2. MAIAC Products (1 km, gridded)
Atmosphere
• Cloud Mask
• Water Vapor
• AOT & fine mode
fraction
3. Surface
MAIAC Products (1 km, gridded)
• Parameters of RTLS BRF
model
• Surface Reflectance (BRF)
• Dynamic Land-Water-
Snow Mask
4. Multi-Angle Implementation of
Atmospheric Correction (MAIAC)
Queue of
K days
New
Granule
2. LTP: Retrieve
Water Vapor
(NIR Algorithm)
3. QB, LTP: Cloud
Mask (Covariance-
Based Algorithm)
1. Grid L1B
Data and Split
in Tiles
NO
Use MODIS Dark Target Algorithm
Is B7 BRF
known?
YES
7. QP: Retrieve BRF and
albedo in reflective bands.
Backup: Lambertian retrieval
6. LTP: Retrieve
Fine Mode Fraction
5. LTP: Retrieve B3 AOT
using known surface BRF,
B=7bB
4. QB: Retrieve Spectral
Regression Coefficients in
Blue band B3 (bB).
QB:Is Snow
Detected?
4a. Retrieve Snow
sub-pixel Fraction and
Snow Grain Size
5. Aerosol Retrieval Algorithm
2
),(
B
TheorMeas
AOTRR
rmse
Compute AOTB and coarse mode fraction
using Blue (B3), Red (B1), SWIR (B7) bands.
Surface BRF: use SRC in blue band, . BRF in B1
and B7 is known from previous retrieval with uncertainty .
Algorithm: Fit Blue band to find AOTB for given , and find
by minimizing
7B
ijij
Blue
ij b
)(ij
8. AERONET Validation, 9 yrs. of TERRA Data
The accuracy of MAIAC and MOD04 is generally similar over green
and relatively dark parts of the world.
MAIACMOD04
AERONET
9. Summary
• Developed a new algorithm MAIAC based on a
time series and imagery processing.
• The algorithm is generic and works over both dark
and bright surfaces.
• The suite of products includes 1 km AOT/coarse
mode fraction and BRF/albedo.
• MAIAC successfully passed NASA HQ ATBD
review and will be tested operationally on
MODAPS.