A Concept For Natural Gas Transmission Pipeline Monitoring

Transcript

Paper for IGRC 2001, Amsterdam 5th – 8th November 1 A CONCEPT FOR NATURAL GAS TRANSMISSION PIPELINE MONITORING BASED ON NEW HIGH-RESOLUTION REMOTE SENSING TECHNOLOGIES CONCEPTION POUR LA SURVEILLANCE DE CANALISATIONS DE TRANSPORT DE GAZ NATUREL BASÉE SUR DES TECHNOLOGIES DE TÉLÉDÉTECTION HAUTE DÉFINITION W. Zirnig Ruhrgas Aktiengesellschaft, Germany D. Hausamann DLR German Aerospace Centre, Germany G. Schreier Definiens Imaging GmbH, Germany ABSTRACT It is in the interest of any gas company to maintain the value of its pipelines and to protect them effectively against damage caused by third parties. As a result of global progress in high-resolution remote sensing and image processing technology, it is now possible to design natural gas pipeline monitoring systems with spaceborne sensors and to enter into targeted negotiations with satellite operators concerning sensor application. The concept developed by Ruhrgas Aktiengesellschaft, the technological leader and user, the German Aerospace Centre DLR, an applied research institute and Definiens Imaging GmbH, a developer of innovative image processing software, for a satellite-borne pipeline monitoring system involves the combination of high-resolution data supplied by various sensors via remote sensing systems and the context-oriented evaluation of these data using innovative image processing techniques. RESUME Toute entreprise gazière a intérêt de conserver la valeur de ses canalisations et de les protéger de manière efficace contre des dangers causés par des tiers. Les progrès réalisés sur le plan international dans le domaine des technologies de télédétection et de traitement d'images haute définition permettent de créer un système de surveillance de canalisations de gaz naturel sur la base de capteurs mis en orbite et d'entrer en contact, de façon ciblée, avec les opérateurs de satellites en vue d'une application pratique. La conception pour un système de surveillance de canalisations mis en orbite, développée par Ruhrgas Aktiengesellschaft en sa qualité de leader technologique et d'utilisateur, par le German Aerospace Centre DLR en tant qu'institut de recherche sur des applications, et par la Definiens Imaging GmbH comme société de développement des programmes de dépouillement d'images, comprend la fusion de données de télédétection haute définition émises par des capteurs différents et leur dépouillement suivant la situation rencontrée à l'aide de méthodes performantes d'analyse d'images. Paper for IGRC 2001, Amsterdam 5th – 8th November 2 BACKGROUND The rules and regulations governing the operation and monitoring of natural gas transmission pipelines vary from country to country. In some cases, the differences are quite significant. Regardless of the details of the individual requirements, it is in the interest of any pipeline owner to maintain the value of its pipelines and to protect them effectively against damage caused by third parties. The monitoring methods most widely used for natural gas transmission pipelines include foot patrols along the pipeline route and aerial surveillance using small planes or helicopters. These patrols prevent developments and events which could place high pressure pipelines, the surroundings of pipelines or security of supplies at risk. Although these methods ensure a high level of safety in pipeline operation, the cost is also very high. Global progress in high-resolution remote sensing technology using both aircraft-borne and satellite-borne sensors, the increase in low-cost computing capacity and the extension of image processing techniques towards automated feature and situation recognition [1] have resulted in considerable potential for the use of this technology for pipeline monitoring tasks. Ruhrgas AG (as the technology leader and user) the German Aerospace Centre DLR (an applied research institute) and Definiens Imaging GmbH (a developer of innovative image processing software) are therefore planning to design and realize a novel remote sensing based monitoring system. As part of the feasibility study presented in this paper, they have identified the remote sensing technologies that currently appear to be best suited for automating gas transmission pipeline monitoring and making monitoring more efficient. THE MONITORING TASKS OF PIPELINE OPERATORS Within the integrated European natural gas transmission system, Ruhrgas operates a highpressure gas pipeline system with a total length of about 11,000 km in Germany, extending from the North Sea to southern Germany and from Bavaria to the border with France. This pipeline system is monitored by regular foot and vehicle patrols and by air patrols carried out using small helicopters. Whereas pipelines can to a large extent be protected against thirdparty damage