Based on the combination ofdata-and model-driven approaches,this study expanded the labels of the training set through model inversion results,and added the model inversion objective function to the deep learning algorithm.By constructing a new loss function,this study proposed a seismic impedance optimization inversion method combining model inversion with deep learning inver-sion.The semi-supervised deep learning network inversion under a pseudo-label was achieved using the RNN network structure.The network inversion results were used as the initial model to participate in the model inversion.The final optimization inversion was com-pleted by continuous iterative optimization of both network and model inversion.The method proposed in this study proves to possess high inversion accuracy and practicability,as demonstrated by the synthesis of the Marmousi model and the actual data.
Owing to the limitations of both the field environment for seismic data acquisition and the performance of instruments, the seismic signals collected in seismic exploration are inevitably mixed with strong noise, thus greatly affecting the subsequent processing and interpretation.In recent years, multi-scale geometric analysis methods have become an important topic in noise suppression owing to their unique advantages.This study proposed suppressing the seismic noise using a non-local mean（NLM） algorithm in the Shearlet domain.First, the non-subsampled Shearlet transform（NSST） was performed for seismic signals.Then, the decomposed coefficient subset was further processed using the NLM method, and the weight function was improved by using eight Sobel operators to approximate the omnidirectional structure.Finally, the inverse Shearlet transform was performed for the coefficients to obtain the denoised seismic signals.Experimental results show that this combined algorithm can effectively suppress the random noise and preserve the weak events, thus showing high practicability in the seismic data processing.
Bioavailable selenium（Se） is an important indicator used to evaluate the supply capacity of Se in the soil to plants. There is a lack of national standard methods for testing bioavailable Se in the soil in China presently. This study used ten different chemical extractants to extract bioavailable Se from the soil. The results show that the extraction capacity of extractants was in the order of NaOH（10.57%） ＞ SOL-Se+EX-Se（8.10%） ＞ AB-DTPA（6.79） ＞ NaHCO 3 （4.99%） ＞ K 2 SO 4 （4.57%） ＞ KH 2 PO 4 -K 2 HPO 4 （4.33%） ＞ KH 2 PO 4 （3.05%） ＞ SOL-Se（2.28%） ＞ EDTA（1.40%） ＞ NH 4 F-HCl（1.22%）. Among them, the bioavailable Se extracted from the soil by KH 2 PO 4 , NaHCO 3 , SOL-Se, KH 2 PO 4 -K 2 HPO 4 , and AB-DTPA was significantly correlated with the Se content of crops. As revealed by the relationship between the percentage content of bioavailable Se in soil and soil properties, the percentage content of bioavailable Se in soil significantly correlated with indicators such as the chemical index of alteration（CIA）, LOI, pH, Eh, P, and SiO 2 . Among them, pH and Eh mainly affect the existence form of Se in the soil, while indicators such as CIA, LOI, P, and SiO 2 reflect the alteration degree of soil and the effect of adsorbates on bioavailable Se in soil. The results of this study can provide data support and a scientific basis for the development of Se-based ecological agriculture and the development and utilization of Se-rich land resources.
在山区进行应用电性源瞬变电磁法勘探易受复杂地形影响,瞬变电磁响应会产生畸变,给数据解释造成困难.为此,基于三维非结构时间域有限元算法,开展了地形效应影响规律和校正方法的研究.首先,利用非结构四面体网格对山峰地形进行精细刻画,计算多种带地形模型响应,分析地形效应的影响规律;然后,基于电磁场线性叠加原理提出地形效应校正方法,根据实际高程数据建立山峰地电模型,通过三维正演计算出山峰模型响应,将山峰模型响应减去平坦大地模型响应得到地形响应,再从总响应中剔除地形响应,获得地形校正后的瞬变电磁响应.研究表明:山峰地形对瞬变电磁响应的影响主要集中在早期,随着时间推移,影响逐渐减弱;地形效应主要集中于山体附近,影响强度取决于测点距山顶的距离;影响范围与响应幅值与山体规模成正比;相对高阻山体的地形影响强度更大.通过设计多种带地形简单规则异常体模型进行仿真,发现经过校正的瞬变电磁响应与直接正演的响应吻合较好,在一定程度上能够有效消除地形效应.对地形效应影响和校正方法的研究,可为复杂地形地区瞬变电磁数据处理解释提供参考.
基于频谱法的瞬变电磁法一维正演策略中,多个计算环节对瞬变电磁响应计算精度有重要影响.为了提高一维正演效率,通常只直接计算数十个频点的频率域电磁响应,再采用三次样条插值将频率域电磁响应扩展至数百个频点.尽管三次样条插值函数计算的数值结果已能满足大多数正演需求,但其计算精度仍有提升空间.本文将高次B样条插值引入至瞬变电磁法一维正演,用于代替传统三次样条插值,并以磁偶源和圆形回线源为例验证了方法准确性.结果表明,对于多个地电模型,基于高次B样条插值计算的瞬变电磁响应精度均优于基于传统三次样条的计算结果.
