DarkSeis: Seismic Imaging Of The Urban Subsurface Using Dark Fibre

我们的城市与光纤电信网络纵横交错。为了实现系统冗余，在技术改进减少了所需带宽之后，安装的光纤有很大一部分没有使用。这种未使用的纤维被称为“暗纤维”。我们可以使用暗纤维来研究我们城市下面的地面属性。分布式声学传感(DAS)系统沿光缆发射激光脉冲，并记录后向散射光。如果电缆被拉伸或压缩，则后向散射光到达的时间会稍晚或更早。通过记录和绘制这些变化，我们可以将任何光缆转变为地球物理传感器，以非常高的分辨率记录光缆沿线的任何移动或振动。地震成像是绘制地下地图的一种成熟的方法。该技术使用震源，如锤击或特殊的振动源，将地震能量传递到地面。通过记录通过地面反射和/或折射产生的振动，我们可以建立地下的图像。这些图像可用于广泛的应用，如地热开发选址、了解地下水和排水、评估山体滑坡的可能性以及探测天坑。然而，传统的地震勘探依赖于在整个勘测区部署大量地震检波器传感器。在城市地区，这往往在后勤上具有挑战性，甚至是不可能的。因此，我们的城镇地下数据，可能是最需要这些数据的地区，往往是缺乏的。使用DAS进行地震成像提供了一种具有巨大潜力的替代方案。暗光纤电缆已经安装在地下的电信网络中，这意味着我们的传感器已经到位，并以最低的成本提供。我们可以访问光纤网络，在安全位置安装DAS设备，并沿电缆长度记录产生的地震数据，而不需要在感兴趣的区域部署地震检波器传感器。因此，使用暗光纤的DAS地震采集提供了改变我们获取城市地区地下图像的方式的潜力。到目前为止，这种方法的巨大潜力才刚刚实现。我们研究的目的是调查城市环境中用于地震成像的暗光纤DAS的性能，目的是找出如何最好地获取和处理这类数据。我们将使用B-Net电信网络获取数据，这是一个贯穿布里斯托尔市的250公里长的光纤网络(由布里斯托尔市议会所有)。我们将使用这些数据来确定如何生成最高质量的DAS地震图像--例如，哪种类型的地震源是最好的，如何最好地设置DAS采集单元，以及如何最好地处理产生的数据。城市地区DAS地震采集的主要挑战之一是背景噪声水平可能很高。这种背景噪音可能会降低我们的成像质量。为了解决这个问题，我们将开发使用最先进的人工智能算法来消除背景噪声，从而提高我们生成的图像的质量。我们获得的所有数据和我们开发的所有机器学习算法都将发布到公开可用的存储库。这将为英国和全球致力于这项技术开发的研究人员和商业地球物理公司提供极其宝贵的资源。

Our cities are criss-crossed with fibre-optic telecommunications networks. For system redundancy and after technological improvements have reduced the required bandwidth, much of the installed fibre is unused. This unused fibre is known as "dark fibre". We can use dark fibre to investigate the properties of the ground below our cities. Distributed Acoustic Sensing (DAS) systems fire laser pulses along fibre-optic cables, and record the back-scattered light. If the cable is stretched or compressed then the back-scattered light will arrive slightly later or earlier. By recording and mapping these changes, we can turn any fibre-optic cable into a geophysical sensor, recording any movements or vibrations along the cable with very high resolution. Seismic imaging is a well-established method to map the subsurface. The technique uses seismic sources such as a hammer strike or a special vibrating source to impart seismic energy into the ground. By recording the resulting vibrations that have reflected and/or refracted through the ground, we can build up an image of the subsurface. These images are useful for a broad range of applications, such as siting geothermal developments, understanding groundwater and drainage, assessing the likelihood of landslips, and detecting sinkholes. However, conventional seismic surveys rely on deploying a large array of geophone sensors across the survey area. This is often logistically challenging or impossible in urban areas. As a result, subsurface data under our towns and cities, perhaps the area where this data is needed most, is often lacking.Seismic imaging using DAS provides an alternative with enormous potential. Dark fibre cables are already installed in buried telecommunication networks, meaning our sensor is already in place and available at minimal cost. We can access the fibre-optic network, install a DAS unit in a secure location, and record the resulting seismic data along the length of the cables, without any need to deploy geophone sensors across the area of interest. Hence, DAS seismic acquisition using dark fibre offers the potential to transform how we acquire images of the subsurface in urban areas. To date, the enormous potential of this method is only just being realised. The objective of our research is to investigate the performance of dark fibre DAS for seismic imaging in urban settings, with the aim of working out how best to acquire and process this type of data. We will acquire data using the B-NET telecommunications network, which is a 250 km-long fibre-optic network running across the city of Bristol (owned by Bristol City Council). We will use this data to identify how to produce the best quality DAS seismic images - for example what types of seismic source are best, how best to set up the DAS acquisition unit, and how best to process the resulting data. One of the major challenges of DAS seismic acquisition in urban areas is that background noise levels are likely to be high. This background noise could degrade the quality of our imaging. To address this, we will develop the use of state-of-the-art artificial intelligence algorithms to remove the background noise, improving the quality of our resulting images. All of the data that we acquire and all of the machine learning algorithms that we develop will be posted to publicly available repositories. This will provide an extremely valuable resource for researchers and commercial geophysical companies, both in the UK and globally, who are working on the development of this technology.