Documentation for MEaSUREs MODIS Mosaic of Greenland 2005 (MOG2005) Image Map, Version 1

Back to Top

Collapse All / Open All

Detailed Data Description

This data set consists of two image maps:

  • Surface morphology derived from high-pass filtered MODIS Band 1 (red light) images;
  • Snow grain-size inferred from the MODIS Band 1 and Band 2 (near-infrared light) normalized difference.

The image maps were assembled from 104 MODIS scenes (five-minute segments of swath data) selected from the MODIS/Aqua and MODIS/Terra Calibrated Radiances 5-Min L1B Swath 250m data sets (MYD02QKM and MOD02QKM). Scenes were mosaicked using: a) a stacking technique (data cumulation) that allows multiple images to contribute to a single grid cell's representation; and b) a weighting scheme that favors near-nadir views and feathers the edges of contributing images.

Both image maps are provided at 100 m and 500 m spacings. In addition, the following ancillary files are available:

  • Scene counts and average weights for grid cells in the 100 m and 500 m image maps;
  • Scene counts, average weights, and standard deviations of unweighted values for grid cells in the 500 m grain size image map¹.

Lastly, this data set includes ESRI shapefiles with outlines of the Greenland coastline and ice edges suitable for overlaying on the 100 m and 500 m MOG image maps. However, these files only contain data for Greenland and not areas west of the mainland including Ellesmere Island, Axel Heiberg Island, Devon Island, Baffin Island, and the Labrador Peninsula.

¹Scene counts and weights are not provided for the 100 m grain size image map. The investigators felt these data offered no additional value compared with the 500 m grid.


The image maps and corresponding ancillary data are available in the Geographic Tagged Image File Format (GeoTIFF, 16-bit).

Maps of the Greenland coastline and ice edges are available as ESRI shapefiles. One complete shapefile consists of four file types: .shp, .shx, .dbf, and .prj. For details about the shapefile format, see the ESRI Shapefile Technical Description white paper (pdf, 124 KB).

Background color on
File and Directory Structure

Data are available on the HTTPS site: directory.

Within this directory, there is one folder which contains:

  • 11 GeoTIFF (.tif) files--4 at the 100 m resolution and 7 at the 500 m resolution (see File Naming Convention section for more information)
  • 4 Shapefiles (.shp) plus auxiliary files (.shx, .dbf, .prj) of the coastlines and ice edges
  • 6 associated metadata (.xml) files
Background color on
File Naming Convention

This section explains the GeoTIFF and Shapefile file naming conventions with examples.


Example File Name


Naming Convention


Refer to Table 1 for descriptions of the variable strings used in the GeoTIFF file naming convetion.


Table 1. Descriptions for Variable Strings in GeoTIFF File Names
String Values Description
mog, _2005_ Mosaic of Greenland, 2005
grid 100 or 500 Grid spacing (100 m or 500 m)
map hp1 Surface morphology (high-pass filtered Band 1) image map
grn Grain size image map
hct Scene count for each cell in the surface morphology (hp1) image map
hwt Average weight applied to each cell in the surface morphology (hp1) image map
gct Scene count for each cell in the grain size image map
gwt Average weight applied to each cell in the grain size image map
gsd Standard deviation of unweighted values for each cell in the grain size image map


Example File Name


Naming Convention



Table 2. Naming Convention Variables for Shapefiles Files
String Values Description
mog Mosaic of Greenland
res 100, 500 Resolution, 100 m or 500 m
src GEUS, GIMP Map source:
  • GEUS: Geological Survey of Denmark and Greenland
  • GIMP: Greenland Ice Mapping Project
See Derivation Techniques and Algorithms for details.
map coastline, iceedge Coastline or ice edges map. Note: these maps do not cover areas in the MOG west of Greenland, including Ellesmere Island, Axel Heiberg Island, Devon Island, Baffin Island, and the Labrador Peninsula
ext .dbf, .prj, .shp, .shx File extension. A complete shapefile comprises four files with the following extensions: .dbf (database file), .prj (projection information), .shp (spatial data), and .shx (shape indices).
Background color on
File Size

The 100 m/500 m surface morphology and grain size GeoTIFFs are approximately 520 MB/22 MB and 235 MB/12 MB, respectively. Ancillary data files range from about 1 MB to 550 MB, depending on grid spacing and parameter of interest. Shapefiles range from approximately 5 MB to 50 MB.

