Content uploaded by Anne Nolin
Author content
All content in this area was uploaded by Anne Nolin on Sep 28, 2017
Content may be subject to copyright.
INTERNATIONAL JOURNAL OF CLIMATOLOGY
Int. J. Climatol. (2015)
Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/joc.4545
Inuence of winter season climate variability on
snow–precipitation ratio in the western United States
Mohammad Safeeq,a,b*Shraddhanand Shukla,cIvan Arismendi,dGordon E. Grant,e
Sarah L. Lewisfand Anne Nolinf
aSierra Nevada Research Institute, University of California, Merced, CA, USA
bUSDA Forest Service, PSW Research Station, Fresno, CA, USA
cDepartment of Geography, University of California, Santa Barbara, CA, USA
dDepartment of Fisheries and Wildlife, Oregon State University, Corvallis, OR, USA
eUSDA Forest Service, PNW Research Station, Corvallis, OR, USA
fCollege of Earth, Ocean and Atmospheric Sciences, Oregon State University, Corvallis, OR, USA
ABSTRACT: In the western United States, climate warming poses a unique threat to water and snow hydrology because much
of the snowpack accumulates at temperatures near 0 ∘C. As the climate continues to warm, much of the region’s precipitation
is expected to switch from snow to rain, causing ashier hydrographs, earlier inow to reservoirs, and reduced spring and
summer snowpack. This study investigates historical variability in snow to precipitation proportion (Sf) and maps areas in the
western United States that have demonstrated higher Sfsensitivity to warming in the past. Projected changes in Sfunder 1.1,
1.8, and 3.0 ∘C future warming scenarios are presented in relation to historical variability and sensitivity. Our ndings suggest
that Sfin this region has primarily varied based on winter temperature rather than precipitation. The difference in Sfbetween
cold and warm winters at low- and mid-elevations during 1916 –2003 ranged from 31% in the Pacic Northwest to 40% in the
California Sierra Nevada. In contrast, the difference in Sfbetween wet and dry winters was statistically not signicant. Overall,
in the northern Sierra, Klamath, and western slopes of the Cascade Mountains Ranges, Sfwas most sensitive to temperature
where winter temperature ranged between −5to5
∘C. Results from our trend analysis show a regional shift in both Sfand
signal-to-noise ratios during 1960–2003 as compared with 1916–2003. Our ndings indicate that natural variability in Sf
over 1916– 2003 across all regions except for the Great Basin most closely resembles the projected 2040-warming scenario
(+1.8 ∘C).
KEY WORDS snow fraction; climate warming; signal-to-noise ratio; trend analysis; western United States
Received 16 March 2015; Revised 28 September 2015; Accepted 29 September 2015
1. Introduction
Mountain snowpacks in the western United States serve
as the primary source of spring and summer runoff, which
supports ecosystems as well as agriculture, industry, and
urban uses. Declines in streamow magnitude (Lins and
Slack, 1999; Luce and Holden, 2009), earlier streamow
timing (Stewart et al., 2005), and altered ood risk (Ham-
let and Lettenmaier, 2007) have been reported for this
region, all of which are primarily attributed to changes
in snowpack. Signicant reductions in snowpack accumu-
lation and earlier snowmelt have been attributed at least
in part to anthropogenic climate warming (Barnett et al.,
2008; Hidalgo et al., 2009). Continuing warming trends
in mid-latitude areas (Intergovernmental Panel on Climate
Change (IPCC), 2007a, 2007b; Adam et al., 2009) would
only intensify changes in snow accumulation and melt rate
across the western United States (Gleick, 1987; Letten-
maier and Gan, 1990; Dettinger et al., 2004; Knowles and
Cayan, 2004; Stewart et al., 2004).
* Correspondence to: M. Safeeq, Sierra Nevada Research Institute, Uni-
versity of California, 5200 N Lake Road, Merced, CA 95343, USA.
E-mail: msafeeq@ucmerced.edu
As temperatures continue to warm, much of this region
is expected to experience a shift from solid to liquid phase
precipitation (Knowles et al., 2006). More precipitation
falling as rain instead of snow, and consequently a lower
total snowfall to precipitation ratio (hereinafter referred
to as snow fraction, Sf), would affect total snow accu-
mulation and the timing of snowmelt and runoff regimes,
potentially leading to higher winter oods and lower ow
in late spring and summer (Safeeq et al., 2013, 2015). In
any given region, however, changes in Sfwould depend
on overall climatic regime as well as the correspond-
ing changes in temperature and precipitation. Specically,
areas where snow accumulates at temperatures near 0 ∘C,
also known as the transient snow zone, are more vulnerable
to warming than areas with snow accumulating at colder
temperatures (Hamlet et al., 2005; Nolin and Daly, 2006;
Sproles et al., 2013).
The transient snow zone is of particular hydrologic
interest, not only from the perspective of climate change
but also for its role in generating large oods through
rain-on-snow events (Harr, 1981; Christner and Harr,
1982; Marks et al., 1998; O’Connor and Costa, 2003;
Sureet and Tullos, 2012). To date, identication of the
© 2015 Royal Meteorological Society
M. SAFEEQ et al.
Figure 1. (a) Drainage boundary and topographic characteristics of the four study regions: Sierra Nevada (SN), Colorado River basin (CRB),
Great basin (GB), Columbia River, and Pacic Northwest coastal basins (PNW); (b) coefcient of determination (R2) and (c) percent bias
(negative PBIAS =underestimation, positive PBIAS =overestimation) between observed snow water equivalent and those empirically derived from
temperature and precipitation using Equation (1).
transient snow zone has always been based on elevation
(Christner and Harr, 1982; Harr, 1986; Sureet and Tul-
los, 2012) and/or temperature thresholds (Hamlet and Let-
tenmaier, 2007; Jefferson, 2011). This is mainly due to
the limited spatiotemporal coverage and record length of
directly relevant climatological [i.e. snow water equiv-
alent (SWE), precipitation, wind speed, and tempera-
ture] measurements (Hamlet et al., 2005). Availability of
high-quality spatially distributed gridded meteorological
data has improved our ability to analyze changes in snow-
pack and controls in a more spatially explicit fashion
(Hamlet et al., 2005; Nolin and Daly, 2006; Das et al.,
2009), as opposed to only a point-based analysis (Karl
et al., 1993; Frei et al., 1999; Mote et al., 2005; Knowles
et al., 2006; Feng and Hu, 2007).
Previous work has shown decreasing trends in snow-
pack in the western United States (Frei et al., 1999; Mote
et al., 2005; Brown and Mote, 2009; Abatzoglou, 2011;
Harpold et al., 2012; Rupp et al., 2013). The relative con-
tributions of changing temperature and precipitation on the
snow fraction (Hamlet et al., 2005; Knowles et al., 2006;
Feng and Hu, 2007) and on hydrologic drought have also
been documented (Mao et al., 2015; Shukla et al., 2015).
To our knowledge, there have been no studies showing
the historical spatiotemporal variability in Sfand how this
relates to past and potential future trends. Studies show-
ing monotonic trends in Sfprovide only a partial view of
future snowpacks, and cannot clarify whether, on average,
future snowpacks would be smaller than extreme years
in the past, or stay within the range of historic variabil-
ity. The answer to this question has direct implications for
water use, especially for reservoir operations in the west-
ern United States. Insight into signal (monotonic trends)
to noise (historical variability) ratios would be useful for
effectively managing water resources and aquatic ecosys-
tems. Our objectives here are to: (1) quantify the historical
variability in Sfduring extreme years and examine how
this relates to future climate warming in a spatially explicit
fashion; (2) quantify the sensitivity of Sfto temperature
as evidenced in the historical record; and (3) determine
the spatiotemporal variability in Sftrends over the western
United States.
2. Datasets and analyses
2.1. Precipitation and temperature dataset
We used gridded, 1/8th degree spatial resolution, daily
precipitation, and temperature datasets for 1916–2003
to compute Sfafter dividing the study domain into four
regions: (a) the California Sierra Nevada (SN); (b) the
Colorado River Basin (CRB); (c) the Great Basin (GB);
and (d) the Columbia River Basin and coastal Oregon and
Washington (PNW) (Figure 1(a)). These gridded datasets
were developed using precipitation and temperature obser-
vations from the National Climatic Data Service’s Coop-
erative Observer (Coop) network (Hamlet and Letten-
maier, 2005). Using gridded temperature and precipitation
datasets for estimating Sfas opposed to using direct tem-
perature and precipitation observations from weather sta-
tions has both advantages and disadvantages. The advan-
tage of using gridded meteorological datasets is that they
provide spatially and temporally consistent meteorological
conditions at places where direct observations are sparse
(Hamlet et al., 2005). The main disadvantage is that grid-
ded datasets are derived through interpolation of available
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
INFLUENCE OF WINTER SEASON CLIMATE VARIABILITY ON SNOW-FRACTION
station data and subjected to a number of potential inaccu-
racies and errors that can propagate to the estimates of Sf
or any other derived spatial data. These errors can be intro-
duced by an irregularly spaced underlying station network,
inaccurate and missing observations, and the interpolation
method used. In many mountainous regions, such as the
western United States, the network of meteorological sta-
tions at middle and high altitudes is sparse. In addition,
changes in number of functioning weather stations and
their location, instrumentation, and land use may cause
inhomogeneity in observed temperature and precipitation
that can propagate into gridded datasets.
