Random Forests Hydrodynamic Flow Classification in a Vertical Slot Fishway Using a Bioinspired Artificial Lateral Line Probe
Ecohydraulic studies rely on observations of fish behavior with hydrodynamic measurements. Most commonly, observed fish locations are compared with maps of the bulk flow velocity and depth. Fish use their lateral line to sense hydrodynamic interactions mediated by body-oriented spatial gradients. To improve studies on fish an artificial lateral line probe (LLP) is tested on its ability to classify either the “slot” or “pool” regions within 28 basins of a vertical slot fishway. Random forests classification is applied using four models based on high-frequency (200 Hz) recordings using 11 collocated pressure sensors and two triaxial accelerometers. It was found that the assigned classification task proved to be reliable, with 100 % correct classification of all four models, across all 28 basins. Preliminary results from the first field study of this new sensing platform show the LLP-random forests workflow can provide robust, highly accurate classification of turbulent flows experienced by fish innatura.
KeywordsRandom forests Lateral line probe Vertical slot fishway Turbulent flow Hydrodynamic classification
This work was supported in part by the Grant-in-aid for Young Scientists A (25712026) and Grant-in-aid for Challenging Exploratory Research (26660190) of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. The research leading to these results has received funding from BONUS, the joint Baltic research and development programme, co‐financed by the European Union’s Seventh Framework Programme (2007–2013) under the BONUS Implementation Agreement. National funding for this work has been provided by the German Federal Ministry for Education and Research (BMBF FKZ:03F0687A) and the Estonian Environmental Investment Centre (KIK P.7254 C.3255). The work has also been partly financed by the EU FP7 project ROBOCADEMY (No. 608096) and the TIWAG research project “Robofish”, Austria.
- 5.Nestler, J. M., Pickens, J. L., Evans, J., Haskins, R. W.: Multiple sensor fish surrogate for acoustic and hydraulic data collection. US5517465 A (1996)Google Scholar
- 16.Tuhtan, J.A., Strokina, N., Fuentes-Pérez, J.F., Muhammad, N., Musall, M., Noack, M., Toming, G., Kämäräinen, J.-K., Kruusmaa, M., Schletterer, M.: Ecohydraulic flow sensing and classification using a lateral line probe. In: Proceedings of 11th International Symposium on Ecohydraulics, Melbourne, Australia (2016)Google Scholar
- 17.Bundesministerium für Land- und Forstwirt-schaft, Umwelt und Wasserwirtschaft (Hrsg.): Leitfaden zum Bau von Fischaufstiegshilfen. Wien (2012)Google Scholar
- 18.Liaw, A., Wiener, M.: Classification and regression by random forest. R News 2(3), 18–22 (2002)Google Scholar
- 19.R Core Team: R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria (2015). http://www.R-project.org/
- 26.Tuhtan, A., Strokina, N., Toming, G., Muhammad, N., Kruusmaa, M., Kämäräinen J.: Hydrodynamic classification of natural flows using an artificial lateral line and frequency domain feature. In: 36th IAHR World Congress (2015)Google Scholar