|Ph.D Student||Sorek-Hamer Meytar|
|Subject||Integration of Satellite Observations and Ground|
Monitoring Data and its Application for Assessing
Exposure to PM in Israel
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor David Broday|
|Full Thesis text|
Particulate matter air pollution (PM) is derived from natural and anthropogenic sources. Recent studies revealed significant associations between exposure to fine particles (diameter <2.5mm) and human health. Until recently, such studies were based on standard ground monitoring data. Global observation from space can give an answer to some ground monitoring limitations, since satellite imagery allows environmental mapping at a large spatial scale. Indeed, application of satellite imagery for air pollution studies and exposure estimation has been well documented in over 100 publications over the last decade. This research explores whether, when and how it is possible to utilize satellite observations for assessing exposure to fine particles in Israel, as an example for an area with limited ground PM monitoring coverage that is characterized by a high albedo.
The work consists of data retrieval from different ground monitoring sources and satellite-based platforms, and applying several statistical methods for linking these databases. The research focused on examining the best satellite aerosol products (AOD, Aerosol Optical Depth) and statistical methods (e.g. linear, MARS, GAM) for estimating PM concentrations in areas that are characterized by highly reflective surfaces. In agreement with studies performed in northeastern USA, the daily variation in the AOD-PM relationships was found to be significant. The later was incorporated in a mixed effects model, which was found to provide the best predictions of daily PM (for both PM2.5 and PM10) based on satellite-borne AOD retrievals at collocated grid cells. The average fit for Israel was R2=0.45 (RMSPE of 12 mg/m3) and the key predictor variable was MODIS AOD Deep Blue product. The improvement over the simple linear regression (R2≈0.2) was dramatic. Similar results were obtained at San Joaquin Valley (SJV), California, USA, which is characterized by similar climate conditions and surface reflectivity.
The results of this study show that in spite of inherent limitations, satellite data can improve PM estimation over bright surfaces and can also be used for dust classification. Furthermore, in areas characterized by bright surfaces the AOD from the Deep Blue algorithm has been found to be more accurate. The results of this work have merit since they enhance our understanding of ambient PM occurrence in both space and time, and provide improved estimates of exposure to PM in Israel. Future highly resolved satellite products are expected to further improve the current results, making them more applicable for environmental health studies.