Nevertheless, the gridded datasets used in this study were
specically developed to overcome some of the aforemen-
tioned issues and facilitate long-term trend analyses of
simulated hydrologic variables (Hamlet and Lettenmaier,
2005). The raw temperature and precipitation data from
Coop stations were rst screened for inaccuracies and then
gridded using 4 and 15 nearest neighbors for temperature
and precipitation, respectively. A higher number of nearest
neighbors were used for precipitation to prevent sharp dis-
continuities in the gridded data as a result of the relatively
low station density. After the gridding process, both tem-
perature and precipitation were subjected to temporal and
altitudinal adjustments to account for changes in number
and locations of Coop stations over time as well as to deal
with any biases associated with the sparse station network
at middle and higher elevations. Detailed description of the
data and gridding algorithms can be found in Hamlet and
Lettenmaier (2005).
2.2. Snow fraction
Daily precipitation (P) and average temperature values
(calculated as the arithmetic mean of daily maximum and
minimum temperatures) were used to calculate the annual
winter season (November-March) Sfand average wet day
(daily precipitation >0) temperature (Tw_avg ). Our deni-
tion of winter season is consistent with Knowles et al.
(2006), who reported 80% of snowfall occurring during
this season. Annual Sffor each grid cell was determined
after partitioning Pinto rain and snow using the lin-
ear empirical relationship developed by the United States
Army Corps of Engineers (1956). Sproles et al. (2013)
evaluated additional computationally complex rain and
snow partitioning algorithms, but the results were identi-
cal. The snowfall equivalent (SFE) (mm), dened as the
amount of Preaching the ground as snow, was calculated
as follows:
SFE =⎧
⎪
⎨
⎪
⎩
P,Tw_avg ≤TS
0,Tw_avg ≥TR
1
TR –TS ×(TR–Tw_avg)×PTS <Tw_avg <TR
(1)
where, Tw_avg is the average daily air temperature (∘C), TR
is the temperature above which all precipitation fell as rain,
TS is the temperature below which all precipitation fell as
snow, and Pis the total daily precipitation (mm day−1).
The annual winter season Sfwas calculated as:
Sf(%)=∑n
i=1SFEi
∑n
i=1Pi
×100 (2)
where nwas the last day of the winter season. The value
of Sfranged between 0 (all rain) and 100% (all snow).
We acknowledge the limitation of an empirically derived
SFE based on a temperature threshold alone, due pri-
marily to the spatiotemporal variation and sensitivity of
the temperature (i.e. TR and TS) for rain to snow tran-
sition. For example, based on the United States Army
Corps of Engineers (1956) study, Sproles et al. (2013)
approximated the precipitation phase transition between
TS =−2andTR=+2∘C in the McKenzie River Basin
of Oregon Cascades. Klos et al. (2014) used TS =−2
and TR =+4∘C, based on the ndings of Dai (2008),
for partitioning total precipitation into snow and rain
across the western United States. For the purpose of
this study, we performed a sensitivity analysis on Sfby
varying the TS between −2.0 and 0 ∘C and TR between
−1.0 and 4.0 ∘Cin0.5
∘C increments. The values of Sf
were calculated for all possible pairs (n=49) of TS and
TR with condition of TS <TR. Among the four regions,
Sfsensitivity (expressed in terms of standard deviation)
to TS and TR varied spatially within a specic region
with highest sensitivity centered (with the exception of
GB) around 50% Sf(Figure 2). In GB, the distribution
of Sfsensitivity was slightly skewed toward higher val-
ues of Sf. This suggested that choice of TS and TR
is particularly critical for characterizing the Sfvariabil-
ity in this region. To identify the representative TS and
TR value for our study domain, we correlated the win-
ter season positive changes (i.e. snow accumulation) in
SWE and empirically calculated SFE using Equation (1)
at 733 Snow Telemetry (SNOTEL) sites (available via:
http://www.wcc.nrcs.usda.gov/snow/). The average length
of the concurrent SWE, precipitation, and temperature
record during 1980–2012 was 25 years. Note that in
this case not all positive changes in SWE were related
to snowfall. In the transitional snow zone, rain-on-snow
could also result in positive SWE. Using TS =−2∘Cand
TR =+2∘C, we found strong agreement between observed
SWE (mm) and calculated SFE (mm), with 78% of the
total SNOTEL sites had an R2>0.5 and 40% of the sites
had an R2>0.8 (Figure 1(b)). In terms of percent bias
(PBIAS), 25% of the total SNOTEL sites had a PBIAS
within 10% of observed SWE and 77% had a PBIAS
within 25% of the observed SWE (Figure 1(c)). The abso-
lute biases of 69% of the sites were less than 5 cm and
89% of the sites were less than 10 cm. Also, the PBIAS
between observed SWE and calculated SFE was con-
sistent among wet (average PBIAS =−13.7%) and dry
(average PBIAS =−14.9%) winters. On average the agree-
ment between observed SWE and calculated SFE was
slightly better during cold (average PBIAS =−3.9%) than
those during warm (average PBIAS =−22.4%) winters.
Irrespective of climate extremes (i.e. wet, dry, cold, and
warm), at the majority of the sites PBIAS was nega-
tive, indicating that TS =−2∘CandTR=+2∘C resulted
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
M. SAFEEQ et al.
Figure 2. Spatial variability in (a) mean and (b) standard deviation of snow fraction (Sf) estimated using varying temperature thresholds (i.e. TR and
TS) for rain to snow transition.
in lower SWE values than those observed at the SNO-
TEL sites. However, most of the sites with large negative
PBIAS were located at lower elevations where using pos-
itive changes in SWE to validate SFE would have been
problematic. At these lower elevation sites, as mentioned
above, rain-on-snow events may result in an increase in
SWE. In addition, because of warmer temperatures at these
lower elevation sites, snowfall and melt may occur simul-
taneously with no net increase in daily SWE. By increas-
ing TR, this negative PBIAS was reduced, but resulted in
more sites with positive PBIAS. This suggested that the
value of TS and TR are site specic, but the lack of spa-
tially distributed SWE measurements limited our ability to
derive such parameters across the study domain. However,
we plan to investigate the variability in TS and TR at the
SNOTEL sites in the subsequent work. For the purpose of
this study, we approximated the precipitation phase transi-
tion as between TS =−2andTR=+2∘C. We considered
the choice of TS and TR values as a potential source of
uncertainty in the estimated Sfand discuss the implications
below.
2.3. Retrospective snow fraction sensitivity and trend
analysis
Spatially averaged November– March total Pwas used
to dene the 10 wettest and driest winters during
1916–2003 for each of the four regions by ranking
the November– March total Ptime series. Similarly, 10
coldest and warmest winters were dened after ranking
the spatially average Tw_avg for each of the four regions.
To capture the natural variability of climatic extremes
during the study period we used 10 as opposed to 5 years
of data to dene the extremes (e.g. Mishra and Cherkauer,
2011). We used a Mann –Whitney –Wilcoxon rank sum
test (p=0.05) to test differences in Sfbetween cold and
warm, and between wet and dry winters, across all the four
regions. Mean Sffor the 10 coldest (cold) and warmest
(warm) winters and corresponding Tw_avg were used to
dene the temperature sensitivity of Sfas:
𝜀T=Sf(cold)−Sf(warm)
Tw_avg (cold)−Tw_avg (warm)(3)
Similarly, precipitation sensitivity of Sfcan be calculated
as:
𝜀P=Sf(wet)−Sf(dry)
P(wet)−P(dry)(4)
The Theil–Sen approach (Sen, 1968) and
non-parametric Kendall’s tau tests (Kendall, 1938)
were used to identify trends in the annual Sftime series.
Monotonic trends were classied as ‘detectable’ when
their signal-to-noise ratios (SNR) were >1. The SNR was
dened as the absolute change in Sfcalculated from the
monotonic trend divided by the standard deviation over
a given period of interest (Déry et al., 2011). To evaluate
spatial and temporal consistency in the Sfchanges, both
trend and SNR analysis were performed for long-term
(1916–2003) and most recent (1960 –2003) time periods.
Observed trends in hydro-climatological data can be
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
INFLUENCE OF WINTER SEASON CLIMATE VARIABILITY ON SNOW-FRACTION
100
SNCRBGBPNW
80
60
40
20
0
100 Legend
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
5 101520
Elevation (×100 m)
Sf (%)
Sf (%)
Sf (cold)
pmean vs cold = 0.028
pmean vs warm = 0.035
pcold vs warm < 0.001
pmean vs wet = 0.732
pmean vs dry = 0.581
pwet vs dry = 0.419
pmean vs cold = 0.094
pmean vs warm = 0.118
pcold vs warm = 0.003
pmean vs wet = 0.528
pmean vs dry = 0.697
pwet vs dry = 0.416
pmean vs cold = 0.016
pmean vs warm = 0.041
pcold vs warm < 0.001
pmean vs wet = 0.617
pmean vs dry = 0.642
pwet vs dry = 0.399
pmean vs cold = 0.037
pmean vs warm = 0.071
pcold vs warm < 0.001
pmean vs wet = 0.867
pmean vs dry = 0.451
pwet vs dry = 0.428
Sf (mean)
Sf (warm)
[] (cold-warm)
25 30 35
100
SNCRBGBPNW
80
60
40
20
0
100 Legend
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
5101520
Elevation (×100 m)
Sf (cold)
Sf (mean)
Sf (warm)
[] (cold-warm)
25 30 35
(a) (b)
Figure 3. Elevation dependence of average snow fraction across the four study regions during 1916– 2003 climatological mean (mean) and (a) the
10 coldest winters (cold), 10 warmest winters (warm), and the difference between them (grayscale bars), (b) the 10 wettest winters (wet), 10 driest
winters (dry), and the difference between them (grayscale bars). Error bars associated with line plot show the standard error of mean Sffor each
100-m elevation bin.
highly sensitive to the time period over which trends are
evaluated; especially when length of record is short and
start/end years fall during episodes of strong large-scale
climate variability (i.e. the Pacic Decadal Oscillation and
El Niño Southern Oscillation). For this reason we choose
1960, an El Niño Southern Oscillation neutral year, as a
starting point for our short-term trend evaluation.
