A MULTIPLE REMOTE SENSOR SYSTEM FOR THE AERIAL SURVEILLANCE OF THE NORTH SEA AND BALTIC SEA*

Transcription

1 A MULTIPLE REMOTE SENSOR SYSTEM FOR THE AERIAL SURVEILLANCE OF THE NORTH SEA AND BALTIC SEA* O. Trieschmann, Th. Hunsänger, U. B. Barjenbruch German Federal Institute of Hydrology Koblenz, Germany ABTRACT Marine pollution in the sensible North Sea and Baltic Sea forces an international aerial surveillance. Within this framework, the German aerial surveillance system operates an advanced instrumentation on board of two Dornier 228 aircrafts. The instrumentation consists of a set of remote sensors within an improved sensor network of the second generation. The aim is to detect oil spills on the water surface, emitted accidentally or illegally. For surveillance purposes the spills have to be located within a large area. If spills were identified, the pollution has to be classified and quantified with a high accuracy. This paper describes the set of instruments and their potential to fulfill these demands. A sideward looking airborne radar (SLAR) operates to locate the discharge, whereas a line scanning infrared/ultraviolet line scanner (IR/UV) allows to examine the polluted area more precisely with a higher resolution. The IR/UV sensor and especially the line scanning microwave radiometer (MWR) on board the aircraft allow to quantify the thickness of the oil film. A laser-fluoro line scanning sensor (LFS) classifies the oil as well as organic material, chlorophyll, and gelbstoff. A forward looking infrared imaging camera completes the set of instruments that allow to convict the polluter. A detailed technical description of the sensors is presented with respect to the specific interests of potential administrative and scientific users. Results of sensor measurements and their synergy effects are given. 1.0 INTRODUCTION During the last two decades, the German Federal Ministry of Transport set up an airborne surveillance system to monitor the German territorial waters in the North Sea and Baltic Sea for oil discharges and marine pollution. These activities are within an international effort culminated in the MARPOL Agreement and HELSINKI Convention defining an international standard on the protection and surveillance of the North Sea and Baltic Sea. The airborne maritime surveillance system consisted of two DO 28 aircraft each equipped with a Side-Looking Airborne Radar (SLAR) for detecting oil slicks over large distances, a UV/IR scanner for mapping the sea surface in the nadir range, and TV and photographic cameras for the identification of the polluter and for collecting evidence. * Presented at the Fifth International Airborne Remote Sensing Conference, San Francisco, California, September 2001.

2 In 1991 and 1996, the German Federal Ministry of Transport upgraded the airborne maritime surveillance system by replacing the Dornier DO 28 by Dornier DO aircraft. The DO aircraft allows a flight time of up to 6 hours. An overview of the upgraded surveillance system in an early stage of its implementation is given by Grüner et al., The instrumentation of one aircraft was supplemented by a Microwave Radiometer (MWR) and a Laser-Fluorosensor (LFS) in 1993 for the more detailed analysis of spills. The second DO 228 aircraft is equipped with the standard instrumentation: SLAR, IR/UV sensor, and TV camera and will be retrofitted by end of The forward looking infrared imaging camera with laser illuminator completes the set of instruments that allows to convict the polluter, but it will not be discussed in detail, since no experiences are available yet. All sensors are mapping devices. With this upgrade, the new network and communication system MEDUSA will be completely established (see conference paper by Hunsänger et al., 2001). Figure 1: Profile of the Do228 LM1 aircraft with measuring equipment Detailed technical descriptions of the sensors will be presented in the Chapters two to five. The interests of administrative and scientific users are described in the following paragraph to better understand the specific selection and design of the instruments. Emphasis will be put on the description of the LFS, since it is the first instrument of its kind that meets operational requirements for long-term use on board an aircraft. Finally, the sensors will be compared to show the potential of the complete sensor system and the synergetic effects of combined data evaluation OPERATIONAL REQUIREMENTS The aim of the German aerial surveillance system is to detect oil spills on the water surface and in the upper water layers, emitted accidentally or illegally. For surveillance purposes the spills have to be located within a large area. If spills were identified, the pollution has to be classified and quantified with high accuracy. As a secondary product, ecological conditions in coastal waters should be determined. To fulfill these demands, remote sensing equipment was selected which is independent of daylight and weather. By taking advantage of the combination of measurements by different physical effects, the requirements can be satisfied. The determination of oil pollution and its classification lead to the following tasks: surveillance of the sea surface over long distances with a long-range sensor (radar) for registering structures which might indicate the presence of oil pollution,