The complete set of all 100 m GeoTIFFs is approximately 1.3 GB. The set of 500 m GeoTIFFs is approximately 100 MB. The total data volume including shapefiles is approximately 1.5 GB.

Background color on
Spatial Coverage

Northernmost Latitude: 85° N
Southernmost Latitude: 57° N
Easternmost Longitude: 11° E
Westernmost Longitude: 109° W

Spatial Resolution

Image maps are provided at 100 m and 500 m spacings. The underlying MODIS Band 1 and 2 swath data have a nominal resolution of 250 m; however, the image stacking scheme used to assemble the mosaics increases the resolution of the final product beyond that of individual MODIS scenes to between 150 m and 250 m, depending on the number of images that were stacked and how the images were weighted. See the Surface Morphology Image Map section of this document for information on Compositing via Data Cumulation.

Projection Description

  • Projection: Polar Stereographic
  • Spheroid: WGS-84
  • Rotation: 45°, that is 45 W longitude extending down from pole
  • Longitude of Central Meridian: 0°
  • Latitude of True Scale: 70° N

Note: This projection is identical to The Digital SAR mosaic of Greenland but on a different grid.

Grid Description

Table 3 lists the dimensions (in pixels) for the 100 m and 500 m grids and the locations (in meters from the origin) of the upper left corner of the upper left cell:

Table 3. Grid Dimensions
Grid x y Upper left corner, upper left cell
100 m 21000 px 28000 px -1200000.0 m from origin
500 m 4200 px 5600 px -600000.0 m from origin

Note: Grids do not include the North Pole.

Background color on
Temporal Coverage

All MODIS scenes were acquired between 12 March and 30 April, 2005.

Background color on
Parameter or Variable

Parameter Description

The MOG image maps report two parameters:

  • Surface morphology, derived from brightness variations in MODIS Band 1 red light images;
  • Snow grain size, inferred from the normalized difference radiance ratio of red to near-infrared light.

Surface Morphology

Many processing steps were required to create a seamless and uniform surface morphology mosaic from the numerous images that contributed to each grid cell. As such, the image values no longer have a clearly quantifiable relationship to the top-of-atmosphere, red light reflectances from which they were derived. Instead, the image map provides a semi-quantitative but highly consistent approximation of the surface shape and reflectivity for the entire continent, as illuminated by the sun across all surface types.

Snow Grain Size

Processing for the grain size mosaic was reduced compared with the surface morphology image map. While this approach sacrifices seamlessness, it produces a truer quantitative map of radiance ratios that approximate mean snow grain size in dust-free, non-shadowed areas with snow, firn, and ice. Values from 10 through 1100 represent mean optical snow grain size in microns; values of 5 and 1105 indicate grain sizes which lie outside of this range or grid cells for which a grain size could not be computed.

Background color on

Software and Tools

Software and Tools

GeoTIFF files are most easily accessed using GIS software such as QGIS and ArcGIS.

Background color on

Data Acquisition and Processing

Note: The MOG2005 image maps were generated using the same procedures as the MODIS Mosaic of Antarctica 2003-2004 (MOA2004) Image Map data set. The following sections outline the key steps used to create the mosaics, ancillary data, and corresponding coastline and ice edge maps for Greenland. Additional detail is available in the Data Acquisition Methods section of the MOA2004 documentation.

Background color on
Theory of Measurements

Since the 1990s, a host of studies have demonstrated that carefully processed satellite radiometry, most notably from the Landsat series and NOAA's AVHRR instruments, can reveal unprecedented details about the surface morphology of ice sheets. Furthermore, subsequent research has demonstrated that the red-infrared normalized difference radiance ratio can be used to map surface snow grain size, because snow reflectivity decreases in the infrared as grain size increases.

