Difference between revisions of "Oil spill monitoring"

From Coastal Wiki
Jump to: navigation, search
Line 42: Line 42:
 
There are two configurations of active side-looking radar systems: [https://en.wikipedia.org/wiki/Synthetic-aperture_radar Synthetic Aperture Radar (SAR)] and side-looking airborne radar (SLAR). SLAR is cheaper and uses a horizontal antenna to produce imagery along the flight path. Synthetic aperture  radar uses the forward motion of the platform to attain spatial resolution. SAR resolution is not dependent on range, but uses extensive electronic processing to produce high resolution images. SAR has larger range and resolution than SLAR and is used in all radar satellites. SLAR is mostly used for airborne oil spill remote sensing, as it is cheaper. The imaging geometry of SAR and SLAR is an oblique projection type. The working wavelength, incidence angle, polarization mode of the radar sensor, surface roughness, and dielectric constant of the ground object all affect the backscattering of the signal. Primary pre-processing of SAR images involves radiometric calibration, geocoding, and land masking.  
 
There are two configurations of active side-looking radar systems: [https://en.wikipedia.org/wiki/Synthetic-aperture_radar Synthetic Aperture Radar (SAR)] and side-looking airborne radar (SLAR). SLAR is cheaper and uses a horizontal antenna to produce imagery along the flight path. Synthetic aperture  radar uses the forward motion of the platform to attain spatial resolution. SAR resolution is not dependent on range, but uses extensive electronic processing to produce high resolution images. SAR has larger range and resolution than SLAR and is used in all radar satellites. SLAR is mostly used for airborne oil spill remote sensing, as it is cheaper. The imaging geometry of SAR and SLAR is an oblique projection type. The working wavelength, incidence angle, polarization mode of the radar sensor, surface roughness, and dielectric constant of the ground object all affect the backscattering of the signal. Primary pre-processing of SAR images involves radiometric calibration, geocoding, and land masking.  
  
−
Winds above 1.5 m/s generate capillary waves at the water surface that appear as so-called 'sea clutter' on the radar image due to Bragg scattering<ref>Phillips, O.M. 1988. Radar Returns from the Sea Surface—Bragg Scattering and Breaking Waves. J. Phys. Ocean. 18: 1065-1074</ref>. As these capillary waves are damped by the oil slick, the slick appears as a dark patch on the SAR image where sea clutter is suppressed. However, other features may produce slick look-alikes, such as fresh water slicks, internal waves, wave shadows behind structures, floating macro-algae such as sargassum and kelp, etc. Several automatic or semi-automatic Radar Image Processing techniques have been developed that can interpret a radar image for oil slicks in a matter of minutes, especially using deep learning algorithms (e.g. decision tree forest<ref>Topouzelis, K. and Psyllos, A. 2012. Oil spill feature selection and classification using decision tree forest on SAR image data. ISPRS J. Photogramm. Remote. Sens. 68: 135–143</ref>, support vector machine classification<ref>Baek, W.-K. and Jung, H.-S. 2021. Performance Comparison of Oil Spill and Ship Classification from X-Band Dual- and Single-Polarized SAR Image Using Support Vector Machine, Random Forest, and Deep Neural Network. Remote Sens. 13, 3203</ref>, convolutional neural network<ref>Fan, Y., Rui, X.; Zhang, G., Yu, T.; Xu, X. and  Poslad, S. 2021. Feature Merged Network for Oil Spill Detection Using SAR Images. Remote Sens. 13, 3174</ref>) for feature extraction.
+
Winds above 1.5 m/s generate capillary waves at the water surface that appear as so-called 'sea clutter' on the radar image due to Bragg scattering<ref>Phillips, O.M. 1988. Radar Returns from the Sea Surface—Bragg Scattering and Breaking Waves. J. Phys. Ocean. 18: 1065-1074</ref>. As these capillary waves are damped by the oil slick, the slick appears as a dark patch on the SAR image where sea clutter is suppressed. However, other features may produce slick look-alikes, such as fresh water slicks, internal waves, wave shadows behind structures, floating macro-algae such as sargassum and kelp, etc. Several automatic or semi-automatic Radar Image Processing techniques have been developed that can interpret a radar image for oil slicks in a matter of minutes, especially using deep learning algorithms (e.g. [[Random Forest Regression|Decision Tree Forest]] <ref>Topouzelis, K. and Psyllos, A. 2012. Oil spill feature selection and classification using decision tree forest on SAR image data. ISPRS J. Photogramm. Remote. Sens. 68: 135–143</ref>, [[Support Vector Regression|Support Vector Machine classification]] <ref>Baek, W.-K. and Jung, H.-S. 2021. Performance Comparison of Oil Spill and Ship Classification from X-Band Dual- and Single-Polarized SAR Image Using Support Vector Machine, Random Forest, and Deep Neural Network. Remote Sens. 13, 3203</ref>, [[Artificial Neural Networks and coastal applications|convolutional neural network]] <ref>Fan, Y., Rui, X.; Zhang, G., Yu, T.; Xu, X. and  Poslad, S. 2021. Feature Merged Network for Oil Spill Detection Using SAR Images. Remote Sens. 13, 3174</ref>) for feature extraction.
 