and potential leaks in open country can mainly be detected from helicopters, the current state of the art is that possible leaks under sealed road services can only be detected at an early stage by patrolling the pipeline route on foot with gas detectors. Monitoring breaks down into object and situation detection, gas leak detection and monitoring of soil movement. These tasks have to be carried out throughout the year at regular intervals, largely regardless of weather conditions. Minor changes to monitoring schedules are possible in extreme weather conditions. Although the areas monitored differ quite significantly in terms of soil characteristics, vegetation and building density, it is important for the remote sensing methods employed to be usable in almost all types of terrain including pipeline sections where the route is not clearly visible (Fig. 1). Paper for IGRC 2001, Amsterdam 5th – 8th November 3 Figure 1. Example of a Pipeline Route with Poor Visibility Most natural gas transmission pipelines are under a soil cover of about 1 m. Along the pipeline routes, the following situations have to be detected in a strip of 20 m on both sides of the centreline: - construction work, earth movement and excavation, laying of cables, sewers, drainage systems and pipes, erection of buildings, foundations, pylons, etc., - soil upheaval, erosion, deep vehicle tracks, water-logged surfaces, - planting of new shrubs and trees, discolouring of vegetation above the pipeline. In addition, any work carried out within a 200 m-wide strip must be reported if there is reason to believe that it may affect the pipeline route at a later stage. A gas leakage detection system must be capable of identifying possible small gas leaks with leakage flow rates of 0.01 - 10 m³/hr at an early stage. Any major gas leaks caused by severe damage to a pipeline are detected and reported directly by other systems. The ability to distinguish between methane from natural gas and other biogenic methane would be an added advantage. APPROPRIATE REMOTE SENSING TECHNOLOGIES AND THEIR PRESENT REALIZATION Table 1 gives a qualitative overview of remote sensing technologies which appear to be well suited for pipeline monitoring on the basis of the feasibility study. It is clear from the table that a complete monitoring system for natural gas transmission pipelines will call for a combination of different sensors. Paper for IGRC 2001, Amsterdam 5th – 8th November 4 Table 1. Suitability of Remote Sensing Systems for Pipeline Monitoring. ü = suitable in principle, (ü) = possibly suitable Sensor system Object recognition Leakage gas detection Earth movement monitoring ü LIDAR Thermography ü High-resolution optical systems ü (ü) ü Hyperspectral sensors (ü) (ü) Imaging SAR systems ü ü Interferometric SAR ü ü Microwave radiometer ü ü LIDAR (Light Detecting and Ranging) LIDAR is a laser light-based optical detection method involving the transmission of laser light in the ultraviolet, visible or infrared range and detecting and analysing the reflected light. LIDAR systems have already been used for many years for the remote sensing of air pollutants and various meteorological parameters. Experimental systems mounted on helicopters for the detection of major leakage from transmission pipelines have been tested in the USA and Russia [2, 3]. In order to measure trace gas concentrations, the DIAL (Differential Absorption Lidar) method is used. This method is based on the Beer-Lambert absorption law and on the absorption properties of the gas to be detected. In order to exclude atmospheric effects and diffuse reflection from the signal, two wavelengths are used for transmission. The first wavelength (λon) is absorbed by the gas, while the second (λoff) is not absorbed and serves as a reference. Figure 2. Schematic Diagram of DIAL Method This method may be used with either a closed or an open measurement path. In the case of an open measurement path (Fig. 2), the pulsed laser light is reflected back by particles or molecules in the atmosphere. Trace gas concentrations can be measured with a certain spatial Paper for IGRC 2001, Amsterdam 5th – 8th November 5 resolution. The result of the measurement is the product of the gas concentration and the absorption cross section, which is a function of the wavelength selected. With a closed measurement path, the radiation emitted by the laser is reflected back to the optical receiver by a topographic target. Measurements of concentrations at specific distances are only possible to a limited extent. The trace gas concentration is the value for the entire measurement path. In order to detect hydrocarbons such as methane or ethane from natural gas leaks, the laser must be set to a wavelength at which these gases have appropriate absorption lines. For a low measurement threshold, the absorption