Implications for Future Water Resources Management · The Evergreen State College Institutional Repository

SNOWMELT HYDROLOGY OF MT. RAINIER, WASHINGTON, RIVERS:
IMPLICATIONS FOR FUTURE WATER RESOURCES MANAGEMENT

by
Andrew E. Marr

A Thesis: Essay of Distinction
Submitted in partial fulfillment
of the requirements for the degree
Master of Environmental Study
The Evergreen State College
June 2010

This Thesis for the Master of Environmental Studies Degree
by
Andrew E. Marr

has been approved for
The Evergreen State College
by

________________________
Martha L. Henderson
Member of the Faculty

________________________
Date

ABSTRACT
Snowmelt Hydrology of Mt. Rainier Rivers: Implications for Future Water
Resources Management
Andrew E. Marr
High elevation snowsheds on Mt. Rainier are critical sources of snowmelt flows
that provide reliable water and power supplies for the rapidly growing
communities of the South Sound. Recent regional scale analyses indicate that
snowmelt flow quantities are declining and runoff timing dates have been
occurring earlier across the western US since the middle of the 20th century.
Observed trends are spatially coherent over large domains, yet finer grained
basin scale analyses are needed to detect the magnitude of regional trend
signals in Mt. Rainier watersheds. The purpose of this study is to establish
quantitative estimates of snowpack volumes, snowmelt flow quantities and peak
runoff timing dates to investigate temporal trends in Mt. Rainier snow hydrology.
Results indicate that snowmelt flows account for 38.8% to 48.8% of annual flow
in Mt. Rainier Rivers illustrating that snowmelt flows largely dictate flow regimes.
Trend analyses reveal significant declines in spring snowpack volumes and
subsequent snowmelt proportions since the middle of the 20th century and
streamflow timing trends indicate that spring runoff peaks are shifting to earlier in
the year. These findings confirm that observed regional hydroclimatic trend
signals are present at Mt. Rainier and suggest that large scale climatic drivers
are influencing declines in snowmelt parameters. Furthermore, these results
present profound implications to water managers as the stresses placed on water
resources resultant of declines in Mt. Rainier snowmelt will be compounded by
population growth and climate change effects. Therefore, this study provides
valuable insights for future water planning applications and basin scale
hydrologic research in the Pacific Northwest.

Table of Contents
Acknowledgements……….............................................................. vii
1. Introduction and Background…................................................. 1
2. Geography, Climate and Water Resources................................ 7
2.1. Regional Hydroclimate................................................... 7
2.2. Regional Physiography…………………………………... 9
2.3. Nisqually River Basin WRIA 11……………………........ 9
2.3.1. Geographic Features…………………………… 9
2.3.2. Water Resources………………………… ……... 10
2.4. Cowlitz River Basin WRIA 26……………………………. 12
2.4.1. Geographic Features…………………………… 12
2.4.2. Water Resources ……………………………….. 13
2.5. Puyallup, White, and Carbon River Basin WRIA 10…. 14
2.5.1. Geographic Features…………………………… 14
2.5.2. Water Resources.............................................. 15
2.6. Aggregated resource Impacts………………………….. 16
3. Methods………………………………………………………………. 18
3.1. Data …………………………………………………………. 18
3.2. Hydrography ………………………………………………. 19
3.3. Quantification of Snowmelt Flows............................... 20
3.4. Snowmelt Runoff Timing………………………………… 22
3.5. Temporal Trend Analyses………………………………. 24
3.6. PDO Phase State, Snowmelt Flow and Snowpack..... 24
3.7. Synthesis……………………………………………………. 25
4. Results………………………………………………………………… 26
4.1. Hydrography................................................................... 26
4.1.1. Nisqually River.................................................. 26
4.1.2. Cowlitz River……………………………………... 27
4.1.3. Puyallup River …………………………………... 28
4.1.4. White River……………………………………….. 29
4.1.5. Carbon River……………………………………... 30
4.2. Snowmelt Contributions to Aggregated Flows………. 30
4.2.1. Spring Snowpack and Runoff Dynamics…… 30
4.2.2. Aggregated Snowmelt Flow Proportions ….. 30
4.2.3. Aggregated Snowmelt Flow Quantities……... 30
4.2.4. Aggregated Trend Analyses………………….. 33
4.2.5. Long Term Snow Water Equivalent Trends… 34
4.3. Snowmelt Contributions to Basin Flows……… …...... 35
4.3.1. Nisqually River……………………………………35
4.3.2. Cowlitz River…………………………………….. 36
4.3.2. Puyallup River…………………………………… 38
4.3.4. White River………………………………………. 39
4.3.5. Carbon River…………………………………….. 40

4.4. Aggregated Snowmelt Timing…………………………. 40
4.4.1. Aggregate Snowmelt Timing………………… 41
4.4.2. 1958-2006......................................................... 42
4.5. Snowmelt Runoff Timing for Individual Basins…….. 42
4.5.1. Nisqually River…………………………………. 42
4.5.2. Cowlitz River……………………………………. 43
4.5.3. Puyallup River…………………………………… 43
4.5.4. White and Carbon Rivers……………………… 43
4.6. PDO Phase State, Snowmelt Flow and Snowpack.…. 44
5. Discussion……………………………………………………………. 46
5.1. Quantification of Snowmelt Flows……………………... 46
5.2. Snowmelt Runoff Timing………………………………… 47
5.3. Hydrologic trend Analyses………………………………. 48
5.3.1. Spring Snowpack……………………………….. 49
5.3.2. AMJJ Flows………………………………………. 50
5.3.3. Streamflow Timing……………………………… 52
5.3.4. Mt. Rainier Hydroclimate and the PDO……… 54
6. Conclusions………………………………………………………….. 56
7. References…………………………………………………………… 61

List of Figures
Figure 1.1: Map of the study area…………………………………… 8
Figure 4.1: Mean annual hydrograph: Upper Nisqually River….. 27
Figure 4.2: Mean annual hydrograph: Upper Cowlitz River……. 28
Figure 4.3: Mean annual hydrograph: Upper Puyallup River…… 28
Figure 4.4: Mean annual hydrograph: Upper White River………. 29
Figure 4.5: Mean annual hydrograph: Upper Carbon River…….. 30
Figure 4.6: Time series of SWE and aggregated AMJJ flow……. 34
Figure 4.7: Snowmelt flow trend: Upper Nisqually River……….. 36
Figure 4.8: Snowmelt flow trend: Upper Cowlitz River………….. 37
Figure 4.9: Snowmelt flow trend: Upper Puyallup River………… 38

List of Tables
Table 3.1: Data parameters for each watershed…………………... 18
Table 4.1: AMJJ and SWE regression results…………………….. 31
Table 4.2: Summary of aggregated snowmelt flow data………… 33
Table 4.3: Mean annual snowmelt quantities……………………… 35
Table 4.4: Mean snowmelt timing dates……………………………. 41

Acknowledgements
This thesis would not have been possible without the years of encouragement
and expertise of Paul Butler. Additionally, the loving support of my family: Jane
and Mike Campbell, Steve and Mara Marr and Lindsey Marr provided much
needed encouragement. No one endured more stress and direct impacts of this
thesis process and my entire MES degree than Kayla Humiston who provided
endless love and understanding. Mark Lacina, Ray Ruiz, and Matt Lebens
deserve special thanks for keeping me employed and always believing in me.
Graham Parrington and Paul Griffith were there from the beginning and have
always helped me to put things into perspective and come along on field trips to
the study area in mid-winter conditions. Furthermore, I owe special thanks to Kurt
Unger, Peggy Clifford, and Brian Walsh for providing me with invaluable
experience in water resources management through an internship with the Water
Resources Program at the Department of Ecology. I applaud Martha Henderson
for taking me on as a thesis student and am grateful for all her help and guidance.

vii

1. Introduction and Background
Montane snowpack serves as a critical natural storage mechanism
for the water resources of the Pacific Northwest. The regional benefits of
natural storage are exemplified during the arid summer and fall seasons
when precipitation inputs to surface water are limited, and the majority of
streamflow is contributed by the gradual melting of annual snowpack.
Spring snowmelt runoff accounts for substantial portions of annual flow in
several major rivers flowing from the Cascade Range into the Puget
Sound Basin with peak runoff periods occurring from March through June.
The timing and magnitude of snowmelt flows form the basis of
regional reservoir operations in which spring inflows are retained to
provide a reliable water supply for the duration of the summer and fall
months. In addition to ensuring dry season water supply, storage facilities
are operated to optimize flood control, hydropower production, recreation,
and instream flow goals. The abilities of each resource sector to meet
reliability goals hinge upon the timely occurrence of snowmelt influxes;
resource productivity is a function of hydrologic variability. Therefore,
snowmelt runoff flows are essential in providing the communities of the
Puget Sound region with the security of reliable water supplies and low
cost hydropower.
Continued population growth and possible climate change effects
threaten to jeopardize the abilities of current water resource regimes to
maintain efficient operations and attainable reliability goals. In order to
accurately prepare for the associated strains on regional water resources
accurate studies detailing the dynamics of basin scale snow hydrology are
required to develop successful adaptive management strategies to aid in
meeting future demands.