2.4. Snow fraction under projected climate
Potential effects of a warmer climate on Sfwere simu-
lated using the average changes in the PNW temperature
from 20 climate models and two greenhouse gas emis-
sions scenarios (B1 and A1B) for the 2020s, 2040s, and
2080s (Mote and Salathé, 2009). The projected changes
in temperature under the A1B scenario for CRB, GB, and
SN were kept the same as those predicted for the PNW.
These warming scenarios were generated after modify-
ing the daily air temperature using the delta method (Hay
et al., 2000) and keeping daily precipitation constant. The
daily temperature values were increased uniformly by 1.1,
1.8, and 3.0 ∘C for 2020, 2040, and 2080 warming scenar-
ios, respectively. In this study, effects of future changes
in precipitation on Sfwere not considered, since both
magnitude and direction of future precipitation, as pre-
dicted by global circulation models, are highly uncertain
for this region. Although recent trend analysis in historical
precipitation showed an increase in spring precipitation
and decrease in summer and autumn precipitation (Abat-
zoglou et al., 2014), there is no agreement in future precip-
itation projections from global circulation models. Mote
and Salathé (2009) reported a small increase (1–2%) in
the annual precipitation of the Pacic Northwest region
of the study domain under B1 and A1B emission scenar-
ios from the Intergovernmental Panel on Climate Change
(IPCC) Fourth Assessment Report (AR4). In contrast,
Ficklin et al. (2014) reported an average increase of 14.4%
in annual precipitation of the Columbia River Basin under
radiative forcing of 8.5 w/m2(RCP 8.5) of the Coupled
Model Inter-comparison Project – phase 5 (CMIP5). This
higher increase in precipitation could have been an artifact
of the differences in the emission scenarios between AR4
and CMIP5 and positive CMIP5 model bias (Mehran et al.,
2014).
3. Results
3.1. Snow fraction variability
To investigate the variability in Sfunder different climate
extremes, we examined Sfas a function of elevation dur-
ing the coldest and warmest (Figure 3(a)) and wettest and
driest winters (Figure 3(b)). For mean and extreme (dry,
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
M. SAFEEQ et al.
50°N
45°N
40°N
35°N
30°N
Climatological mean
(a) (b) (c)
(d) (e)
(g) (h)
(f)
Cold winter
Wet winter Dry winter Warming-2020
Warming-2040 Warming-2080
Warm winter
50°N
45°N
40°N
35°N
30°N
50°N
100
90
80
70
60
50
40
30
20
10
45°N
40°N
35°N
30°N
50°N
45°N
40°N
35°N
30°N
50°N
45°N
40°N
35°N
30°N
50°N
45°N
40°N
35°N
30°N
50°N
45°N
40°N
35°N
30°N
50°N
45°N
40°N
35°N
30°N
125°W120°W115°W105°W110°W125°W120°W115°W105°W110°W125°W120°W115°W105°W110°W
125°W120°W115°W105°W110°W125°W120°W115°W105°W110°W125°W120°W115°W105°W110°W
125°W120°W115°W105°W110°W125°W120°W115°W105°W110°W
Sf (%)
Figure 4. Average snow fraction (Sf%) during (a) climatological mean 1916 –2003, (b) 10 coldest winters, (c) 10 warmest winters, (d) 10 wettest
winters, (e) 10 driest winters, (f) 2020 warming scenario, (g) 2040 warming scenario, and (h) 2080 warming scenario.
wet, cold or warm) winters, Sfshowed a logistic relation-
ship with elevation across all four regions. As expected,
there were contrasting differences in Sfat similar eleva-
tions between the four regions. PNW showed signicantly
higher Sfat lower elevations when compared with SN,
CRB, and GB, which can be attributed to latitudinal dif-
ferences. Sfreached 100% at elevations just above 2000 m
in PNW as compared with nearly 2500 m in SN, CRB, and
GB (Figure 3). The highest Sfregions were along the Cas-
cade and Rocky Mountains in PNW, the southern part of
SN, and the Wasatch Range in CRB and GB during clima-
tological mean (Figure 4(a)), cold (Figure 4(b)) and warm
(Figure 4(c)), as well as during wet (Figure 4(d)) and dry
winters (Figure 4(e)). The Sfin northern Cascades, north-
ern Rockies, southern SN, and Wasatch Range remains
elevated during all climate extremes.
A two sample non-parametric Mann–Whitney–
Wilcoxon rank sum test showed statistically signi-
cant differences (p<0.05) between cold and warm winter
Sfacross all four regions. In PNW, the difference in
Sfbetween the historical mean and those during warm
winters was not statistically signicant (p>0.05). This
indicated that in PNW the Sfdistribution was skewed
toward warm winters. In CRB, there were no statisti-
cally signicant differences in mean Sfand those during
extreme cold and warm winters. However, the difference
in Sfduring warm and cold winters was statistically sig-
nicant (p<0.05). The difference in Sfbetween cold and
warm winters was large at mid-elevations (Figure 3(a)).
Geographically, there were large differences in cold
(Figure 4(b)) and warm (Figure 4(c)) winter Sfin the
northern part of SN, central part of CRB, both eastern and
western edges of GB, and the Columbia Plateau, Snake
River Plain and much of the eastern Oregon in PNW.
The greatest change in Sfbetween cold and warm winters
across elevation (shown as the bar chart in Figure 3)
occurred in SN (40%) followed by GB (39%), CRB
(34%), and PNW (31%). These ndings were inconsistent
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
INFLUENCE OF WINTER SEASON CLIMATE VARIABILITY ON SNOW-FRACTION
Table 1. Variability of winter season (November –March) precipitation (P) and average wet day temperature (Tw_avg) and their
inuence on snowfall to precipitation ratio (Sf) for the 10 coldest and 10 warmest years, selected based on domain-averaged Pand
Tw_avg for the period 1916– 2003 in the Sierra Nevada (SN), Colorado River Basin (CRB), Great Basin (GB), and Pacic Northwest
(PNW).
SN CRB GB PNW
Year P
(mm)
Tw_avg
(∘C)
Sf
(%)
Year P
(mm)
Tw_avg
(∘C)
Sf
(%)
Year P
(mm)
Tw_avg
(∘C)
Sf
(%)
Year P
(mm)
Tw_avg
(∘C)
Sf
Cold
1917 390 2.7 28.9 1917 128 −2.5 63.3 1917 140 −4.1 78.9 1916 644 −4.4 70.4
1922 452 4.0 25.2 1930 116 −0.6 49.8 1923 138 −2.7 73.0 1917 476 −5.3 76.4
1923 336 3.8 26.8 1932 176 −0.7 50.4 1930 118 −2.9 59.9 1922 502 −4.6 68.8
1932 402 3.2 27.9 1933 106 −3.0 59.7 1932 152 −3.9 71.5 1923 472 −4.4 70.6
1933 349 2.2 33.7 1937 177 −0.9 49.8 1933 118 −5.7 77.2 1929 375 −4.2 65.8
1937 442 3.0 30.2 1949 162 −1.8 54.4 1937 172 −4.6 71.4 1937 435 −5.5 72.8
1949 391 2.5 31.5 1952 197 −0.3 48.6 1944 145 −2.5 63.8 1949 532 −5.1 68.7
1950 433 4.1 23.6 1955 117 −0.5 49.5 1949 165 −5.3 72.3 1952 520 −4.2 70.0
1952 611 3.4 25.7 1974 129 −0.3 48.3 1952 215 −3.0 67.6 1969 574 −4.5 68.2
1969 644 4.1 23.6 1979 265 −0.4 43.1 1955 148 −3.1 57.8 1979 444 −4.5 65.8
Mean 445 3.3 27.7 –157 −1.1 51.7 –151 −3.8 69.3 –497 −4.7 69.8
War m
1934 316 6.5 12.8 1934 96 3.1 30.9 1934 113 1.4 35.1 1934 573 −0.1 45.2
1940 548 6.9 10.8 1938 168 2.5 32.7 1961 132 0.8 38.7 1940 540 −0.7 47.3
1963 364 7.0 11.0 1961 113 2.5 38.9 1963 127 0.4 36.7 1958 527 −1.0 53.6
1970 518 6.8 11.4 1978 218 2.4 34.0 1970 158 0.6 36.6 1961 592 −0.7 50.4
1978 591 6.6 14.4 1981 134 3.2 31.4 1978 188 0.9 37.5 1963 479 −1.1 48.4
1980 536 6.6 13.4 1986 161 2.4 35.4 1981 158 1.2 35.2 1981 517 −0.1 44.0
1981 400 6.5 13.4 1995 206 2.8 32.8 1986 195 0.4 41.6 1983 631 −0.6 50.6
1992 350 6.6 15.3 1996 128 2.5 31.9 1992 132 1.3 38.3 1992 436 −0.2 46.9
2000 450 6.4 15.1 2000 115 2.6 36.9 2000 150 0.4 43.9 2000 578 −1.2 56.0
2003 489 7.4 8.9 2003 148 2.7 31.0 2003 126 1.1 28.5 2003 563 0.4 42.9
Mean 456 6.7 12.7 – 149 2.7 33.6 – 148 0.9 37.2 – 544 −0.5 48.5
Cold–Warm −11 −3.4 15.0 –8−3.8 18.1 –3−4.7 32.1 –−47 −4.2 21.3
Gray shading indicates wettest and driest year for each region.