SAR observations do not depend on weather (clouds and sunshine), which enables the detection of illegal discharges that most frequently appear during night. SAR can also survey areas during storms, where accident risks are increased. An example of an image is shown in Fig. 2.     
 
SAR observations do not depend on weather (clouds and sunshine), which enables the detection of illegal discharges that most frequently appear during night. SAR can also survey areas during storms, where accident risks are increased. An example of an image is shown in Fig. 2.     
  

Revision as of 21:46, 9 February 2024

The development of remote sensing techniques allows the detection and monitoring of oil spills. This article describes the possibilities, techniques and requirements to detect oil spills using remote sensing.

Introduction

The ability to remotely detect and monitor oil spills at sea is important due to the constant threat posed to marine wildlife and the ecosystem. Remote sensing can allow for early detection of slicks, provide size estimates, and help predict the movement of the slick and possibly the nature of the oil. This information will be valuable in aiding clean-up operations, and will not only help save wildlife and maintain the balance of the local ecosystem, but will also provide damage assessment and help to identify the polluters.

Remote sensing can use two types of sensor systems:

  • Passive systems: Passive systems make use of sensors that detect the reflected or emitted electro-magnetic radiation from natural sources (visible spectrum, reflective infrared and thermal infrared).
  • Active systems: Active systems detect reflected responses from objects that are irradiated from artificially-generated sources, such as radar or laser systems.

Remote sensing platforms for oil spill monitoring are satellites, aircrafts and drones.

Satellite monitoring

Remote-Sensing Satellites are characterized by their altitude, orbit and sensor. They cover vast areas and have a repetition time ranging from several to 16 days. In the past, times from the tasking of the satellite to image delivery were as long as 12 h, currently, times of 4 h are possible. A further consideration is overpass time. Several satellites cover an area once per day. Satellite constellations such as Cosmo (Constellation of Small Satellites for Mediterranean basin Observation) give revisit times of a few hours compared to the present one-day. Satellite-carried radars with their frequent overpass, high spatial resolution and their day–night and all-weather sensors are essential for detecting large spills and monitoring ship and platform oil releases.

Airborne monitoring

Fig. 1. Aerial image of an oil slick. Photo credit The Norwegian Coastal Administration/NOFO/Sundt Air.

Airborne oil spill monitoring has become a global practice over the last three decades. In the 1970s and 1980s a major effort has been directed toward developing sensors with enhanced oil spill monitoring capabilities based on various techniques like infrared/ultraviolet line scanners, microwave radiometers, laser fluorosensors, and X-band radar systems. Currently a multitude of specialized airborne remote sensing systems are operated, especially for the deterrence of potential polluters and support to oil spill clean-up activities.

Using satellite platforms to monitor oil spills is more cost effective than using airborne monitoring techniques but operation of aircraft is still the only possible way to perform a spatio-temporally flexible surveillance, so airborne monitoring can be seen as complementary to satellite monitoring. Users advocate a combined satellite and airborne monitoring service.