value of the gas should be as high as possible. However, especially in the case of gases such as methane with low background concentrations (~ 1.7 ppm), it is important to avoid saturation effects. Potential absorption curves for the detection of methane are located in the spectral range from 1.6 µm to 4.0 µm, with three significant bands at about 1.6 µm, 2.3 µm and 3.3 µm. The strongest absorption lines are located at about 3.3 µm. Ethane also has absorption lines in this area. However, overlapping with the absorption lines of water vapour must also be taken into consideration. The design calculations made in the course of work on the feasibility study indicate that appropriately selected LIDAR systems would be in a position to detect the required very small leaks of below 0.1 m³/hr from altitudes of up to 300 m. This method therefore appears to be suitable for gas leakage detection during regular aerial patrols with small helicopters. Thermography Thermography relies on imaging detectors which pick up the infrared radiation emitted by a body and convert it into a visible image of that body. Normally , the wavebands from 3 µm to 5 µm and 8 µm to 12 µm are used for thermography. Thermography has a variety of applications, including industrial quality assurance, the assessment of the thermal insulation of buildings, the location of fires and environmental monitoring. Various airborne camera systems are available on the market and can be integrated into monitoring systems. Fig. 3 shows a recording system combined with an optical camera installed on an aircraft. In the case of the automated monitoring of natural gas pipelines using radar or photographic systems, a combination of these systems with thermographic methods would allow more precise image evaluation and would increase the probability of detection or reduce the number of false alarms. Figure 3. Combined Airborne Thermographic and Photographic System Paper for IGRC 2001, Amsterdam 5th – 8th November 6 High-Resolution Optical Systems High-resolution optical systems are available for any platform. Digital images are normally recorded using linear arrays of photosensitive semiconductors (CCD = Charged Coupled Devices). Up to 12,000 of these semiconductors are required for an image line. The earth's surface is then scanned in lines at the speed of travel of the carrier. If semiconductor arrays tuned to different sections of the spectrum are connected in parallel, a multispectral image can be recorded. Earth’s surface images are commercially available. The most recent examples are the pictures taken by the IKONOS satellite operated by Space Imaging which was launched in September 1999 (www.spaceimaging.com) and, at 1 m in black and white (panchromatic), offers what is at present the highest commercially available resolution. As IKONOS orbits some 680 km from the earth, each semiconductor represents an area of 1 m x 1 m on the earth's surface. 11,000 semiconductors connected in series (each for 1 m on the earth's surface) allow an image with a width of 11 km. The information from these electronic components is digitized and can either be stored on the satellite or transmitted directly to a ground station in the line of sight. As an example of the image quality which can be obtained, Fig. 4 shows a snow-covered landscape in southern Germany, with long shadows cast by the low winter sun. Vehicles of different sizes can clearly be seen on the motorway in the picture. Figure 4. Section from an Image Recorded by the IKONOS Space Imaging Satellite. Resolution: 1 Pixel = 1 m x 1 m. However, an even higher resolution of about 0.5 m would be needed for pipeline monitoring. This resolution has already been licensed by the US authorities and is planned by Space Imaging for 2004 and by its competitor EarthWatch for the next satellite launch at the end of 2001. Of course, optical images of the earth's surface can only be recorded from space over cloudless skies. The use of these techniques for routine pipeline monitoring is therefore only conceivable if several satellite systems could be employed jointly to achieve high repetition rates and thus compensate for any monitoring limitations imposed by poor weather Paper for IGRC 2001, Amsterdam 5th – 8th November 7 conditions. Even in the unfavourable climatic zone of central Europe, it can be assumed that adequately sized sections of the earth's surface would be visible at intervals of several days. Hyperspectral Sensors Hyperspectral sensors measure the degree of reflection of natural and artificial objects with high spectral resolution, allowing a variety of surface types and objects to be identified. Many elements on the earth's surface (vegetation pigments, minerals, rock, artificial surfaces) show specific absorption characteristics in defined wavelength bands and thus allow a