High elevation snowsheds on Mt. Rainier are the primary sources
of snowmelt runoff for water resource sectors supporting the communities
of the South Sound region, including the rapidly growing cities of Tacoma
and Olympia. The Nisqually, Cowlitz, Puyallup, White, and Carbon Rivers
originate on Mt. Rainier and all support regional demands for municipal
water supplies, irrigation water, and hydroelectric generation.

Similar to

other Puget Sound regions, the communities of the South Sound face
challenges arising from population growth and climate change implications.
Thus reliable and accessible data pertaining to the hydrologic
characteristics and temporal variations in the snowmelt flows of Mt.
Rainier Rivers is necessary for future water resource planning efforts in
the South Sound. Therefore, the hydrologic characteristics of each Mt.
Rainier river form the basis of this report.
Increasing concerns surrounding the implications of climate change
and population growth to water resources of the American West initiated
the growth of a distinct hydroclimatic discourse to investigate the interface
between snowmelt flows and hydroclimatic variation. Numerous studies
have analyzed the dynamics of snowmelt hydrology in the Pacific
Northwest (Fountain and Tangborn, 1985; Pelto, 1993; Hamlet et al., 1995;
Miles et al., 2000; Mote, 2003, 2005, 2008; Bach, 2002) and the greater
western United States (Mote et al., 2005; Mote, 2006; Regonda et al.,
2005; Stewart et al., 2004, 2005; Cayan et al. 2001; Dettinger and Cayan,
1995 ) in which quantifications and temporal trend analyses of snowpack,
snowmelt runoff, and spring streamflow timing parameters are central foci.
The majority of research detailing snowmelt runoff hydrology
assesses hydroclimatic variability as a function of historical climate to
identify temporal trends in seasonal snowmelt flows (Hamlet et al., 1995;
Miles et al., 2000; Mote, 2003, 2005, 2008; Mote et al., 2005; Mote, 2006;
Regonda et al., 2005; Stewart et al., 2005). The results of these studies

are regionally coherent and indicate that spring snowpack and snowmelt
flows have been steadily declining across western North America since
the mid 20th century. Mote et al. (2005, 2003) found that spring snow
water equivalencies (SWE) have declined from 10-60% in snowmelt
dominated basins across the West, with the highest observed declines
occurring in Pacific Northwest watersheds (Mote et al. 2003, 2005;
Stewart et al. 2004, 2005; Regonda et al. 2005). Recent analyses
assessing temporal SWE variations in the Washington Cascades indicate
pronounced negative trends corresponding to losses of annual snowpack
ranging from 15-35% of mid 20th century means (Mote et al. 2008). High
trend coherence between large and regional scales suggest that declines
in snowpack are not spatially isolated; spring snowpack volumes are
decreasing across large domains and suggest that snowpack volumes in
Mt. Rainier watersheds are also diminishing.
Subsequent declines in snowmelt runoff (Wahl, 1992; Aguado et al.
1992; Pupacko, 1993; Dettinger and Cayan, 1995; Stewart et al. 2004,
2005; Regonda et al. 2005) coupled with earlier occurrences of annual
snowmelt runoff periods (Stewart et al. 2004, 2005; Regonda et al. 2005;
Cayan et al. 2001) have been observed in hundreds of basins across
western North America. Stewart et al. (2005) found that snowmelt flows
occurring from April to July are decreasing in 81% of 241 rivers sampled in
western North America. Again, the largest decreases were observed in
low elevation basins across the Pacific Northwest.
Regonda et al (2005) and Stewart et al. (2005) have established
statistically significant negative trends for spring snowmelt timing where
the annual snowmelt pulse is occurring earlier in the water year and
observed trends correspond to shifts on monthly scales. Stewart et al.
(2005) sampled the same 241 rivers discussed above and found negative
timing trends for 71% of the rivers where nearly half (49%) are statistically

significant (α=.10) indicating that linkages between flow timing and
snowmelt flow variability are inherent.
Decreasing snowmelt flows and earlier timing pulses are becoming
commonplace across the American West and threaten to decrease water
resource productivity in several snow-dominant systems. Similar to trends
in SWE, snowmelt flow and timing trends exhibit high degrees of spatial
coherence indicating that negative runoff and timing trends are inherently
related to variations in observed seasonal snowpack volumes.
Widespread declines in snowmelt runoff timing and flow quantities across
large spatial domains, and the Pacific Northwest in particular, suggest that
the regionally declining trends in snowmelt parameters are likely occurring
at Mt. Rainier and warrants further localized study.
Increasing temperature trends have been established as the
primary climatic driver influencing declines in snowmelt parameters across
the American West (Miles et al., 2000; Mote, 2003, 2005, 2008; Mote et al.,
2005; Regonda et al., 2005; Stewart et al., 2005), yet uncertainties remain
as to the cause of increasing temperatures, especially in the Pacific
Northwest (Hamlet and Lettenmaier, 1999; Miles et al., 2000; Mote, 2003,
2005, 2008; Mote et al., 2005; Mote, 2006; Regonda et al., 2005; Stewart
et al., 2005). Long-term historical variation in Pacific Northwest
hydroclimatic conditions corroborate well with the fluctuations of the
Pacific Decadal Oscillation (PDO) (Mantua et al. 1997; Hamlet and
Lettenmaier, 1999; Miles et al. 2000; Payne et al. 2004; Mote, 2003, 2005,
2008). The PDO affects regional weather as a result of shifts in sea
surface temperature in the North Pacific Ocean, and alternates between
warm (positive) and cool (negative) phases on a cycle of 20-30 years.
Positive PDO regimes are characterized by warm and dry conditions
whereas negative regimes display cold and moist weather patterns
(Mantua et al. 1997). Subsequently, the effects of PDO greatly influence

snowmelt hydrology at Mt. Rainier on decadal timescales and may
partially mask the effects of climate change. To assess the relationships
between local hydrology and global scale climate further studies detailing
trends associated with the PDO are necessary to assess the sensitivity of
Mt. Rainier snowpack and streamflow to phase state oscillation and large
scale climate variability.
The results of the aforementioned studies clearly indicate that
snowmelt runoff parameters are declining over vast spatial domains and
that profound impacts to regional water resources are imminent if current
trends continue. Providing accurate information to water managers
operating in the Puget Sound region will require fine-grained analyses to
isolate the local effects of regionally significant trends. Therefore, the
purpose of this study is to apply the techniques of previous regional
analyses to the watersheds of Mt. Rainier to quantify the magnitudes of
snowmelt influence for each river and to detect and assess the effects of
observed regional trends on local watershed scales.
In order to produce meaningful results for water management
purposes the analyses included in this study are driven by the following
central questions: (1) How do snowpack, snowmelt, and streamflow timing
parameters interact to shape the annual hydrologic variability for each Mt.
Rainier river? (2) Are regional hydroclimatic trend signals of the past
century present in the hydrologic records of Mt. Rainier rivers? (3) How
does the PDO influence snowmelt hydrology at Mt. Rainier? (4) What are
the implications of hydrologic shifts to water resource sectors that rely
upon Mt. Rainier snowmelt to fulfill resource demands in the South Sound?
The results of this study will provide water managers and
communities in the South Sound region with accurate information detailing
the importance of snowmelt flow, timing, snowpack, and large scale

climatic drivers to local water resource sectors. Findings will aid in critical
adaptation strategy development to combat the implications of population
growth and climate change to water scarcity and demand. Additionally,
this study will provide a firm baseline for further fine-grained studies
pertaining to snowmelt hydrology and water resources of the Pacific
Northwest.
Furthermore, this study is an important contribution to the field of
environmental studies as a whole by integrating the implications of social
and environmental change. Linking variations in hydrologic conditions to
critical resource sectors employs central methods of interdisciplinary
research to arrive at conclusions that are useful to both the natural and
social sciences. Thus, the results of this study will aid in the development
of both environmental and social adaptations to the increasingly
challenging hardships of a changing world.

2. Geography, Climate, and Water Resources
An understanding of watershed characteristics is fundamental to
exploring the importance of snowmelt contributions and runoff timing as
each basin possesses unique physiographic settings and water resource
regimes that respond differently to variations in snowmelt flows. The
purpose of this section is to illustrate the pertinent characteristics of each
watershed in the study area to better understand the susceptibility of both
natural and social systems to snowmelt variability. Each basin is
characterized by Water Resource Inventory Area (WA Dept. of Ecology) to
more efficiently outline water resources activities and provide a helpful
format for future watershed planning applications. A map of the entire
study area with WRIA divisions is provided on page 8.

