Servicios Económicos

The scheduling of passes over the RSA was carried out at about 04:00 UTC35 _each

day starting with a download of the Two–Line Element file (TLE). The TLE file includes the orbital elements that describe the designated orbit of Terra and Aqua satellites. Accordingly, four nadir passes were selected (that is a day and night pass for each of the two satellites), covering most of or all of RSA. To maintain the accuracy of the generated geolocation (GEO) file, the two ancillary files utcpole.dat and leapsec.dat, were updated on weekly intervals. After which, the following process was automated

32 _{https://lpdaac.usgs.gov/lpdaac/tools/modis_reprojection_tool_swath} 33 _{http://www.brockmann-consult.de/cms/web/beam/welcome} 34 _{http://gretl.sourceforge.net/}

Chapter 4: Experimental procedure using a custom–written script: sending the telemetry data to the antenna motion controller; processing all MODIS levels to L2; and generating the false and natural geotiff RGB colour composites. Table 4.4 outlines the automated tasks and packages that were used together with their sources.

Table 4.4. The automated processing tasks with the packages used and their sources.

Data level Software & packages used Source of packages

Raw telemetry data to L0 RT–STPS NASA Goddard Space Flight Centre's (GSFC) Direct Readout Laboratory (DRL)

The antenna system and its associate hardware configuration used for its control and the processing of data are shown in Figure 4.5.

Figure 4.5.The hardware configuration for the MODIS antenna.

The 250 m/pixel resolution false RGB colour composite geotiff file corresponding to bands λ2=859, λ1=645 and λ1=645 nm respectively is transferred to a remote server by

ftp for manual analysis through a visual investigation during off–site periods. This file was chosen over other products to carry out the preliminary image processing for three reasons; firstly it is geo–referenced (pixel location is identified in terms of latitude and longitude); it is of maximum spatial resolution; and the RGB colour composite will

Chapter 4: Experimental procedure immediately highlight all surface–water features including oil spills and their look–alike patterns such as surface floating blooms. In this RGB colour combination, the water– surface blooms will appear reddish due to the “red edge” effect (the greatest reflectance slope between the maximum absorption in the red, due to photosynthetic pigments, and the maximum reflection in the IR) (Tucker, 1979; Jackson et al., 1983). BEAM is then used to analyse the two–band geotiff file, by producing a new grey scale image created using the NIR band (L2=859 nm) according to the formula shown in Equation 4.22:

1

L₂2 (4.22)

The reason for performing this step was to invert the data contrast in order to improve the visual discrimination efficiency in detecting surface water features such as oil platforms, algal mats, clouds and oil spills from the generally dark water appearance in single-band images of the whole RSA (covering ~ 240,000 km2_{). Once a suspected}

patch is discovered, further spectral analysis was conducted, the details of which will be discussed in the following chapters. Additional L2 data such as Chlor-a concentration and SST were also investigated manually to see whether any relationship exists between the suspected spill and the biophysical/geophysical data. For spatially–large oil spills (in the order of 100's km), the SST4 data (Equation 4.20) of the previous night pass were also analysed to see whether any temperature difference between the spill and the water can be detected. This analysis is based on the presumption that the spill may have started a day earlier.

A GIS map was also used to identify the nearest oil production facilities to the suspected spills, such as offshore oil platforms.

An oil spill detected in MODIS Aqua on 19th _{June, 2010 at 09:03 UTC is shown}

under different RGB colour composites including the transformation in Equation 4.6 as demonstrated in Figure 4.6.

Figure 4.6. (A) The swath of MODIS Aqua on 19th _{June, 2010 at 09:03 UTC under which an oil spill was}

observed within the sun–glint. (B) An image of the spill produced using the transformation shown in Equation 4.6, in which the spill's negative contrast has been reversed to a positive one; (C) the natural 500 m/pixel resolution RGB colour composite corresponding to bands λ1=645 (aggregated from the 250

m resolution band group), λ4=555 and λ3=469 nm respectively. Clouds and their shadows are visible; and

(D) the 500 m/pixel resolution false RGB colour composite corresponding to bands (λ2=859, λ1=645 and

λ1= 645 nm respectively (λ2 and λ1 were oversampled from the 250 m/resolution bands) (Courtesy of

ROPME). The image is centred at the position N27° 00' 57" E51° 03' 40".

In terms of validation, in situ verification of oil spills remains the main source for validation. Visual inspection of spills was usually conducted by the local environmental authorities that belong to the countries where the spill was located. This is carried out in liaison with the Marine Emergency Mutual Aid Centre (MEMAC)36 _{– the Bahrain–}

based ROPME associate organization, whose role is to combat pollution caused by oil and other harmful substances in case of marine emergencies in the region. The

Chapter 4: Experimental procedure verification task is normally undertaken by an aerial surveillance or boat dispatches. An alternative indirect method was also used for validation which involved comparing the spill's appearance in MODIS image with the corresponding image of ASAR data, whenever the ASAR data are available. In which case, it becomes possible to trace the evolution of the spill in both MODIS and ASAR images, due to the time difference between them.

Figure 4.7 summarizes the whole processing procedure starting from scheduling of the passes through to the validation process.

Chapter 5: The spectral contrast shift (SCS)