quantitative analysis. The use of imaging systems makes it possible to identify objects by a combination of their spectral signatures and their three-dimensional characteristics. In this way, it should be possible to detect changes in vegetation caused by natural gas escaping under the ground surface. Although the geometrical resolution of airborne sensors should be adequate for pipeline monitoring purposes, the planned satellite-mounted hyperspectral sensors (such as OrbView 4) only reach a resolution of 8 m. Because of US government security requirements, the resolution may even be reduced to 20 m before the data are made available. Imaging SAR (Synthetic Aperture Radar) Systems SAR systems provide a holographic image of the area scanned by the radar. As a result of the wavelengths selected, the radar can even penetrate clouds of water vapour, allowing the earth's surface to be monitored largely unaffected by weather conditions. However, image resolution depends on the wavelength used and on the size of the aperture or antenna. With antenna lengths of a few metres, the radar waves (with a wavelength of a few decimetres) could only detect paths with a length of several kilometres on the earth's surface. The antenna is therefore extended synthetically to a length of several kilometres by computer-aided addition of the signals received at the various antenna points. However, even a very small metal reflector gives a strong reflection signal and objects with metal edges can therefore be detected very effectively. As a result of the complex image processing functions involving the wavelength and phase of the active signal, SAR can be used to detect features invisible to the human eye. Interferometry can be used to obtain three-dimensional images. Image resolution depends on the system parameters and the processing methods used and is now about 6 m for satellite-borne systems, while airborne systems offer resolutions down to 0.5 m. Significant increases in the image resolution of commercially available satellite systems can be expected over the next three to four years. Radarsat II (planned for 2003) will have a resolution of about 3 m while TerraSAR (launch scheduled for 2004, www.astrium-space.com) is expected to reach 1 m. The technology is largely independent of the time of day and of weather conditions. However, because of system-inherent features, the earth's surface is always viewed obliquely and shading or distortion effects may therefore occur [4]. Interferometric SAR Interferometric SAR uses the phase information contained in the radar waves of two or more SAR images to develop terrain models and detect ground surface movements in the centimetre range. With tandem operation of identical SAR satellites such as the combined flights of the European ERS-1 and ERS-2 and the planned operations of Radarsat II and Radarsat III, images of the same area can be recorded with very short intervals of one day (ERS) or even only a few minutes (planned by Radarsat). As regards pipeline monitoring, this method could conceivably be used for detecting subsidence following water abstraction and the collapse of subterranean hollows or for monitoring slopes subject to slippage. Paper for IGRC 2001, Amsterdam 5th – 8th November 8 Microwave Radiometers Microwave radiometers use a scanning antenna to detect radiation in the microwave range and allow a vertical view of the earth's surface. The microwave radiation received is proportional to the so-called apparent temperature of the surface observed, which is in turn a function of the emission and radiation properties of the surface. A number of experimental systems with promising results have been developed. They could be used for further investigations in connection with pipeline monitoring tasks. From altitudes between 100 m and 3,000 m this method could be used almost irrespective of the time of the day and weather conditions to obtain information on objects along the pipeline route. DATA FUSION AND IMAGE PROCESSING APPROACHES In order to manage and interpret the extremely large volumes of data contained in highresolution optical and SAR images, it will be necessary to combine the most advanced information and knowledge systems available with object and signal analysis methods. The objective must be to automatically extract easy-to-handle warnings about pipeline hazards with a very low proportion of false alarms from the data available. Advanced data management procedures which will need to be adapted to the specific requirements of pipeline monitoring include data mining, change detection and feature detection. Data Mining Data mining is a term used to refer to automatic searching for previously specified signals, objects and features in large volumes of graphic data. In contrast to databases, the data in an image are not available in a