Oil monitoring sensors and techniques

Monitoring requirements

Figure 2. The Lebanese oil spill accident of July 2006. SAR image at 7:51 GMT 6 Aug 2006. Courtesy of ESA, INGV and JRC.

Requirements for oil slick monitoring are:

  • High temporal resolution, due to the changing nature of the oil and its immediate threat to the ecosystem
  • The ability to image a given area regardless of cloud cover and prevailing weather conditions (even time of day)
  • High spatial resolution, to identify individual small oil patches (windrows)
  • Capability to distinguish the oil from the adjacent water

Monitoring with radar

There are two configurations of active side-looking radar systems: Synthetic Aperture Radar (SAR) and side-looking airborne radar (SLAR). SLAR is cheaper and uses a horizontal antenna to produce imagery along the flight path. Synthetic aperture radar uses the forward motion of the platform to attain spatial resolution. SAR resolution is not dependent on range, but uses extensive electronic processing to produce high resolution images. SAR has larger range and resolution than SLAR and is used in all radar satellites. SLAR is mostly used for airborne oil spill remote sensing, as it is cheaper. The imaging geometry of SAR and SLAR is an oblique projection type. The working wavelength, incidence angle, polarization mode of the radar sensor, surface roughness, and dielectric constant of the ground object all affect the backscattering of the signal. Primary pre-processing of SAR images involves radiometric calibration, geocoding, and land masking.

Winds above 1.5 m/s generate capillary waves at the water surface that appear as so-called 'sea clutter' on the radar image due to Bragg scattering[1]. As these capillary waves are damped by the oil slick, the slick appears as a dark patch on the SAR image where sea clutter is suppressed. However, other features may produce slick look-alikes, such as fresh water slicks, internal waves, wave shadows behind structures, floating macro-algae such as sargassum and kelp, etc. Several automatic or semi-automatic Radar Image Processing techniques have been developed that can interpret a radar image for oil slicks in a matter of minutes, especially using deep learning algorithms (e.g. Decision Tree Forest [2], Support Vector Machine classification [3], convolutional neural network [4]) for feature extraction. SAR observations do not depend on weather (clouds and sunshine), which enables the detection of illegal discharges that most frequently appear during night. SAR can also survey areas during storms, where accident risks are increased. An example of an image is shown in Fig. 2.

Band Frequency [GHz] Wavelength [cm]
L 1-2 15-30
C 4-8 3.75-7.5
X 8-12 2.5-3.75


The lengthy availability of oil during the Deepwater Horizon spill provided several researchers with the opportunity to study radar remote sensing as well as band relationships. Overall, X-band is superior to other bands, but C-band radar and even L-band radar, to a degree, can provide useful oil spill data[5]. The contrast between oil and water is highest in X-band, moderate in C-band, and lowest in L-band. Signal polarizations using vertical (V) and horizontal (H) electromagnetic wave propagation can be used to provide further information. Polarimetric SAR yields information to aid in the discrimination between slicks and look-alikes[6]. A suitable SAR radar configuration for oil pollution study is C-band radar frequency with VV polarization, with a 20 to 45° incident angle. This is the case for ERS , RADARSAT, Envisat and Sentinel-1. ERS and Sentinel SAR imagery provide the option to recover the surface wind characteristics which important for oil weathering (see Oil spill pollution impact and recovery). Wind direction can be retrieved from an orientation of the organized structures in the atmospheric boundary layer (convective rolls) that clearly manifest themselves in the SAR images[7], as well as can be estimated from backscatter variation near islands and capes.