3 subsequent analysis of the extent and type of a pollution incident with short-range sensors (IR/UV line scanner, MWR, LFS) which, for example, allow to distinguish the film thicknesses within the affected area. The following general requirements had to be met: state-of-the art data processing and display concept ensuring the best possible operation for the detection and combating of marine pollution as well as the collection of evidence to convict identified polluters, transmission of the data measured in the aircraft to the off-shore or land-based operation centers, a mission time of approximately 2 to 4 hours for routine surveillance flights and a mission time of 3 to 4 hours above the area of operation for airborne guidance of combating vessels, a maximum flight route of approximately 1,300 km, a minimum of approx. 800 flight hours per year (per aircraft, without approach and departure) to perform the main task. 2.0 SIDE-LOOKING AIRBORNE RADAR (SLAR) The SLAR is the primary sensor for long-range detection of oil pollution on the sea surface. The radar transmits high-frequency pulses in the X-band perpendicular to the flight direction to both sides of the aircraft via a cylindrical antenna mounted at the bottom of the aircraft. The pulses are scattered back by the rough, undulating sea surface or by ships. Pronounced azimuthal beam compression, together with the forward movement of the aircraft make two dimensional scanning possible which is visualized as gray-level image. Even very thin oil films smooth the sea surface and thus prevent radar return. If there is no radar return, it is permissible to conclude that an oil film is present. It is, however, also possible that other phenomena, like natural substances floating on the sea surface, local changes in the wind field or variations of the bottom topography smooth the water surface making it impossible to distinguish these areas from oil slicks. This procedure is, to a large extent, independent of weather and visual conditions and allows the detection of oil pollution through a cloud cover. 3.0 INFRARED/ULTRAVIOLET SCANNER (IR/UV) The IR/UV scanner scans the sea surface below the aircraft line by line in the ultraviolet and in the thermal infrared. The scan area is about ±250 m perpendicular to the aircraft at a flight level of 300 m with a resolution of a few meters. The UV channel (λ = 0.32 to 0.38 µm) records the sunlight reflected on the water surface. Due to the short wavelengths, even very thin oil films of less than 0.1µm can be detected. However, it is not possible to deduct from the data of the UV channel the film thickness of an oil spill. Since it is dependent on sunlight, the application of the UV channel is limited to daylight conditions and requires good visibility. The IR channel (λ = 8.5 to 14 µm) measures the thermal emission of the sea surface. Its sensitivity limit regarding the oil film thickness is around 10µm. Due to its somewhat lower emissivity, the oil seems to be colder than the water surface. However, this effect is not unambiguous, since oil films with a thickness of more than 0.5 mm absorb sunlight and can, therefore, appear to be warmer than the surrounding water surface on sunny days. Because of this ambiguity, it is not possible to obtain from the IR channel data any absolute information on layer thickness. The IR channel is not dependent on daylight and can, consequently, also be used at night, but then, due to the cooling of the oil the black/white contrast can be inverted again. Due to the very high absorption of water vapor it is not possible to detect

4 oil through a cloud cover. But as opposed to satellite measurements the aircraft is able to fly below the cloud layer. Figure 2: IR/UV pictures of an oil polluted water surface of 1200x500m capturing 503l heavy crude oil. Also LFS data for oil classification is shown on the left. 4.0 MICROWAVE RADIOMETER (MWR) The microwave radiometer is a short-range sensor. Like the IR/UV scanner, it measures natural radiation. Its sensitivity, however, is in the cm to mm wavelength range of the electromagnetic spectrum. Therefore it is a system which is almost weather-independent (penetration of clouds and fog) and which is capable of making a quantitative analysis of heavy oil slicks. The MWR relies on wavelength-selective interference phenomena for layer thicknesses in the cm or mm ranges, comparable to color iridescence in the case of very thin oil films. Thus, the modular multiple-frequency radiometer allows the unambiguous determination of film thicknesses in the extended range from 0.05 to 2.5 mm with an improved geometrical resolution of 5 m at a flight level of approx. 300 m. The measuring frequencies are selected so as to comply with the following aspects: avoidance of interference by artificial microwave radiation (18.7, 36.5 and 89 GHz), thickness and volume determination for thicker layers (18.7 GHz) with reduced geometrical resolution (22 m) high geometrical resolution (5 m) and sensitive detection of thin layers or of whitecaps (89 GHz), best-possible compromise regarding geometrical resolution (11 m) and all-weather capability (36.5 GHz), an additional 89 GHz radiometer (zenith radiometer) picks up atmospheric radiation with a fixed aerial in an angle range of 8 times 80 (8 in flight direction). The sea surface is scanned with the help of two parabolic mirrors which rotate around their common axis at 10 revolutions per second. Two identical receiver groups with three detection channels