Table 2. Variability of winter season (November –March) precipitation (P) and average wet day temperature (Tw_avg) and their
inuence on snowfall to precipitation ratio (Sf) for the 10 wettest and 10 driest years, selected based on domain-averaged Pand
Tw_avg for the period 1916– 2003 in the Sierra Nevada (SN), Colorado River Basin (CRB), Great Basin (GB) and Pacic Northwest
(PNW).
SN CRB GB PNW
Year P
(mm)
Tw_avg
(∘C)
Sf
(%)
Year P
(mm)
Tw_avg
(∘C)
Sf
(%)
Year P
(mm)
Tw_avg
(∘C)
Sf
(%)
Year P
(mm)
Tw_avg
(∘C)
Sf
(%)
Wet
1938 649 5.6 16.5 1916 219 0.8 45.1 1916 194 −2.4 61.4 1916 644 −4.4 70.4
1952 611 3.4 25.7 1920 203 1.2 44.7 1952 215 −3.0 67.6 1938 638 −1.9 57.6
1956 603 4.6 19.0 1941 210 1.9 38.3 1969 197 −1.9 60.3 1956 696 −4.0 66.6
1969 644 4.1 23.6 1952 197 −0.3 48.6 1980 204 0.4 38.6 1971 660 −2.7 58.6
1978 591 6.6 14.4 1978 218 2.4 34.0 1982 213 −0.4 50.8 1972 684 −3.0 61.4
1982 611 5.8 16.6 1979 265 −0.4 43.1 1983 216 0.3 42.7 1974 775 −1.7 53.7
1983 744 5.7 15.6 1980 233 2.2 36.3 1984 216 −1.2 57.2 1982 677 −2.3 55.1
1986 586 6.2 13.8 1983 218 2.0 36.1 1986 195 0.4 41.6 1996 690 −1.5 53.6
1995 681 5.5 16.5 1993 256 1.0 37.1 1995 200 −0.1 43.6 1997 760 −2.4 61.0
1998 670 5.8 16.8 1995 206 2.8 32.8 1997 192 −0.5 46.3 1999 746 −1.4 56.7
Mean 639 5.3 17.9 –223 1.4 39.6 –204 −0.8 51.0 –697 −2.5 59.5
Dry
1920 302 4.6 23.5 1933 106 −3.0 59.7 1924 116 −2.0 58.4 1924 418 −2.6 61.1
1924 242 4.9 20.0 1934 96 3.1 30.9 1926 126 −0.6 46.1 1926 408 −1.4 57.5
1929 304 4.3 22.6 1946 106 0.4 44.7 1930 118 −2.9 59.9 1929 375 −4.2 65.8
1931 276 6.2 17.0 1959 102 1.4 42.8 1931 100 −1.6 60.2 1930 426 −3.6 62.9
1934 316 6.5 12.8 1964 108 0.5 47.7 1933 118 −5.7 77.2 1931 396 −2.1 58.9
1948 314 4.6 20.6 1972 93 0.0 44.7 1934 113 1.4 35.1 1941 389 −1.2 55.3
1976 249 4.6 21.7 1977 96 0.3 45.5 1977 93 −1.5 53.2 1944 331 −2.4 60.3
1977 210 4.8 20.7 1990 104 1.1 43.7 1987 123 −0.2 46.3 1977 291 −2.0 60.4
1990 256 4.3 22.0 1999 101 2.2 35.8 1990 107 −1.4 52.5 1994 404 −1.8 60.7
2001 322 4.9 21.5 2002 74 1.4 43.3 1991 119 −1.1 47.7 2001 297 −2.8 65.1
Mean 279 5.0 20.2 –99 0.7 43.9 –113 −1.6 53.7 –374 −2.4 60.8
Wet– Dry 360 0.3 −2.3 –124 0.7 −4.3 –91 0.8 −2.7 –324 −0.1 −1.3
Gray shading indicates wettest and driest year for each region.
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
M. SAFEEQ et al.
Figure 5. (a) Spatial distribution of temperature sensitivity of snow fraction (𝜀T), (b) elevation dependence of 𝜀T, and (c) elevation dependence of
Tw_avg.The𝜀Tand Tw_avg values were binned in 100-m elevation intervals and are shown as the mean (solid lines) and standard error (shading) for
each bin.
Figure 6. Temperature dependence (Tw_avg) of (a) temperature sensitivity of snow fraction (𝜀T), (b) trends in snow fraction (Sf) and (c) signal-to-noise
ratio (SNR) across regions. The 𝜀T,Sf, and SNR values were binned in 1 ∘CTw_avg intervals and are shown as the mean (solid lines) and standard
error (shading) for each bin.
with the corresponding change in Tw_avg between cold
and warm winters (Table 1). Although the average change
in temperature across all elevation grids between cold
and warm winters was smallest (−3.4 ∘C) in SN and
largest (−4.7 ∘C) in GB, both showed a similar change
in Sf. This was attributed to overall warmer temperatures
in SN, where a small change in temperature led to a
larger change in Sf. In contrast, the change in Sfin PNW
between cold and warm winters was the smallest despite a
larger (−4.2 ∘C) change in temperature, because of overall
colder temperatures during both cold and warm winters
(Table 1). Winter season precipitation during cold and
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
INFLUENCE OF WINTER SEASON CLIMATE VARIABILITY ON SNOW-FRACTION
Figure 7. Comparison of long-term (1916 –2003) and recent (1960 –2003) trends with 1 : 1 line (solid black) in (a) average winter wet day temperature
(Tw_avg), (b) snow fraction (Sf), and (c) corresponding signal-to noise ratios (SNR) in Sf. Spatial trends (d and e) and spatial shifts (f) are reported as
the difference between values for recent and long-term periods.
warm years was similar across the four regions (Table 1).
During warm winters, precipitation was higher by as much
as10%inPNWandlowerbyasmuchas5%inCRBas
compared with precipitation during cold winters. Across
all four regions the majority of cold winters occurred prior
to 1950 and the majority of warm winters occurred after
1970 (Table 1).
A nearly twofold increase in precipitation during
extreme wet years (Table 2) showed no inuence on Sf
across all the four regions (Figure 4(d) and (e)). In fact,
Sfduring dry winters was marginally higher (Figure 3(b))
compared with wet winters, particularly in SN and CRB,
which can be attributed to relatively cold temperatures in
those regions (Table 2), although this was not statistically
signicant (p>0.05). Similar to cold and warm winters,
the majority of the 10 driest and wettest winters dur-
ing the period 1916–2003 were prior to and post 1950,
respectively, except in CRB where both extremes occurred
predominantly after 1950s (Table 2). Year 1977, one of
the driest on record, ranked higher in terms of Sfbecause
of colder temperatures. In general, Sfwas lower when
a dry or wet year was associated with warm tempera-
tures (e.g. 1934 in SN, CRB, GB, and 1941 in PNW),
suggesting that temperature trumped precipitation in
determining Sf.
3.2. Retrospective snow fraction sensitivity and trends
The temperature sensitivity of Sf(𝜀T) varied signicantly
among regions (Figure 5). The Sfalong the Cascade and
Olympic mountain ranges in PNW, Klamath and north-
ern Sierra in SN, and part of the lower CRB was most
sensitive to temperature (Figure 5(a)). An increase in tem-
perature by 1 ∘C could result in a decrease of 10 –15% Sfin
these regions. In terms of elevation, the medium and high
sensitivity (𝜀T<−5) regions were between 1000– 2200 m
in SN, 1500–2000 m in CRB, 1200 –2200 m in GB, and
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
M. SAFEEQ et al.
0
–10
–20
–30
–40
0
–10
–20 Warm winter
Warm winter vs
2020: p = 0.010
2040: p = 0.576
2080: p < 0.001
Warm winter vs
2020: p = 0.175
2040: p = 0.215
2080: p < 0.001
Warm winter vs
2020: p < 0.001
2040: p = 0.528
2080: p < 0.001
Warm winter vs
2020: p < 0.001
2040: p = 0.142
2080: p < 0.001
2020
2040
2080
–30
–40
0
–10
–20
–30
–40
0
–10
–20
–30
–40
5–10 10–15 15–20 20–25
Elevation (×100 m)
25–30 30–35 35–40 >40
–
ΔSf (%)
SNCRBGBPNW
Scenario
Figure 8. Decline in Sfwith respect to climatological mean under historical warm winter, 2020, 2040, and 2080 scenarios. The line inside the box
represents the median value, the box itself represents the interquartile range (IQR) (25th– 75th percentile range) and the whiskers are the lowest and
highest values within 1.5 ×IQR of the 25th and 75th percentiles.
300–1500 m in PNW (Figure 5(b)). These highly sensi-
tive Sfregions were very similar to ‘at-risk’ snow mapping
developed by Nolin and Daly (2006) for the Pacic North-
west. The pattern in elevation–𝜀relationship (Figure 5(b))
can be characterized by the elevation-Tw_avg prole or
lapse rate among the four regions (Figure 5(c)).
Most of the medium and high 𝜀Tregions had Tw_avg
between −5and5
∘C (Figure 6(a)) and followed mono-
tonic trends (Figure 6(b)) and SNR (Figure 5(c)) in Sfvery
closely. However, despite higher Sfsensitivity between −5
and 5 ∘C temperature range in SN, the monotonic trend and
SNR in SN were comparable with those in PNW and GB.
