Data Fusion: MODIS Terra

1
Data Fusion MODIS Terra Aqua Aerosol Optical
Thickness
Viktor Zubko, Gregory Leptoukh and Arun Gopalan,
GES DISC Code 610.2
Realistic gaps
Continuous Original scene

• Data
• MODIS Terra AOT (T)
• MODIS Aqua AOT (A)
• Combining Options
• Merging (M)
• Interpolation (I)
• Simultaneous M and I (MI)
• Used Combinations
• im(t,a) M(T, A)?I
• mi(t),i(a) I(T), I(A)?M
• mi(t,a) MI(T,A)
• mmi(t),mi(a) MI(T), MI(A)?M
• i(t) I(T)
• i(a) I(A)
• mi(t) MI(T)
• mi(a) MI(A)

Initial AOT
Merged AOT
Terra Aqua AOT
Interpolated AOT

• Data Fusion (DF) is a method of combining
  near-coincident (in time and space) satellite
  data to produce complete global or regional maps.
  This is important because satellite data contain
  gaps arising from Sun glint, clouds, interorbital
  areas, or retrieval algorithm failures.
• DF usually includes data merging and
  interpolation steps.
• The goal of the work is to find a fast and
  accurate DF method for implementation into
  Giovanni.
• We investigated various methods and limitations
  of data merging with and without interpolation
  using MODIS Terra and Aqua Daily Level 3 Aerosol
  Optical Thickness (AOT) data as a prototype.
• To assess the merger accuracy, we introduced a
  merging confidence function, which is the
  percentage of the merged AOT pixels as a function
  of the relative deviation of the merged AOT from
  the initial Terra and Aqua AOTs.
• We worked with eight different combinations of
  merging and interpolation. The best method is to
  first merge the Terra and Aqua AOTs, and then to
  fill in the remaining gaps by interpolation.