Monitoring and Forecasting Air Pollution Levels by Exploiting Satellite, Ground-Based, and Synoptic Data, Elaborated with Regression Models

Abstract

This paper presents some of the results of a project that aimed at the design and implementation of a system for the spatial mapping and forecasting the temporal evolution of air pollution from dust transport from the Sahara Desert into the eastern Mediterranean and secondarily from anthropogenic sources, focusing over Cyprus. Monitoring air pollution (aerosols) in near real-time is accomplished by using spaceborne and in situ platforms. The results of the development of a system for forecasting pollution levels in terms of particulate matter concentrations are presented. The aim of the present study is to utilize the recorded PM10 (particulate matter with aerodynamic diameter less than 10 μm) ground measurements, Aerosol Optical Depth retrievals from satellite, and the prevailing synoptic conditions established by Artificial Neural Networks, in order to develop regression models that will be able to predict the spatial and temporal variability of PM10 in Cyprus. The core of the forecasting system comprises an appropriately designed neural classification system which clusters synoptic maps, Aerosol Optical Depth data from the Aqua satellite, and ground measurements of particulate matter. By exploiting the above resources, statistical models for forecasting pollution levels were developed.