Benthic habitat mapping and coral bleaching detection using quickbird imagery and Kd algorithm

Remotely sensed satellite images have shown their abilities for mapping the benthic habitats in shallow waters. In this regard, estimating depth values is typically a key phase in benthic habitat mapping using multi-spectral satellite imagery. The overall objective of this thesis was applying QuickBird imagery for benthic habitat mapping (with specific attention to the coral reefs) and detecting bleaching in coral reef communities. First objective of this research is to apply a general linear model to compute the parameters with unknown values of the Stumpf’s methodology for depth estimation utilizing multi-spectral satellite images. In this regard, two QuickBird satellite images (2005 and 2008) were corrected geometrically and radiometrically prior to accomplishing the main analysis. A total number of 80 and 74 points with known depth values were available for the years 2005 and 2008 images respectively. Half numbers of these points were selected to determine the mentioned parameters using minimizing the sum of the squared residuals, and the other points were used for accuracy assessment of the applied method. On the other hand, diffuse attenuation coefficient (kd) is another critical parameter for benthic habitat mapping using remotely sensed data. In this regard, the second objective of this research was to develop an approach to estimate kd in blue and green bands of QuickBird satellite image. To do this, the Lyzenga’s method was utilized to determine the ratio of kd in different bands of satellite image. Afterwards, NASA-kd 490 algorithm was modified and then applied to determine kd 490 in each pixel of image. After determining the kd 490 values as kd for blue band, the kd values for green and red bands were subsequently obtained by using results obtained from Lyzenga’s method. It is a well-known fact that, Sea Surface Temperature (SST) is an important parameter which can affect the health of coral reef communities. Therefore, as third objective of this research, it is focused on the studies of fluctuations of SST values in near-shore waters around the studied area between 1985 and 2009 using NOAA-Pathfinder-5 dataset and fieldobservations. Subsequently, the results were analyzed to assess the impacts of these fluctuations on main coral reef communities. Finally, as fourth objective of this thesis, the feasibility of using multi-temporal satellite images for detecting bleaching events was investigated. Two aforementioned QuickBird images were selected to detect 2007 bleaching incident in the study area and then, the accuracy of results was compared with in situ observations. In the proposed procedure, pre- and postbleaching images were classified and changes in reflectance values within coral classes were interpreted as bleaching areas. Final results of the applied methodology for depth estimation revealed correlation values equal to ~0.84 and ~0.83 between estimated and measured depth values, while their mean values were ~2.06±1.44 m and ~1.75±1.33 m for 2005 and 2008, respectively. Additionally, the results obtained for determination of kd values proved to be consistent in the areas deeper than 2m between estimated values using the proposed approach and the results obtained from Mumby and Edwards’methodology. The results also demonstrated an increment of ~12% in the overall accuracy of classification. Moreover, statistical analyses on SST data showed a meaningful and positive correlation between the SST values and bleaching in coral reefs. Specifically, in August 2007, this phenomenon was clearly observable when most of the corals were affected by the bleaching phenomenon while the SST values were ~1.5°C higher than the normal situation. Finally, by using the applied method for detecting bleached corals, results confirmed that it is possible to eliminate the effect of misclassification between bleached corals and sand as well as algae-covered corals and live corals while the proposed technique could detect ~28% of algaecovered bleached corals. It is concluded that a positive anomaly more than 1°C in SST values, may lead to bleaching in the coral reef communities. Additionally, to detect this event, it is necessary to having at least two high resolution satellite images for pre- and postbleaching date. In this regard, applying correction on substrate reflectances needs to have knowledge about depth and kd values and consequently, may increase the final results of image classifications