SNOW SEASON FRACTIONAL FLOW 6 Understanding the sensitivity of Eurasian Arctic runoff changes to snow cover-related surface energy flux variations Amanda Tan 1, Jennifer Adam 2, Mark Serreze 3, Dennis P. Lettenmaier 1 1.Department of Civil and Environmental Engineering, Box , University of Washington, Seattle 2.Department of Civil and Environmental Engineering, Washington State University, Pullman, WA,Department of Civil and Enviromental Engineering 3.National Snow and Ice Data Center, University of Colorado, Boulder, CO, National Snow and Ice Data Center, 449 UCB University of Colorado Boulder Pronounced land surface process changes have occurred in the Arctic and sub-Arctic in recent decades. Satellite data have confirmed that average snow cover has decreased, especially in the spring and summer. One implication of the ablation of snow cover earlier in spring is the potential for a positive snow albedo feedback, where patchy snowcover leads to reduced effective albedo, increasing net radiation, and thereby making energy available for advection from snow free to snow covered areas, further accelerating melt. Furthermore, some of this increased surface energy is available to melt permafrost. Changes in timing of snowmelt and permafrost melt have been linked to increased riverine discharge to the Arctic Ocean which has the potential to affect the thermohaline circulation and in turn the formation of North Atlantic Deep Water and the northward-flowing Gulf Stream. To date, however, causality for observed hydrologic trends and changes in climatic forcings to the land surface system remain elusive, primarily because linkages between hydrologic and climatic sensitivities are not well established. This study explores the role of snow on the sensitivities of annual and seasonal Arctic river discharge through analysis of in- situ and satellite-based observations and a large-scale hydrologic model applied over Lena River basin in the Eurasian Arctic. Through analysis of satellite-based snow cover data for the period of 1972 — 2001 and observed river discharge for 1958 — 1999, we attempt to draw associative relationships between changes in snow cover over time and their effect on runoff timing. Streamflow gauges chosen for analysis are relatively uninfluenced by anthropogenic changes and we find that while average annual snow cover has decreased slightly, this apparently does not account for observed discharge trends – winter discharge increase in 75% of gauges, and June discharge decrease for 82% of the gauges. 1 ABSTRACT CHANGES IN STREAMFLOW TIMING 5 2 METHODOLOGY There are 417 streamflow gauges in the Lena basin of which only 28 had full-length, uninterrupted records for 1958 – The 28 gauges in the Lena were analyzed over the same 1958 – 1999 period of record. The 28 stations were selected to be relatively uninfluenced by land-use changes and anthropogenic factors – including effects from the Vilyui reservoir. Since only monthly streamflow data were available, the Variable Infiltration Capacity (VIC) hydrologic model was used to generate daily streamflow through use of the model’s ration of daily to monthly streamflow, which was applied to the observed monthly streamflows to produce reconstructed daily discharge time series for each gauge, which are constrained to the sum of monthly observed totals. Forcings for VIC: Temperature - Wilmott & Matsura, 2005; Precipitation - Adam et al., 2007; Groisman, 2005 Daily streamflow was analyzed for three measures outlined in Stewart et al., 2005 (a)Spring Pulse Onset : The first day of spring or snowmelt season (b)Centroid of Timing (CT): Center mass of annual flow (c)Fractional flows: Ratio of monthly or seasonal flow to the annual flow Each measure was tested for linear trends and significance using Seasonal Mann Kendall test. Correlation analysis between snow cover and spring pulse onset and snowmelt was conducted using the Log-Pearson method and Kendall’s Tau method. Correlation analyses between different months and interseasonal flow was also conducted. streamflow for the snowmelt period of May and June show an increase in May monthly fractional flows at 79% of the gauges, whereas June discharge is decreasing for 82% of the gauges. Significant trends are detected on the higher latitudes