The twin MODIS instruments on board NASA's Terra and Aqua satellites provide an opportunity to exploit two active sensors during the compilation period with a higher spatial and radiometric resolution than AVHRR. In addition, this data set improves the accuracy, detail, and seamlessness of the final products by combining new methods with several preexisting techniques.

Background color on
Data Acquisition Methods

The MOG2005 image maps were composited from MODIS scenes acquired between 12 March and 30 April 2005. Scenes were selected from a specific time window to restrict solar illumination to a range of azimuths and to ensure that all scenes were illuminated from approximately the same direction.

Background color on
Derivation Techniques and Algorithms

Geolocation and Processing

Band 1 and Band 2 scenes from MYD02QKM and MOD02QKM, together with illumination and viewing angles from MYD03 and MOD03¹, were geolocated and resampled onto the projection grid using NSIDC's MODIS Swath-to-Grid Toolbox (MS2GT). The software interpolated the MYD03 and MOD03 latitude/longitude data from 1 km resolution to 250 m and then resampled the MODIS/Aqua and MODIS/Terra Level-1B calibrated radiances to the grid using a forward elliptical weighted average algorithm (Greene et al. 1986).

¹MODIS Level 1A Geolocation Fields from EOS Aqua and MODIS Level 1A Geolocation Fields from EOS Terra

Destriping of MODIS Image Data

The MS2GT algorithm was modified to remove MYD02QKM and MOD02QKM striping artifacts, a known problem with all Terra and Aqua MODIS 250 m Level-1B data, by adding a Lambertian solar zenith angle normalization on the swath data for both bands. Telemetry noise and line drops, which have the appearance of chads in the projected images, were reset to zero (treated as masked cloud areas). This procedure is discussed in detail in the Destriping of MODIS Image Data section of the MOA2004 documentation.

Cloud Masking

The geolocated scenes were manually masked to remove clouds, cloud shadows, fog, blowing snow, and heavy surface frost. Refer to the Cloud Masking section of the MOA 2004 documentation. The final image maps are nearly perfectly cloud-cleared, except for some areas of thin clouds, cirrus cloud shadows, and fog or low-lying small clouds.

Surface Morphology Image Map

Geolocated, destriped Band 1 images were high-pass filtered to reduce non-Lambertian illumination and to reset the mean grayscale range to a common value for compositing. Then for each gridded image the investigators created a corresponding "weight image" in which each non-masked pixel is assigned a scalar value (or weight). Weights were computed based on proximity to the nadir track, favoring near-nadir areas, and proximity to an image or mask edge to feather the edges of the component images. Weight images were then combined using stacking techniques called image super-resolution or data cumulation. These techniques allow multiple images to contribute to how a single grid cell is represented in the final composite.

The algorithms used to compute and combine the weight images into the final mosaics are provided in the MOA2004 Compositing the Image Swaths documentation.

Snow Grain Size

The snow grain size image map was generated from MODIS Band 1 and Band 2 imagery using the following normalized difference ratio:

(Band1 - Band2)/(Band1 + Band 2)

This ratio exploits the decreasing reflectivity of snow in the infrared range to create an image that is sensitive to grain-size variations (Warren 1982) (Fily et al. 1997). To maintain the quantitative relationship, images were not pre-processed beyond geolocation, calibration, and destriping.

Intermediate images of grain size were generated by applying a model-derived lookup table to computed normalized differences and solar zenith angles. These images were then composited using the same approach as the surface morphology image map, except that computed grain sizes of <10 µm and >1100 µm were treated like missing data and set to 5 µm and 1105 µm, respectively. Count and weight images for the grain size composites were slightly different from the corresponding surface morphology count and weight images because of these additional missing values. In cases where all images gave out-of-range grain size results, the grid cell value was set to either 5 µm or 1105 µm. For additional details, including algorithms and snow grain size validation measurements, see the Optical Mean Snow Grain Size section in the MOA2004 documentation.

