INTEGRATING CITIZEN SCIENCE AND INTRUMENTAL MONITORING FOR THE RECOGNITION OF ODOR CLASSES NEARBY A WASTEWATER TREATMENT PLANT

Abstract

Odor emissions are very complex sources of annoyance to the resident population to be effectively characterized with the monitoring protocols currently applied for common environmental parameters. The extreme complexity lies both in the high temporal variability of the odour perception, related to atmospheric dispersion and the orography of the territory and the presence of compounds characterized by a very low odor threshold that require very sensitive analytical tools for real-time identification.

For such reason, the use of instrumental odor monitoring systems (IOMS) is becoming more and more widespread as systems that integrate an instrumental monitoring section with different sensors and a software section which applies algorithms to determine the type or concentration of the odor starting from the signals received from the sensors. In the context of odor nuisances there is also the contribution of citizens, who can be a determining factor for the control of the environment: citizen science is today considered an additional tool for the intelligent management of environmental monitoring, able to carry out an in-depth analysis of the pollution problem. This paper presents a continuous monitoring study of odours at the fenceline of an urban wastewater treatment plant, integrating instrumental monitoring with IOMS and citizens’ reports. The training of the IOMS was carried out in the field, using odor classification and quantification algorithms obtained through modern machine learning techniques, such as neural networks and random forest, in order to obtain a system capable of automatically recognizing either the class of odour or its concentration. Afterward, citizen science was used to collect and analyze reports made by citizens, who had an App for monitoring odour annoyance: the App was designed to make the citizens enter type and odor intensity they detected, associating each signal with position, time and weather data. The monitoring campaign was carried out over 5 months, and then the data obtained from the IOMS were analyzed along with citizens' reports. At first, an analysis was carried out taking into account the results of the algorithms deriving from the trained IOMS, in order to identify the most frequent odor classes. Subsequently, the results of IOMS were correlated with the reports and with the weather data, in order to verify whether the reports were connected to the plant.

The description of the odours perceived by the population, together with the identification of the appropriate wind direction, allowed to isolate the events that showed a correlation with the emission trends of the plant. The results of the study were therefore useful for establishing the real contribution of the monitored plant with respect to all the olfactory nuisance events perceived by citizens.