Cyber-social-physical approaches for effective detection of cyanobacterial blooms.

Cyanobacterial Harmful Algal Bloom (CyanoHABs) is a wide geographic water infestation phenomenon negatively impacting ecology, human and animal health, recreational activities and economy. While traditional monitoring methods are in-place, they are infeasible for large scale frequent monitoring of fresh waterbodies. In this dissertation, we present cyber, social and physical approaches that enables increase in the frequency of spatio-temporal monitoring of the CyanoHABs activity. First, we study the popular microbloggingplatform, Twitter for understanding the opportunities it provides in large scale monitoring. We evaluate the efficacy of different keywords and present experiments to show concept drift in the keywords. Second, we acquire 10 years of multispectral satellite data for large lakes in Southeastern United State. We study the inherent issues of cloud-cover in these satellite imagery and propose data mining technique to address it. Third, we build solar power based wireless sensor framework to regularly capture and transfer the CyanoHABs data from remote lake to our cloud server.