and towards the eastern part of the basin. This is the Aldan basin which is underlain by 90% continuous permafrost (Brown et al., 2000). Summer flow is also decreasing with some gauges registering as much as a 20% decrease in discharge. The spring pulse onset and snowmelt season flow is correlated at 65% of gauges. The snowmelt season fractional flow is decreasing in 86% of the gauges with some of the highest numbers reaching 30% decreases, particularly in the eastern basin of Aldan at Lena (which is underlain by continuous permafrost). WINTER SEASON FRACTIONAL FLOW of the gauges for all of the winter months and the winter seasonal flow as a whole. Trends are statistically significant at roughly half of the gauges. The correlation between spring pulse onset and winter discharge is significant at 30% of the gauges. A recent study of the Arctic and particularly the Lena basin have shown that winter temperatures are increasing (Yang et al, 2002). Furthermore, the cold and threshold basins generally have positive correlation coefficients, indicating that as temperature increases, streamflow also increases, possibly due to the melting of ground ice (Adam et al.,2007). (a) (d)(c) (b) (a) December (b) January (c) February (d) Seasonal Fractional Flow during Winter (Dec, Jan, Feb) (d) (c) (b)(a) 7 Spring Pulse Onset (Fig a): 0 – 16 day shifts towards earlier dates in 61% of stations Trends occur largely towards eastern part of basin Due to interannual variability, no trends are significant at p = 0.10 Stations with increasing trend are linked largely to increased snow cover (see Section 4). Centroid of Timing (Fig b): Equal shifts in earlier and later trends, no coherent trends CT is significantly correlated with spring pulse onset for 36% of gauges A shift of CT later in the year may indicate that the melt season is being prolonged while the maximum peak discharge is decreasing. Summary There is modest evidence that snowmelt is shifting earlier in the water year (about 60% of the gauges) Winter season fractional flow is increasing in 90 % of the gauges in the basin Summer season flow is decreasing at most stations The most significant changes occur during the months of May, July and all through winter Snow cover disappearance influences spring pulse timing, but the linkage is not as strong as might be expected. (a) May (b) June (c) July (d) Seasonal Fractional Flow during Snowmelt (May, June, July) 7 THE DOMAIN 3 The Lena Basin The Eurasian Arctic region was chosen because the most significant increases in river discharge to the Arctic Ocean have been observed there (Peterson et al. 2002). We chose the Lena River basin because 80% of its drainage area is underlain by permafrost (Brown et al.,1998; Zhang et al., 1999) and is therefore more sensitive to changes in temperature and precipitation. The Lena basin outline based on drainage characteristics is shown in green, whereas the R-ArcticNet definition of the Lena basin is shown in black lines. Some of the stations selected therefore fall outside of, but are close to the Lena basin boundary, because of inconsistencies in the two basin definitions. −Adam, J.C., and D.P. Lettenmaier, 2008: Application of new precipitation and reconstructed streamflow products to streamflow trend attribution in Northern Eurasia J. Climate 21(8): −Stewart I., D. R. Cayan, and M. Dettinger, 2005: Changes toward earlier streamflow timing across western North America. J. Climate, 18, 1136– SNOW COVER CHANGES Figure 4(a-f) show anomalies relative to the long-term mean over the entire basin (i.e. in space and time). The average snow cover disappearance day starts on Day 99 of the calendar year (April 1), snow cover onset is Day 228 (August 15). Snow cover disappearance dates correlate significantly (r = 0.1 – 0.5) with spring pulse onset dates at 36% of the gauges. (a) (f)(e) (d) (c)(b) (a) The combined seasonal fractional flow for the winter season is taken as December, January and February. The trends show an increase in 75% Snowmelt occurs during May and June in the Arctic, with peak discharge usually occurring in mid-June. Analysis of fractional 5-year anomalies of number of snow cover days over the Lena basin : (a) 1972 – 1976 (b) 1977 – 1981 (c) 1982 – 1986 (d) 1987 – 1991 (e) 1992 – 1996 (f) 1997 – 2001.