5 each are coupled to these mirrors via level deflection reflectors (Figure 4). In this way, it is possible to achieve a rate of 20 lines per second. The calibration is based on hot/cold measurements for every complete revolution. Figure 3: Measurement example of an crude oil field with approx. 5m 3 oil impact; maximum layer thickness 1.8mm The picture above (Figure 3) shows the MWR image of water polluted by oil in direct comparison with an IR image and an UV one. The MWR shows the areas of higher thickness, opposed to the IR and especially the UV, which also indicate the spatial extent of the thin oil film. 5.0 LASER FLUOROSENSOR (LFS) The basis of the application of the LFS for the detection of oil pollution on the water surface are the spectroscopic properties of oil. Mineral oils consist of a complex mixture of various hydrocarbon compounds with widely differing fluorescence-spectroscopic properties. Due to the differences in the composition of mineral oils, the fluorescence spectra of different oils show variations with respect to the spectral form as well as to the intensity of the fluorescence observed. The laser-induced fluorescence signal, therefore, contains a great deal of substance-specific information, which allows the differentiation between different types of oil as well as an estimation of the oil quantity released. Similarly, marked differences are also shown by the spectra of substances which may occur naturally on the sea surface (e.g. fish oil or excretions by algae) or of substances whose discharge is deemed harmless (e.g. edible oil), so that it is also possible to distinguish between these substances and mineral oils. Therefore, a large number of mineral oil samples were systematically examined and used to establish a catalogue of optical signatures which provides the basis of classification (Reuter, 1992). Based on intensive research and development activities during the 1980s, a new generation of imaging LFSs was created. Two high-powered pulse lasers, an XeCI excimer laser (308nm, 150mJ/20ns) for the analysis of oils, fulvic acid (gelbstoff), and organic pollutants; and a dye laser (253, 383nm, 20 mj/15 ns) for chlorophyll-fluorescence excitation generates very short intensive light pulses in the ultraviolet or visible spectral ranges. The laser light is partially returned by elastic and Raman scattering

6 and fluorescence on the water surface or in the upper water layers. Some of this light is picked up by a telescope (type Schmidt-Cassegrain with 20cm aperture) with high throughput and registered by a 12- channel optical spectrograph. This spectrograph consists of 12 identical modules with photomultiplying, range-gated detectors spectrally separated by dicroitic and interference filters and a numerical resolution of 11 bit by the ADC. Each channel has a typical bandwidth of 10nm. The spectral channels were selected by the constraints to be aligned to the Raman scattering and to fluorescence signals of oil and organic substances, which should be independent from each other. Since these fluorescence effects are spectrally broad features, a main-component analysis was performed by Hengstermann, 1992 to be able to separate the species: Table 1: Detector channels of the LFS Wavelength Optical effect Comments 330 nm Fluorescence of oil and gelbstoff for derivation of the baseline of the Raman peak 344 nm Raman scattering of water (Excimerlaser) 3400 cm -1 shifted from laser wavelength 365 nm, 382 nm, 410 nm Fluorescence of oil and gelbstoff for derivation of the baseline of the Raman peak 440 nm Raman scattering of water (dye-laser) 470 nm Basic Raman scattering 500 nm, 550 nm, 600 nm, 650 nm Fluorescence of oil and gelbstoff 685 nm Fluorescence of oil and chlorophyll a aligned to peak of chlorophyll fluorescence The elastic scattering is blocked, since it is superimposed by the reflection of the water surface which is not predictable. Using a conical scanner allows two-dimensional mapping of the sea surface in the nadir range. Maps of oil film thickness and hence the released volume as well as substance classes are derived from these measurements. For hydrographic and biological applications, the sensor is used to measure the concentration of gelbstoff and algae (phytoplankton) by their fluorescence and the attenuation by the water Raman signal. The phytoplankton distributions, e.g. for forecasting algae blooms, are measured by their chlorophyll fluorescence excitation. If the visibility is sufficient, the LFS can be used during day and night and is, within certain limits, practically independent of the sea state. The detection of oil and organic compounds is also possible underneath the water surface. As a secondary product, the turbidity of the water can be determined through analyzing the backscatter signal of the water surface. The measuring procedure is applied at flight levels between 160m to 500m. The LFS can make a very sensitive determination of the layer thickness between 0.1 and 20 µm of an oil film on the sea surface by using characteristic signals from the water column which are drastically attenuated through absorption by oil on the water surface. The absorption of light by the oil depends on the specific type of the oil, but in all cases it is so high that the laser beam is completely absorbed in the oil layer, even with a layer thickness as low as 100µm, and signals from the water column are no longer registered. The magnitude of the attenuation of the signals from the water column is a direct measure of the thickness of the oil layer. Clear water has a very characteristic signal which, due to its narrow bandwidth, can be easily isolated from the fluorescence of substances in the water and of mineral oil. By determining the signal of the oil-free water surface in the immediate vicinity of an oil spill, it is possible to directly determine the layer thickness from the attenuation of the signal within the area polluted by oil.