sorted form but only in the form of pixels. Data mining has given rise to the investigation of new data access and distribution methods such as information mining, scene understanding, synergetic decompression and classification, data and information visualization, user adaptation and the semantic modelling of information extraction. Studies on these subjects and test systems are available [5, 6, 7]. The classic task in the processing of remote sensing data is the interpretation of an individual image and the radiometric information recorded in the various spectral channels. Interpretation can be made significantly more precise by combining information from a number of different sources, e.g. by fusing data from different sensors or data recorded by the same sensor at different times (multitemporal analysis) or by combining these data with information from geographic information systems. Change Detection and Feature Detection Change and feature detection are the main strategies for detecting changes in the vicinity of a pipeline route. In change detection, image data are compared with the corresponding data from an earlier image pixel by pixel. Changes in the scenery are reflected by differences between the corresponding pixels. This very direct method results in problems with natural changes in vegetation, lighting or surface conditions (snow or rain) and the resulting radiometric changes in the pixels. Comprehensive corrections are required to reduce the very high proportion of false alarms generated by automated change detection. In addition, pixelbased change detection calls for comprehensive processes and reference data for geoencoding. Paper for IGRC 2001, Amsterdam 5th – 8th November 9 Feature detection is better suited for analysing changes in complex scenery. This method includes the identification and generation of objects from the original pixel-based files and the establishment of semantic links between these objects and known features, for example in the form of a feature database. Image features such as vehicles or pits are classified on the basis of radiometric, geometric and other links between the image objects and placed in relation to neighbouring objects and known information from geographic information systems. Any vehicles detected not on defined roads but in open country near to the coordinates of a pipeline route are therefore identified as potential hazards. By assessing the area concerned, assumptions can be made concerning a possible distinction between agricultural vehicles and construction equipment. This combination of object identification and a semantic knowledge network would appear to be an image processing procedure which is especially well suited for pipeline monitoring. eCognition [8] (www.definiens.com), a commercial analysis system which has recently been introduced, allows data from different sensors (optical, radar, infrared) to be merged and combined with graphic information system data for object identification. SYSTEM CONCEPTS AND FURTHER DEVELOPMENT STEPS Any system to be used for the monitoring of pipelines on the basis of remote sensing must meet two major operational requirements; • The data of the routes monitored must be available at regular intervals shortly after recording. • The specification and quality of the data must be such as to allow the definite identification of objects and situations representing hazards to the pipeline on the basis of a rapid automatic evaluation process. The results of the feasibility study confirm that the present state of the art in remote sensing and evaluation technology makes it possible to implement airborne sensor concepts featuring: • new imaging sensors and image evaluation techniques for automatic object detection, • SAR systems for pipeline monitoring with a high degree of independence from weather conditions, • laser systems for detecting minor leaks. Moreover, in view of the growing fleet of commercial high-resolution optical and SAR satellites in space (Table 2 and 3), it is quite conceivable that certain pipeline monitoring tasks could be transferred to spaceborne sensors, at least for situation and object detection purposes. Prospects are promising for optical systems with resolutions better than 1 m and future SAR systems in the 1 to 3 m range [9]. As a fundamental prerequisite for the automated evaluation of image data, it will be essential for the position of natural gas transmission systems to be defined with sufficient accuracy in digital form in a geographic information system for transfer to the evaluation system. The image processing programmes available for automatic object and situation Paper for IGRC 2001, Amsterdam 5th – 8th November 10 detection will need to be developed still further and adapted to the application. Finally, the resulting data need to be bind into