The Predictability of Oceanic Circulations via FFT-ANFIS Spectral Adaptive Fuzzy Network

Oceanic currents and circulations are vital components of the marines on Earth for their physical and biological existence [1–6]. In the literature, many different data processing techniques are proposed and employed for their analysis [2–6]. In this paper, we examine the predictability of oceanic circulations using spectral fuzzy logic. More specifically, we propose the possible implementation and usage of the Adaptive Neuro-Fuzzy Inference System (ANFIS) fuzzy network [7, 8] approach for the spectral (FFT) analysis of oceanic circulation data. The data set investigated in this study was acquired by the National Oceanic and Atmospheric Administration (NOAA) between October 2002 and February 2003 in Massachusetts Bay, US. We show that the current speed time series and its spectrum in 2D horizontal directions, namely u and v, can be successfully predicted using the FFT-ANFIS approach. We present prediction error in terms of the coefficient of determination (R²) and the root-mean-square error (RMSE). The effect of the training data set on the prediction error and the spectral properties of predictions are also analyzed. We show that depending on the temporal and spatial resolution of the data, the prediction times and distances can vary. Depending on their resolution in some cases, they can be very beneficial for the early prediction of the ocean current parameters. Our results can find many important applications including but not limited to predicting the statistics and characteristics of tidal energy variation and controlling the current-induced vibrations of structures in the marine environment.