2.1 Regional Hydroclimate
Pacific Northwest climate is characterized in the Koppen climate
classification system as a Marine West Coast climate with wet and mild
winters and warm and dry summers. However, average annual
temperature and precipitation values within the study area vary
considerably with differences in elevation. All of the watersheds analyzed
in this study have a minimum elevation of sea level and a maximum
elevation of 14,411 ft. at the top of Mt. Rainier. Annual precipitation values
range from 35 inches in the lowlands to more than 150 inches in the
higher mountainous terrain. Temperature also varies as a function of
elevation with warmer temperatures in the lowland valleys and cooler
temperatures in the mountains. Thus, elevation dictates the type of
precipitation (snow vs. rain) and subsequent hydrologic response to
precipitation inputs. Significant snowfall occurs on the slopes of Mt.
Rainier each winter; snowfall records range from 330 inches to 1,122
inches recorded at Paradise. As a result, much of the winter precipitation

in the study area is stored in the snowpack until the spring thaw when
snowmelt is gradually released to rivers and aquifers.

Figure 1.1: Mt. Rainier watersheds, rivers, reservoirs and gauging stations.

2.2 Regional Physiography
All five rivers originate on the flanks of Mt. Rainier and flow
westward to Puget Sound with the exception of the Cowlitz which is a
tributary to the Columbia River. Each basin is split by two physiographic
regions: the Cascade province in the east and Puget Lowlands in the west
(WDNR, 2009). The Cascade province is mountainous with steep
topographic gradients that confine the rivers to narrow forested valleys or
canyons. Hydraulic gradients are high in the upper reaches, especially in
tributary basins. The impacts of human development are scarce as basin
areas in the Cascade province lie in forest lands and Mt. Rainier National
Park. The Puget Lowland is a low elevation basin characterized by gentle
topographic gradients and broad floodplains dominated by agriculture and
urban development. Thus, the majority of socially important implications of
snowmelt variation are encountered in the lower portions of each basin
where water use is abundant.

2.3 Nisqually River Basin: WRIA 11
2.3.1 Geographic Features
The Nisqually River originates on the southern flank of Mt. Rainier
at the terminus of the Nisqually Glacier and flows 78 miles to southern
Puget Sound draining an area of 711 square miles. Headwater tributaries
contribute flow from high altitude snowsheds and glaciers including the
Kautz, Van Trump, and South Tahoma Glaciers. The river flows freely
through the Cascade province in steep walled mountainous valleys until
the hydraulic gradient begins to ease once the river leaves the national
park. The first gauging station downstream of headwaters exists at river
mile (RM) 57.8 near the town of National (USGS 12082500) and serves as
the best gauge for analyzing snowmelt derived flows.

The river becomes impounded by Alder Dam, which forms Alder
Reservoir at RM 44.2. Alder Reservoir serves as a storage facility for
hydroelectric production at Alder Dam and has a storage capacity of
231,900 acre feet. La Grande Dam impounds the river at RM 42.5 and
forms a second storage reservoir, La Grande Reservoir, which is has a
storage capacity of 2,700 acre feet. Storage releases from La Grande
Dam are utilized for power generation through the use of a diversionary
canal. The Nisqually flows freely through the lowlands until RM 26.2 where
the Centralia Canal diverts flows from the river to a power generation
facility 14 miles downstream. For the remaining 12.6 miles, the river flows
through lowland terrain until reaching the Nisqually Estuary, a biologically
rich tide flat area, which is now designated as Nisqually National Wildlife
Refuge.

2.3.2 Water Resources
Hydropower production is the predominant water resource sector
operating within the basin and largely dictates current water management
structures. Cumulatively, the three hydroelectric projects within the
Nisqually Basin produce 573.012 gigawatt hours of electricity annually.
The City of Tacoma owns and operates both Alder and La Grande Dams
which collectively produce 573 gigawatt hours annually to power 40,500
homes (Tacoma Public Utilities, 2010). The third project is a run-of-river
dam owned and operated by the City of Centralia which produces 12
megawatt hours annually. Large scale diversions such as the Alder Dam
have greatly altered the natural flow regimes of the river below the storage
reservoirs, thus flows below the dams more accurately reflect hydropower
production schedules than that of hydroclimatic variation.
Increasing population and urban development within the basin have
resulted in widespread allocations of water for municipal, domestic, and

irrigation uses. As of 2002, there were 938 active water right permits,
applications, and certificates for the lower Nisqually Basin, and 2,677
claims for new allocations (Watershed Professionals Network (WPN),
2002). Of the 2,677 claims 351 are surface water claims with the
remainder requesting groundwater rights. 2,452 claims are for domestic
and municipal consumptive use, 64 for irrigation, and 109 for stock
watering (WPN, 2002). The total allocated amount for the lower basin in
2002 was 63,078 acre feet/year, excluding an additional 58,000 acre
feet/year strictly for hydropower production (WPN, 2002). Major municipal
right holders include the Cities of Olympia, Lacey, Yelm, DuPont, and
Centralia; Tacoma Public Utilities holds nearly all of the hydropower rights.
Consequences of hydropower operations have been dire for
Nisqually salmon, as dams have limited the upstream extent of rearing
habitat and altered sedimentation rates downstream of all hydroelectric
projects. In an effort to curtail the degradation of salmon habitat and to
enhance water availability instream flow regulations are mandated by the
Washington Department of Ecology (WA DOE) and are supplemented by
storage releases required by the Federal Energy Regulatory Commission
(FERC) for hydropower projects. WRIA 11 was one of the first watersheds
to implement instream flow regulations and support the adoption of
instream flow rules through watershed planning efforts; flow regulations
were enacted into law for the Nisqually River and associated tributaries in
1981 and continue to the present, although watershed planning in the
basin is continuously evolving amongst utilities, water right holders, the
Nisqually Tribe and regulatory bodies (WA DOE, 2008).

2.4 Cowlitz River Basin: WRIA 26
2.4.1 Geographic Features
The Cowlitz River originates at the confluence of Clear Fork and
the Ohanepecosh River in northeastern Lewis County and flows
southwest for 133 miles to the Columbia River draining a total area of
2,840 square miles. The Clear Fork drains high mountain terrain and
provides significant snowmelt to the upper Cowlitz, whereas the
Ohanepecosh River originates amongst several tributary glaciers on the
southeastern flank of Mt. Rainier. Major tributary glaciers include the
Cowlitz, Ingraham, Whitman, and Ohanepecosh Glaciers. The first
gauging station on the Cowlitz is near the town of Packwood at RM 126.5
(USGS 14226500) and provides the most accurate flow data for snowmelt
quantification purposes.
The upper Cowlitz flows freely for the first 44.5 miles until the first
impoundment structure, Cowlitz Falls Dam, at RM 88.5. The resultant
Cowlitz Falls Reservoir (Scanewa Lake) is the smallest storage facility in
the basin with a capacity of 11,000 acre feet covering a geographic area
of 610 acres. The second impoundment structure is Mossyrock Dam
which supports Riffe Lake, the basin’s largest storage reservoir. Riffe Lake
extends upstream to RM 65.6 and has a storage capacity of 1,298,002
acre feet and occupies a geographic area of 11,335 acres. The third, and
last, major storage reservoir is Mayfield Lake which is directly downstream
of Mossyrock Dam. Mayfield Lake has a storage capacity of 133,720 acre
feet and a surface area of 2,200 acres. Beyond the major storage
reservoirs the Cowlitz continues southeast towards the Columbia and
accepts the mainstem’s primary downstream tributary, the Toutle River, 20
miles before reaching the Columbia River at the community of Longview.

