Can We Trust Snow Extent Data?
By Paul Homewood
h/t Billy Liar/Ben Vorlich
According to the Rutgers University Global Snow Lab, snow extent across the Northern Hemisphere in autumn and winter has been on growing trend since records began in 1967.
Yet, snow cover in spring is showing a stark decline.
This decline is stronger in Eurasia than in North America.
There may be logical reasons for this phenomenon, though I have not seen any convincing explanation.
One is entitled to ask how relevant these statistics are, given they start in 1967, coinciding with a much cooler period in the NH. As with a lot of these things, we had no satellites prior to the 1960s to tell us what conditions were like back in, say, the 1930s and 40s.
Nevertheless, my attention was brought to a paper by Libo Wang et al, which was published in 2005 and entitled “Evaluation of spring snow covered area depletion in the Canadian Arctic from NOAA snow charts”.
This is the abstract:
The National Oceanic and Atmospheric Administration (NOAA) weekly snow cover dataset (1966–) is the longest available record of snow cover extent (SCE) over the Northern Hemisphere (NH). This dataset has been used extensively to derive trends in continental SCE and in climate-related studies, but it has received only limited validation, particularly in high latitude areas of the NH. This study evaluated spring snow cover depletion in the NOAA dataset over a study area in the Canadian Arctic mainland north of the tree line. The evaluation used four sources of information: (1) surface snow depth and snow survey observations, (2) snow cover extent produced from the Advanced Very High Resolution Radiometer (AVHRR), (3) snow cover extent derived from Special Sensor Microwave/Imager (SSM/I), and (4) Landsat 5 TM browse images. Six spring seasons from the period 1981–2000 with low (1984, 1988, and 1998) and high (1985, 1995, and 1997) spring snow cover extent were evaluated. The evaluation revealed that the NOAA weekly dataset consistently overestimated snow cover extent during the spring melt period, with delays of up to 4 weeks in melt onset. A number of possible reasons for this delay were investigated. The most likely causes for the delayed melt onset were frequent cloud cover in the spring melt period, and the low frequency of data coverage over higher latitudes. The results suggest that caution should be exercised when using this dataset in any studies related to the timing of snowmelt in the high latitudes of the Northern Hemisphere.
Although the paper talks about snow extent being overestimated, remember that the years examined were all in the 1980s and 90s, and that the paper was itself published in 2005. Therefore, the findings don’t relate to measurements in the last 19 years.
Do we still have the same measurement problems now, or has satellite technology made it more accurate? We don’t know, but, as the paper emphasises, we need to treat the data going back to 1967 with a large shovel of salt.
Spring, of course, is the time of year when it is most difficult to measure snow, as it is continually melting. So the figures for autumn and winter may be more reliable.