GLIMS Databasing Guide for the

University of Alberta Regional Center

 

            This document provides a brief overview of the methods used at the University of Alberta to define glaciers and their associated characteristics for the GLIMS (Global Land Ice Measurements from Space) project. The University of Alberta is the Regional Center responsible for the Canadian High Arctic, so we frequently deal with the task of subdividing large ice caps into meaningful glacier-size units for input to the GLIMS database. Our base imagery is provided by orthorectified Landsat 7 imagery from 1999 and 2000, and our base digital elevation information is provided by Canadian Digital Elevation Data (CDED) with 100 m grid spacing. Some of the CDED data was purchased from the Canadian government (Geomatics Canada), while the remainder was acquired from an access agreement between the University of Alberta data library and Geomatics Canada. The CDED data was derived from the digitization of 1:250,000 topographic map sheets, which in turn were derived from early 1960’s aerial photography. Consequently, some of this elevation data may be in error compared to present day conditions, but the general patterns remain correct and the data set still provides the highest resolution information that is currently available for the entire region.

            In sequence, the steps that we take to define a glacier and its associated characteristics are:

 

Orthorectifying imagery and elevation data

1.      1:250,000 topographic maps of the area of interest are scanned at 300 dpi on a large format scanner as tiff images. These maps are then loaded into ERDAS Imagine as RASTER imagery, and orthorectified using the Geometric Correction tool. Points are selected around the map corners and in the center to act as GCPs (Ground Control Points). The map is then resampled with the Imagine nearest neighbor algorithm based on the location of these GCPs. The working coordinate system for all of our datasets is UTM zone 17, with a NAD 83 datum.

2.      CDED data is mosaicked in ERDAS Imagine using the Data Prep/Mosaic Images function (DEMs are input in ‘USGS DEM’ format). Once the images are mosaicked, they are converted from their original lat/long coordinates to UTM 17 / NAD 83 and output in their new format with 100 m grid spacing (background elevation set to -150m)

3.      To orthorectify the base Landsat 7 imagery, we use PCI OrthoEngine and create separate files for panchromatic and multispectral modes (since panchromatic has 15 m resolution and multispectral has 30 m resolution). ‘Satellite orbital modeling’ is used as the math modeling method in OrthoEngine, and the relevant bands are loaded from the raw Landsat 7 HDF file. GCPs are then collected from the orthorectified topographic map and CDED data described above. Points on both the map and image are then chosen at a variety of elevations and throughout the image. Ideally >40 points are chosen per image, and any points with residuals >4 pixels (60 m in panchromatic) are removed from the analysis. The original image is then orthorectified based on these points and output as a new file.

4.      Orthorectified images are mosaicked

5.      Any additional Landsat 7 images are georeferenced to the original Landsat 7 imagery using ‘Thin Plate Spline’

 

Defining drainage basins for glaciers

1.      Load the orthorectified imagery and DEMs into ArcView

2.      Load the ‘Hydrov11’ and ‘Basin1’ extensions into ArcView (version 3.2). Hydrov11 is provided in the examples directory provided with software, while the Basin1 extension has to be downloaded from the ESRI website at www.esri.com

3.      Fill the sinks in the DEM using the Hydro/Fill Sinks option in ArcView, and calculate Flow Direction and Flow Accumulation using the Hydro Extension.

4.      Activate the Basin extension by selecting the filled DEM and clicking on the ‘Initiate’ button. Select the Flow Direction and Flow Accumulation grids as requested

5.      Define the Stream Network for the DEM by clicking on the ‘RIV’ button. Define the threshold for the number of upstream cells needed to define a stream channel. We typically repeat this process twice (or more) to obtain and save the stream network at different resolutions (e.g., using 50 and 250 upstream cells). Very similar results seem to be produced from either the ‘Shreve’ or ‘Strahler’ options.

6.      Define the approximate drainage basin for each glacier by clicking on the ‘Basin’ button and then clicking over the main stream outlet from each glacier. This creates a separate shape file for each glacier.

7.      Modify each shape file in ArcView (using Theme/Start Editing, and then clicking on the ‘Vertex Edit’ button) to make sure that the basin outlines fit the observed topography and surface features on the imagery. The final shape file for a particular glacier basin covers all areas which feed ice/water to the glacier terminus (i.e., hydrological definition). Using this definition, all tributaries and surrounding rock covered areas are included in the outline and are considered as one glacier unit.

