Absolute Airborne Thermal SST Measurements and Satellite Data Analysis From the Deepwater Horizon Oil Spill

Abstract

Rapid assessment and continuous monitoring was critical to addressing the changing conditions in response to the Deepwater Horizon Oil Spill. Airborne, satellite, shipborne, and underwater sensors were all used with data being assimilated as actionable reports. To assist with the recovery effort and evaluate the utility of new sensors for oil spill response, Ball Aerospace sent a team of scientists and engineers to the Gulf in July 2010. The team deployed on a Twin Otter aircraft with a suite of sensors including a thermal imaging radiometer, ultra-violet to visible hyperspectral imaging radiometer, and a visible high dynamic range context imager. All three sensors were operated at the same time with overlapping fields of view to assist with targeting and characterization of the oil. Data was also gathered from satellite synthetic aperture radar (SAR) and optical sensors during the same time and analyzed to determine additional capabilities from these sources. The ultra-violet and thermal imaging capabilities demonstrated were unique when compared with other airborne sensors flown over the spill. In the optical spectral range, imagery from the WorldView 2 satellite operated by DigitalGlobe was utilized while SAR imagery was collected by the TerraSAR-X, COSMOSkyMed (also X-band) and the ENVISAT Advanced SAR (ASAR, C-band). This combination of optical and radar imagery proved very useful in being able to map oil features on the surface of the Gulf of Mexico. Results of the thermal measurements are presented along with a discussion of the other sensor data used to further characterize the spill. 1. Background Oil spill monitoring and characterization requires a suite of sensors with varying capabilities. Several sources [Fingas and Brown, 2000; Jha et al., 2008] have reviewed the strengths and weaknesses of different sensors ranging from ultraviolet and visible sensors to passive microwave and SAR. SAR systems have been shown to be very effective at locating and mapping oil spills with the capabilities to see through clouds, but if the wind in the area is too light or too strong, the signal can be corrupted. SAR is also generally unable to detect oil thickness parameters. Visible and UV sensors are capable of detecting oil thickness as well as location but are blocked completely by clouds or sunglint conditions. Thermal sensors also can offer advantages for detecting oil location and thickness while operating at night as well as during the day. As with UV-Vis sensors though, thermal sensors cannot measure through clouds. SAR is well known for being able to sense oil on the sea surface due to the damping effect of the oil on the capillary waves at the ocean‟s surface which reduces the radar backscatter [Gade et al., 1998]. Thus, oil features appear as “dark” relative to the brighter SAR sea clutter due to the presence of the wind driven capillary waves. This damping effect is most effective at shorter radar wavelengths such as C or X band and the oil versus open water contrast ratio increases with increasing frequency [Gade et al., 1998; Wismann et al., 1993]. As a result we have only considered C and X band SAR data. The advantage in using SAR imagery to sense oil slicks is that it is all-weather and functions at nighttime as well as during the day. This is very different from the optical imagery, which requires daytime illumination and low cloud cover to be able to view the sea surface. In this way these different satellite data sets prove to be very complimentary. The visible-near infrared (VIS-NIR) data will provide high spatial resolution multichannel information while the SAR will supply all-weather and day/night sensing capability. As the sensing capabilities are slightly different it can be expected that each data source will give a slightly different form of the oil slick. The possibility of detecting and mapping oil spills in optical satellite images has been demonstrated by Hu et al., [2000]. Multi-spectral optical sensors have the potential for detailed identification of the oil spill type (light or crude oil) and estimations of their abundance/thickness. In addition, since oil spills have a higher thermal conductivity they become heated faster from above and become warmer than the surrounding seawater during the daytime making them distinguishable in infrared satellite imagery. At night the oil loses heat faster than the seawater becoming cooler than its surroundings [Tseng and Chiu, 1994]. Mapping oil slicks with optical imagery is more difficult than with SAR due to the problems of day only sensing and then only under ideal weather conditions (low cloud cover). The need for atmospheric corrections over and above the cloud filtering also complicates the retrieval of oil slick conditions with