A comparative analysis of soft computing techniques used to estimate missing precipitation records
Kajornrit, J., Wong, K.W. and Fung, C.C. (2012) A comparative analysis of soft computing techniques used to estimate missing precipitation records. In: International Telecommunications Society 19th Biennial Conference, ITS 2012 , 18 - 21 November, Bangkok, Thailand.
|PDF - Authors' Version |
Download (630kB) | Preview
Estimation of missing precipitation records is one of the most important tasks in hydrological and environmental study. The efficiency of hydrological and environmental models is subject to the completeness of precipitation data. This study compared some basic soft computing techniques, namely, artificial neural network, fuzzy inference system and adaptive neuro-fuzzy inference system as well as the conventional methods to estimate missing monthly rainfall records in the northeast region of Thailand. Four cases studies are selected to evaluate the accuracy of the estimation models. The simultaneous rainfall data from three nearest neighbouring control stations are used to estimate missing records at the target station. The experi-mental results suggested that the adaptive neuro-fuzzy inference system could be considered as a recom-mended technique because it provided the promising estimation results, the estimation mechanism is trans-parent to the users, and do not need prior knowledge to create the model. The results also showed thatfuzzy inference system could provide compatible accuracy to artificial neural network.In addition, artificial neural network must be used with care becausesuch model is sensitive to irregular rainfall events.
|Publication Type:||Conference Paper|
|Murdoch Affiliation:||School of Information Technology|
|Item Control Page|