1. Jakobshavn Isbrae Glacial Retreat
Darin Erickson
Aaron Gaither

2. Project Goals Time lapse imagery showing glacial transformation
Area change
Retreat distance
Link glacial retreat with increasing temperatures
Correlate temperature with glacial retreat
Provide information for future analysis
Comparison to other glacial studies

3. Jakobshavn Glacier Location Characteristics Area Studied
West Greenland
Near the city of Ilulissat
Coordinates: ° N, ° W
Characteristics
Drains 6.5% of Greenland’s ice sheet
Produces 10% of Greenland's icebergs
One of the worlds fastest moving glaciers
Area Studied
Main Glacier
Northern Outlet
Image
Information

4. Data/Acquisition 5 Landsat Images from Landsat 4,5, 7 and 8
Less than 20% cloud cover
Acquisition from:
USGS Earth Explorer
Path 9 – Row 11
Temperature data from ILULISSAT (Neighboring town to Jakobshavn glacier)
Image ID
Landsat #
Image Acquisition
# of Bands
LT KIS00
Landsat 4-5 TM
July 3rd, 1985 7
LT KIS00
June 15th, 1990
LE EDC00
Landsat 7 SLC-on
July 7th, 2001 9
LE EDC00
Landsat 7 SLC - off
July 29th, 2009
LC LGN00
Landsat 8 OLI
July 3rd, 2014 12

5. Software Used Erdas Imagine ArcGIS 10.2 Google Earth

6. Preprocessing Layer Stack
Set Spectral Bands for Images (Landsat 7)
Red 4
Green 3
Blue 2
Set Spectral Bands for 2014 Images (Landsat 8)
Red 5
Green 4
Blue 3
Subset
No Rectification needed due to use of Landsat images
Layer Stacking- We then selected each individual Tiff file to be stacked in ascending order. After completing these steps for each of the five images, we were able to clearly view our area of interest.
All Images had to undergo these steps
No Rectification needed due to use of Landsat images because they were all from the same path 9 and row 11

7. Subset Images
Cut down our study area size

8. Processing/Classification
Initially used unsupervised classification
7 classes, 50 max iterations, .98 convergence threshold to identify cover types
Snow
Open Water
Cloud Cover
Fjord Ice
Glacier
Vegetative cover
Non-Vegetative Cover
Classification Failure
Why we tried unsupervised- It was a quick and easy algorithm to see if we could identify all cover types accurately
Failed because it did not successfully identify the cover types to the precision that we wanted.
So we tried supervised

9. Processing/Classification
Supervised Classification
7 classes
Snow
Open Water
Cloud Cover *
Fjord Ice
Glacier
Vegetative cover
Non-Vegetative Cover
Created 10 polygons within each cover type (To identify our training data)
Classification Success
Cloud cover was only classified in one image so we didn’t feel it was necessary to do a radiometric rectification of this image to eliminate the cloud cover because it did not cover our study area and was not significant.
NON Parametric decision rule
Considers range of DN values
Two dimensional polygon is drawn around training pixels
Advantages- Fast, simple consider variance
Picture on next slide

10. Supervised Classification
Snow was not classified because the DN values were so similar to the Fjord so they were just classifies as fjord.

11. Accuracy Assessment Search Count 1024
50 point of stratified random distribution
2014 Accuracy Assessment
By choosing stratified random, we were able to have an equal distribution of points across the image while still maintaining randomization within each classified pixel value
Aarons Slide

12. Basic Digitizing Digitized glacial edge across entire study area
Shape files overlaid on 2014 subset image
Distance measurements
Area Measurements
We computed values of change relative to 1985 and change relative to the previous study year.
Picture on next slide

13. Darin Ends HERE

14. Measurement Relative to Previous Study Year
Main Glacier
North Outlet
Year
Retreat Distance (km)
Change in Area (km2)
-1.15
-7.88
-0.745
-1.64
2.57
22.98
0.957
2.11
12.52
144.35
2.05
6.19
2.27
31.6
0.453
2.29
Explain the Negative for glacial advance
Explain the positive for glacial retreat
We noticed a significant retreat distance and change in area in years 2001 – We calculated 75% of the total glacial retreat occurred between the years of 2001 – 2009.
75% of the total glacial retreat occurred between the years of 2001 – 2009

15. Measurements Relative To 1985
Main Glacier
North Outlet
Year since 1985
Retreat Distance since 1985 (km)
Area lost since 1985 (km2)
Distance since 1985 (km )
Area since 1985 (km2) 5
-1.15
-7.88
-0.75
-1.64 16
1.42
15.1
0.20
0.47 24
13.94
159.45
2.25
6.66 29
16.20
191.05
2.70
8.95
Total
Average retreat distance of 0.56km/yr
Total retreat distance 16.2 km
Area lost 6.6km2/yr
Total of 8.95 km2

16. This helps to visualize the 75% of the total glacial retreat occurred between the years of 2001 – 2009

17. Temperature Comparison
We noticed a slight increase in average yearly temperature from 1985 – While an increase of only 1-2 deg. Celsius may not
appear to be significant, it warrants a correlation between glacial retreat and rising temperature. We also noticed a substantial
increase in temperature of 9.1 deg. Celsius between the years 2003 – This is also the time frame of the major glacial retreat.

18. Discussion Points Glacial advance between 1985-1990
Glacial retreat
Major glacial retreat during
Correlated with increase in temperature
Several factors we didn’t acquire data for
Several factors we didn’t acquire data for
-sea temperatures
-precipitation amounts
-sediment flux and erosion
-previous data before 1985

19. Further Study Data collection of surrounding Labrador Sea temperature
Change detection of vegetation and exposed land area
Comparison to similar glaciers
Fjord ice and open water distribution
Further understanding of historical measurements prior to 1985
With more research and data collection of the area, we might be able to present a stronger correlation between temperature and glacial retreat.

20. Questions