Acoustic Doppler Current Profiler Data Analytics Using a Hybrid of Machine Learning Methods

This article presents a hybrid of machine learning techniques for predicting and analyzing Acoustic Doppler Current Profiler (ADCP) data. ADCP is a commonly used instrument for measuring ocean, river, and lake water currents. However, the data generated by ADCP instruments makes it challenging to analyze and extract valuable insights. The study uses three datasets with 25 features, including horizontal velocity components essential for environmental monitoring and management. In this study, we propose a hybrid of machine learning approaches that combines the strengths of multiple algorithms, including Convolutional Neural Networks (CNN), Logistic Regression, Principal Component Analysis (PCA), and Fuzzy c-Means (FCM). Firstly, we predict ADCP data and analyze its accuracy compared to conventional prediction methods. Secondly, we used P-value and correlation analyses to perform sensitivity analysis to identify the patterns of ADCP data based on the proof of concept discovered by the domain experts. The results show that the CNN model provides better accuracy than the results before features extraction. Furthermore, the sensitivity analysis helps identify the presence of horizontal velocity components deemed crucial by domain experts. Overall, this study highlights the potential of machine learning and data analytics techniques for improving ADCP data analysis and prediction in various water monitoring applications.

I. Introduction

A Broadband Acoustic Doppler Current Profiler (BB-ADCP) is used to measure ocean currents by utilizing the Doppler effect (the change in frequency of a wave caused by the motion of a source or a reflector) to measure the speed and direction of water currents. The ADCP instrument is used to measure water velocity, ADCP device is mounted on a moving platform, like a ship as shown in Fig. 1.A BB-ADCP operates by emitting high-frequency sound pulses into the water and measuring the frequency shift of the returning echoes [1]. By analyzing the Doppler shift of the echoes, the velocity of the water particles in the direction of the sound wave propagation can be determined using the ADCP instrument [1]–[3]. BB-ADCPs are commonly used in oceanographic research and marine engineering projects to obtain data on ocean currents, tides, and water circulation patterns. BB-ADCPs are also used in hydrological studies and for monitoring coastal erosion and the movements of oceanic fronts. BB-ADCPs have several advantages over their narrowband counterparts, including improved accuracy, higher resolution, and the ability to operate in shallow or turbid waters. Fig. 1

Acoustic doppler current profiler