MISR Georegistration Overview

1
MISR Geo-registration Overview
Brian E. Rheingans Jet Propulsion Laboratory,
California Institute of Technology AMS Short
Course on Exploring and Using MISR Data Long
Beach, California 9 February 2003
2
MISR Background
Four MISR images over Appalachain
Mountains Nadir, 45.6 deg, 60.0 deg, 70.5 deg
forward viewing cameras To make use of angular
as well as spectral information all (9 cameras X
4 bands 36) pixels must be accurately
co-registered
3
MISR Background
Each MISR camera eventually Views one ground
point at a Slightly different time from
a Different angle as the spacecraft Passes over
that point. Data is resampled from each
channel Onto a common map projection, Called
Space Oblique Mercator (SOM).
4
SOM Background
The Space Oblique Mercator (SOM) map projection
was developed to support LandSat which covers the
same large geographic extent as MISR. SOM was
designed to minimize the shape distortion and
scale errors throughout the length of the MISR
swath near the satellite ground track.
5
SOM Background

• Terra follows a pattern of orbits which repeats
  after 233 unique orbits
• Each of the 233 possible orbits is called a path
• SOM defines a separate projection for each of
  these paths
• For MISR, a path begins at a particular longitude
  as the satellite
• crosses the ascending node.
• This path implies a specific longitude of
  ascending node, which implies
• a specific SOM projection applicable to that path

6
MISR Orbital Paths
7
SOM Background

• SOM coordinates are X,Y in meters
• X axis points in the approx. direction of the
  satellite groundtrack
• Y axis is perpendicular to the X axis
• The origin of X is at the ascending node equator
  crossing
• Once ascending node is reached, X values start
  over at 0 for the next path/proj.
• The sign of Y is relative to the sinusoidal X axis

8
SOM Background

• Line/Sample
• Pixels are arranged in a regular 2-D array in SOM
  space
• The indices are absolute line and sample
• Given the SOM(X,Y), pixel resolution, and
  absolute line/sample offset
• you can compute the SOM coordinate of any other
  point in the swath
• Blocks
• MISR path is cut-up into a series of pre-defined,
  uniformly-sized SOM
• boxes along the ground track, called blocks
• For MISR, block-relative lines and samples
  restart at 0,0 at the top
• left of each block and thus MISR has a block
  number
• MISR files are BLOCK-RELATIVE, not absolute.

9
HDF-EOS Background

• Grid
• All MISR products are in HDF 4 format
• HDF-EOS HDF 4 GCTP
• HDF-EOS data type for map projected data is
  called Grid
• HDF-EOS Grid metadata stores orbital parameters
  and map
• projection parameters

• Stacked Block Grid
• Problem was MISR data (w/ large extent) didnt
  fit the Grid model
• Extension to the grid model to store MISR blocks
• Actually, it is a 3-D array (block, line, sample)
• Blocks are stack on top each other with a
  constant vertical offset (1 block)
• The lateral offset is not constant

10
HDF-EOS Background

• Block Offset
• Blocks may be shifted left or right by an
  integral multiple of 17.6 km
• Shifts are pre-defined, as is the block locations
  to encompass MISR footprint
• MISR stack-block grid is defined for a given path
  by the coord. of the first
• block, along with standard projection metadata
  and an array of block offsets
• to define the locations of all subsequent
  blocks.

11
HDF-EOS Background
HDF-EOS routines do NOT assemble the Blocks.
That is left for the user. 180 blocks are
defined for every MISR Product to make block
index in absolute. However, roughly 142 blocks
have data for Any given orbit. The extra blocks
are to Allow for seasonal variation.
12
MISR Product Characteristics

• MISR L1B2 and above product characteristics
• Space Oblique Mercator (SOM) map projection
• Stacked-block HDF-EOS Grid format
• Large geographic extent
• Determining latitude and longitude of a MISR
  pixel (2 methods)
• 1) Look up in Ancillary Geographic Product (AGP)
• One AGP for each Terra orbital path
• Separate product
• Stored at 1.1km
• 2) Convert coordinate systems
• Orbital param. and projection info. are in all
  MISR products
• HDF-EOS library access routines used to read
  metadata
• GCTP map projections library used (w/ HDF-EOS)
• Algorithm to map between SOM(X,Y) and
  MISR(b,l.l,s.s)
• See Appendix A of the MISR DPS

13
Coordinate Conversions

• Forward Conversion
• Lat/Lon -gt SOM(X,Y) -gt MISR(b,l.l,s.s)
• Useful for resampling MISR data to another
  projection
• Notice line and samples can be fractional
• Inverse Conversion
• MISR(b,l.l,s.s) -gt SOM(X,Y) -gt Lat/Lon
• Useful for determining lat/lon of a pixel
• Orbit path must be known a priori (metadata is
  required)
• Coordinate conversions are reversible with
  reasonable numerical
• Precision only for positions near the satellite
  ground track.
• Positions not near the ground track are
  better-described on another
• Path and SOM projection.

14
SOM lt-gt Lat/Lon

• Forward Lat/Lon -gt SOM
• Given a pixels position in Lat/Lon coordinates
  assuming a particular
• MISR orbit path, SOM(X,Y) in meters can be
  computed by
• Retrieving projection parameter metadata.
• Calling SOM forward routines from the GCTP
  library.

• Inverse SOM -gt Lat/Lon
• Given a pixels position in SOM(X,Y) meters,
  assuming a particular
• MISR orbit path, Lat/Lon can be computed by
• Retrieving projection parameter metadata
• Calling SOM inverse routines from the GCTP
  library.

15
MISR(b,l.l,s.s) -gt SOM(X,Y)Inverse Conversion
16
SOM(X,Y) -gt MISR(b,l.l,s.s)Forward Conversion
17
Coordinates in other Map Projections

• Forward
• AnyProj(X,Y) -gt Lat/Lon -gt SOM(X,Y) -gt
  MISR(b,l.l,s.s)
• Useful for reprojecting to another map projection

• Inverse
• MISR(b,l.l,s.s) -gt SOM(X,Y) -gt Lat/Lon -gt
  AnyProj(X,Y)

18
Interrupted Goode Homolosine
19
Geometric Parameters Product
The following Parameters are Stored in the
Geometric Product (GMP)
View Zenith per camera View Azimuth per
camera Solar Zenith Solar Azimuth Scatter Angles
per camera Glitter Angles per camera