Long-duration, consistent records of snow are rare owing to many challenges in making accurate and representative measurements. Although weather stations in snowy inhabited areas often report snow depth, records of snowfall are often patchy or use techniques that change over time (e.g., Kunkel et al., 2007). The density of stations and the choice of metric also varies considerably from country to country. The longest satellite-based record of SCE is the visible-wavelength weekly product of the National Oceanic and Atmospheric Administration (NOAA) dating to 1966 (Robinson et al., 1993), but this covers only the NH. Satellite mapping of snow depth and SWE has lower accuracy than SCE, especially in mountainous and heavily forested areas. Measurement challenges are particularly acute in the Southern Hemisphere (SH), where only about 11 long-duration in situ records continue to recent times: seven in the central Andes and four in southeast Australia. Owing to concerns about quality and duration, global satellite microwave retrievals of SWE are of less use in the data-rich NH than in the data-poor SH.
By blending in situ and satellite records, Brown and Robinson (2011) have updated a key indicator of climate change, namely the time series of NH SCE (
Figure 4.19). This time series shows significant reductions over the past 90 years with most of the reductions occurring in the 1980s, and is an improvement over that presented in AR4 in several ways, not least because the uncertainty estimates are explicitly derived through the statistical analysis of multiple data sets, which leads to very high confidence. Snow cover decreases are largest in spring (Table 4.7), and the rate of decrease increases with latitude in response to larger albedo feedbacks (Déry and Brown, 2007). Averaged March and April NH SCE decreased 0.8% [0.5 to 1.1%] per decade over the 1922–2012 period, 1.6% [0.8 to 2.4%] per decade over the 1967–2012 period, and 2.2% [1.1 to 3.4%] per decade over the 1979–2012 period. In a new development since AR4, both absolute and relative losses in June SCE now exceed the losses in March–April SCE: 11.7% [8.8 to 14.6% per decade or 53% [40 to 66%] total over the 1967–2012 period and 14.8% [10.3 to 19.3%] per decade over the 1979–2012 period (all ranges very likely). Note that these percentages differ from those given by Brown and Robinson (2011) which were calculated relative to the mean over the 1979–2000 period, rather than relative to the start-ing point. The loss rate of June SCE exceeds the loss rate for Coupled Model Intercomparison Project Phase 5 (CMIP5) model projections of June SCE and also exceeds the well-known loss of September sea ice extent (Derksen and Brown, 2012). Viewed another way, the NOAA SCE data indicate that, owing to earlier spring snowmelt, the duration of the snow season averaged over NH grid points declined by 5.3 days per decade since winter 1972–1973 (Choi et al., 2010).
Over Eurasia, in situ data show significant increases in winter snow accumulation but a shorter snowmelt season (Bulygina et al., 2009). From analysis of passive microwave satellite data since 1979, signif-icant trends toward a shortening of the snowmelt season have been identified over much of Eurasia (Takala et al., 2009) and the pan-Arctic region (Tedesco et al., 2009), with a trend toward earlier melt of about 5 days per decade for the beginning of the melt season, and a trend of about 10 days per decade later for the end of the melt season.
[...]
AR4 stimulated a review paper (Brown and Mote, 2009) that synthesized modelling results as well as observations from many countries. They showed that decreases in various metrics of snow are most likely to be observed in spring and at locations where air temperatures are close to the freezing point, because changes in air temperature there are most effective at reducing snow accumulation, increasing snowmelt, or both. However, unraveling the competing effects of rising temperatures and changing precipitation remains an important challenge in understanding and interpreting observed changes.