Open AccessModeling of Precipitable Water Vapor Using an Adaptive Neuro-Fuzzy Inference System in the Absence of the GPS Network
Author(s) -
Wayan Suparta,
Kemal Maulana Alhasa
Publication year - 2016
Publication title -
journal of applied meteorology and climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.079
H-Index - 134
eISSN - 1558-8432
pISSN - 1558-8424
DOI - 10.1175/jamc-d-15-0161.1
Subject(s) - adaptive neuro fuzzy inference system , global positioning system , mean squared error , artificial neural network , multilayer perceptron , computer science , precipitable water , neuro fuzzy , relative humidity , environmental science , meteorology , statistics , data mining , artificial intelligence , fuzzy logic , precipitation , mathematics , fuzzy control system , geography , telecommunications
This paper constructs an adaptive neuro-fuzzy inference system (ANFIS) model to estimate precipitable water vapor (PWV) in Southeast Asia, particularly in the Peninsular Malaysia, Sabah, and Singapore region. The input to the model is developed using the surface pressure, temperature, and relative humidity. The models are trained and tested using PWV values derived from the global positioning system (GPS). The data used are for May 2012 taken at the Nanyang Technology University of Singapore (NTUS) and Universiti Malaysia Sabah, Kinabalu (UMSK); and for February 2009 taken at the Universiti Kebangsaan Malaysia Bangi (UKMB). The validation process is conducted using June 2012 data for NTUS and UMSK and March 2009 data for UKMB. The performance the ANFIS model is compared with a multilayer perceptron (MLP), Elman neural networks, and multiple linear regression (MLR) models. Results from validations at the three stations showed that the ANFIS model performed well as compared with MLP, Elman neural networks, and MLR, with a mean absolute error of 0.015 mm, a percent error of 0.028%, and root-mean-square error of 0.019 mm. These results suggest that the ANFIS model is a promising approach for estimating PWV values that is cost effective, continuous, and potentially useful for meteorological applications.