Retrogressive thaw slumps (RTSs) are thermokarst landforms that develop on slopes in permafrost regions when thawing permafrost causes the land surface to collapse. RTSs are an indicator of climate change and pose a threat to human infrastructure and ecosystems in the affected areas. As the availability of ready-to-use high-resolution satellite imagery increases, automatic RTS mapping is being explored with deep learning methods. We employed a pre-trained Mask-RCNN model to automatically map RTSs on Banks Island and Victoria Island in the western Canadian Arctic, where there is extensive RTS activity. We tested the model with different settings, including image band combinations, backbones, and backbone trainable layers, and performed hyper-parameter tuning and determined the optimal learning rate, momentum, and decay rate for each of the model settings. Our final model successfully mapped most of the RTSs in our test sites, with F1 scores ranging from 0.61 to 0.79. Our study demonstrates that transfer learning from a pre-trained Mask-RCNN model is an effective approach that has the potential to be applied for RTS mapping across the Canadian Arctic.

This thesis applies a novel method of change detection, the Landsat Image Stack Trend Analysis method to the monitoring of retrogressive thaw slumps in the Richardson Mountains, NWT. This method represents a significant improvement upon previous methods of thaw slump monitoring, which utilized air photos and high-resolution satellite imagery. This method applies Tasseled Cap brightness, wetness and greenness indices to Landsat TM/ETM images acquired between 1985 and 2011 and analyzes the temporal change of each pixel for the different indices values. This method is useful in retrogressive thaw slump monitoring in two ways. First, by creating a map showing the linear change over time from 1985 to 2011, retrogressive thaw slumps can be easily identified, as they are more dynamic than the surrounding tundra. In total, 251 thaw slumps were identified within an area of roughly 18 000km2. Second, thaw slump activity, from initiation, growth and stabilization can be studied by plotting the annual vegetation index pixel values of adjacent pixels in a thaw slump. This method allows for the efficient extraction of annual thaw slump headwall retreat rates, provided the availability of cloud-free imagery. The retreat rates of 16 slumps were extracted, which were found to have an average annual retreat rate of 11.8 m yr-1.

This thesis applies a novel method of change detection, the Landsat Image Stack Trend Analysis method to the monitoring of retrogressive thaw slumps in the Richardson Mountains, NWT. This method represents a significant improvement upon previous methods of thaw slump monitoring, which utilized air photos and high-resolution satellite imagery. This method applies Tasseled Cap brightness, wetness and greenness indices to Landsat TM/ETM images acquired between 1985 and 2011 and analyzes the temporal change of each pixel for the different indices values. This method is useful in retrogressive thaw slump monitoring in two ways. First, by creating a map showing the linear change over time from 1985 to 2011, retrogressive thaw slumps can be easily identified, as they are more dynamic than the surrounding tundra. In total, 251 thaw slumps were identified within an area of roughly 18 000km2. Second, thaw slump activity, from initiation, growth and stabilization can be studied by plotting the annual vegetation index pixel values of adjacent pixels in a thaw slump. This method allows for the efficient extraction of annual thaw slump headwall retreat rates, provided the availability of cloud-free imagery. The retreat rates of 16 slumps were extracted, which were found to have an average annual retreat rate of 11.8 m yr-1.