Visible light sensors

The visible part of the electromagnetic spectrum ranges from 400 to 700 nm. Thin layers of oil or sheen appear silvery to the human eye and reflect light over a broad spectrum range – up to blue. Thick oil layers appear to be the same color as bulk oil, typically brown or black (Fig. 1). However, the spectral information reflected by oil and by the water on which the oil floats is fairly similar[5]. Three other obstacles hinder the use of visible light: darkness, cloud cover and sun glitter. Sun glitter, which is often confused with oil sheen, is problematic in visible remote sensing. Sun glitter can be attenuated through signal processing techniques. Although inspection of different spectral regions does not yield strong discrimination, multispectral observations of optical images can help to distinguish actual oil spills from other features such as algal blooms. Infrared oil spill detection uses thermal infrared with wavelengths of 8–14 µm. Oil with a thickness greater than about 10 µm absorbs light in the visible region and re-radiates some of it in the infrared spectrum. However, this does not provide information about the thickness of the oil layer. In addition, natural objects can resemble oil, such as seaweed, sediment, organic matter, coastlines and oceanic fronts[5]. On the other hand, infrared sensors are cheap and easily available.

Laser fluorosensors

Laser fluorosensors employ a UV laser operating between 308 and 355 nm. Laser fluorosensors take advantage of the phenomenon that aromatic oil compounds interact with ultraviolet light, absorb light energy and release the extra energy as visible light. The absorption and emission wavelengths are unique to oil. Crude oil fluoresces from 400 to 650 nm with the 308 nm excitation. Distinguishing different oil classes is possible because different types of oil have different fluorescent intensities and spectral properties[5]. The laser fluorosensor is best at distinguishing between light, medium and heavy oils. In some fluorescent sensors, the detector is activated at the exact time the light returns from the target surface. This technique is called 'gating'. This gating technique enhances the differentiation of the oil from other interfering phenomena. Some fluorosensors can also gate their detectors to target areas below the sea surface. This enables oil detection in the water column[8]. Laser fluorosensors can also be useful for detecting oil pollution because they provide a method to detect oil on coastlines and to distinguish between, for example, oiled and unoiled seaweed[5].


Related articles


References

  1. ↑ Phillips, O.M. 1988. Radar Returns from the Sea Surface—Bragg Scattering and Breaking Waves. J. Phys. Ocean. 18: 1065-1074
  2. ↑ Topouzelis, K. and Psyllos, A. 2012. Oil spill feature selection and classification using decision tree forest on SAR image data. ISPRS J. Photogramm. Remote. Sens. 68: 135–143
  3. ↑ Baek, W.-K. and Jung, H.-S. 2021. Performance Comparison of Oil Spill and Ship Classification from X-Band Dual- and Single-Polarized SAR Image Using Support Vector Machine, Random Forest, and Deep Neural Network. Remote Sens. 13, 3203
  4. ↑ Fan, Y., Rui, X.; Zhang, G., Yu, T.; Xu, X. and Poslad, S. 2021. Feature Merged Network for Oil Spill Detection Using SAR Images. Remote Sens. 13, 3174
  5. ↑ 5.0 5.1 5.2 5.3 5.4 Fingas, M. and Brown, C.E. 2018. A Review of Oil Spill Remote Sensing. Sensors 18, 91
  6. ↑ Chen, Y. and Wang, Z. 2022. Marine Oil Spill Detection from SAR Images Based on Attention U-Net Model Using Polarimetric and Wind Speed Information. Int. J. Environ. Res. Public Health 19, 12315
  7. ↑ Wang, C., Vandemark, D., Mouche, A., Chapron, B., Li, H. and Foster, R.C. 2020. An assessment of marine atmospheric boundary layer roll detection using Sentinel-1 SAR data. Remote Sensing of Environment 250, 112031
  8. ↑ Brown, C.E. 2017. Laser fluorosensors. In Oil Spill Science and Technology, 2nd ed. (Fingas, M., Ed.) Gulf Publishing Company, Cambridge, MA, USA. Chapter 7: 402–418


The main authors of this article are Renata Archetti and Job Dronkers
Please note that others may also have edited the contents of this article.
Citation: Renata Archetti; Job Dronkers; (2024): Oil spill monitoring. Available from http://www.coastalwiki.org/wiki/Oil_spill_monitoring [accessed on 24-11-2024]
Retrieved from "https://www.coastalwiki.org/w/index.php?title=Oil_spill_monitoring&oldid=80701"