Virtual Sensor to Estimate Air Pollution Heavy Metals Using Bioindicators

Abstract

The main objective of this work is to demonstrate that a set of bioindicators linked to the lichen Lobaria Pulmonaria and the bryophyte called Leucodon Sciuroides are adequate predictors of air pollution heavy metals (HM). A study case was performed in Oran, a port and coastal city in northwestern Algeria, located on the coast of the Mediterranean Sea. Each of the HM has been modelled using a machine learning procedure and in the experiments, the artificial neural networks (ANN) produces always better and more accurate results than multiple linear regression (MLR). Furthermore, good obtained results (R correlation coefficient greater than 0.9) demonstrate the main hypotheses and could be used as a virtual sensor.