2.4.2 Water Resources
The Cowlitz River supports the largest hydropower generation
sector in the study area with three major production facilities. The City of
Tacoma owns and operates Mossyrock and Mayfield Dams which
collectively produce 1,904 gigawatt hours annually to supply 136,000
homes in the Tacoma area with low cost electricity (Tacoma Public
Utilities, 2010). Cowlitz Falls Dam is owned and operated by Lewis Public
Utility District (LPUD) and generates 260 gigawatt hours annually to
supply 1/3 of the total electricity demand for the service area of the LPUD.
Both Mossyrock and Mayfield Dams impound large reservoirs which are
controlled to provide flood mitigation and numerous recreation
opportunities such as fishing, camping, and boating. Subsequently, large
scale impoundments on the Cowlitz have resulted in drastic alterations to
natural flow regimes resulting in increased scarcity for downstream uses.
In response, water resource management in the Cowlitz is currently
evolving to accommodate the various needs of multiple sectors including
environmental objectives.
Watershed planning efforts in WRIA 26 have not produced state
regulated instream flow rules despite the fact that quantitative flow
requirements have been proposed by a diverse group of interests in the
current watershed plan (WDOE, 2010). During the writing period of this
study, draft rules existed for WRIA 26 that would set instream flows,
reserves, and basin closures for the watershed. Importantly, the
watershed planning process has yielded a recommendation for closing
much of the basin to further appropriations. Ecology expects to adopt the
final rules for WRIA 26 in September, 2010 (WDOE, 2010). FERC flow
requirements apply to each of the hydroelectric facilities operating in the
basin and provide a source of flow for environmental applications (Tacoma

Public Utilities, 2010; WDOE, 2008). Further state regulated flows will
occur in conjunction with federally mandated FERC flows providing much
needed flow augmentation for ecosystem restoration.
Several consumptive users hold water rights in the Cowlitz Basin.
As of 2010, there were 900 permits and certificates to appropriate surface
water in WRIA 26. Predominant water right classifications include irrigation,
stock water, and domestic uses. Major municipal rights are held by the
cities of Kelso and Longview in which the Cowlitz provides the majority of
the domestic water supplies for each respective city (WDOE, 2010).
Increases in population will ultimately lead to an increase in water
demands throughout WRIA 26 and add further tension between
hydropower entities, water right holders, and environmental interests.

2.5 Puyallup, White, and Carbon River Basins: WRIA 10
2.5.1 Geographic Features

WRIA 10 (Puyallup-White) includes the watersheds of the Puyallup,
White and Carbon Rivers in a single planning unit. All three rivers
converge in the Puget Lowland near Tacoma before entering Puget Sound
through a common mouth. WRIA 10 encompasses an area of 948 square
miles that drains the entire northern half of Mt. Rainier. The White River
originates at the termini of the Emmons and Inter Glaciers and flows
northwest for 68 miles draining 464 square miles. The Puyallup River
flows northeast from the Puyallup Glacier for 45 miles until draining into
Puget Sound at Tacoma. The Puyallup River basin covers 405.1 square
miles of WRIA 10. The Carbon River is the shortest river in WRIA 10
running 30 miles from the terminus of the Carbon Glacier where it meets
the Puyallup near Orting. The Carbon drains 78.9 square miles, most of

which is forest land including one of the only remaining temperate
rainforest stands in the Southern Cascades.
Impoundments have been constructed in each basin with the
largest structures in place on the White and Puyallup Rivers. Mud
Mountain Dam on the White River provides flood control for the
communities in the lower reaches of WRIA 10 and is owned and operated
by the US Army Corps of Engineers. The White River flows for
approximately 10 miles past Mud Mountain Dam where up to 2,000 cubic
feet/second of White River flows are diverted to form Lake Tapps reservoir
with a storage capacity of 46,700 acre feet (Cascade Water Alliance,
2010). The Electron Dam in the upper reaches of the Puyallup Basin
diverts river flows for hydroelectric production although the project does
not impound flows. The largest anthropogenic alteration in WRIA 10 is the
straightening of the Lower Puyallup near Tacoma where the river is
contained between concrete walls to prohibit meander and flood damage
and to improve the navigability of the mouth.

2.5.2 Water Resources
WRIA 10 has been a regionally important watershed for decades
due to its close proximity to major population centers. High demands for
hydropower, consumptive uses and irrigation have resulted in severe
over-allocations which necessitated a basin closure in 1980. Legally (WAC
173-510), no further water appropriations are to be made to new users in
WRIA 10 unless obtained through the procedures of transfers and leases
(WDOE, 2008) Therefore, further development in the basin is hindered by
the relative unavailability of water resources and further growing water
scarcity.

Predominant water resource sectors operating within the basin are
domestic and municipal water suppliers, irrigation, and hydropower. There
are currently 1211 total water rights issued in WRIA 10 with 687 rights to
surface water bodies (WDOE, 2010). Municipalities hold the rights to the
largest volumes of water and include Auburn, Orting, Puyallup, Bonney
Lake, and Tacoma making WRIA 10 one of the largest sources of drinking
water in the South Sound. In comparison to the other WRIAs in the study
area, the White-Puyallup basin is not a significant source of hydropower
with only one major project. The Electron Dam owned by Puget Sound
Energy is capable of meeting peak electricity demand for 17,000 regional
homes (PSE, 2009).
Despite the basin closure, progress is being made to increase the
municipal water supplies in WRIA 10. In 2004, Puget Sound Energy (PSE)
ceased power production at Lake Tapps and subsequently sold the
storage facility and associated water rights to Cascade Water Alliance
(CWA), a non-profit water supplier. CWA plans on utilizing stored White
River flows in Lake Tapps to secure a water supply for the next 60 years
for 8 communities serving 370,000 people (CWA, 2010; WDOE, 2006).
Therefore, the implications of snowmelt variability to WRIA 10 water
resources are stark, especially in regards to further water supply
development.

2.6 Aggregated Resource Impacts
Snowmelt flows from Mt. Rainier dictate the productivity of water
resource operations for a large geographic area stretching from Tacoma
to the Columbia River. Cumulatively, hydroelectric projects in the study
area provide 193,500 regional homes with power supplies, and Mt. Rainier
runoff supports 176 municipal water supply rights. Additionally, each river
fulfills numerous domestic, irrigation, and environmental demands

throughout each WRIA. Basin closures in WRIA 10 and near certain
closures in WRIA 26 provide visible evidence that population growth is
already beginning to strain current water resource availability in the Puget
Sound region, and with an expected population of nearly 7.7 million
people by 2010 (WOFM, 2007) it is imperative that resource management
adaptation begin to address the challenges of future demands.
Furthermore, population effects are compounded by uncertainties
introduced from climate change effects. Therefore, impacts to snowmelt
hydrology need to be understood in order to accurately prepare water
managers for the implementation of adaptive measures.

3. Methods
The following analytical methods were chosen due to their high
degree of applicability to the central research questions of this study.
Importantly, the techniques used here have been accepted in the greater
discourse and provide a means to detect regional hydrologic trend signals
at Mt. Rainier. Secondly, the methodologies employed herein allow for
reliable estimates of snowmelt flows and timing to be calculated and easily
transferred to water managers for planning applications.
3.1 Data
Flow data for each river was obtained from the United States
Geological Survey (USGS) surface water data network. Discharge data
from USGS is reported and analyzed as monthly mean discharge values
measured in cubic feet per second (cfs). Gauging stations were selected
to achieve maximum accuracy for snowmelt flow estimates and therefore
were required to be upstream of all obstructions and diversions and be
proximal to snowmelt sources. Each gauging station for all five basins is
the first station downstream of headwaters, and exist above all major
diversions or impediments to natural flow. A comprehensive list of gauging
station attributes is given in table 3.1. After gauge selection and data
importing was complete, each dataset was re-organized in terms of water
year (October-September). For example, water year 2007 began in
October of 2006 and concluded in September of 2007.

River

USGS Station

USGS Station ID

Drainage Area (mi.2)

Period of
Record

Nisqually

National

12082500

133

1944-2006

White

Greenwater

12097000

216

1930-1975

Puyallup

Electron

1209200

92.8

1957-2008

Cowlitz

Packwood

14226500

287

1948-2008

Carbon

Fairfax

1209400

78.9

1930-1977

Table 3.1: Data parameters for each watershed

April 1st snow water equivalency (SWE) data was obtained from the
Natural Resource Conservation Service for the SNOTEL station located at
Paradise. The Paradise station was selected as it is the only accurate and
regularly maintained snowpack station in the study area. SWE is reported
in inches of liquid water equivalent and all analyses herein remain in units
of inches. Monthly mean temperature and precipitation data for each
gauging station location was gathered to enhance understanding of basin
hydrography and compare climatic variation in the study area. Additionally,
data was obtained for Tacoma, the study area’s low point, and for
Paradise (National Park Service, 2007) to illustrate the magnitude of
spatial variation on a large scale. Precipitation is reported in inches and
temperature is recorded in degrees Fahrenheit.
Prior to any analysis, all datasets were screened for outliers and
normality. No severely anomalous river flow, SWE, or climatic data points
were observed and flow density curves well approximate a normal
distribution. Assumptions of normality were confirmed with histograms,
quantile plots skewness tests and kurtosis tests.

3.2 Hydrography
The first step in identifying the magnitude of snowmelt influence for
a watershed is to understand the major hydrographic characteristics of the
basin. Hydrographs for each river were created using mean monthly flow
data in order to assess annual patterns of flow. Precipitation data for each
gauging station area was plotted along with the hydrograph to gain
insights into the average hydrologic response of each river in regards to
average precipitation. Once hydrographs were completed, dominant
hydrologic regimes were delineated for each basin as either transient or
snow-dominant systems. Rivers displaying two prominent runoff peaks (bi-

modal) were delineated as transient and rivers displaying only one runoff
peak in the spring (right skewed) were delineated as snowmelt dominated.