 

Deriving GLIMS database products from shape files

Once the shape file for a given glacier drainage basin has been defined, this shape file can be modified to provide the data products and information required for the GLIMS database:

1.      Thumbnail imagery: ENVI 3.5 can use a shape file as a ‘cookie cutter’ to cut out a particular part of an image:

a.      Use ‘File/Open Image File’ to load in the orthorectified Landsat 7 image files that you want to cut images from

b.      Use ‘File/Open Vector File/ArcView Shape File’ to open the shape file (the one with the .shp suffix) that you want to use as a cookie cutter. Select the correct projection (UTM), datum (North America 1983), units (meters) and zone (17N) for the input. Choose ‘Output Result to Memory’.

c.       The loaded shape files are shown in the ‘Available Vectors List’ – click on ‘Load Selected’ to overlay a shape file on an image. To use a shape file as a cookie cutter, it has to be converted to a region of interest (ROI) file. Do this by selecting ‘File/Export Layers to ROI’ in the Available Vectors List box. Select an image file that you want to associate with the ROI (i.e., select the image that you want to cut).

d.      Now that the ROI has been defined, select ‘Basic Tools/Subset Data via ROIs’ to cut the image using the selected shape file. Select ‘Yes’ to ‘Mask the pixels outside of ROI’ and select a value of –999 for the ‘Mask Background Value’. Save the Output Result to Memory (not to a file as ArcView won’t be able to read this!).

e.      View the subsetted image by loading it from the ‘Available Bands List’, and save it as a tif image using the ‘File/Save Image As/Image File’ menu. Remember to include the year the image was taken in the filename.

f.        The georeferenced tif image of each basin can now be loaded back into ArcView.

2.      Digital elevation information: the Map Calculator in ArcView can use a shape file as a ‘cookie cutter’ to chop out sections of a DEM:

a.      In ArcView, select the shape file that you want to use as the cookie cutter. Convert this shape file to a grid by using the command ‘Theme/Convert to Grid’, and select the location to save the grid (this can just be a temporary file). When prompted, set the output grid extent to the same as the input shape file, and set the grid cell size to the same as the DEM (typically 100 m). Select ‘Id’ as the field for cell values, and ’No’ to join feature attributes to grid.

b.      Now use ‘Analysis/Map Calculator’ to multiply the gridded shape file by the elevation file you want to subdivide. Since the values are 1 within the basin and missing data outside of the basin, only the elevation information from within the basin will be saved.

 

GLIMS file formats for data transfer

For a particular ice cap, several shapefiles will be produced which will contain all the information necessary to both populate the central GLIMS database at NSIDC, and to populate the University of Alberta ArcIMS GLIMS server. Note that each shapefile actually consists of three files with the same prefix but different suffixes: .shp contains points, lines or polygons; .dbf contains attributes; .shx is an index file. In addition, orthorectified ‘mugshots’ of the individual basins will also be included in the databases. These will be in tiff world format, and each basin will contain two files:

Glacier_id.tif – the raw imagery

Glacier_id.tfw – the tiff world file with the georeferencing information

 

Note that for a particular submission to the central GLIMS database for an ice cap, all the files for the ice cap should be zipped together and submitted as a single file.

Files will be available for download by end-users by following the hyperlinks included in the attribute files (which will be displayed in the map service in ArcIMS). One zipped file will be available for download for each basin, and will contain all the information available for that basin (i.e., glacier outline, rock outline, image description, raw imagery). See example at:

http://www.tnris.org/website/CHSProject/viewer.htm

 

The shapefiles which will be produced for each ice cap (in this example, Manson) are:

 

 

1. Manson_session.shp

-         holds information that pertains to the entire analysis session for an ice cap

-         geographic coverage = point in centre of analysis area

 

Attributes

Used by GLIMS?

Value

Numeric/

text?

Comments

Rc_id

Y

3

Numeric

Regional center id: 3 = Canada

Rc_name

N

University of Alberta

Text

Name of regional center

Analy_time

Y

mm/yyyy

Text

Time when analysis of ice cap completed

Data_src

Y

Landsat 7

Text

Name of primary data source

Data_url

N

http://icedata.eas.ualberta.ca/manson/manson_landsat7.zip

Text

URL of location where orthorectified imagery for entire ice cap (zipped) may be downloaded

Proc_desc

Y

Orthorectified

Text

Description of processing applied to data source (Landsat 7)

Anlst_surn

Y

Copland

Text

Surname of analyst

Anlst_givn

Y

Luke

Text

Given name(s) of analyst

3D_desc

Y

CDED DEM data

Text

Description of how 3D information was derived

2. Manson_glaciers.shp

-         holds information that pertains to the characteristics of individual glaciers

-         many fields will not be completed (Null) because the available options are too simplistic: a better indication of the glacier’s true characteristics is achieved by the end user viewing the included ‘mugshots’.

-         geographic coverage = point at the centre of each glacier basin

 

Attributes

Used by GLIMS?

Value

Numeric/

text?

Comments

ID

Y

GxxxxxxEyyyyyN

(Eg 79.894W,77.146N=

G259894E77146N)

Text

Based on geographic centre of glacier with 3dp. xxxxxx is longitude (000.000-359.999); yyyyy is latitude (00.000-90.000)*

Name

Y

Glacier name

Text

Name of glacier (if one exists)

Prim_class

Y

Number

N/A

Primary classification

Form

Y

Number

N/A

Form

Front_char

Y

Number

N/A

Frontal characteristics

Long_char

Y

Number

N/A

Longitudinal characteristics

Mass_src

Y

1

Numeric

Dominant mass source: 1=snow

Tongue_act

Y

Number

Numeric

Tongue activity. If surge-type: 6=possible surge; 7=known surge

Width_m

Y

Width

Numeric

Representative width in metres

Length_m

Y

Length

Numeric

Representative length in metres

Area_km2

Y

Area

Numeric

Area of basin (i.e., glacier + rocks) in km2

Abarea_km2

Y

Null

N/A

Size of ablation area: not used

Speed_myr

Y

Null

N/A

Surface velocity: not used – information included in vec_sets if available

Snwln_elev

Y

Elevation

Numeric

Elevation of snowline in imagery (i.e., not equivalent to ELA) if available

Wgms_id

Y

WGMS code

Numeric

WGMS name of glacier (if available)

