Estimation of particulate matter from simulation and measurements

The particulate matter is a typical indicator of small particles in the atmosphere. In addition to providing impacts on climate and environment, these small particles can bring adverse effects on human health. Then an accurate estimation of particulate matter is an urgent subject. We set up SPM sampler attached to our AERONET (Aerosol Robotics Network) station in urban city of Higashi-Osaka in Japan. The SPM sampler provides particle information about the concentrations of various SPMs (e.g., PM10 and PM2.5) separately. The AEROENT program is world wide ground based sunphotometric observation networks by NASA and provides the spectral information about aerosol optical thickness (AOT) and Angstrom exponent (α). Simultaneous measurements show that a linear correlation definitely exists between AOT and PM2.5. These results indicate that particulate matter can be estimated from AOT. However AOT represents integrated values of column aerosol amount retrieved from optical property, while particulate matter concentration presents in-situ aerosol loading on the surface. Then simple way using linear correlation brings the discrepancy between observed and estimated particulate matter. In this work, we use cluster information about aerosol type to reduce the discrepancy. Our improved method will be useful for retrieving particulate matter from satellite measurements.