Aerosol Retrieval over Urban Area in MODIS Dark Target Land Algorithm: Implication to Surface Air Quality Monitoring
Friday, 19 December 2014
With amplified urbanization and industrialization during the last few decades, now more than half of the world’s population lives in urban areas. With surface particle matter (PM) concentration five or ten times higher than World Health Organization guidelines in some cities, it is very critical to accurately monitor PM air quality for global cities on a daily basis. The new version (C6) of MODIS Dark Target Land Aerosol Algorithm (MDT) provides near-daily aerosol optical depth (AOD) retrievals at 10km2 and 3km2 spatial resolutions, which can be used to estimate surface PM. However, initial validation efforts during DISCOVER-AQ field campaign over Baltimore-DC area showed that MDT overestimates AOD over urban areas, primarily because the bright and complex urban surface does not meet MDT assumptions. Surface characterization can be challenging and small error (~0.01) can produce large errors in retrieved AOD (~0.1), which can further lead to errors of around 5 μgm-3in surface PM2.5 air quality in highly urban areas. Here, we combined the urban percentage information from the MODIS Land Product (MCD12Q1) with MODIS land surface spectral reflectance product (MOD09A1) to derive a new urban surface reflectance relationship to be used within the MDT algorithm framework. We applied the new surface characterization to the MDT algorithm, and compared the retrieved AOD with AOD observed the ground-based AERONET network. AOD retrievals both in 10km and 3km spatial resolution show significant improvement over urban areas over the U.S. The bias in AOD reduced to -0.01 from 0.07, percentage of retrievals within uncertainty window increased to 85% from 62%. The improvement in AOD retrieval in other parts of the world is also observed but more analysis and research is required to apply these surface corrections globally. We will also present analysis on the impact of particulate matter air quality estimation over US while utilizing new retrievals as compared to existing retrievals over the urban areas.