3.3 Quantification of Snowmelt Flows
In order to accurately estimate snowmelt contributions to annual
flow, all discharge recorded during the months of April, May, June, and
July (AMJJ) is considered to be of snowmelt origin. Although the use of
AMJJ flows as proxies for snowmelt flow is quite general and inherently
noisy with non-snowmelt inputs (rainfall and baseflow) its use here is
highly appropriate. Firstly, studies that are highly relevant to this paper
utilize AMJJ fractional flows to delineate snowmelt contributions (Stewart
et al. 2004, 2005). Secondly, the hydroclimate of the study area lends
itself well to the use of AMJJ as substantial spring snowpack, perennial
snowfields, and glacier ice likely prolong the snowmelt runoff season at Mt.
Rainier. Additionally, low summer precipitation strictly limits non-snowmelt
sources from entering surface water systems.
To alleviate some of the uncertainties associated with the use of
AMJJ fractions, dynamics between snowpack and AMJJ flow were
addressed. Relationships between April 1st SWE and AMJJ flows were
assessed through linear regression analyses for each river using the
following equation:

y AMJJ = b 1 X SWE + ε 1SWE
Here y AMJJ is the mean annual AMJJ flow for each river, b 1 is the
linear coefficient corresponding to April 1st SWE , and ε 1SWE represents
noise introduced by non-snowmelt derived inputs. Regression results are
utilized to determine how well spring snowpack volumes correlate with
AMJJ flow and therefore provided a basis for validating the use of AMJJ

flows as an effective proxy for estimating snowmelt contributions.
Furthermore, SWE and AMJJ regressions provided insights into major
sources of flow variation for each basin.
Two metrics are applied to snowmelt flow estimates: proportions of
annual flow, and AMJJ discharge volumes. Proportions represent the
fraction of mean annual flow that was observed during the AMJJ period,
and results are given in terms of percentage of mean annual flow. AMJJ
flow volumes were calculated by converting mean monthly discharge
values to acre feet using the following algorithm:
cfs  cfd (cfs x 86,400)  cf/period  af/period ((cf/period)/43,560)  af AMJJ
af AMJJ = Σaf /AMJJ

where:
cfs= cubic feet per second
cfd= cubic feet per day
86,400 = seconds per day
cf/period = cubic feet per month
43,560 = ft3 per acre feet
af/period = acre feet per month
af AMJJ = the sum of af/period for AMJJ
Quantifying river flows in terms of proportions and volumes
provides for a more in depth analysis of temporal trends. AMJJ
proportions and volumes are distinctly different measures in which
inconsistencies in their coherence can be effectively used to identify shifts
in hydroclimate (Bach, 2002). Secondly, AMJJ fractional flows were
converted to acre feet to increase the applicability of the findings of this
study to water resource management applications.
Snowmelt contributions are estimated for individual basins and for
the study area as a whole (aggregate). Aggregated flows were calculated
to establish a baseline estimate of total mean annual snowmelt
21

proportions and volumes originating from Mt. Rainier snow sources. Data
gaps in flow records for each river necessitated the analysis of three
differing datasets in order to obtain accurate estimates of aggregated
snowmelt volumes. The three datasets detail the following time periods:
1958-1975, 1958-2006, and 1992-2006. The 1958-1975 period is the only
period when all five rivers have consistent records and was selected for
the purpose. The 1958-1975 period serves to provide a basic baseline
estimate for the entire aggregate. The 1958-2006 time period was
selected as it is the longest period of record with the most rivers
consistently represented (Nisqually, Puyallup, and Cowlitz). The latter
period (1992-2006) provides the best estimate of current mean snowmelt
flows; all rivers besides the White are represented. Additionally, the same
methodology for quantification of snowmelt flows discussed above is
applied to each river individually in order to illustrate the importance of
snowmelt to each basin. All results are reported with a 95% confidence
interval (α =.05) with the exception of the Carbon and White Rivers in
which means are reported with a 90% confidence interval (α=.10).
3.4 Snowmelt Runoff Timing
Snowmelt runoff timing is quantified for both the aggregate and
individual basins in order to establish mean snowmelt timing dates and to
provide a basis for temporal trend analyses pertaining to the seasonality of
AMJJ fractional flows. A majority of studies analyzing trends in streamflow
timing throughout the western United States (Stewart et al. 2004, 2005;
Regonda et al. 2005; Cayan et al. 2001; Dettinger and Cayan, 1995; Wahl,
1992) employ calculations of snowmelt pulse onset date (Cayan et al.
2001) or center of mass flow timing date (CT) (Stewart et al. 2005, 2004)
to assess snowmelt timing. The calculation of CT date was selected for
use in this analysis to be consistent with similar studies of western United
States runoff timing (Stewart et al. 2005, 2004) and for it’s compatibility
with the AMJJ proxy.

CT is defined as the date marking the center of mass of annual flow
for each water year (Stewart et al. 2005), or the date when half of the flow
for each water year has occurred. Therefore, CT dates do not directly
pinpoint snowmelt runoff timing dates with the accuracy of the Cayan et al.
(2001) onset pulse algorithm, although high correlations ( r = 0.5 – 0.8)
have been established between the two measures (Stewart et al. 2005).
Stewart et al. (2005) also find that the CT date provides a “time integrated
perspective of the timing of this pulse and the overall distribution of flow
for each year, and is less noisy than the spring pulse onset date” (p.1139).
Given the empirical success of the CT date and applicability to monthly
datasets, it is clear that the use of CT methodology will yield the most
accurate results for Mt. Rainier rivers. The CT date is calculated from the
following equation:

CT = ∑(t i q i ) / ∑q i

Here t i is the time in days since the beginning of the water year
(wyd), and q i is the corresponding streamflow for month i. The CT
equation was adopted from Stewart et al. (2004, 2005).
Streamflow timing for the aggregate was calculated using the time
periods in section 3.3 to correct for inconsistencies in flow data. The
aggregated mean CT date for the entire period of record (1930-2008) was
calculated by compiling every CT date for each river during their
respective periods of records. Once all CT dates were compiled, the
average date was calculated to give the long-term mean aggregate CT
date. Snowmelt timing for each individual basin was calculated by
applying the CT equation to each river’s flow records. Mean annual CT
dates were then calculated from the results of the CT equation and
reported in units of water year day (wyd).
23

In an effort to better understand the relationship between snowmelt
timing (CT) and snowpack, linear regressions were performed on
corresponding CT and SWE time series to assess CT trend significance
as a function of SWE variation. The following regression was performed
for each river’s historical record:

y CT = b 1 X SWE + ε 1SWE
Here y CT is the mean annual CT date for each river, b 1 is the
linear coefficient corresponding to April 1st SWE , and ε 1SWE represents
noise introduced by non-snow variables.

3.5 Temporal Trend Analysis
Hydrologic trends are of great interest to water resource managers
and as of late are receiving increasing attention by climate researchers. In
response to this increase, temporal trend analyses were employed to
assess changes over time in snowmelt contribution, overall water year
discharge, snowmelt timing (CT), and SWE. Trends in snowmelt
contribution, annual discharge, CT, and SWE well approximate linear
functions, thus simple regression analyses were utilized to gain insights
into the significance of temporal variations. Trend significance is assessed
through the use of t and F tests at the 95% confidence level (α = .05).
3.6 PDO Phase State, Snowmelt Flow and Snowpack
Numerous studies have identified PDO phase state as a significant
driver of snowmelt flow and SWE variation (Hamlet et al., 1995; Miles et
al., 2000; Mote, 2003, 2005, 2008; Stewart et al., 2004, 2005, Regonda et
al., 2005).To investigate the influence of the PDO on Mt. Rainier
hydroclimate, two-sample t-tests are employed to investigate the

hypotheses that there are significant differences (α=.05) in snowmelt
runoff and SWE volumes respective of PDO phase state. The null
hypothesis states that there is no difference in flow or SWE volumes due
to phase state alterations. Oscillations of the PDO are categorized as
either positive or negative; the last negative phase occurred from 19471976 and the latest positive phase occurred from 1976-1998 (Mantua et al.
1997). The Nisqually and Cowlitz datasets were the only records that
contained enough data points to span both phases and were the only two
rivers analyzed for PDO related differences. Flow data and SWE records
for each watershed were categorized by phase state prior to analysis.