Local_id

Y

Canadian Glacier code

Numeric

Code from Canadian Glacier Inventory, if available and different from wgms_id

Parent_id

Y

Null

N/A

Parent id: not used

Image_id1

Y

Id of imagery

Numeric

Landsat 7 image id used to define this glacier: full details provided in ‘image’ shapefile

Image_id2

Y

Id of imagery

Numeric

As above, if >1 image used for a basin

 

* to determine centrepoint of a shapefile (glacier basin) in ArcView:

-         load shapefile into ArcView

-         select information button () and click on shapefile to bring up attribute table

-         go to tables, and select the one you want to find the centre of

-         open the table, and select (from the menu) Table / Start Editing

-         go to Edit / Add Field to add a new number field with 3dp with name xcenter or ycenter

-         select Field/Calculate and use the following statements to find the center:

[shape].ReturnCenter.Getx

[shape].ReturnCenter.Gety

-         to add a point to the view with these coordinates, select Table / Stop Editing, and then select the view and go to View / Add Event Theme

 

3. Manson_basins.shp

-         contains the polygons (segments) of the outlines of the glacier basins

-         uncertainty in locations cannot be determined in our processing methods

-         features to left and right of segment only valid when lines are used (not polygons)

 

Attributes

Used by GLIMS?

Value

Numeric/

text?

Comments

Category

Y

Basin_outline; or

Rock_outline; or Ice_outline

Text

Description of the feature that the polygon encloses

ID

Y

GxxxxxxEyyyyyN

Text

Glacier id that polygon relates to

Type

Y

Measured

Text

Whether feature is measured or arbitrary: in our case, all features are measured

Label

Y

Null

N/A

Label: not used

Loc_unc_x

Y

Null

N/A

Local location uncertainty in x: not used

Loc_unc_y

Y

Null

N/A

Local location uncertainty in y: not used

Glob_unc_x

Y

Null

N/A

Global location uncertainty in x: not used

Glob_unc_y

Y

Null

N/A

Global location uncertainty in y: not used

Left_mat

Y

Null

N/A

Material on left: not used

Right_mat

Y

Null

N/A

Material on right: not used

Left_feat

Y

Null

N/A

Feature on left: not used

Right_feat

Y

Null

N/A

Feature on right: not used

 

 

 

4. Manson_rocks.shp

-         same format as the manson_basins shapefile, but for the exposed rock areas within basins

-         placed in separate shapefile so that the rock outlines can be viewed independently of the basin and ice outlines

 

 

5. Manson_ice.shp

-         same format as the manson_basins and manson_rocks shapefiles, but for the ice outline within a basin

-         created by subtracting the manson_rocks polygon from the manson_basins polygon with the ‘Map Calculator’ in ArcView

 

 

NB: a ‘filelist’ file for the _basins, _rocks and _ice shapefiles has to be included in the submission to the central GLIMS database as these shapefiles fall into the same ‘segments’ category. The file must be named ‘segments.list’, and would contain the lines:

manson_basins

manson_rocks

manson_ice

 

 

6. Manson_images.shp

-         contains information about the images used in the analysis of the ice cap

-         relates to image id(s) given in manson_glaciers.shp

 

Attributes

Used by GLIMS?

Value

Numeric/

text?

Comments

Image_id

Y

Id of imagery

Numeric

Landsat 7 image id

Inst_id

Y

Landsat

Text

Spacecraft name

Inst_name

Y

ETM+

Text

Instrument name

Orig_id

Y

Null

N/A

Original id: same as image id

Imglocurl

Y

http://edcsns17.cr.usgs.gov/EarthExplorer/

Text

URL of image source

Acq_time

Y

mmddyyyy, hh:mm:ss

Text

Time of image acquisition

Imgctrlon

Y

xxx.xxx W

Text

Longitude of image centre

Imgctrlat

Y

xx.xxx N

Text

Latitude of image centre

Imglon_unc

Y

Null

N/A

Uncertainty of image centre: not known

Imglat_unc

Y

Null

N/A

Uncertainty of image centre: not known

Path

N

xxx

Numeric

WRS Path #

Row

N

xxx

Numeric

WRS Row #

Image_azim

Y

Null

N/A

Image azimuth: not known

Cloud_pct

Y

xx

Numeric

% cloud cover in image

Sun_azim

Y

xxx.xxx

Numeric

Solar azimuth (degrees east of North)

Sun_elev

Y

xx.xxx

Numeric

Solar elevation (degrees)

Inst_zen

Y

Null

N/A

Instrument zenith: not known

Inst_azim

Y

Null

N/A

Instrument azimuth: not known

Projection

Y

UTM 17N, NAD83

Text

Name of projection