Climate%20change%20and%20the%20cryosphere:%20Part%20Deux

1
Climate change and the cryospherePart Deux
2
Recent Northern Hemisphere Snow Cover Extent
Trends

• Stephen Dery and Ross Brown

3
Background

• Northern Hemisphere snowcover extent (SCE) varies
  between 4-46 x 106 km2.
• Its distinct properties makes snow a key
  component of global climate.
• Snow responds to changes in surface air
  temperatures precipitation, thus providing
  another indicator of climate change.

4
Motivation Goals

• In light of near-record warmth in 2006 the
  recent changes observed in the cryosphere, there
  is an urgent need to better understand SCE
  trends.
• Objective To develop interpret weekly trends
  in Northern Hemisphere (NH), North American (NA)
  Eurasian (EU) SCE for the period 1972-2006.

5
Data Methods

• Weekly values of SCE from January 1972 to
  December 2006 from Rutgers University.
• Monotonic trends in weekly SCE assessed with
  Mann-Kendall test (MKT) over NH, NA (excluding
  Greenland) EU.
• MKT assumes a linear trend in the form
• S mt b (1)
• Where S is SCE, t is time (year) m is the slope
  of the linear trend given by

6
mk (Sj Si)/(tj ti), k 1, 2, ,
n(n-1)/2 i 1, , n-1 j 2, 3, , n
All of the slopes mk are then ranked, with the
median value representing the slope m of the
linear trend. The coefficient b is found by
substituting the median values of SCE time in
Eq. (1) solving for b.
7
Strength of the MKT
8
Strength of the MKT
9

• Trends expressed in absolute values ( 106 km2),
  as a from initial (1972) values, in
  standardized units, insolation-weighted
  anomalies.
• Time series of weekly SCE data (Si) are
  standardized by

SSi (Si Si)/si , (i 1-53)
Insolation-weighted anomalies are computed by
multiplying the absolute values of SCE by the
ratio of the average maximum weekly incoming
solar radiation at 60oN.
10
(No Transcript)
11
Data Issues

• Continental snowcovers exhibit temporal
  persistence.
• This implies positive autocorrelation of SCE
  values, meaning that time series of subsequent
  weekly SCE values do not form independent
  datasets.
• Thus methodologies must be developed to
  reduce/remove the effect of serial correlation on
  trend analyses.
• Trends correlations are considered
  statistically-significant when p lt 0.05.

12
Number of weeks with significant autocorrelations
of continental SCE
13
Year-to-year autocorrelation in continental SCE
14
(No Transcript)
15
(No Transcript)
16
Summary of Trend Analysis
Statistic N.H. N.A. Eurasia
Mean SCE (106 km2) 23.8 8.7 15.1
SCE Trend (106 km2)/35 years -1.28 -0.78 -0.48
Positive significant trends 2 0 4
Negative significant trends 24 23 20
Source Déry and Brown (2007), GRL.
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
Coherent variability signal

• Correlation between standardized NA EU weekly
  SCE is r 0.41 (p lt 0.001).
• Standardized weekly SCE are of the same sign 64
  of the time (88 when greater than 1 standard
  deviation).
• Correlation between NA EU trends in
  standardized SCE is r 0.83 (p lt 0.001).
• This implies a hemispheric-scale process may be
  acting on continental snowcovers.

25
Poleward amplification of trends

• Linear regressions on standardized SCE trends
  (January to early August) yield correlation
  coefficients of -0.89 to -0.96.
• This suggests a poleward amplification of SCE
  anomalies owing to persistence in the cryospheric
  system.
• Negative trends in early spring SCE amplify
  during late spring summer, with implications to
  the growing season, vegetation growth, species
  composition,

26
Poleward Amplification
27
Snow-albedo feedback

• Trends in insolation-weighted SCE values show
  greatest changes near the summer solstice.
• This feature, in addition to the spatial
  coherence of the intercontinental snowcovers
  temporal persistence on weekly annual time
  scales, are possible manifestations of
  snow-albedo feedback.

28
Conclusions

• Strong negative trends in NH, NA, EU weekly SCE
  (1972-2006) are observed.
• These trends are influenced by temporal
  persistence (i.e. serial correlation) in the
  cryospheric system.
• Similar behaviour in NA EU snowcovers,
  including covariability, persistence, amplified
  trends in spring/summer provides evidence of the
  snow-albedo feedback acting on a
  hemispheric-scale.

29
Postscript Autocorrelation
30
(No Transcript)
31
Observational evidence of an intensifying
hydrological cycle in northern CanadaStephen
DéryMarco Hernandez, Jason Burford, Eric Wood
IPY collaborators
32
Motivation Arctic rivers form a vital link
between the atmosphere, the pan-Arctic land
surface, and the Arctic Ocean
climate change may thus alter this natural
pathway for freshwater, leading to significant
environmental and societal change in the Arctic
and beyond.
33
International Polar Year (IPY)

• This work is a contribution to the IPY project
  Arctic Freshwater Systems.
• Research will lead to new knowledge information
  on the eco-hydrology of northern freshwater
  ecosystems.
• Our component seeks to better understand recent
  trends variability of river discharge in
  northern Canada.

34
Pan-Arctic domain
River Basin Discharge (km3 yr-1)
Lena 532
Yenisey 630
Ob 530
Yukon 205
Mackenzie 309
HJUB 714
Pan-Arctic 5250
Hudson, James Ungava Bays gauged area only
35
Arctic Ocean freshwater budget
Source term Input (km3 yr-1)
Net precip. 2000 (26)
River discharge 3200 (42)
Bering Strait 2500 (32)
Source Serreze et al. (2006), JGR.
36
Observed 20th century changes in pan-Arctic river
discharge

• text

• Try

Source McClelland et al. (2006), GRL.
37
Data Methods

• Daily river discharge is from the online Water
  Survey of Canadas HYDAT.
• Recent daily data for rivers in Québec obtained
  from Environnement Québec Hydro-Québec.
• A total of 45 rivers spanning gt 5 million km2 in
  northern Canada over 1964-2007 are used.

38
Detection of trends

• Compute mean, standard deviation coefficient of
  variation (CV) using 11-year moving windows of
  annual discharge.
• Linear trends of CV then determined from
  Mann-Kendall Test after pre-whitening of time
  series (Yue et al. 2002).
• Trends are significant when p lt 0.05.
• Results shown when lt 10 of the data are missing

39
?Increasing
?Decreasing
40
?Increasing
?Decreasing
41
?Increasing
?Decreasing
42
?Increasing
?Decreasing
43
Discussion/Summary

• Recent data show a reversal to increasing river
  discharge in northern Canada.
• This accompanies a trend toward greater
  streamflow variability (or in hydrological
  extremes) over this domain.
• These changes may be related to rising air
  temperatures that have induced changes in
  atmospheric and land surface processes.

44
Future Work

• Establish seasonality timing of river discharge
  changes.
• Attribute these changes including the role of
  large-scale teleconnections.
• Provide an updated record of river discharge time
  series trends in northern Canada to the IPY
  community.