In contrast, 𝜀Tbetween −5and5∘C temperature range in
CRB was similar to those in PNW and GB, but showed
much higher monotonic trend than SNR. These contrast-
ing patterns indicated that in CRB the monotonic trend in
Sfsupersedes the historic variability in Sf, whereas, in SN
historical variability in Sfsupersedes the monotonic trend.
Because there was no statistically signicant difference in
Sfbetween wet and dry years, precipitation sensitivity of
Sf,𝜀P, was excluded from further analysis.
Comparative analyses of long-term (1916– 2003) and
recent (1960–2003) monotonic trends indicated an
increase in warming indicated by Tw_avg during the most
recent time period (Figure 7(a)). Since 1960, Tw_avg
has increased on average by 1.4 (SN), 2.5 (CRB), 1.3
(GB), and 1.3 (PNW) times faster as compared with
the long-term trend. In terms of absolute magnitude, the
recent rate of warming in Tw_avg has been higher by as
much as 1.0 ∘C decade−1(Figure 7(d)). The effect of this
recent rapid warming in Tw_avg was clearly reected in
Sftrends. An overall downward trend in Sfwas observed
during both long-term and recent periods (Figure 7(b)).
The long-term (1916–2003) downward trend in Sfwas
statistically signicant for 37, 45, 76, and 63% of the grid
cells in SN, CRB, GB, and PNW, respectively. Regionally,
the average signicant decreasing trend ranged from 1.3%
decade−1in SN to 1.9% decade−1in GB. The spatial
extent of the statistically signicant trend during the
recent period (1960–2003) was reduced to 12, 42, 19,
and 23% of the grid cells in SN, CRB, GB, and PNW,
respectively. However, the regional average signicant
decreasing trend ranged from 1.6% decade−1in PNW
to 3.9% decade−1in GB, showing a more rapid decline
during the recent period. This large decline in Sfduring
1960–2003 as compared to 1916 –2003 was 2.1 and 2.3
times higher in CRB and GB, respectively (Figure 7(e)).
Similar to the trend analyses, the SNR analyses showed
a greater reduction in Sfduring the period 1960–2003
as compared with the period 1916–2003 (Figure 7(c)).
Trends in 16, 27, 54, and 31% of the grid cells in SN, CRB,
GB, and PNW, respectively were detectable (|SNR| >1)
during the period 1916–2003. During the more recent
period 1960–2003, the spatial extent of detectable trend
decreased to 7, 21, and 16% of the grid cells in SN, GB, and
PNW, respectively, but increased to 40% of the grid cells in
CRB. This indicates that not only did a greater percentage
of CRB grid cells show a decreasing trend in Sfduring
the recent period but also that the magnitude of the trend
was exceeded by the natural annual variability. Also, in the
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
INFLUENCE OF WINTER SEASON CLIMATE VARIABILITY ON SNOW-FRACTION
Table 3. Watershed average snow fraction (Sf) under historical climate (climatological mean), 10 warmest winters (warm winter),
and three future warming scenarios (i.e. 2020, 2040, and 2080).
HUC4 watershed Watershed average Sf(%)
Climatological mean Warm winter 2020 2040 2080
Bear 79 (5) 69 (7) 72 (6) 67 (6) 59 (7)
Black Rock Desert-Humboldt 56 (8) 38 (8) 46 (7) 41 (7) 32 (6)
Central Lahontan 50 (7) 33 (7) 41 (7) 36 (6) 28 (6)
Central Nevada Desert Basins 52 (7) 35 (8) 43 (7) 38 (6) 30 (6)
Colorado Headwaters 87 (3) 83 (4) 82 (4) 79 (4) 72 (5)
Escalante Desert-Sevier Lake 61 (7) 47 (8) 52 (7) 47 (7) 38 (6)
Great Divide-Upper Green 87 (4) 83 (5) 81 (5) 77 (5) 70 (6)
Great Salt Lake 54 (7) 39 (8) 46 (7) 40 (6) 32 (6)
Gunnison 88 (3) 83 (4) 83 (4) 80 (4) 74 (5)
Klamath-Northern California Coastal 30 (7) 19 (6) 23 (6) 19 (5) 13 (4)
Kootenai-Pend Oreille-Spokane 79 (6) 72 (6) 72 (6) 66 (7) 57 (7)
Little Colorado 42 (7) 30 (7) 33 (7) 28 (6) 21 (5)
Lower Colorado-Lake Mead 31 (6) 19 (6) 24 (5) 20 (5) 14 (4)
Lower Columbia 27 (6) 18 (5) 20 (5) 16 (5) 11 (3)
Lower Green 73 (5) 65 (6) 67 (5) 62 (5) 55 (6)
Lower Snake 70 (5) 62 (6) 63 (6) 58 (6) 49 (6)
Middle Columbia 44 (7) 32 (7) 35 (7) 30 (6) 23 (5)
Middle Snake 63 (7) 48 (8) 54 (7) 48 (7) 39 (7)
North Lahontan 51 (9) 31 (9) 40 (8) 34 (8) 25 (6)
Northern Mojave-Mono Lake 23 (4) 15 (3) 19 (3) 16 (3) 13 (2)
Oregon Closed Basins 58 (8) 40 (9) 48 (8) 42 (8) 32 (7)
Oregon-Washington Coastal 14 (4) 8 (3) 10 (3) 8 (2) 5 (2)
Puget Sound 32 (5) 26 (5) 26 (5) 23 (4) 17 (4)
Sacramento 27 (6) 18 (5) 21 (5) 17 (4) 12 (4)
Salt 24 (6) 15 (5) 18 (5) 15 (4) 10 (3)
San Joaquin 25 (2) 22 (2) 23 (2) 21 (2) 18 (2)
San Juan 54 (7) 43 (8) 45 (7) 40 (7) 31 (6)
Tulare-Buena Vista Lakes 22 (2) 18 (2) 19 (2) 18 (2) 15 (2)
Upper Colorado-Dirty Devil 55 (7) 44 (7) 47 (6) 42 (6) 34 (6)
Upper Colorado-Dolores 64 (6) 54 (7) 56 (6) 52 (6) 43 (6)
Upper Columbia 72 (5) 64 (6) 65 (6) 60 (6) 52 (6)
Upper Gila 20 (5) 15 (4) 15 (4) 12 (3) 8 (3)
Upper Snake 74 (5) 63 (7) 67 (6) 63 (6) 55 (6)
White-Yampa 80 (5) 74 (6) 73 (5) 69 (6) 61 (6)
Willamette 23 (5) 15 (5) 17 (5) 14 (4) 9 (3)
Yakima 59 (7) 48 (8) 49 (8) 44 (8) 34 (7)
Parenthetical values are one standard deviation and express variability in the estimates due to the plausible range of TR (−1to+4∘C) and TS (−2
to 0 ∘C) threshold values (n=49).
majority of the grid cells the detectable decreasing trends
(SNR <−1) during 1916–2003 and 1960 –2003 did not
overlap, indicating a geographical shift towards a decline
in Sfduring recent time (Figure 7(f)).
3.3. Effects of climate warming on snow fraction
The projected Sffor 2020 (Figure 4(f)), 2040 (Figure 4(g)),
and 2080 (Figure 4(h)) warming scenarios showed a
signicant decline, particularly along the Cascade and
Olympic mountains in the PNW region. However, higher
elevations in southern SN, northern Rockies, and Wasatch
Range (Figure 1(a)) would continue to be dominated by
higher Sf. Much of the southern Cascade and Olympic
Mountains in PNW and GB would experience more rain
than snow under the 2080 warming scenario. On average,
the decrease in Sfunder warm winters and future warm-
ing scenarios in the SN, CRB, GB, and PNW regions
ranged from 7 to 16%, 4 to 9%, 6 to 14%, and 10 to
22%, respectively. The declines in Sfunder warm winters
place results from future warming scenarios in the con-
text of historical climate variability. GB and SN showed
the largest and smallest changes, respectively, under all
four scenarios. However, at mid-elevations, PNW and SN
showed the greatest decrease in Sf, followed by GB and
CRB (Figure 8). The difference in Sfbetween the histor-
ical average and warmest winters most closely resembles
the 2040 warming scenario in all the four regions except
GB. In GB, the difference in Sfunder warm winters is
higher at lower elevations and lower at higher elevations
compared with the 2040 warming scenario. An increase
of temperature by 3 ∘C under the 2080 warming scenario
resulted in Sfeven lower than current extreme warm win-
ters across all the four regions. As expected, the warming
showed greater impact on Sfat mid-elevations. However,
the zone of inuenced elevation expands with increasing
temperature.
The comparison of average Sfat the USGS Hydrologic
Unit Code sub-regional scale (HUC4) under historical
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
M. SAFEEQ et al.
Sf > 25% Sf > 50%
Warm winter
100
(a) (b)
90
80
70
60
50
40
30
20
10
2020
2040
2080
Scenario
Warm winter
2020
2040
2080
Scenario
SN CRB GB
Re
g
ion
PNWSN CRB GB
Re
g
ion
PNW
Percentage area, [100 × (historic mean – scenario) / historic mean]
100
90
80
70
60
50
40
30
20
10
Percentage area, [100 × (historic mean – scenario) / historic mean]
Figure 9. Average decline in snowfall dominated (a) Sf>25% and (b) Sf>50% area under warm winter conditions (showing historical range) and
under three warming (2020, 2040, and 2080) scenarios. The error bars show ±one standard deviation across the plausible range of TR (−1to+4∘C)
and TS (−2to0∘C) threshold values (n=49).
climate, warm winter, and three future warming scenar-
ios showed signicant declines across all the watersheds
(Table 3). The uncertainty in Sfestimates (expressed in
terms of standard deviation) associated with the choice of
TR and TS was far smaller than projected changes in Sf.By
2020, 8% of the watersheds showed Sflower than the aver-
age Sfduring the 10 warmest winters, which increased to
58% by 2040 and 100% by 2080. This further conrms that
by 2080 Sfmay probably surpass the historical variability
associated with climatological extremes (i.e. 10 warmest
winters).
