The image below is a selected area of the image file: DMS_1000109_01938_20091016_17423840.tif

Figure 1. Sample Data Record

The listing below shows the information embedded in the GeoTIFF image file DMS_1000109_01938_20091016_17423840.tif displayed above. GeoTIFF file information can be displayed using tools such as gdalinfo from the Geospatial Data Abstraction Library (GDAL). Visit the GDAL Web site for more information.

This DMS data processing package is based on a rigorous camera model for perspective scene generation and mapping from image plane to the ground. In order to transform from 2D image frame to 3D world coordinates, camera focal length, pixel size, principal point, lens distortion coefficients, and camera position and attitude (Roll, Pitch, and Heading) must be known. These parameters are used in a photogrammetric rotation between image-plane and target coordinates process to orthorectify images into a real world coordinate on a pixel-by-pixel basis.

Two instruments are used in the airborne data acquisition: an Applanix POS/AV navigation system and Canon EOS II digital camera. The POS/AV is a combined inertial measurement unit (IMU) and Global Positioning System (GPS) and provides precise position and attitude information.

The basic photogrammetry relation between a point (x, y) on the image frame to the ground surface point (Xg,Yg,Zg) can be expressed as Equation 1:

(Equation 1)

Where:

Rcam rotation matrix in Equation 1 uses a grid north coordinate system with three rotation angles as Omega, Phi, Kappa. However, the navigation system collects Roll, Pitch, and Heading angles with a true north coordinate and differ from Omega, Phi, and Kappa angles.

Assuming the navigation system and camera offset can be measured and adjusted, a rotation correction (Rb) between IMU and camera angular misalignment can be estimated with a boresight angle correction procedure. Rcam rotation matrix can add the correction matrix (Rb) as in Equation 2:

(Equation 2)

Where:

The orientation of the DMS camera is 90 degrees counter clockwise from the flight/navigation direction. Rnav in Equation 2 is therefore rewritten as in Equation 3:

(Equation 3)

Where:

DMS acquires standard JPEG imagery, which are stored via Ethernet on a supporting laptop computer. This computer also runs the EOS camera utility software to monitor and control image exposure. Frame capture is controlled using an external intervalometer. The intervalometer provides a Transistor-Transistor-Logic (TTL) pulse to the navigation system which enables precise timing for each image capture.

1. DMS Camera Calibration

Lens calibration is fundamental to rigorous camera modeling. It is used to eliminate geometric distortion. Calibrated camera interior correction parameters include:

1. Focal length
2. Principal point (xp, yp)
3. 3rd-order (k1), 5th-order (k2), and 7th-order (k3) terms of radial distortion correction
4. Coefficients of decentering distortion (p1, p2)

The corrected image frame coordinates (xc, yc) can be calculated from the measured coordinates (xm, ym) by using the Equations 4 and 5:

(Equation 4)

(Equation 5)

Where:

x and y are now with respect to the principal point:

(Equation 6)

(Equation 7)

(Equation 8)

(Equation 9)

Where:

2. Offset Adjustment between Navigation System and Camera

The three dimensional distances between camera principle point, IMU, and GPS antenna are integrated into the final navigation data solution. This enables the rotation matrix Rb to properly represent the boresight correction angles.

The measured vector (dx, dy, dz) are fully integrated in the post flight navigation data process. Therefore, the camera position is adjusted to the same origin of navigation system.

3. Boresight Correction

If we combine Equations 2 and 3, Rcam can be rewritten as:

(Equation 10)

Therefore, the angular misalignment (Rb) between navigation system and camera can be estimated if Pg and Pf are known from equation (1).

Pg vector can be derived from GPS measured ground control points or a high resolution orthorectified reference map with DEM. Pg related Pf vector can be located in the raw image frame. Rotation matrix Rc is a constant matrix for each DMS configuration. Rotation matrix Rg is a constant matrix for a targeted area with the converged angle difference between grid north and true north. However, rotation matrix Rb can vary with physical and environmental conditions as well as navigation system errors. Rb is assumed to be relatively stable and errors from physical and environmental factors are minimized if the system installation is fixed under the same condition.

In the boresight correction process, we select three to four continuous frames in multiple flight directions to cover a targeted area. Numerous ground control points, either from a high resolution reference map or GPS surveyed points, are picked for each frame. Boresight angles are derived based on Equations 1 and 10 from Rb for each selected frame. Finally, boresight angles are averaged from all frames.

4. DMS Production Software Package

The DMS software package developed for data production is an integration of public domain software with an in-house user interface to facilitate the entire production process. The public domain software includes:

1. Open Source Software Image Map (OSSIM) (http://www.ossim.org). OSSIM is a publically available high performance engine for photogrammetry, remote sensing, image processing, and geographical information systems.
2. Geospatial Data Abstraction Library (GDAL). GDAL is a translator library for raster geospatial data formats and processing. It is available under an X/MIT style Open Source license by the Open Source Geospatial Foundation (http://www.gdal.org).

OSSIM is written in C++ with object-orientated modules and classes. A customized sensor model is integrated in the OSSIM modules and recompiled for use in DMS production. The front end interface is developed in Perl to make the OSSIM and GDAL libraries and utilities more accessible.

Special processing notes regarding each deployment to date are provided below.

