifer Based on Correlation Scale Error

相关长度误差对于研究含水层渗透系数反演精度的提升起着非常重要的作用。基于数值实验,运用连续线性估计方法(SLE)对含水层渗透系数K进行反演,通过增加K实测值数据量来分析不同相关长度误差条件下,渗透系数模拟值与真实值之间的差值,从而分析得出K实测值数据量和渗透系数反演结果误差之间的关系。研究结果表明:K实测值数据量较少时,改变相关长度数值,会对反演结果造成误差。在相关长度误差存在的基础上,随着K实测值数据量的增加,渗透系数K的反演精度增加的幅度越来越慢。表明过多的K实测值数据量会造成数据冗余,在设置的数值实验中,K实测值数应小于30个。因此,在实际工程中可以根据其精度要求选取较适宜的K实测值数据量。

The correlation scale error plays a very important role in the improvement of the inversion accuracy of the hydraulic conductivity of the aquifer.Based on numerical experiments,this paper used the successive linear estimation algorithm(SLE)to invert the hydraulic conductivity K.By increasing the number of K measured data to analyze the difference between the estimated value and the real value under the different correlation scale error conditions,and obtained the relationship between the data volume of K measured data and the error of the inversion result.The results show that,when the volume of K measured data is less,changing the correlation scale will cause errors to inversion results.On the basis of the correlation scale errors,with the increase of the measured value of K,the increase in the inversion accuracy of the hydraulic conductivity K becomes slower and slower.It shows that too much K measured data will cause data redundancy.For the numerical experiment set in this paper,the number of K measured value should be less than 30.Therefore,in actual engineering,a more appropriate amount of K measured value data can be selected according to its accuracy requirements.