Landsat snow

Another way to look at snow cover is by satellite remote sensing. The longest and best-known series by far is that from the US Landsat satellites, now up to Landsat 8, launched in early 2013. Thanks to US government generosity and improved cloud computing, historical Landsat data has become much more accessible in recent years. For this effort I’m using processed false colour imagery from the US Geological Survey’s LandsatLook viewer. Focusing on the New South Wales snowfields¹ and taking the best available image mosaic for each year near the usual time of the season peak² leads to an interesting animation beginning way back to 1977:

 

Snow extent 1977 - 2014, New South Wales snowfields, Australia

Near-peak snow cover extent 1977 – 2014, New South Wales snowfields, Australia

The decline in cover over the years is clear enough if you look carefully, but perhaps less obvious than that in the peak depth or season integral depth data from ground measurements. The areal extent of cover near the peak of the snow season is only one part of the story. Snowpack decline is much more marked during the melt leg of the season (season shortening):

Spencers Creek snow depths by decade

Spencers Creek snow depths by decade

 

Notes:

1. The Australian Alps, southern New South Wales, Australia. Perisher Valley resort is near the centre of the image. Thredbo is at the head of the deep, SW-NE trending valley immediately south of the main snow covered area. (Small patches of artificial snow near the valley floor mark its location in later images.)

2. Given the vagaries of winter cloud cover and of satellite / instrument availability over the years, it’s not practical to make a highly focused choice of image date to display for each year. I’ve just chosen a mosaic that (mostly) gives a decent view, somewhere around the average time of the peak depth in late August (which may not, of course, coincide with peak snow cover extent).

3. The date shown appears to be for the middle of the mosaic; other parts may be older.  [For reasons I don’t understand (viewer defects or viewer update lag?), some parts of some mosaics appear to be much older.]

4. The multiple diagonal black lines across some later images are due to the failure in May 2003 of a mechanical image scan correction device on Landsat 7.  Despite the scanning defect, LandsatLook appears to favour the higher resolution Landsat 7 ETM imagery over Landsat 5 TM imagery, which continued to be available until November 2011.  There are products that attempt to remove the scan gaps by interpolation, but LandsatLook does not offer them.