In terms of decline in areal extent of current Sfregions
under each warming scenario, GB and SN were most
sensitive (Figure 9(a)). For example, spatial extent in
GB and SN where Sfis currently greater than 25% may
decline under the 2080 warming scenario by 53 ±9and
37 ±11% of the historical mean, respectively. When
this threshold of Sfwas increased to 50%, the reduc-
tion in spatial extent with respect to the historical mean
is even larger (Figure 9(b)). Analysis of areal extent
reduction at the regional watershed scale showed very
similar patterns (Table 4). By 2080, 35 and 64% of the
watersheds where Sfis currently greater than 25 and
50%, respectively, would probably experience a 50–95%
reduction in proportional area. As expected, reduction
in proportional areal extent is inversely correlated with
the watershed average Sf. As noted by Klos et al. (2014),
loss of currently snow-dominated areas occurred mainly
in watersheds with moderate relief and elevation. These
types of watersheds are typically warmer and received
most precipitation in the form of rain (i.e. lower Sf). In 33
of our 36 study watersheds, reduction in proportional area
with Sf>50% under the warm winter scenario was higher
than those under the 2020 warming scenario. However,
was reduced to only ve watersheds by 2040, which
further conrms warming of 1.8 ∘C and higher would
surpass the historical variability Sf.
As the climate warms, one of the immediate implica-
tions from decreasing Sfis a potential increase in winter
ooding as the type of precipitation changes from snow
to rain. We illustrated this based on historical records that
showed ashier streamow during low as compared with
high Sfyears (Figure 10). The Richard–Baker ashiness
index (Baker et al., 2004) measures oscillations in daily
streamow relative to total streamow over a time period.
Based on 93 USGS stream gage stations (Figure 1(a)), the
ashiness index was higher by, on average, 17% during
low as compared with high Sfyears. These differences
are only from the recent period (1950–2003) and may not
necessarily reect the range of stream ashiness over the
period of record (1916–2003) for which variability in Sfis
presented. Nonetheless, this highlights the strong coupling
between the stream hydrograph and type of precipitation
(rain or snow) occurring in the entire study region.
4. Discussion and conclusions
This study provides new information on Sfvariability
and trends across the western United States in a spatially
explicit fashion using gridded meteorological data. Our
ndings reveal underlying geographic patterns in Sf,and
provide a foundation for anticipating changes in Sfunder
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
INFLUENCE OF WINTER SEASON CLIMATE VARIABILITY ON SNOW-FRACTION
Table 4. Current snow extent (% area with Sf>25% and Sf>50%) and reduction under historical 10 warmest winters (warm winter)
and three future warming scenarios (i.e. 2020, 2040, and 2080).
HUC4 watershed Current
proportional
area (%)
Reduction in
proportional area (%)
Current
proportional
area (%)
Reduction in
proportional area (%)
War m
winter
2020 2040 2080 Warm
winter
2020 2040 2080
Sf>25% Sf>50%
Bear 100 (0) 8 (4) 1 (2) 3 (3) 8 (4) 91 (0) 12 (4) 6 (2) 11 (3) 25 (4)
Black Rock
Desert-Humboldt
96 (7) 30 (12) 10 (6) 19 (8) 38 (13) 61 (7) 57 (12) 34 (6) 55 (8) 80 (13)
Central Lahontan 89 (14) 48 (10) 21 (11) 36 (12) 57 (9) 42 (14) 46 (10) 32 (11) 48 (12) 69 (9)
Central Nevada Desert
Basins
86 (5) 29 (11) 9 (5) 17 (8) 35 (11) 56 (5) 52 (11) 30 (5) 49 (8) 76 (11)
Colorado Headwaters 100 (0) 2 (2) 0 (1) 1 (2) 5 (4) 94 (0) 7 (2) 5 (1) 9 (2) 16 (4)
Escalante Desert-Sevier
Lake
99 (2) 23 (16) 5 (8) 13 (14) 33 (17) 63 (2) 37 (16) 27 (8) 40 (14) 60 (17)
Great Divide-Upper Green 100 (0) 0 (0) 0 (0) 0 (0) 0 (0) 100 (0) 0 (0) 0 (0) 1 (0) 6 (0)
Great Salt Lake 95 (8) 34 (12) 12 (10) 24 (12) 43 (10) 52 (8) 44 (12) 28 (10) 45 (12) 69 (10)
Gunnison 100 (0) 3 (3) 1 (2) 2 (3) 6 (4) 93 (0) 6 (3) 4 (2) 7 (3) 12 (4)
Klamath-Northern
California Coastal
53 (8) 41 (18) 22 (11) 38 (16) 66 (15) 26 (8) 62 (18) 50 (11) 70 (16) 89 (15)
Kootenai-Pend
Oreille-Spokane
99 (1) 3 (3) 2 (3) 5 (4) 12 (4) 90 (1) 9 (3) 10 (3) 18 (4) 33 (4)
Little Colorado 75 (13) 33 (8) 21 (5) 34 (8) 58 (11) 35 (13) 59 (8) 50 (5) 72 (8) 92 (11)
Lower Colorado-Lake Mead 52 (10) 48 (12) 25 (7) 41 (9) 63 (7) 22 (10) 51 (12) 39 (7) 56 (9) 78 (7)
Lower Columbia 38 (9) 31 (6) 27 (5) 41 (5) 64 (8) 21 (9) 43 (6) 36 (5) 55 (5) 82 (8)
Lower Green 100 (1) 6 (5) 2 (3) 5 (4) 11 (5) 83 (1) 23 (5) 13 (3) 23 (4) 37 (5)
Lower Snake 90 (5) 11 (3) 7 (3) 13 (4) 23 (4) 73 (5) 13 (3) 12 (3) 18 (4) 31 (4)
Middle Columbia 77 (12) 30 (11) 20 (6) 34 (9) 58 (12) 37 (12) 56 (11) 43 (6) 64 (9) 85 (12)
Middle Snake 96 (5) 16 (7) 6 (3) 11 (4) 24 (10) 73 (5) 39 (7) 24 (3) 41 (4) 65 (10)
North Lahontan 91 (10) 41 (15) 15 (8) 29 (12) 55 (17) 51 (10) 67 (15) 47 (8) 70 (12) 91 (17)
Northern Mojave-Mono
Lake
31 (5) 36 (2) 19 (2) 30 (2) 45 (2) 18 (5) 28 (2) 20 (2) 30 (2) 46 (2)
Oregon Closed Basins 100 (1) 20 (19) 3 (6) 10 (13) 31 (24) 70 (1) 68 (19) 44 (6) 66 (13) 89 (24)
Oregon-Washington Coastal 17 (6) 40 (5) 32 (4) 45 (5) 63 (3) 8 (6) 37 (5) 31 (4) 48 (5) 73 (3)
Puget Sound 40 (5) 17 (3) 16 (3) 26 (4) 42 (4) 29 (5) 20 (3) 20 (3) 32 (4) 50 (4)
Sacramento 46 (6) 39 (16) 19 (10) 34 (14) 61 (15) 25 (6) 57 (16) 46 (10) 66 (14) 86 (15)
Salt 39 (12) 50 (3) 33 (5) 50 (4) 69 (3) 15 (12) 50 (3) 39 (5) 58 (4) 83 (3)
San Joaquin 29 (2) 12 (3) 7 (2) 12 (3) 21 (2) 25 (2) 12 (3) 10 (2) 15 (3) 25 (2)
San Juan 95 (7) 20 (14) 13 (10) 25 (15) 50 (17) 49 (7) 39 (14) 38 (10) 53 (15) 69 (17)
Tulare-Buena Vista Lakes 27 (3) 19 (2) 11 (2) 18 (2) 28 (2) 21 (3) 15 (2) 11 (2) 17 (2) 28 (2)
Upper Colorado-Dirty Devil 89 (7) 25 (12) 12 (6) 22 (10) 42 (11) 52 (7) 36 (12) 28 (6) 40 (10) 55 (11)
Upper Colorado-Dolores 97 (5) 16 (7) 7 (5) 13 (6) 26 (9) 69 (5) 26 (7) 20 (5) 32 (6) 50 (9)
Upper Columbia 97 (2) 9 (6) 5 (4) 9 (6) 20 (7) 77 (2) 15 (6) 13 (4) 21 (6) 35 (7)
Upper Gila 30 (6) 29 (7) 22 (2) 35 (5) 59 (13) 14 (6) 57 (7) 54 (2) 76 (5) 95 (13)
Upper Snake 98 (2) 9 (4) 3 (2) 6 (2) 13 (3) 85 (2) 24 (4) 12 (2) 21 (2) 37 (3)
White-Yampa 100 (0) 0 (0) 0 (0) 0 (0) 0 (1) 98 (0) 9 (0) 6 (0) 12 (0) 29 (1)
Willamette 33 (7) 33 (9) 25 (7) 40 (11) 66 (11) 18 (7) 53 (9) 44 (7) 65 (11) 86 (11)
Yakima 96 (6) 17 (11) 9 (7) 18 (10) 38 (14) 60 (6) 31 (11) 28 (7) 45 (10) 71 (14)
Parenthetical values are one standard deviation and express variability in the estimates due to the plausible range of TR (−1to+4∘C) and TS (−2
to 0 ∘C) threshold values (n=49).
climate warming at a daily resolution. Consistent with pre-
vious analyses (i.e. Knowles et al., 2006; Das et al., 2009;
Serquet et al., 2011), we show that winter temperature, as
opposed to precipitation variability, is the primary con-
trol on Sf. The difference in Sfbetween cold and warm
winters during 1916–2003 at low- and mid-elevations has
ranged between 31% in the PNW to as much as 40% in
the SN. In contrast, the difference in Sfbetween wet and
dry winters during 1916–2003 at all elevation ranges is not
signicantly different across all four regions. Interestingly,
the majority of the 10 wettest (at least 7 of 10) and warmest
(at least 8 of 10) winters during the period 1916–2003
occur post 1950s.
