This study tries to develop an early warning system for fishery chilling disaster and coral bleaching in Penghu waters. Data mining technology was mainly applied to explore the relationship between marine and atmosphere conditions and will provide the real-time web-site information for decision making to reduce the loss of cage-cultural fishery in winter time. Results of this year include:
1. Maintained 5 WSN water-quality monitoring stations for 15 times to calibrate and validate accuracy of various sensors. Over 70000 multi-parameter water-quality data were real-time transmitted and stored this year.
2. Three increasing transect lines at each coral reef monitoring site was investigation every season. In order to watch community variation and bleaching, coral reef photos were took to analyze the coverage rate by software CPCe4.1. ‘Penghu WSN Water Quality Monitoring System’ Web also add DHWs and DCWs information every Sunday for early coral reef warning function.
3. We used the multiple regressions method of data mining to build a forecast model for water temperature by using average wind velocity and air temperature data. This model seemed to present more consistency for long-term trend.
4. Designed and tested several new prototypes of the small WSN floating ball by improving there electronic hardware and mechanism for multi-layer (surface, 5m and 10m depths) water temperature observation and cost-down purposes.