Urban aerosol retrieval in MODIS dark target algorithm

Collection 6 Aerosol Products
Becoming Available
•C6 Aerosol Product Includes:
• Dark Target AOD at 10 km2
• Deep Blue AOD at 10 km2
• Deep-Dark Merged AOD
• Dark Target AOD at 3 km2
Urban Aerosol Retrieval in MODIS
Dark Target Algorithm: Implications
to Air Quality Monitoring
Pawan Gupta1,2 , Rob Levy2, Shana Mattoo2,3, and Leigh Munchak2,3
Universities Space Research Associations
NASA Goddard Space Flight Center, Greenbelt, MD, USA
Air Quality Applied Sciences Team 6th Semi-Annual Meeting (AQAST 6)
January 15-17, 2014
Motivation: Why Urban?
 PM2.5 pollution levels in many mega cities exceeds the WHO standards
by 5 to 10 times.
 Satellite observed aerosol information has been increasingly in use for
air quality monitoring efforts at local to regional to global scales.
 MODIS Dark Target AOD validation studies have shown a bias over
urban areas due to surface assumptions (Oo et al., 2010; Castanho et
al., 2007, Munchak et al., 2013, Gupta et al., 2013)
 Urban areas comprise 0.5% of the Earth’s surface, but will contain 2/3
of the Earth’s population by 2025, thus addressing urban bias in AOD
is critical for obtaining accurate air quality information from space.
City Center Appears as ‘HOT SPOT’ in
Distance (deg)
New Delhi, India
Gupta et al., 2012
Aerosol and Pollution in Mega Cities
Mega cities appeared as hot-spots in
MODIS AOD images with high gradient
from center to outside the city area.
Percent of pixels identified as
urban in same 15 km box
around AERONET station
Mean bias (MODIS 3 km AERONET) averaged over the
campaign duration at each
AERONET location
Land identified as urban by MODIS
land cover product at 500 m resolution
Munchak et al., 2013
Surface Characterization in MODIS
Dark Target (MDT) Retrieval
•MDT assumes a relationship between
the visible (VIS) and shortwave-IR
(SWIR) surface reflectance, based on
statistics of dark-target (primarily
Levy et al., 2007, 2013
vegetated) surfaces.
Over brighter and more variable surfaces (e.g. urban),
the assumed VIS/SWIR relationship breaks down (Oo et
al., 2010; Castanho et al., 2007)
Accounting for
Urban Bias
Here, we use MODIS Land
surface product (“MOD09”,
Vermote et al.) to derive a new
VIS/SWIR ratio versus Urban%
VIS/SWIR surface relationship
for urban areas where urban %
> 20%.
RVIS = f (RSWIR, Angles, NDVISWIR, Urban%)
10 km2
3 km2
C6 vs C6_Urban – Aqua, April 18, 2010
C6, C6_Urban
(0.41±0.14, 0.26±0.09)
Washington DC
(0.21±0.02, 0.17±0.01)
(0.18±0.04, 0.15±0.03)
Spring Time Reduction in AOD over
Urban Regions
Aerosol Retrieval Improvements over Large
Urban Corridors of Eastern USA
Spring 2010
Philadelphia / New York
Aerosol Retrieval Improvements over Large
Urban Corridors of Eastern USA
Spring 2010
Washington DC / Baltimore
Aerosol Retrieval Improvements over Large
Urban Corridors of Eastern USA
Spring 2010
Implication to
Surface PM Air
Ancillary Data
Driving surface PM from column AOD measurements is challenging
problem, having more reliable AOD over urban areas will improve
PM estimation skills of statistical/physical models
MODIS land surface reflectance and land cover classification
data sets have been used to define a VIS/SWIR surface
reflectance relationship to be used over urban surfaces (urban
percentage > 20%). The standard C6 MODIS Dark-Target
surface reflectance relationship was replaced.
Reduced AOD is seen over urban areas. Compared to AERONET
observations, these new retrievals remove some of the high bias
normally seen over large urban areas.
Ongoing/Future Work
Evaluating the urban surface relationship over global cities,
testing over longer time series.
Evaluating Aerosol Models used over Urban Areas in the DTA.
Implementing into the MODIS Dark Target Land algorithm?
Exploring impacts of new AOD retrieval on regional and global
studies of air quality, PM2.5 and health

similar documents