The Snow Darkening Effect and the Simulation of Extremes over Eurasia

We have recently completed an updated ensemble of NASA GEOS-5 simulations with a snow-darkening module (now officially named GOddard SnoW Impurity Module, or GOSWIM, and summarized in the published paper by Yasunari et al., SOLA, 2014; see at: https://www.jstage.jst.go.jp/article/sola/10/0/10_2014-011/_article). This ensemble ("snow-darkening case (SDC)"), consisting of ten parallel simulations (differing only in their initial conditions) spanning 2002-2011, is compared here to a corresponding ensemble with all snow-darkening effects disabled ("non-SDC"). We focus particularly on the production of extremes associated with snow darkening. To identify regions of interest over Eurasia, we first rank the 100 separate spring (MAM) or summer (JJA) values of a given quantity in each combined 100-yr data (i.e., 10-yr x 10-ensemble), and then compute the differences of the 90th percentile values between SDC and non-SDC. For spring, large differences are seen in a specific area of Europe and Central Asia (ECA), and for summer, they are seen for an area in the Russian Arctic (RA). The next step in our analysis addresses the month-by-month variation of the percentile differences within these identified regions - for each month, and for a given meteorological or hydrological variable, we determined the SDC percentile that corresponds to the 90th percentile value found for the non-SDC ensemble. For example, in the RA domain, the surface air temperature corresponding to the 90th percentile in the non-SDC ensemble has a consistently lower percentile in the SDC data - not only during spring and summer through the increased absorption of radiation by snow polluted with dust, black carbon, and organic carbon, but also in the post-snow season through some form of memory in the system. The temperature extremes in the SDC ensemble thus exceed those of the non-SDC ensemble throughout the year. This analysis supports the idea that the consideration of snow darkening effect in global models is important for simulating key aspects of climate variability and extremes, even during snow-free seasons.