Abstract

Developing methods to monitor harmful algae is a current research “hot-topic.” One type of algae, the blue-green algae or Cyanobacteria, cause blooms that can lead to a health threat to humans and animals. This research will test the use of a cost effective and temporally efficient method using multispectral remote sensing system, WASPLITE, as a monitor of algal blooms. This airborne system will be optimized to the specific application of detecting Cyanobacteria on optically complex waters. Attempts have been made in the past using existing instruments, e.g., SeaWiFS and Landsat, to provide these data, but our solution can provide more information by using optimally selected bands with very high spatial resolution. To analyze these algal blooms, standard multispectral techniques (such as band ratio, spectral curvature and principal component analysis) were used on the airborne data. These results were compared with ground truth collected concurrently with the airborne over flight. Because of the very high spatial resolution of the system, (0.7 m), compared to many commonly used satellite systems (~30m to 1km), it could be seen that the patchiness of the algae was very high. Difficulties in applying the ground truth were both technical shortcomings and were due to the nature of the algal blooms. Technical issues include the time lag between the ground sample collect and the airborne collect (the water and algae move with time), the drift of the boat during ground sampling (there was no anchor), and the error in the GPS units in both the boat and the plane. The issues due to the nature of water and algae include, sun glint in the imagery, white foam lines created by waves and wind, and most importantly, the patchiness of the algae in the water. Because the ground truth of one sample point per location was not adequate, we could not correlate the ground truth to the imagery. Qualitatively, the images did show a large variation of algae concentration in the water through the principal component analysis. Further, flow-through data from another vessel taken from the same week this research was performed, suggests that the variation that is seen in the imagery is real. Overall, this research shows the difficulties in effectively and accurately performing ground truth measurements to be used to test algorithms and methods that are applied to detecting harmful algae using remotely sensed data. The traditional ground sampling methods failed to capture the spatial variation observed in the image data. With improved techniques we are confident these methods can be used to effectively monitor algal blooms using the high spatial and temporal resolution.

Library of Congress Subject Headings

Environmental monitoring--Remote sensing; Cyanobacterial blooms--Monitoring; Algal blooms--Monitoring; Remote sensing

Publication Date

8-1-2007

Document Type

Thesis

Department, Program, or Center

Chester F. Carlson Center for Imaging Science (COS)

Advisor

Kremens, Robert

Advisor/Committee Member

Rhody, Harvey

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: GE45.R44 M36 2007

Campus

RIT – Main Campus

Download

Share

COinS