Greenland Coastlines and Ice Edges

Greenland coastline shapefiles, suitable for overlaying on the 100 m and 500 m MOG image maps, were derived from 500,000:1 and 2,500,000:1 (complex UTM) shapefiles, respectively, which were graciously provided by Willy Lehmann Weng of the Geological Survey of Denmark and Greenland (GEUS). The investigators opted to approach GEUS to address the apparent gelocation problem over northern Greenland found in all public domain coastline data sets (Henriksen 2000). The GEUS source files were imported into Esri ArcMap 10.0 and reprojected to the MOG projection.

The ice edge shapefiles were derived from 90 m and 180 m ice masks developed by the Greenland Ice Mapping Project (GIMP) at the Byrd Polar and Climate Research Center. The 90 m and 180 m masks (Version 1, downloaded 02 December, 2011) were first nearest-neighbor resampled to the 100 meter and 500 meter MOG grids, respectively. Using the ENVI raster-to-vector function, the resampled masks were then converted to ENVI vector files which were exported to the GIMP Greenland ice edge shapefiles (Howat 2014).

Error Sources

Wolfe et al, 2002 estimated the accuracy of the MYD03/MOD03 Level-1A geolocation data to be 50 m, considerably better than the MYD02QKM/MOD02QKM ground-equivalent nadir pixel size of 250 m. The accuracy and precision of this geolocation was also tested for the MOA2004 using known surface sites, such as South Pole Station, Vostok Station, Siple Dome camp and traverse trail, and areas of well-mapped coastline such as Ross Island and the northern Antarctic Peninsula. The investigators did not find discrepancies greater than 125 m in the projected location of any fixed object.

In general, the MOG quality is higher in areas with both high counts and high weights. The mean image weight and image count ancillary data are provided for users to assess image quality in various regions of the MOG image maps.

The MOG snow grain size error is approximately ±50 µmeters, as estimated for the MOA by comparing in situ spectra of varying snow grain sizes with near-simultaneous MODIS images processed in the same manner. However, snow grain size varies greatly over the period of image acquisition. As such, large ranges of snow grain sizes were averaged together in certain areas. Additional details are provided in the Error Sources section in the MOA2004 documentation.

Background color on
Sensor or Instrument Description

The MODIS instruments collect 12-bit radiometric data in 36 spectral bands, ranging from 0.4 µm to 14.4 µm in wavelength. Bands 1 and 2 are imaged at a nominal resolution of 250 m at nadir. The Terra satellite crosses the equator from north to south (descending node) at 10:30 a.m. local time; Aqua crosses from south to north (ascending node) at 1:30 p.m. local time. Both satellites occupy sun-synchronous, near-polar, circular orbits at an altitude of 705 km. The MODIS instruments' ±55 degree scanning pattern produces a 2330 km cross-track by 10 km along-track swath with nearly complete global coverage every one to two days.

Background color on

References and Related Publications

Contacts and Acknowledgments

Terry Haran
National Snow and Ice Data Center
449 UCB, University of Colorado
Boulder, CO 80309-0449 USA

Jennifer Bohlander
Polar Science Consulting
200 Merry Hill Drive
Cary, NC 27518 USA

Ted Scambos
National Snow and Ice Data Center
449 UCB, University of Colorado
Boulder, CO 80309-0449 USA

Thomas Painter
Jet Propulsion Laboratory
4800 Oak Grove Drive
Pasadena, CA 91109

Mark Fahnestock
Institute for the Study of Earth, Oceans, and Space
University of New Hampshire
Morse Hall
8 College Road
Durham, NH 03824-3525 USA


This work was supported by NASA grants NNG04GM10G and NAG5-7760, and by USGS funds supporting J. Bohlander (instigated by J. Ferrigno) during image selection and cloud clearing. GIMP ice masks were developed by the Greenland Ice Mapping Project Two: Measuring Rapid Changes in Ice Flow (MEAS-06-0071). The investigators also wish to thank R. Bindschadler, W. Abdalati, and J. Ferrigno for their interest and support initiating this project.

Document Information


November 2015