Disaster Preparedness and Recovery using Spectral Classification and Change Detection Technology
When Hurricane Harvey made landfall in August 2017, it left unprecedented rain, flooding, and destruction over Houston, one of America’s most populated cities and most critical energy regions. Harvey was one of the costliest natural disasters in modern American history, with an economic cost estimated as high as $108 billion.
Disasters can wreak havoc on infrastructure anywhere, but a hurricane’s unpredictable nature makes preparation and recovery infinitely more difficult. Oil and gas companies with operations in the pathway of a major hurricane are particularly vulnerable to the unpredictable and often devastating effects. With hundreds of miles of pipeline and millions of barrels per day in refining capacity in its path, Harvey’s devastation caused effects on energy companies that were felt worldwide.
How it Works
Remote sensing technology has been around since the late 1800s; however, modern advancements have made more sophisticated analyses possible through the advent of higher resolution sensors capable of being deployed on various types of platforms. Sensors can be affixed to many different platforms, such as satellites, nanosatellites, aircraft, drones, balloons or fixed stations. By utilizing complex data analysis techniques such as machine learning and artificial intelligence, data derived from spectral sensors can be manipulated to reveal characteristics within the environment that are not visible to the naked eye.
Imagery used in remote sensing science ranges in terms of radiometric, spatial, spectral, and temporal resolutions. Passive optical sensors for object identification, classification and change detection typically provide high spatial resolution multispectral data; however, the exact type of data depends on the type of investigation being completed.
Remote sensing science measures electromagnetic radiation from the sun that is reflected off objects on the surface of the Earth. Because everything on earth has a unique spectral composition, data is extracted from spectral imagery to identify features of interest. The spectral makeup of concrete is different from water, which is different from hydrocarbons, which is different from buildings or vehicles. This can also be referred to as the spectral signature or spectral fingerprint of an object. Identifying the unique spectral signature of a feature of interest allows object identification and change detection software to map locations of features on the surface of the Earth.
The characteristics of the object are contained in digital data composed of picture elements, or pixels. A pixel is the measuring element of the sensor noted as a representative square that makes up the sensor’s information. Resolution is a factor of the pixel size and the type of sensor used to create the image.
Preparing for Natural Disasters
Companies in high hurricane risk areas require additional preparedness to cope with wind and rain damage that often occurs with severe storms. Flooding after a major hit can intensify problems in an oilfield, port or refinery. As storms approach, standard preparedness plans call for operators to evacuate personnel, shut down production, and move monitoring systems to a command center well away from the storm, and to monitor remotely.
After the threats have cleared, companies instigate remobilization to get their operations back online. One element of preparedness that change detection technology can provide is a benchmark for oil and gas operators assisting in the decision-making process guiding remobilization efforts. In the case of Hurricane Harvey, having data prior to the storm allowed for an expedient assessment of where resources were needed, when, and to what extent. This technology can help operators understand surface water location, riparian zones, watersheds and floodplains: where water is, where it wasn’t before and its path over time.
Why Big Data?
The information contained in spectral imagery is much more than just a pretty picture. By taking a deep dive into spectral data with knowledge and software, insightful information is revealed. An image that is seen by the eye (in person or in a regular photograph) does not relay the same kind of information that is present in a spectral image. As electromagnetic radiation from the sun interacts with materials on the Earth’s surface a portion of the energy is absorbed by the feature and a portion is reflected towards the sensors.
By measuring the differences in intensity of reflected electromagnetic radiation in discrete bands from the visible to the infrared, important actionable information can be extracted from the imagery. Vegetation shows stress when hydrocarbons affect the plant biochemistry. Hydrocarbons on water have a different composition than water alone. Changes along a pipeline or right of way could be hard to discern with the naked eye, but spectral imagery and built-in machine learning techniques illustrate these areas of concern easily. Imagery is collected over an area of interest and analyzed using complex algorithms, providing valuable insight over the entire area of interest.
After a hurricane event, recovery can be very difficult. If spectral imagery of the affected area is captured regularly over time, comparison of operations during hurricane incidents and recovery can add crucial actionable knowledge. For example, if a pipeline goes through an area routinely affected by water, periodic examination with spectral imagery will help to reveal if current (hurricane-related) problems were based on something that was previously overlooked. Spectra-based change detection can illustrate where and when a leak may have begun. A portion of an oil refinery could have had undetected issues that were made worse by the after-flooding but may have been more easily mitigated with periodic change detection before the event.
Terabytes of spectral satellite imagery can be processed within specialized software to identify oil leaks on land and water; categorize significant changes such as erosion, landslide, and vegetation loss; and reveal damaged or downed structures and equipment. Viewing each analysis via mobile or email alerts, web dashboards and APIs that can connect with other software enables better decisions and faster reaction to events. Big data is only helpful with the means to analyze it into actionable information.
Spectral imagery of an area of interest next to the port and along the right-of-way of a pipeline that fell directly within the pathway of the devastation left by Hurricane Harvey was obtained and examined. Visual inspection of assets during hurricanes and during the immediate recovery phase can be impossible. Utilizing satellite or aerially acquired imagery provides visualization of wide areas of interest. Among hundreds of miles of pipeline, the software identified numerous areas where significant water impact had occurred, including erosion, standing water and other potential risks that would undoubtedly require follow up and mitigation. The images below show various datasets as they would be displayed in the software change detection interface.
Act Faster. Speed Up Recovery.
This specialized software can detect and display important changes related to weather damage and flooding. As illustrated in the above images, moving water and erosion can have serious effects on critical infrastructure. Furthermore, it can detect oil and natural gas leaks over land or water and illustrate critical changes that affect infrastructure stability. It is often unsafe and nearly impossible to visualize infrastructure manually in the recovery period after a hurricane. The software provides the whole picture over a set of oil and gas infrastructure remotely, and helps operators prioritize and narrow down large areas of interest into just the critical areas in need of manual inspection or remediation.
In today’s oil and gas industry there has been a remarkable stabilization of aging infrastructure with modern techniques. Change detection and object identification software applies complex statistical algorithms to significantly improve early detection, often long before the naked eye can see the problem. The ability to solve critical business problems by utilizing satellite data (or spectral data from other platforms) processed with proprietary algorithms to alert for damage, changes, leaks or third-party encroachments can be accomplished within hours of capturing data. Monitoring remediation efforts of the infrastructure and surrounding environment will ensure efforts and resources are efficiently and effectively
The remote sensing industry is beginning to see a paradigm shift in availability of satellite imagery. Decades ago, government satellites (such as the Landsat program) were the only option to procure satellite data. For the last 20 years though, more and more earth observing satellites have become available. In addition to more availability of conventional satellites, the next iteration of this remarkable tool is nanosatellites. These small satellites will be arranged into vast constellations and are expected to rapidly decrease the amount of time (and cost) that it takes to revisit an area of interest anywhere on Earth. This exciting prospect will enable assessment of threats, risks, and changes even more quickly than before. The industry will take advantage of the increasing volume of data and quicker revisit times to give operators the most up-to-date analyses over their infrastructure. Nanosatellite technology will provide even faster assessment of threats, risks and changes to major ports to minimize consequences, financial, environmental, regulatory, and product loss.
- Date October 22, 2018
- Tags 2018 October