3.7 Synthesis
Although each technique described in this section provides
important information on particular hydrologic parameters, the combination
of techniques acts to synthesize sources of variation resulting in a concise
and convenient analysis of Mt. Rainier snow hydrology. Therefore, the
results of this report will provide water managers with a versatile tool for
planning purposes in which data for individual basins and parameters may
be isolated or integrated to suit further research needs.

4. Results
4.1 Hydrography
Regional rivers are classified as rain dominant, transient, or snow
dominant systems. Rain dominant systems occur in low elevation basins
where the primary input to surface water is rain and hydrologic response is
initiated almost immediately following a precipitation event. Hydrographs
of rain dominated systems display a characteristic peak flow period during
the winter months when regional precipitation is at a maximum and exhibit
a left skewed flow distribution. Transient (rain and snow dominated)
systems occur at moderate elevations where precipitation type may
alternate between snow and rain several times a year. Typically,
accumulated snow does not linger for the duration of the winter as
snowpack and is input to the river system several times as temperatures
fluctuate. Transient hydrographs are typified as having two peaks: one in
the winter (rain) and another in the spring (snowmelt) resulting in a bimodal flow distribution. Snow dominant systems occur at high elevations
where winter temperatures remain below freezing and snowfall acts as the
primary precipitation type. Subsequently, runoff is delayed as the majority
of winter precipitation is stored in the snowpack until the snowmelt begins.
Snow dominant hydrographs display peak flows occurring during the
spring and summer months and follow a right skewed flow distribution.

See Also

2008 Summary (PDF) - [PDF Document]

Month

Jul

Aug

Sep
Discharge
Precipitation

Figure 4.5: Mean annual hydrograph: Upper Carbon River

4.2 Snowmelt Contributions to Aggregated Annual Flow
4.2.1 Spring Snowpack and Runoff Dynamics
Regression analyses were utilized to investigate the relationships
between spring runoff and annual snowpack for all five rivers. Relatively

large distances between gauges and snowmelt sources exist for each
river in the study area which inherently introduces noise to each snowmelt
dataset from non-snowmelt inputs. Therefore, regression analyses
provided insights into how much variability in each streamflow record can
be explained by variations in annual snowpack quantities. Additionally,
regression results serve to further clarify the effectiveness of using AMJJ
fractional flows as proxies for estimating overall snowmelt contributions.
Aggregated AMJJ discharge was regressed on April 1st snow water
equivalency (SWE) data recorded at Paradise. Regression results indicate
very strong relationships between April 1st SWE and AMJJ discharge for
each river. SWE coefficients for each regression are statistically significant
(α=.05) with R2 values ranging from .42 to .79, indicating that substantial
portions of variation in AMJJ flows for each river can be explained by
snowpack variability (Table 4.1).
River

p(t)

p(f)

Adj.R2

Nisqually
Cowlitz
Puyallup
Carbon
White

10.017
15.3
6.79
4.96
8.29

<.001
<.001
<.001
<.001
<.001

233.2
233.2
46.1
24.6
68.75

<.001
<.001
<.001
<.001
<.001

0.64
0.79
0.49
0.42
0.69

Table 4.1: AMJJ and SWE regression results.

Paradise April 1st SWE best describes flow variation for the
Nisqually, Cowlitz, and White Rivers during the snowmelt runoff period.
Paradise is located proximal to the watershed divide for the Nisqually and
Cowlitz hence the strong explanatory power for each respective river. The
high correlations affirm that there is minor variability in snowpack
formation in the study area allowing Paradise SWE data to serve as
accurate proxies for estimating annual snowpack influences for all 5
watersheds.

4.2.2 Aggregated Snowmelt Flow Proportions
Estimates of total snowmelt runoff proportions from Mt. Rainier
were calculated using three datasets. The first dataset includes records
from all five rivers from 1958-1975 which is the only time period when
concurrent records exist for all five watersheds. The second dataset
includes records from 1958-2006 and utilizes flow records from the
Nisqually, Cowlitz, and Puyallup Rivers. This time period was selected for
analysis as it provides the longest time period of consistent records for the
greatest number of rivers. The third dataset takes advantage of renewed
data collection for the Carbon River and includes records for each river
with the exception of the White, from 1992-2006. Estimates derived from
the 1992-2006 dataset provide the best insights into modern snowmelt
quantities. Mean aggregated snowmelt quantities during the 1958-1975
period are 44.78% ±2.8% (α=.05) and range from 33.3% to 55.47% of
annual flow. Aggregated snowmelt contributions from 1958-2006 are
nearly identical to that of the complete aggregate estimate with mean
annual snowmelt flows of 44.80% ±2.82% (α=.05). However, snowmelt
flows during the 1958-2006 period are more variable ranging from 25.46%
to 67.69% of annual flow. Mean annual snowmelt contributions during the
1992-2006 period are slightly lower at 41.58%±3.61% (α=.05) and range
from 25.02% to 51% of annual flow.
4.2.3 Aggregated Snowmelt Flow Quantities
Aggregated snowmelt volumes were quantified to better extend
these findings to water resources applications. Therefore, all snowmelt
flow quantities were converted from cubic feet per second (cfs) to acre
feet (AF). The mean snowmelt volume in Mt. Rainier rivers from 19581975 is 1.42 million acre feet ± 135.89 acre feet (α=.05) and ranges from
867,621 to 1,990,147.2 acre feet. Meltwater quantities for the 1958-2006
period range from 531, 260 to 1,420,317 acre feet with a mean volume of

895,551.63 acre feet ± 55,128 acre feet (α=.05). Mean snowmelt volume
for the recent period (1992-2006) is 969,292 acre feet ± 111,087.5 acre
feet (α=.05). Mean annual snowmelt proportions and quantities for each
time period are summarized in Table 4.2.

Period

Rivers

Annual Snowmelt Proportion

Quantity
(af)

1958-1975

N,C,P,Ca. W

44.78%

1,419,738

1958-2006

N, C, P

44.80%

895,551

1992-2006

N,C,P,Ca.

41.58%

969,296

Table 4.2: Summary of aggregated snowmelt flow data. N= Nisqually; C= Cowlitz;
P= Puyallup; Ca. = Carbon; W= White.

4.2.4 Aggregate Trend Analysis
Temporal trend analyses were conducted for the 1958-2006 period
in order to identify significant changes in snowmelt flow quantities,
proportions and overall annual (12 month water year) discharge. Trend
analyses were also conducted for April 1st snow water equivalencies
(SWE) at Paradise in an effort to corroborate trends in snowmelt runoff
with SWE variability. 1958-2006 aggregated snowmelt flows are highly
correlated with Paradise SWE (R2 = .789, t= 13.35, p=<.001).
Subsequently, temporal trends in SWE and snowmelt flows are highly
coherent (Fig. 4.6).
Snowmelt quantities and 1958-2006 Paradise SWE were found to
be declining over time at similar rates, which is shown graphically in
Figure 4.6. Although clear negative trends are observed for both SWE and
runoff, neither trend is statistically significant at the 95% confidence level:
SWE (t= -1.35, F= 1.83, p = .18), aggregated snowmelt flow (t= -1.45, F=
2.1, p = .154). However, aggregated mean annual snowmelt proportions
were found to be significantly decreasing (t= -2.194, F= 4.815, p=.0332)
suggesting increases in non-snow precipitation inputs over time.

Importantly, overall annual discharge for the aggregate was found to be
trend neutral with no significant gains or losses (p= .174). The potential
causes and implications of significantly declining proportions of annual
flow in relation to trivial fluctuations in overall annual discharge are
addressed in Chapter 5.

120

25000

100

20000

15000

60
10000

5000

20
0
1958

Discharge (cfs)

SWE (in)

Paradise April 1st SWE and Aggregated Snowmelt
Discharge: 1958-2006

0
1963

1968

1973

1978

1983
Year

1988

1993
SWE

1998

2003
AMJJ Flow

Figure 4.6: Time series of SWE and aggregated AMJJ flow.

4.2.5 Long Term Snow Water Equivalent Trends
Paradise April 1st SWE records date back to 1944 and extend to the
present, therefore longer term trend analyses detailing SWE variability
were possible. Results indicate that spring snowpack totals have been
significantly (α=.05) decreasing for the entire duration of record (t= -2.62,
F= 6.846, p= .01). These findings are consistent with decreases in both
mean annual snowmelt quantities and snowmelt flow proportions outlined
above.

4.3 Snowmelt Contributions to Annual Flow in Individual
Basins
Snowmelt quantities in all five rivers were calculated individually to
examine runoff variability on basin scales. Mean annual snowmelt flows
were estimated in terms of proportions of mean annual discharge and
through the calculation of AMJJ flow volumes. Temporal trend analyses
were applied to quantification results in order to identify trends in runoff
over time and provide insights into possible spatial variation in the study
area. Table 4.3 provides snowmelt proportions and flow quantities for
each river:

River

Mean Annual Snowmelt
Proportion

Mean AMJJ Flow
Volume (af)

Nisqually

39.8%

221,394

Cowlitz

49.1%

564,744

Puyallup

39.0%

108,492

White

48.6%

304,114

Carbon

41.6%

127,380

Table 4. 3: Mean annual snowmelt quantities for Mt. Rainier Rivers.

