EcoInformatics International

EcoInformatics International

Home Services Contact

Monitoring Permafrost with Satellites


Melting of Peat Plateaus and Palsas in the Subarctic Hudson Bay Lowlands, near Churchill, Manitoba


Summary: High Resolution Google Earth images (2004-6) are compared with 1972-3 high resolution airborne colour and colour infrared images to assess melting of permafrost in wetlands.

Figure 1:

A typical example of the type of permafrost landforms in wetlands is provided by the photograph below taken from a Beaver aircraft in 1972. Area A shows a collapse scar where permafrost has melted. The small black spruce trees are slowly being submersed in the waterlogged scar. The area around B is typical of the non-frozen wetland. Sparse growth of Tamarack (Larix Laricina)-light green tree- can be seen here in a poorly developed ribbed fen. Drainage is from right to left. Around D is a typical growth of stunted Black Spruce (Picea Mariana) can be seen. Light coloured ground cover between the spruce is a combination of lichen and mosses.

Figure 2:

The same area is show in a high resolution Google Earth image 'taken' at about 140 meters above ground level. Date of image 2003- Aug 27.

There is a significant 31 year time lapse between the two images but initially no significant differences are apparent. It is interesting to look at these two images combined on Google Earth: try this KMZ link

Since 2003 a number of high resolution satellite images are available through Google Earth, Appland Bing Maps. Even some winter images were availabel for the 2013-15 year period (Bing and Apple). The image below from Digital Globe was available through Bing Maps a few years ago. The small peat plateau (purple arrow) is distictly visible. It is interesting to note that the most stable peat plateau edges have clear dark 'lines' of black spruce - see blue arrows, while melting or unstable edges lack this dark 'tree line'- yellow arrows. The peat plateau with the purple arrow shows cclearly the dark mostly stable edge of black spruce, with in the center a melting area, with spruce trees sinking in the melting 'scar'. The time difference between the images above, show that this melting was relatively slow between 1972 and 2003.

  1. It can be concluded that high resolution satellite winter images are ideally suited to 'map' active melting of peat plateaus in Manitoba's northern wetlands. This would be applicable to all ecodistricts and landscapes where peat plateaus and palsa are covered by semi-open black spruce tree cover. [+]
  2. peat plateaus and palsas melting

  3. The rate of melting can be aproximated by using the unique environmantal baseline of aerial images available through Canada's National and Provincial Airphoto Libraries. To demonstrate this we will look at a series of aiphotos available for this study area dateing back as far as 1930, providing us with an 85 year plus time frame.

The 2013 Digital Globe winter image (right) compared with the 1930 Airborne photo taken on August 1 (left) shows clearly the significant increase of the melting scars (m) in the two permafrost peat plateaus. The maximum width (M1) is 55 M, the maximum length 79 M. On the 1930 photo the melting scar is about 12M by 27 M. Based on these measurements, melting edges of this peat plateau have receded 2.5- 0.3 meters per year. Such a small amount is difficult to measure and monitor on an annual basis with satellites (Figure 4). Looking at Figure 1 above, it appears that most of the melting occurred before 1972. Using the melting rate of the collapse scar in reverse, we can approximate time the melting of this peat plateau started. Our backcast is that melting started in the centre of this peatplateau in around 1880.

The images below also show a number of areas, marked with blue dotted lines (and blue arrows) where tree cover has expanded into the surrounding wetland, implying aggradation of permafrost. The extended peat plateaus on the right shows a continuing process of coalescing of three smaller plateaus into 'one' probaly the result of the expansion of this permafrost under tree cover.

Figure 3



Figure 4:

This figure shows a time interval of 10 years. The right image was taken on 2003-08-27, the left one on 2013-09-13. Together they form a stereopair which can be sudied in 3 d with a stereoscope. While the distinct impression exsist that the melting scar has increased, it seems impossible to measure using these two images. The steromodel shows changes but tree shadows and surface water (2003) make it difficult to compare actual edges. Using the melting rate estabished by comparison of the 1930 and 2013 images above, the edges should have receded by 2.5-3 meters over this 10 year period.



The 1973 summer and winter Landsat images give a good perspective of the lack of tree cover in this area. The winter image on the right enhances the visibility of the very limited and sparse tree cover in this area. The peat plateau of the slide and the GE image is located in the top part of the white circle. The high red on the summer (false colour ) image shows an abundance of stunted tamarack, black spruce, dwarf birch, willow and alder on peat plateaus fens and stream beds. the more yellow tones around A show sparsely covered or bare peat plateaus. Bare peat polygons or polygonal peat plateaus are common around B. D shows some of the former strand lines, again enhanced by larger trees and denser coverage. Around C treed peat plateaus are predominant.


1973 Landsat winter summer image