Arts and Social Sciences

Permanent URI for this collection

http://openscience.utm.my/handle/123456789/35

Browse

Browsing Arts and Social Sciences by Author "Althawadi, Jamal Jasim Abdulla"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Variants of oil spills spectral signature and characterisation in Arabian Gulf using optical remote sensing satellite system
    (Universiti Teknologi Malaysia, 2020) Althawadi, Jamal Jasim Abdulla
    Remote sensing technology is an important tool for detecting, mapping, monitoring and analyzing oil spills contamination in the Arabian Gulf (AG). Studies have shown the usage of digital imagery from multi-sensor systems acquired from various platforms and operating in various spectral regions of the electromagnetic spectrum. Satellite-based optical systems have found to be widely applied due to long profound understanding in the image formation and its related image analysis at fine, medium and even regional scales. However, despite of the robustness of the applications, the results obtained are varying due to inherent randomness in the data acquisition systems. Hence, this study focused on the variants of the oil spill signatures within the short visible spectrum regions (blue, green and red bands) of the Landsat-7 Enhanced Thematic Mapper (ETM+), Landsat-8 Operational Land Imager (OLI) and Sentinel-2 Multispectral Instrument (MSI) satellite systems, respectively. This study hypothesized that the variants of the oil spill signatures in AG despite being pre-processed, still possess inherent randomness that need further minimizations due to local effects of image scene-illumination-geometry, environmental parameters, the age oil of spill and spill film thickness at the time of image acquisition. The minimization approach introduced in this study using a 3-stepwise data processing with specific objectives to : (i) examine the variants of the spectral signatures of oil spills for optimizing detectability; (ii) assess and create the calibration matrices for oil spill variants because of the satellite scene illumination geometry and corresponding environmental factors; (iii) create a model and map oil spills thickness for estimating remnant volume; and (iv) evaluate the relationship between the age of oil spill to the corresponding spectral reflectance of spills. A total of 1,235,765 pixels equivalent to an area of 254 km2 from 7 oil spill areas during the period of 2016 - 2017 located throughout the AG region were used in this study, where all the corresponding satellite image sets were examined for variant effects. The location of spills within AG were identified in the satellite image set, initially assigning the segmentation using region growing algorithm to identify all the spill pixels using all the visible sites. Results indicated that the pre-processed and calibrated spill pixels in all visible bands have a relatively small, ranging from 1% to less than 2 % in the blue, increased to 2% in the green and 5% in the red band for Landsat-7 ETM+, Landsat-8 OLI and Sentinel-2 MSI systems, respectively (R2 > 0.85, p < 0.05). However due to the small variations between spill and non-spill pixels including look-alikes, this absolute characterization was crucial. The environmental effects including the illumination geometry although shown local errors variation but the magnitude is relatively small (R2 = 0.75, p<0.05) hence indicating non-inherent environmental effects in the medium resolution satellite data used. Spill thickness was found best model with multiple regression of all bands, and best band is blue for ETM+, blue for OLI, and red for MSI (R2 > 0.9, p<0.05). The age of spills also shown agreement with the thickness trend. Hence, the spectral signatures of slick thickness and different spill ages could be identified and modelled to map the spill thickness distribution and estimate the spill remnants volume. It is therefore concluded that the creation of variants minimization matrix tool has been successfully created for oil spill detection and mapping in AG region using visible bands of Landsat-7 ETM+, Landsat-8 OLI and Sentinel-2 MSI satellite data.