Optimization of rain gauge network in Johor using hybrid particle swarm optimization and simulated annealing

• Main Author: Mohd. Aziz, Mohd. Khairul Bazli
• Format: Thesis
• Language: English
• Published: 2017
• Subjects: QA Mathematics
• Online Access: http://eprints.utm.my/id/eprint/79392/1/MohdKhairulBazliPFS2017.pdf; http://eprints.utm.my/id/eprint/79392/
• Institution: Universiti Teknologi Malaysia

An optimal design of rain gauge network is important as it produces fast, accurate and important rainfall data used in designing an effective and economic hydraulic structure for flood control. The use of inaccurate rainfall data may result in significant design errors in other water resources project. The objective of this study is to determine the optimal number and location of the rain gauge network in Johor by using geostatistical method integrated with hybrid method consisting of simulated annealing and particle swarm optimization. This study also explored and compared the results of all existing methods namely the coefficient of variations, the maximum covering location problems and geostatistical method with the proposed model. The use of methods such as maximal covering location problem and coefficient of variation can only provide the figure for number of rain gauge stations but not the optimal location for the stations. The geostatistics method however, can provide the optimal number of rain gauge station and its location through the minimum variant value. The integration of geostatistics with hybrid methods comprised of simulated annealing method and particle swarm optimization is successful in providing the optimum number and location of the stations. In order to identify the effect of rain gauge station locations toward rainfall data, this study considered the repositioning of the existing rain gauge into new locations to improve their effectiveness and reduce the error. The analysis analysed the density of the rain gauge, daily rainfall data from 1977 to 2008, latitude and longitude of the rain gauge location, elevation, humidity, wind speed, temperature and solar radiation to determine the new optimal network design for the rain gauge network. The minimum value of estimated variance produced by the proposed method indicates that the method is successful in determining the optimal rain gauge network from the existing 84 rain gauges in Johor. Relocation of all 84 rain gauge stations to new locations give better results in terms of the estimated variance value but, it is not necessary to relocate all of the stations due to the expensive costs. Therefore, the location of the station also influences the result. In this study, hybrid simulated annealing and particle swarm optimization as an optimization method successfully determined the optimal rain gauges network in Johor. In conclusion, this study has shown that a well-design rain gauge network will help to provide essential input for effective planning, designing and managing of water resources project such as flood frequency analysis and forecasting.