4.3.1 Nisqually
In general, snowmelt contributions to the Nisqually River are low
compared to the other rivers in the study area. The average proportion of
snowmelt flows in the upper basin is 39.8%±1.5 %( α=.05) of annual flow
at National, and ranges from 24.2% to 55.1% (n=61, σ=6.3) throughout
the period of record. Therefore, snowmelt flows rarely account for more
than half of total annual flows, results which are consistent with the
transient hydrography of the basin. Mean annual snowmelt volumes range
from 131,582 acre feet to 332,406 acre feet (σ=45,532) and average
221,394 ± 11,243 acre feet.

Nisqually River trend analyses indicate that both snowmelt
proportions and AMJJ volumes are steadily decreasing over time, and
represent one of this study’s key findings. Annual proportions of snowmelt
flow were found to be significantly decreasing (t= -2.28, F= 5.22, p= .026)
concurrently with significant decreases in AMJJ flow quantities (t= -2.19,
F= 4.824, p= .032). Trends are shown graphically in Figure (4.7).

Upper Nisqually River Snowmelt Time Series:
1944-2006

325

275

225

175

125

75
1944

Proportion of
Annual Flow (%)

Thousand Acre Feet

375

0
1954

1964

1974

1984

1994

2004

Year

Volume
Proportion

Figure 4.7: Snowmelt flow trend: Upper Nisqually River

4.3.2 Cowlitz
The Cowlitz River comprises of the largest snowmelt dominated
basin originating on Mt. Rainier and contributes substantial spring and
early summer flows to regional hydrology. Mean annual snowmelt
contributions account for 49.1%±1.7% (α=.05) of mean annual flows at
Packwood. Snowmelt proportional flows display a high degree of
variability with annual proportions ranging from 26.4% to 71.4% (n=79, σ=
8.89) of annual discharge. These figures corroborate well with the
observed snow dominant nature of Cowlitz hydrography by essentially
accounting for half of the total annual discharge for the upper reaches of
the basin. Additionally, the large range and high variance of proportional
values are indicative of sensitive climatic response, especially in regards
36

to spring snowpack; correlations between Paradise SWE and Cowlitz
discharge are the highest in the study area (R2= .79). Mean AMJJ
discharge volumes range from 280,504 acre feet to 908,592 acre feet
(n=79, σ= 150,443) and average 564,744 acre feet ± 33,175 acre feet
(α=.05).

1000

900
800

700
600

500

400
300

200
100

19
3
19 0
3
19 5
4
19 0
45
19
5
19 0
5
19 5
6
19 0
6
19 5
70
19
7
19 5
8
19 0
8
19 5
9
19 0
95
20
0
20 0
05

Proportion of Annual
Flow (%)

Thousand Acre Feet

Upper Cowlitz River Snowmelt Timeseries (1930-2008)

Year

Volume
Proportion

Figure 4.8: Snowmelt flow trend: Upper Cowlitz River

Temporal trend analyses for the Cowlitz River are similar to those
detailing Nisqually River flow as both snowmelt fractional flows and melt
season discharge quantities are steadily decreasing over time. Again,
these findings are amongst the most important observations in this study.
Snowmelt proportions are significantly (α=.05) decreasing (t= -2.27,
F= 5.15, p=.026) resultant from significant declines in snowmelt season
discharge volumes (t= -2.03, F=4.12, p=.04), which are displayed
graphically in Figure 4.8. The similarities in temporal trend behavior
between the Nisqually and Cowlitz Rivers raise questions about spatial
influences on long term trends primarily due to the close proximities of
their respective high elevation snowsheds, and the fact that both rivers are

the only systems displaying statistically significant negative trends in the
study area.
4.3.3 Puyallup
The Upper Puyallup River contributes the least amount of snowmelt
flows to the region in terms of both snowmelt proportion and quantity.
Snowmelt flow proportions range from 24% to 53.8% of annual flow (n= 51,
σ= 5.77) with an average snowmelt fraction of 38.96%±1.58 (α=.05). Mean
annual snowmelt quantities range from 76,596.1 acre feet to 151,123 acre
feet (n=51, σ= 15839.4). The mean annual snowmelt volume for the basin
is 108,492 acre feet ± 4,347.2 acre feet. The aforementioned
quantification results are consistent with a transient dominated hydrograph.

160

140

50
40

120

30
100

Proportion of Annual
Flow (%)

Thousand Acre Feet

Upper Puyallup River Snowmelt Time Series

60
0
19581963 19681973 197819831988 19931998 20032008
Year

Quantity
Proportion

Figure 4.9: Snowmelt flow trend: Upper Puyallup River

The Puyallup River is also experiencing declines in mean snowmelt
proportions and quantities over time, although, trend analyses indicate
that the decline of Puyallup snowmelt flows is not as pronounced as those
identified for the Nisqually and Cowlitz (Fig. 4.9). Statistically significant
negative trends were not identified for snowmelt proportions (t= -.637,
F= .406, p=.527) or snowmelt quantities (t= -.202, F= 1.03, p= .841).

Puyallup snowmelt quantity coefficients regressed on time are nearly zero
(-30.74) indicating that snowmelt flow quantities are essentially stagnant
and are not experiencing any significant deviations from observed long
term means. Variability in snowmelt proportion accurately follows quantity
fluctuations further elucidating the stability of temporal snowmelt trends for
the Puyallup River.
4.3.4 White
Incomplete records for the upper reaches of the White River
prevented analysis beyond 1975. However, observed long term trends in
similar basins are not strong enough to drastically alter current means.
Therefore, taking into consideration observed trends, the assumption is
made that White River flows have not increased or decreased by
magnitudes that render available White River data useless for snowmelt
estimation purposes. In order to help mitigate for temporal uncertainties,
all means for the White River are reported with a 90% confidence interval.
No time series regressions were utilized due to the shortness of the
available record and inapplicability to current and future water resource
goals.
The White River is the second largest river originating on Mt.
Rainier in terms of drainage area (216 mi.2) and snowmelt contribution.
Annual snowmelt quantities range from 186,653 acre feet to 452,818 acre
feet (n= 45, σ= 59,106.8) with an average of 304,114 acre feet ± 23,614
acre feet. 48.6% ± 2.73% of annual flow is attributable to snowmelt inputs
where Snowmelt proportions range from 28.26% to 65.31% of mean
annual flow (n=45, σ= 6.84). The high degree of variability is likely due to
annual variations in snowpack volumes as the White River is highly snow
dominant; correlation of flow quantity to Paradise SWE is the second
highest in the study area (R2= .69). Close similarities between the White

River and the Nisqually and Cowlitz Rivers in terms of snowpack
correlation suggest that White River snowmelt trends are likely decreasing.

4.3.5 Carbon
Accurate flow records for the upper reaches of the Carbon River
ended in 1977 therefore trend analyses were not conducted for Carbon
River snowmelt proportions or quantities. Trend strengths of neighboring
rivers were again used to support the assumption that current Carbon
River flows are not substantially different from that of the existing period of
record even if temporal variations are occurring. As in the case with the
White River, Carbon flows are reported with a 90% confidence interval.
Additionally, the Carbon River’s low correlation with Paradise SWE
(R2= .42) compared to the higher correlations of rivers experiencing
negative trends suggests that Carbon River flows are not experiencing
significant declines in snowmelt flow and are likely similar to trends
observed for the Puyallup system.
The Carbon River Basin is the smallest catchment studied (78.9 mi.2) and
total snowmelt contributions rank as the second smallest in the study area.
The mean annual snowmelt proportion is 41.55% ± 2.65% of annual flow
and is consistent with the basin’s transient hydrography. Carbon River
snowmelt proportions range from 24.09% to 58.42% (n=48, σ=6.84),
which serves to supplement the aforementioned assumption regarding
snowmelt and streamflow relationships. Snowmelt flow quantities range
from 183,843 acre feet to 76,011 acre feet (n= 48, σ= 23,943) with an
average annual volume of 127,380 acre feet.

4.4 Aggregated Snowmelt Runoff Timing
The timing of snowmelt flows are critical for sustaining water
resource reliability goals and are therefore equally as important as

snowmelt flow proportions and quantities for resource planning purposes.
For the purposes of this study, center of mass flow timing dates (CT) are
calculated for use as a proxy in identifying the starting date of the
snowmelt season. Snowmelt timing dates are given for the study area as a
whole (aggregate) and for each individual basin. Temporal trend
regressions were performed for the aggregate dataset (1958-2006) and
for each basin to identify shifts in snowmelt peaks over time. CT variation
was also analyzed in terms of SWE variability to indentify linkages
between CT shifts and spring snowpack volumes.

