Snow Cover Detection {Methodology Using Landsat}
NASA MODIS Land February 2004
This methodology is far from finished in the sense that it an be applied to locations all over the world with a couple simple changes in the tools and steps taken. A future goal would be to have a database set up which contains water body basemaps categorized by province or state or country, depending on the size. These can then quickly be applied to any Landsat image along with the Methodology for snow cover detection.
Future steps for this method may be to apply it to areas of landscape terrain other than mountains to assess its accuracy and usefulness. This may require adjusting the particular threshold values and masking parameters in order to create a set of models for regions based on landscape type.
A higher accuracy of snow cover awareness can be useful for many fields of research:
Potential agricultural land in the higher latitudes is a new area of research which is interested in quantifying the future land available for agriculture mostly as a result of climate change resulting in increasing temperatures and a decrease in permafrost. With global populations increasing every year, there is recently a greater interest in this field or research.
This methodology can be used for local flood forecasting and risk assessments due to snow melt. As shown in the Seymour Watershed example, snow cover results complimented by snow depth assessments will give a estimation of spring snow melt and therefore river discharge volumes and possible floods.
Future of the Method
Limitations of the Method
Water bodies - The limitations of the method include the requirement of having a shapefile of waterbodies for the area under study, if water is present. For British Columbia this data was readily available, however this may not be the case in other areas of the World. The final result is also limited to the accuracy of this water body shape file, however considering the current MODIS snow area detection the error of this in the study image here is negligible.
Cloud Cover - Snow cover detection is limited to areas with thin or no cloud cover and as Landsat only takes images every 16 days of a specific area, this can be a limiting factor to the use of this method.
Terrain - This method was developed in mountainous terrain in British Columbia and the use of thresholds in the Hillshade and Band 1 filters may vary slightly based on landscape terrain. It would require a visual validation early on in snow detection in order to assure accurate results in future analysis.
Data size - At 30 meters spatial resolution Landsat images are very large in size, requiring a lot of computer storage and processing. This may be a limitation when large areas are being studied and the length of time spent on the analyzation process will increase. To avoid this it is important to only work with images of the bands being used in the method and to avoid importing the other bands.