optical imagery. Still there is the potential for estimating the thickness of the oil slick that is not possible with a single band SAR image. One important aspect of optical remote sensing of oil on the ocean is the accurate calibration of the optical sensor. This calibration step converts the raw reflectances into radiance values that represent the radiation upwelling from the ocean‟s surface. In our study we used spectral radiance to take advantage of the multispectral images collected by WV2. These spectral radiances have the units of Wsr -1 m -3 . Due to the varied capabilities of the sensors, it is recommended to employ a suite of sensors for an oil spill region [Jha et al., 2008]. Both satellite and airborne sensors are often used. Satellites provide broad coverage while airborne assets provide high temporal and spatial resolution. It is a combination of the strengths of all these systems that provides the full picture in the response to an oil spill. 2. The Flight System Ball Aerospace deployed a suite of instruments on a Twin Otter aircraft contracted from Twin Otter International over the Gulf oil spill on July 9-10, 2010. The instruments, shown mounted on the plane in Figure 1, collected simultaneous data during each flight. Details regarding spectral range, fields of view, and other key parameters are shown in Table 1 with a more detailed discussion of each instrument in the following sections. Footprint and swath width parameters are listed for the actual flight altitude of 1,070 meters (3,500 ft) MSL. The Ball Experimental Sea Surface Temperature (BESST) radiometer was designed and built by Ball Aerospace to provide a well calibrated measure of sea surface temperature (SST) from an airborne platform. The Ball sensor suite provided a complementary and spectrally unique data set to other sensors flown over the region. 2.1 Ball Experimental Sea Surface Temperature (BESST) Radiometer The BESST is a low-mass, low-power SST sensor designed for rapid airborne deployment and shown in Figure 2. It uses a 324 x 256 array of VOx microbolometers to sense infrared radiation from the ocean „skin‟ temperature. The microbolometers eliminate the need for power hungry cryocoolers or cryogenic fluids which greatly simplifies flight requirements for operation. The measured noise equivalent change in temperature (NEDT) of the BESST instrument is 0.1 K providing the resolution required to distinguish oil and perform remote scientific thermal analyses. Laboratory calibrations of BESST indicate an absolute temperature accuracy of 0.3 K for individual pixels. The key to maintaining this accuracy in the field with varying instrument conditions is onboard calibration blackbodies. During operation, the instrument periodically views the onboard blackbodies to adjust for temperature changes within the instrument and detector. BESST includes three thermal bands covering the 8-13 um range of the sensor. There is a broadband 8-13 micron window for maximum signal to noise measurements and two narrow 1.0 micron wide bands centered at 10.5 and 11.5 microns. The narrow band filters are used to characterize the atmospheric transmission of the thermal signal due to water vapor in between the sensor and the surface of the water. 2.2 GLobal IMager for Marine Ecosystem Research (GLIMMER) GLIMMER is a compact, UV-enhanced, hyperspectral ocean color imager also developed by Ball Aerospace. It uses diffraction gratings for dispersion and covers the 350 nm to 700 nm spectral range. GLIMMER, shown in Figure 1, is a pushbroom imager providing swaths of spectral data that are accumulated into spatial images as the plane flies. The instrument includes a unique time-domain polarization scrambler that reduces systematic errors associated with ocean scene polarization variability. In addition to the data shown, results from the flight measurements with GLIMMER will be presented at the SPIE Optics & Photonics Conference in August 2011. 2.3 Low Light Imager (LLI) LLI is a visible broadband context imager that operates over a dynamic range greater than seven orders of magnitude with SNR > 10 at the low end of the dynamic range, 2.85 x 10 -9 W/cm 2 -sr [Osterman et al., 2010]. This performance enables imaging in broad daylight as well as scenes illuminated only by a quarter moon. It is capable of saturationfree imaging of scenes spanning the earth‟s terminator. The LLI was used to provide context imagery of the oil region and to help discern oil sheen areas that are not detectable by thermal IR imagery. 2.4 Multi-spectral Satellite Imagery The optical imagery used in this study was high-resolution imagery from DigitalGlobe‟s WorldView 2 (WV2) satellite. It has a 50 cm resolution panchromatic band as well as 1.8 m resolution 8 spectral channels from 450 nm visible to 1050 nm in the near infrared (Figure 3). Geolocation accuracy was close to the 1.8 m spatial resolution.