This study investigated how a February 20, 2025 Chinook event reshaped snow cover across the Alberta Foothills using terrain analysis, weather station data, and InSAR-derived digital surface models. By combining wind exposure modelling with DSM differencing, patterns of snow erosion and deposition were mapped and compared to underlying terrain conditions. Results showed strong terrain control over snow redistribution, with significantly more wind-driven transport than expected thermal ablation.
<aside> 🛰️
This project focuses on distinguishing wind-driven redistribution vs thermal ablation, a key challenge in snowpack interpretation using remote sensing.
Aeolian snow morphology occurs when snow is redistributed by wind, creating formations such as dunes, ridges, hummocks, and mass wasting features in cold, dry environments.
Chinook winds are warm, dry downslope winds originating from the Rocky Mountains that can rapidly raise temperatures while generating strong wind speeds across southern Alberta.
While Chinooks are often associated with rapid snowmelt, this project explored whether wind-driven redistribution might be a more significant driver of short-term snow surface change than ablation alone.
How did wind–terrain interactions during the February 20, 2025 Chinook event structure patterns of snow erosion, deposition, and net change?
What proportion of observed snow-surface volume change reflected redistribution versus ablation?
Does terrain roughness amplify the magnitude and spatial variability of snow-surface change?
Hypothesis
During the February 20, 2025 Chinook event, wind-driven snow redistribution will be evident with deposition in leeward terrains, with more erosion/deposition and heterogeneity found in areas with higher underlying terrain roughness.
The workflow combined atmospheric characterization, terrain modelling, and remote sensing-based surface change detection to identify relationships between wind exposure and snow redistribution patterns.
The Chinook event was identified through abrupt temperature increases and changes in prevailing wind direction using hourly weather station records from four locations across the study area.
Prior to the event, winds were generally weaker and northerly. During the Chinook, winds shifted to predominantly westerly directions and increased substantially in speed.
.png)
A digital terrain model was mosaicked from 30 m InSAR-derived elevation tiles and used to derive slope and aspect layers within Google Earth Engine.
Terrain variables were used to evaluate how underlying topography influenced snow redistribution patterns during the wind event.
Technical Challenge: Projection & Terrain Derivatives
Initial terrain derivatives failed due to projection inconsistencies introduced during DEM mosaicking. Terrain products produced invalid outputs, including zero-value slope layers and constant hillshade values. Reprojecting the DEM into EPSG:3978 at 30 m resolution resolved these artifacts and enabled terrain extraction.
Initial terrain derivatives failed due to projection inconsistencies introduced during DEM mosaicking. Terrain products produced invalid outputs, including zero-value slope layers and constant hillshade values. Reprojecting the DEM into EPSG:3978 at 30 m resolution resolved these artifacts and enabled terrain extraction.
Two InSAR-derived digital surface models were aligned and differenced to quantify snow-surface elevation change over the course of the event.
The DSMs represented surface elevation above base terrain, allowing snow redistribution patterns to be isolated through pre/post-event comparison.
Raster Alignment Challenges
DSM differencing proved highly sensitive to raster alignment and reprojection consistency. Early subtraction attempts produced edge artifacts and systematic offsets. Manual raster correction and alignment were required before reliable change detection could be performed.
Terrain was classified into windward and leeward surfaces using the median Chinook wind direction (~280°). This classification was then combined with slope angle to generate a slope-weighted wind exposure index.
This exposure model acted as a spatial proxy for aerodynamic forcing, allowing observed snow-surface changes to be interpreted relative to likely zones of erosion and deposition.