We illustrate a high sensitivity of Sfunder average wet
daily temperature values (November– March) between 5
and −5∘C. In the Northern Sierra, Klamath Mountains,
and western slopes of the Cascade Mountain Range, a 1 ∘C
increase in winter temperature would result in a decrease
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
M. SAFEEQ et al.
Figure 10. Comparisons of January– March Richard– Baker ashiness index (Baker et al., 2004) between low (10th percentile) and high (90th
percentile) Sfyears at 93 USGS gages during 1950– 2003, with linear t (solid line) and 1: 1 line (dashed line).
of 10–15% in Sf. Although our results further conrm a
signicant decline in Sfacross the entire region we show
that this reduction is more pronounced during the recent
decades (1960–2003) as compared with the longer his-
torical period (1916–2003). There has also been a recent
(1960–2003) regional shift in Sftrends and SNR. The
warming scenario analysis indicates that natural variabil-
ity in Sfover 1916– 2003 most closely resembles the
2040 warming scenario, except for the GB region. This
suggests that under the future scenarios examined here,
the long-term trend in Sf(with inter-annual variability
removed) would begin to exceed the historical range of Sf
variability around 2040, corresponding here to a warming
of 1.8 ∘C relative to historical temperature. In addition, we
demonstrated a differential decline in Sfacross the four
geographic units SN, CRB, GB, and PNW indicating a
non-homogenous effect of the recent (1960–2003) warm-
ing climate across the region.
Our trend analysis approach yields Sfsensitivities that
are in agreement with those reported by Nolin and Daly
(2006), and expands this sensitivity mapping to include
geographies with lower Sfsensitivity. However, the high Sf
sensitivity regions show a strong SNR indicating that the
trend in Sfhas already exceeded the historical variability.
Given that many of the high Sfsensitivity regions fall
in the mid-elevation range, a continued warming climate
(Mote and Salathé, 2010) would push future Sfeven fur-
ther outside its historical variability. Our warming sce-
narios also show signicant shifts in Sfacross the four
regions, some more than others. The actual magnitude
and changes in Sfpresented in this study are subjected to
uncertainties; especially at lower elevations where agree-
ment between empirically derived Sfand those measured
at SNOTEL sites were poor. Uncertainties associated with
temperature threshold-based Sfportioning, however, are
far smaller than the projected change in Sfand correspond-
ing snow extent areal reduction; giving condence in the
results. Also, uniform increases of 1.1, 1.8, and 3.0 ∘C may
not accurately represent the uncertainties associated with
future warming that could vary in both space and time.
Nonetheless, this simplistic approach provides a relative
comparison of future Sfin terms of sensitivity to tempera-
ture across the region. Our ndings also imply that uncer-
tainties associated with future changes in precipitation,
which show mixed trends across the domain (Dominguez
et al., 2012; Reclamation, 2013) and strong disagree-
ment between new-generation CMIP5 and old-generation
CMIP3 models (Reclamation, 2013), should not affect
temperature-driven Sfsensitivity.
Runoff from mountain snowmelt is the primary source
of water in much of the western United States. A decline
in Sfwould increase winter runoff, reduce snowpack, and
subsequently reduce summer runoff, with widespread
implications for water management, including reservoir
operations (Barnett and Pierce, 2009; Brekke et al.,
2009; Danner, 2013) and irrigation regimes (Benson
and Williams, 2013; McDonald and Girvetz, 2013). In
addition to a seasonal shift in the timing and magnitude
of streamow (Barnett et al., 2005; Regonda et al., 2005;
Stewart et al., 2005; Luce and Holden, 2009), decline in
Sfis likely to reduce total annual streamow. A recent
study by Berghuijs et al. (2014) reports a signicant
linear increase in precipitation-normalized streamow of
0.37/U−1increase in Sf. We show that a warming of only
1.8 ∘C under the 2040 scenario would result in a Sflower
than the warm winters during the period 1916–2003,
and may require changing the way water infrastructure is
currently managed (Danner, 2013).
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
INFLUENCE OF WINTER SEASON CLIMATE VARIABILITY ON SNOW-FRACTION
References
Abatzoglou JT. 2011. Inuence of the PNA on declining mountain snow-
pack in the western United States. Int. J. Climatol. 31(8): 1135– 1142.
Abatzoglou JT, Rupp DE, Mote PW. 2014. Seasonal climate variability
and change in the Pacic Northwest of the United States. J. Clim.
27(5): 2125– 2142.
Adam JC, Hamlet AF, Lettenmaier DP. 2009. Implications of global
climate change for snowmelt hydrology in the twenty-rst century.
Hydrol. Process. 23(7): 962–972.
Baker DB, Richards RP, Loftus TT, Kramer JW. 2004. A new ashi-
ness index: characteristics and applications to Midwestern Rivers and
streams. J. Am. Water Resour. Assoc. 40(2): 503–522.
Barnett TP, Pierce DW. 2009. Sustainable water deliveries from the
Colorado River in a changing climate. Proc. Natl. Acad. Sci. U. S. A.
106(18): 7334– 7338.
Barnett TP, Adam JC, Lettenmaier DP. 2005. Potential impacts of a
warming climate on water availability in snow-dominated regions.
Nature 438: 303– 309.
Barnett TP, Pierce DW, Hidalgo HG, Bonls C, Santer BD, Das T,
Bala G, Wood AW, Nozawa T, Mirin AA, Cayan DR, Dettinger MD.
2008. Human-induced changes in the hydrology of the western United
States. Science 319(5866): 1080– 1083.
Benson A, Williams R. 2013. Cost of early snowmelt in terms of reduced
irrigation deliveries. J. Nat. Resour Policy Res. 5(2–3): 79–89.
Berghuijs WR, Woods RA, Hrachowitz M. 2014. A precipitation shift
from snow towards rain leads to a decrease in streamow. Nat. Clim.
Change 4(7): 583– 586.
Brekke LD, Maurer EP, Anderson JD, Dettinger MD, Townsley ES,
Harrison A, Pruitt T. 2009. Assessing reservoir operations risk
under climate change. Water Resources Research 45: W04411, doi:
10.1029/2008WR006941.
Brown RD, Mote PW. 2009. The response of northern hemisphere snow
cover to a changing climate. J. Clim. 22(8): 2124– 2145.
Christner J, Harr RD. 1982. Peak streamows from the transient snow
zone, Western Cascades, Oregon. In 50th Annual Meeting Western
Snow Conference, Harr RD (ed). Colorado State University: Reno,
NV.
Dai A. 2008. Temperature and pressure dependence of the rain-snow
phase transition over land and ocean. Geophys. Res. Lett. 35: L12802,
doi: 10.1029/2008GL033295.
Danner AG. 2013. Will We Need to Change the Rules: Assessing the
Implications of Climate Change for Dam Operations in Oregon’s
McKenzie River Basin. MS thesis, Oregon State University: Corvallis,
Oregon.
Das T, Hidalgo HG, Pierce DW, Barnett TP, Dettinger MD, Cayan
DR, Bonls C, Bala G, Mirin A. 2009. Structure and detectability
of trends in hydrological measures over the western United States. J.
Hydrometeorol. 10(4): 871–892.
Déry SJ, Mlynowski TJ, Hernández-Henríquez MA, Straneo F. 2011.
Interannual variability and interdecadal trends in Hudson Bay stream-
ow. J. Mar. Syst. 88(3): 341– 351.
Dettinger MD, Cayan DR, Meyer MK, Jeton AE. 2004. Simulated
hydrologic responses to climate variations and change in the Merced,
Carson, and American River basins, Sierra Nevada, California,
1900– 2099. Clim. Change 62(1–3): 283–317.
Dominguez F, Rivera E, Lettenmaier D, Castro C. 2012. Changes in
winter precipitation extremes for the western United States under a
warmer climate as simulated by regional climate models. Geophys.
Res. Lett. 39(5): L05803.
Feng S, Hu Q. 2007. Changes in winter snowfall/precipitation ratio in
the contiguous United States. J. Geophys. Res. 112: D15109, doi:
10.1029/2007JD008397.
Ficklin DL, Barnhart BL, Knouft JH, Stewart IT, Maurer EP, Letsinger
SL, Whittaker GW. 2014. Climate change and stream temperature
projections in the Columbia River basin: habitat implications of spa-
tial variation in hydrologic drivers. Hydrol. Earth Syst. Sci. 18(12):
4897– 4912.
Frei A, Robinson DA, Hughes MG. 1999. North American snow extent:
1900– 1994. Int. J. Climatol. 19(14): 1517 –1534.
