Artificial neural network onto eight bit microcontroller for Secchi depth calculation

Autores UPV
Revista Sensors and Actuators B Chemical


In this paper we present a model to predict Secchi depth in water bodies by means of an artificial neural network application onto eight bit microcontroller. Water turbidity data were collected by both a Secchi disk and a new patented device (named LUZEX) that uses commercial photodiodes with not monochromatic sensitive band as a basis to perform "in situ" measurements for sunlight extinction coefficients. In order to have a wide range of turbidity data three different water bodies were selected to do the measurements. The developed neural network model is able to relate well the data obtained by these methods and the obtained value for regression coefficient (R) is 0.998. Secchi depth measure is a reference method to determine turbidity in continental and coastal water bodies, especially in the Mediterranean Sea region, but sometimes there are particular cases that makes difficult the use of the Secchi disk (e.g. shallow water bodies), the authors propose LUZEX as a substitute for Secchi disk when it is difficult or impossible to use. © 2011 Elsevier B.V.