Special Processing Notes

2012 Antarctica Campaign

GeoTIFF files were produced using the RADARSAT 200 m DEM. GEOID reference is established by DEM. Sea Ice imagery referencing EGM2008 are identified with a COMMENT tag in the GeoTIFF. There are areas were DEM and aircraft converge and where DEM exceeds aircraft height. These frames were projected with zero elevation as input and altitude above terrain fixed to 250 m. Affected frames are identified with COMMENT tags in the GeoTIFF, where each data issue is explicitly noted. All output reference NSIDC Antarctic stereographic projections. There is one exception: imagery acquired over California reference standard geographic lat/long and were produced using the National Elevation Dataset 10 m DEM.

2012 Greenland Campaign

Geotiff files were produced using the GLAS/ICESat 1 km Laser Altimetry DEM. GEOID reference is established by DEM. Sea ice imagery referencing EGM2008 are identified with a COMMENT tag in the geotiff. There are areas were DEM and aircraft converge and where DEM exceeds aircraft height. These frames were projected with zero elevation as input and alitude above terrain fixed to 230 m. Affected frames are identified with COMMENT tags in the geotiff, where each data issue is explicitly noted. All output references NSIDC Antarctic stereographic projections. There is one exception: imagery acquired over Virginia's East Coast reference standard geographic lat/long and were produced using the National Elevation Dataset 10 m DEM.

2011 Antarctica Campaign

GeoTIFF files were produced using the RADARSAT 200 m DEM. GEOID reference is established by DEM. Sea Ice imagery referencing EGM2008 are identified with a COMMENT tag in the GeoTIFF. There are areas were DEM and aircraft converge and where DEM exceeds aircraft height. These frames were projected with zero elevation as input and altitude above terrain fixed to 230 m. Affected frames are identified with COMMENT tags in the GeoTIFF, where each data issue is explicitly noted. All output references NSIDC Antarctic stereographic projections. There is one exception: imagery acquired over Southern California reference standard geographic lat/long and were produced using the National Elevation Dataset 10m DEM.

2011 Greenland Campaign

GeoTIFF files were produced using the GLAS/ICESat 1 km Laser Altimetry Digital Elevation Model of Greenland. There are areas were DEM and aircraft converge. Affected frames are identified with a COMMENT tag in the GeoTIFF. Frames where DEM exceeds aircraft height were projected with zero elevation as input. Frames where Aircraft-DEM delta was less than 65 m were projected using a 2D model.

2009 Antarctica Campaign

A refined camera model was implemented, as well as enhanced boresight angles. In addition RADARSAT 200 m DEM was used as ground reference input. A new metadata field has been embedded in each GeoTIFF referencing version differences. There are areas were DEM and aircraft converge. Affected frames are identified with an additional COMMENT tag in the GeoTIFF. Frames where the DEM exceeds aircraft height were projected with zero elevation as input. Frames where Aircraft-DEM delta was less than 65 m were projected using a 2D model. Most imagery were acquired in panchromatic mode, but all provided output is 3-banded imagery. Flights where some imagery were collected in both modes were processed as "color" as identified in the GeoTIFF embedded metadata.

DMS is a commercial off the shelf camera. The time stamp in the Level-0 jpeg imagery is from the internal camera clock. The camera's intervalometer is tied to the Applanix. For the Level-1B data, a series of linear regressions is used to map out the time lag and sync the camera shutters to the nav data. That synched time is embedded into the GeoTIFF along with other metadata.

The following processing steps are performed by the data provider.

DMS geo-rectified TIFF imagery is supplied on a best effort basis. Processing steps include (1) time sync between the DMS internal clock and POS/AV, (2) definition of frame geometry to include bore-sight adjustment, lens distortion and digital elevation model (DEM) inputs, and (3) GeoTIFF production with ancillary documentation and a variety of map projection allowances.

The camera's internal clock drifts over time. Alone it cannot be used to sync to corresponding navigation data. Thus an intervalometer is used to trigger each shutter function. This is captured via a TTL pulse as an Applanix event.

The cameras are calibrated prior to each deployment using the Photometrix Australis package. The calibration provides focal length, principal point, terms of radial distortion, as well as coefficients of de-centering distortion.

Boresight angles are calculated using High Resolution Ortho-rectified images (HRO) or other precise surveys and best available elevation data. Three or four continuous frames from multiple flight headings are used to derive the alignment differences between the camera's focal plane and the POS/AV. Ten to twelve broadly spaced control points are selected for each frame.

Once the boresight and lens distortion corrections are applied, a 3D ray trace is performed to generate the final GeoTIFF. The quality and resolution of DEM is a key factor in geo-location accuracy. The distance between the focal plane and ground determines the footprint of each frame. Over varying terrain the frame-edge location will be severely degraded if DEM resolution significantly differs from pixel geometry.

• ASTER Global Digital Elevation Model (GDEM)
• GLAS/ICESat 1km Laser Altimetry Digital Elevation Model of Greenland
• IceBridge DMS L0 Raw Imagery
• IceBridge DMS L3 Photogrammetric DEM
• IceBridge CAMBOT L1B Geolocated Images
• National Elevation Dataset 1/3 Arc-Second (NED 1/3)
• Radarsat Antarctic Mapping Project Digital Elevation Model Version 2

• NASA Digital Mapping System Web page (http://asapdata.arc.nasa.gov/dms/).
• IceBridge Data Web site at NSIDC (http://nsidc.org/data/icebridge/index.html).
• IceBridge Web site at NASA (http://www.nasa.gov/mission_pages/icebridge/index.html).
• ICESat/GLAS Web site at NASA Wallops Flight Facility (http://glas.wff.nasa.gov/).
• ICESat/GLAS Web site at NSIDC (http://nsidc.org/daac/projects/lidar/glas.html).