Gleick PH. 1987. Regional hydrologic consequences of increases
in atmospheric CO2and other trace gases. Clim. Change 10(2):
137– 160.
Hamlet AF, Lettenmaier DP. 2005. Production of temporally consistent
gridded precipitation and temperature elds for the continental United
States. J. Hydrometeorol. 6(3): 330 –336.
Hamlet AF, Lettenmaier DP. 2007. Effects of 20th century warming and
climate variability on ood risk in the western U.S. Water Resour. Res.
43: W06427, doi: 10.1029/2006WR005099.
Hamlet AF, Mote PW, Clark MP, Lettenmaier DP. 2005. Effects of
temperature and precipitation variability on snowpack trends in the
Western United States. J. Clim. 18(21): 4545 –4561.
Harpold A, Brooks P, Rajagopal S, Heidbuchel I, Jardine A, Stielstra C.
2012. Changes in snowpack accumulation and ablation in the inter-
mountain west. Water Resour. Res.48: W11501, doi: 10.1029/2012
WR011949.
Harr RD. 1981. Some characteristics and consequences of snowmelt
during rainfall in western Oregon. J. Hydrol. 53(3): 277–304.
Harr RD. 1986. Effects of clearcutting on rain on snow runoff in Western
Oregon: a new look at old studies. Water Resour. Res. 22: 383 –392.
Hay LE, Wilby RL, Leavesley GH. 2000. A comparison of delta change
and downscaled GCM scenarios for three mountainous basins in the
United States. J. Am. Water Resour. Assoc. 36(2): 387–397.
Hidalgo HG, Das T, Dettinger MD, Cayan DR, Pierce DW, Barnett TP,
Bala G, Mirin A, Wood AW, Bonls C, Santer BD, Nozawa T. 2009.
Detection and attribution of streamow timing changes to climate
change in the Western United States. J. Clim. 22(13): 3838 –3855.
Intergovernmental Panel on Climate Change (IPCC), 2007a. Sum-
mary for policymakers. In Climate Change 2007: The Physical Sci-
ence Basis. Contribution of Working Group I to the Fourth Assess-
ment Report of the Intergovernmental Panel on Climate Change,
Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB,
Tignor M, Miller HL (eds). Cambridge University Press: Cambridge,
UK and New York, NY.
Intergovernmental Panel on Climate Change (IPCC). 2007b. Climate
Change 2007: Impacts, Adaptation And Vulnerability. Contribution of
Working Group II to the Fourth Assessment Report of the Intergovern-
mental Panel on Climate Change, Parry ML, Canziani OF, Palutikof
JP, van der Linden PJ, Hanson CE (eds). Cambridge University Press:
Cambridge, UK and Geneva, Switzerland, 976 pp.
Jefferson AJ. 2011. Seasonal versus transient snow and the elevation
dependence of climate sensitivity in maritime mountainous regions.
Geophys. Res. Lett. 38: L16402, doi: 10.1029/2011GL048346.
Karl TR, Jones PD, Knight RW, Kukla G, Plummer N, Razuvayev V,
Gallo KP, Lindseay J, Charlson RJ, Peterson TC. 1993. Asymmetric
trends of daily maximum and minimum temperature. Bull. Am. Mete-
orol. Soc. 74(6): 1007– 1023.
Kendall MG. 1938. A new measure of rank correlation. Biometrika
30(1/2): 81– 93.
Klos PZ, Link TE, Abatzoglou JT. 2014. Extent of the rain-snow tran-
sition zone in the western U.S. under historic and projected climate.
Geophys. Res. Lett. 41: 4560– 4568, doi: 10.1002/2014GL060500.
Knowles N, Cayan DR. 2004. Elevational dependence of projected
hydrologic changes in the San Francisco estuary and watershed. Clim.
Change 62(1– 3): 319–336.
Knowles N, Dettinger MD, Cayan DR. 2006. Trends in snowfall versus
rainfall in the Western United States. J. Clim. 19: 4545 –4559.
Lettenmaier DP, Gan TY. 1990. Hydrologic sensitivities of the
Sacramento-San Joaquin River Basin, California, to global warming.
Water Resour. Res. 26(1): 69– 86.
Lins HF, Slack JR. 1999. Streamow trends in the United States.
Geophys. Res. Lett. 26(2): 227– 230.
Luce CH, Holden ZA. 2009. Declining annual streamow distributions
in the Pacic Northwest United States, 1948– 2006. Geophys. Res.
Lett. 36: L16401, doi: 10.1029/2009GL039407.
Mao Y, Nijssen B, Lettenmaier DP. 2015. Is climate change implicated
in the 2013–2014 California drought? A hydrologic perspective.
Geophys. Res. Lett. 42: 2805– 2813, doi: 10.1002/2015GL063456.
Marks D, Kimball J, Tingey D, Link T. 1998. The sensitivity of snowmelt
processes to climate conditions and forest cover during rain on snow:
a case study of the 1996 Pacic Northwest ood. Hydrol. Process. 12:
1569– 1587.
McDonald RI, Girvetz EH. 2013. Two challenges for US irrigation due to
climate change: increasing irrigated area in wet states and increasing
irrigation rates in dry states. PLoS One 8(6): e65589.
Mehran A, AghaKouchak A, Phillips TJ. 2014. Evaluation of CMIP5
continental precipitation simulations relative to satellite-based
gauge-adjusted observations. J. Geophys. Res.-Atmos. 119(4):
1695– 1707.
Mishra V, Cherkauer KA. 2011. Inuence of cold season climate vari-
ability on lakes and wetlands in the Great Lakes region. J. Geophys.
Res.-Atmos. 116: D12111, doi: 10.1029/2010JD015063.
Mote, P. and Salathé, E., 2009. Future climate in the Pacic North-
west (Chapter 1). In The Washington Climate Change Impacts
Assessment, Elsner MM, Littell J, Binder LW (eds). Center for
Science in the Earth System, Joint Institute for the Study of
the Atmosphere and Oceans, University of Washington: Seattle,
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
M. SAFEEQ et al.
WA. http://www.cses.washington.edu/db/pdf/wacciareport681.pdf
(accessed 23 October 2015).
Mote PW, Salathé EP. 2010. Future climate in the Pacic Northwest.
Clim. Change 102(1– 2): 29–50.
Mote PW, Hamlet AF, Clark MP, Lettenmaier DP. 2005. Declining
mountain snowpack in Western North America. Bull. Am. Meteorol.
Soc. 86(1): 39– 49.
Nolin AW, Daly C. 2006. Mapping “at risk” snow in the Pacic North-
west. J. Hydrometeorol. 7: 1164 –1171.
O’Connor JE, Costa JE. 2003. Large oods in the United States: where
they happen and why. Circular 1245. US Geological Survey, Reston,
VA .
Reclamation. 2013. Downscaled CMIP3 and CMIP5 climate and hydrol-
ogy projections: release of downscaled CMIP5 climate projections,
comparison with preceding information, and summary of user needs,
prepared by the U.S. Department of the Interior, Bureau of Reclama-
tion, Technical Services Center, Denver, CO, 47 pp.
Regonda SK, Rajagopalan B, Clark M, Pitlick J. 2005. Seasonal cycle
shifts in hydroclimatology over the western US. J. Clim. 18: 372– 384.
Rupp DE, Mote PW, Bindoff NL, Stott PA, Robinson DA. 2013. Detec-
tion and attribution of observed changes in Northern Hemisphere
spring snow cover. J. Clim. 26(18): 6904–6914.
Safeeq M, Grant G, Lewis S, Tague C. 2013. Coupling Snowpack and
groundwater dynamics to interpret historical streamow trends in the
Western United States. Hydrol. Process. 27(5): 655– 668.
Safeeq M, Grant GE, Lewis SL, Staab B. 2015. Predicting landscape
sensitivity to present and future oods in the Pacic Northwest, USA.
Hydrol. Process., doi: 10.1002/hyp.10553.
Sen PK. 1968. Estimates of the regression coefcient based on Kendall’s
tau. J. Am. Stat. Assoc. 63(324): 1379– 1389.
Serquet G, Marty C, Dulex JP, Rebetez M. 2011. Seasonal trends
and temperature dependence of the snowfall/precipitation-day ratio
in Switzerland. Geophys. Res. Lett. 38: L07703, doi: 10.1029/2011
GL046976.
Shukla S, Safeeq M, AghaKouchak A, Guan K, Funk C. 2015. Temper-
ature impacts on the water year 2014 drought in California. Geophys.
Res. Lett. 42(11): 4384– 4393.
Sproles E, Nolin A, Rittger K, Painter T. 2013. Climate change impacts
on maritime mountain snowpack in the Oregon Cascades. Hydrol.
Earth Syst. Sci. 17(7): 2581–2597.
Stewart IT, Cayan DR, Dettinger MD. 2004. Changes in snowmelt runoff
timing in western North America under a business as usual climate
change scenario. Clim. Change 62(1– 3): 217–232.
Stewart I, Cayan DR, Dettinger M. 2005. Changes toward earlier
streamow timing across western North America. J. Clim. 18:
1136– 1155.
Sureet CG, Tullos D. 2012. Variability in effect of climate change on
rain on snow peak ow events in a temperate climate. J. Hydrol. 479:
24– 34.
U.S. Army Corps of Engineers. 1956. Snow hydrology; Summary report
of snow investigations. U.S. Department of Commerce Ofce of
Technical Services, Washington, D.C., 437 pp.
© 2015 Royal Meteorological Society Int. J. Climatol. (2015)