7 The hydrographic fluorescence Lidar technology is described extensively in the literature (e.g. the review by Measures 1984). Layout and development of the LFS described in this paper was performed at the University of Oldenburg. Specific properties of the LFS system are indicated in Reuter et al., Figure 4: Schematic layout of MWR construction Figure 5: Schematic layout of LFS construction (Excimer laser) 5.1. DATA ACQUISITION AND PROCESSING First of all, the data have to be corrected for background illumination and scan geometry (see Reuter, 1993). Identification and classification of the oil is based on a number of pre-defined oil classes (natural and mineral oil classes) by using a main-component analysis (Hengstermann,1992). Signals from natural organic compounds in seawater, like gelbstoff and phytoplankton pigments, generally interfere with the fluorescence emission of oils and need to be considered when interpreting data. The fluorescence signals from gelbstoff and algae are normalized to the Raman signal for compensation of attenuation effects. The attenuation coefficient can be estimated from the inverse Raman scattered signal, if the data are compensated for geometrical (i.e. flight altitude and geometry of scan) and sea-state effects at the air-water boundary. The background signal from daylight can be a problem when intense sun glitter directly reaches the detector or when the fluorescence signals are very weak. For routine operation, algorithms have been derived to recognize faulty data and exclude them from further processing steps. Problems with sun glitter are minimized by selecting footprints of the scan pattern in the opposite direction of the azimuth of the sun. The signal-to-background ratio can be further improved by selecting lower flight altitudes. All data are evaluated online during flight, and all results are directly available to the operator. Kriging interpolation between flight tracks is used to generate maps of these parameters. An online test of the data quality during measurements is important to ensure the operation of the sensors with an optimal set-up under changing conditions (mainly a change of water type).