the pipeline operators’ communication systems. Table 2. Existing and Planned Commercial High-Resolution Satellites. Year Satellite 1999 IKONOS-1 2001 EROS-A1 2001 2001 2001 Quickbird-2 Orbview 3 Cartosat 2001 EROS-A2 2002 SPOT 5 2004 IKONOS-2 Resolution 1 m pan, 4 m multi-spectral 1.8 m pan, 4 m multi-spectral 0.6 m pan 1 m pan 1 – 3 m pan, simultaneous stereo 1.8 m pan, 4 m multi-spectral 3 m pan, simultaneous stereo 0.5 m pan Operator Space Imaging International Imagesat Country USA Israel/USA EarthWatch Orbimage Indian Space Agency/ANTRIX Imagesat USA USA India Israel/USA Spot Image France Space Imaging International USA Table 3. Existing and Planned Non-Classified Radar Satellites. Year 1991 1995 1995 2000 2001 2003 2003 2004 2004 Satellite ERS-1, ERS-2 Radarsat I SRTM ENVISAT ALOS Radarsat II TerraSAR Cosmo Skymed Radar band & max. resolution C-Band, 25 m Operator Country ESA Europe C-Band, 8 m L,C & X-Band, 25 m C-Band, 25 m L-Band, 10 m C-Band, 3 m L & X-Band, 3 m X-Band, 1 m Radarsat Intl., CSA NASA, DLR ESA NASDA Radarsat Intl., CSA InfoTerra/ASTRIUM ASI, Alenia Canada US & D Europe Japan Canada D & UK Italy The next step will be to draw up a detailed specification for the interaction of different satellite-borne sensor systems, suitable image processing programs and communication interfaces and to verify system interaction using suitable test cases. For this purpose, data from targeted aerial patrol campaigns and commercial satellite images could be linked and evaluated using the software modules referred to above. At the beginning of this year, European gas pipeline operators, national space agencies, sensor developers and system integrators formed a working group with a view to specifying a satellite-supported pipeline monitoring system and to quantifying the potential for improving the effectiveness of such a system. There are good prospects that a satellite-supported pipeline monitoring system may be ready for operational use within the next ten years and that certain parts of such a system may even be implemented earlier. Paper for IGRC 2001, Amsterdam 5th – 8th November 11 CONCLUSIONS In-depth correlation of the requirements of pipeline operators with the possibilities offered by remote sensing technologies will allow targeted practical discussions with satellite operators. In addition, the international space station, ISS, which is currently being built up, will provide a suitable platform for possible future tests without requiring the launch of a special satellite. In this way, it could prove possible to develop a focussed pipeline monitoring system based on remote sensing technology which would provide effective, reliable support to pipeline operators throughout the world and help to detect hazards to pipeline systems and the environment at an earlier stage and more comprehensively than has previously been the case. ACKNOWLEDGEMENTS The authors wish to thank the remote sensing experts in the specialist departments of the DLR Institutes at Oberpfaffenhofen, the Institute of Photogrammetry and Remote Sensing of the University of Karlsruhe, Aero-Sensing Radarsysteme and Astrium and the experts of a number of gas companies whose know-how and willingness to engage in discussions have been instrumental in developing this system concept. REFERENCES 1. Bähr, H.-P.; Vögtle, T. editors: Digitale Bildverarbeitung. Heidelberg: Wichmann Verlag 1998. 2. McRae, T.G.: Evaluation of a new aerial leak survey approach. GRI Summary Report 1996 (GRI-96/0376) 3. Philippov, P.G., et al.: DIAL-Infrared lidar for monitoring of main pipelines and gas industry objects. In: Optical remote sensing for industry and environmental monitoring, Proceedings of Beijing meeting 1998 (SPIE Proceeedings, Vol. 3504), p.119-127. 4. Schreier, G. editor: SAR geocoding – data and systems, Karlsruhe 1993. 5. Datcu, M.; Seidel, K.; Schwarz, G.: Information mining in remote sensing image archives. In Kanellopoulos, I.; Wilkinson, G.; Moons, T., editors, Machine Vision and Advanced Image Processing in Remote Sensing (MAVIRIC), pages 199-212. Springer, 1999. 6. Schröder, M.; Rehrauer, H.; Seidel, K; Datcu, M.: Interactive learning and probabilistic retrieval in remote sensing image archives, IEEE Trans. on Geoscience and Remote Sensing, Vol. 38, pp. 2288-2298, 2000. 7. Datcu, M.; Seidel, K.: Bayesian methods: applications in information aggregation and data mining. International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-43 W6, pp. 68-73, 1999. 8. Baatz, M., Schäpe, A.: Object-Oriented and Multi-Scale Image Analysis in Semantic Networks. In: Proc. of the 2nd International Symposium on Operationalization of Remote Sensing August 16th – 20th 1999. Enschede. ITC., 1999 9. Petrie, G.: Commercial High Resolution Space Imagery, a very long gestation & a troubled birth; GeoInformatics, March 2001.