8 5.2. EXPERIMENTAL RESULTS OF LFS Since November 1993 the LFS has been operated as a component of the sensor package in one of the maritime surveillance aircraft. The prototype has been checked in mechanical, electronic, and climatic endurance tests (crash calculations, eye safety of the laser at the operational altitude, and qualification of the gas reservoir) and is now certified for permanent aircraft use. Regular flights are carried out over the coastal waters of Germany, i.e. the German Bight and the Western Baltic Sea in the framework of the aerial surveillance. In Figure 2, the LFS data are compared with the IR and UV mappings. Whereas the IR and especially the UV channel can determine the lateral extent of the pollution with high resolution, the LFS determines the quantity (e.g. 503l) and type of the oil pollution (e.g. heavy crude oil with a probability of 98%). Additional flights with the LFS operating in the hydrographic mode have been performed in the Canary Islands region (Reuter et al.,1997). Operational hydrographic measurements from the winter and spring periods of 1999/2000 over the coastal waters of Germany were analyzed by Zielinski, Due to bad weather and instrument conditions only few days of measurement remain for data evaluation. But it could be shown that the LFS is capable to measure continuously gelbstoff and phytoplankton along the flight track and therefore proves to be an important sensor for monitoring these species in larger aereas over extended time periods. 6.0 COMBINATION OF ALL SENSOR DATA The potential of the surveillance measurements consists in the combination of several sensor data. While the SLAR is the only wide-range sensor covering a large area to find any potential oil pollutions, the narrow-range sensors can qualify, quantify, and identify the oil or organic material. While the IR/UV scanner has its potential in quantifying the spatial extent of the spill, the MWR and the LFS evaluate a wide range of oil film thicknesses. Table 2 identifies the characteristic features of the individual sensors and allows a comparison between them. Table 2: Characteristic properties of the sensors. SLAR UV IR MWR LFS Range wide narrow narrow narrow narrow Classification capabilities no no no no yes Sensitivity to oil film thicknesses N.A. >0.1µm >10µm 50µm to 2.5mm 0.1 µm to 20 µm Horizontal range from flight altitude ±30km ±250m ±250m ±250m ±75m Spatial resolution 10m (flight track) 75m (perp.) 3.5m 3.5m >5m Detection of oil spills underneath the surface no no no no yes Operating at night yes no yes yes yes 10m pixel-to-pixel distance Film thickness determination no no no 50µm to 2.5mm 0.1 µm to 20 µm Measuring geometry Line-by-line Line-by-line, 160Hz Line-by-line, 160Hz Sensitivity no On clouds On clouds no Line-by-line, 20Hz Conical, 5Hz (20Hz max.) On clouds, flight altitude

9 The underlined features of each sensor in Table 2 justify to call this sensor as a indispensable element of the system. The functionality of one sensor cannot be replaced by the functionality of another single sensor or a combination of other sensors. However, there are, except for the classification by the LFS, comparable measurements by different sensors which allow a reliability check of the results. 7.0 CONCLUSION The upgraded aerial surveillance system described in this paper is equipped with an enhanced set of sensors. The combination of them on an aircraft has proved to be successful in terms of aviation compatibility, instrumental reliability, and in terms of the potential of the measurements. It could be shown that additionally to the standard results as the spatial oil extends, the spilled volume and type can be determined. The selected instruments are complementing each other so that the task of aerial surveillance can be fulfilled under different weather conditions and during day and night. During the measurements over the North Sea and Baltic Sea it could be shown that the system and especially the LFS is capable to support monitoring programs e.g. of phytoplankton. Nowadays, monitoring is performed by sampling water at different sites. The routine flights of the aerial surveillance could support the monitoring program by continuous measurements along the flight track. Still the measurement accuracy of concentrations in the samples is better than the derived concentrations from LFS data. However, the data obtained by sampling could serve as ground truth for calibration the LFS data. Especially the ability of the LFS to determine even very small quantities of oil quantitatively and to identify the type of oil or organic material makes it an indispensable sensor of state-of-the-art sensor equipment for aerial marine surveillance. 8.0 ACKNOWLEDGEMENTS This work was supported by the German Federal Ministry of Transport, Building, and Housing. We are grateful to the staff of the "Marinefliegergeschwader 3 - Graf Zeppelin" who operate the aircraft and to the Optimare GmbH for their co-operation. 9.0 REFERENCES K. Grüner et al., A New Sensor System for Airborne Measurements of Maritime Pollution and Hydrographic Parameters. GeoJournal, 24; Th. Hengstermann, Untersuchungen zur Laserfernerkundung mariner Ölverschmutzungen, Ph.D. Thesis, Univ. of Oldenburg, Oldenburg, 1992 Th. Hunsänger et al., Advanced Operational Data Processing in a Distributed System for Maritime Surveillance. Fifth International Airborne Remote Sensing Conference, San Francisco, California, September R.M. Measures, Laser Remote Sensing: Fundamentals and Applications, John Wiley & Sons, New York, 1984, 510 pp R. Reuter et al., "Hydrographic Laser Fluorosensing: Status and Perspectives" In Operational Oceanography. The Challenge for European Co-operation, eds. J.H. Stel et al., Elsevier Science B.V., 1997 R. Reuter et al., Catalog of optical signatures of oils, Univ. of Oldenburg, Germany, 1992 O. Zielinski et al., Operational Airborne Hydrographic Laser Fluorosensing, 4th Workshop on Lidar Remote Sensing of Land and Sea on the 20th EARSeL Symposium, Dresden, June 2000