4.4.1 Aggregated Snowmelt Timing
In order to accurately estimate the beginning of the snowmelt
season for Mt Rainier rivers four datasets were utilized to capture CT
dates. The same datasets used to calculate aggregated snowmelt
proportions and quantities are used for snowmelt timing estimation in
order to alleviate uncertainties due to inconsistent records (see section
4.2.2 for explanation). Aggregated CT results are given in Table 4.4.
Period

Rivers

Mean CT (d)

Date Range (α=.05)

1930-2008
1958-1975
1958-2006
1992-2006

All
N,C,P,Ca. W
N, C, P
N,C,P,Ca.

201 ± 2.5
205 ± 4.2
197 ± 3
192 ± 5.2

April 17-April 23
April 20-April 28
April 13th-April 19
April 6- April 16

Figure 4.4: Mean snowmelt timing dates for Mt. Rainier Rivers

Mean CT dates indicate that the annual snowmelt pulse begins in
early to late April and has been decreasing over time. During the 19581975 period, in which consistent records for all five rivers exist, snowmelt
flows began in late April. In contrast, during the 1992-2006 period in which
all rivers except the White are included, mean CT dates occur no later
than mid-April. Although the White River is a significant source of
snowmelt flows, and likely would delay the 1992-2006 CT date, these

findings nonetheless suggest that shifts in CT date towards earlier in the
spring are occurring for Mt. Rainier Rivers. Furthermore, shifts in CT
appear to be intrinsically linked to declines in Mt. Rainier SWE; regression
results indicate statistically significant negative trends (p=.008) when CT is
regressed on SWE.
4.4.2 1958-2006 Temporal Trend Analysis of CT Date
A regression analysis was performed on the 1958-2006 dataset in
order to indentify long-term trends in aggregated snowmelt runoff timing.
The 1958-2006 time period was selected as it is the longest period of time
where there are consistent records and for the overall applicability of the
trend analysis to current water resource goals. Snowmelt timing was found
to be decreasing over time, although the trend is not significant at the 95%
confidence level (t=-1.608, F= 2.82, p= .09). However, the trend is
significant at the 90% confidence level suggesting that the mean CT day
for the Nisqually, Cowlitz, and Puyallup Rivers is advancing and snowmelt
runoff in Mt. Rainier Rivers is occurring earlier in the year. Regression
results are supported by comparing the mean CT date for the 1992-2006
period (April 11th) to that of the 1958-2006 period (April, 16th) where the
difference in mean CT date is 5 days earlier.

4.5 Snowmelt Runoff Timing for Individual Basins
4.5.1 Nisqually
CT dates for the Nisqually River are the earliest in the study area
with a mean date of April 7th and CT of water year day (wyd) 188± 5.3.
Considerable variability exists for the basin with CT dates ranging from
240 wyd to 140 wyd (σ= 20.14) which is equivalent to February 18th to
May 29th. Linear trend analyses indicate that snowmelt runoff timing in the
Upper Nisqually basin is trending towards earlier dates as evidenced by a

highly negative and significant trend in CT (t= -3.2, F= 10.22, p= <.001).
CT trends are consistent with negative trends in Nisqually River snowmelt
proportions and quantities further validating that CT variation is influenced
by spring snowpack volumes.
4.5.2 Cowlitz
The Cowlitz River displays the least variability in runoff timing
throughout the study area with a mean CT date of wyd 201±3.6 and a
range of 238 wyd to 151 wyd (σ= 15.9) which corresponds to a mean
calendar date of April 20th which ranges from March 1st to May 27th.
Temporal trend analyses indicate that snowmelt runoff in the Cowlitz River
basin is occurring early in the year, and significantly advancing over time
(t= -2.16, F= 4.6, p= .03). Concurrent with trends observed for the
Nisqually River, snowmelt timing advances in the Cowlitz Basin are
consistent with declines in snowmelt runoff proportions and quantities and
further suggest that snowpack volumes are influential upon runoff timing in
snow dominant river systems.
4.5.3 Puyallup
The mean CT date for the Puyallup River is the second latest in the
study area at wyd 207± 5.12wyd. The calendar date of the long term
Puyallup CT mean is April 26th. Additionally, the Puyallup has a large
range of mean annual CT dates spanning from 154 wyd (March 4th) to 243
wyd (June 1st) (σ=18.04). Trend analysis yielded a slightly negative and
non-significant trend (t= -.20, F= .04, p= .80). Puyallup CT trend intensity
is reflected in Puyallup flow relationships with SWE (R2= .49) and is
consistent other SWE and CT relationships in the study area.
4.5.4 White and Carbon
No CT trend analyses for the White or Carbon Rivers were
performed due to data limitations, however observed trends for other

rivers in the study area are not strong enough to greatly alter historical
means. Therefore, substantial deviations in CT since the late 1970’s are
unlikely for the White and Carbon Rivers which permits CT calculations
from available records to provide reliable estimates current snowmelt
timing.
The mean annual CT date for the White River is the latest in the
study area at wyd 218 (May 7th) ± 6.7 wyd. CT dates for the White River
have the greatest range in the study area spanning from 146wyd to 260
wyd (February 24th to June 18th respectively) (σ= 22.5). Similarities
between the White River and the Cowlitz River, namely high correlations
between flow and SWE, indicate that CT trends for the White River are
most likely declining at rates comparable to those found for the Cowlitz.
The Carbon River has the second earliest CT date in the study area
at wyd 199 (April 18th) ± 6.4 wyd, and a range of 137 wyd to 243 wyd
(February 15th to June 1st respectively) (σ= 22.5). The Carbon River is a
transient system like the Nisqually and Puyallup, although mixed results in
trend analyses exist for the Nisqually and Puyallup rivers. Therefore it is
difficult to extend trend results from other transient basins to the Carbon.
However, Carbon River runoff is least correlated with SWE, suggesting
that CT trends are similar to that of the Puyallup which also displays a low
correlation with snowpack.
4.5.5 PDO Phase State, Snowmelt Flow and Snowpack
Results of two sample t-tests indicate that significant differences in
AMJJ snowmelt proportions and SWE exist between positive and negative
phases of the PDO. Due to limitations in period of record, only the
Nisqually and Cowlitz Rivers had enough observations to adequately span
both phases. During negative phases of the PDO, snowmelt contributions

of the Nisqually (t=3.65, p<.01) and Cowlitz (t=4.74, p<.01) were
significantly greater than contributions observed during positive phases.
Results comparing phase state and winter snow accumulations are
consistent with the observed differences in AMJJ flow where Paradise
SWE values are significantly higher during negative PDO phase states
(t=3.8, p<.01). Therefore, more snow accumulation and subsequent AMJJ
flows occur in response to cooler and wetter conditions for the Nisqually
and Cowlitz Rivers. Although it is difficult to confidently estimate PDO
effects for the Puyallup, White, and Carbon Rivers, the convincing
differences observed for the Nisqually and Cowlitz suggest that the PDO
has profound effects over the entire study area.

5. Discussion
The results of this study indicate that the hydrologic regimes of Mt.
Rainier rivers are heavily influenced by the magnitude and timing of
snowmelt inputs. The role of snow in annual hydrologic variability is
elucidated by the high degree of coherence observed between streamflow
variations and fluctuations in spring snowpack; large April 1st SWE
volumes generate high AMJJ fractional flow proportions and quantities.
Snowmelt flow magnitudes also influence annual streamflow timing for Mt.
Rainier Rivers, where later CT dates are associated with large spring
snowpacks and early dates occur during years of low snow volume.
Temporal trend analyses of AMJJ flow, annual streamflow timing, and
snow water equivalencies provide the most important findings of this study
as numerous negative trends for all three parameters were identified,
several of which are statistically significant. Additionally, the role of the
Pacific Decadal Oscillation as a substantial driver of hydrologic variation
has been confirmed for the study area through the identification of
significant differences in SWE and AMJJ flow in regards to PDO phase
state. Therefore, the findings outlined in this study present numerous
implications for future water resource planning goals and provide insights
into potential hydrologic responses of Mt. Rainier Rivers to climate change
effects.

5.1 Quantification of Snowmelt Flows
Snowmelt runoff accounts for substantial portions of annual flow
quantities for all five rivers in the study area with proportions ranging from
39% to 49.1% of total annual discharge. Although only the Cowlitz and
White Rivers are delineated as snow dominated systems, the remaining
three transient stream hydrographs display pronounced AMJJ peaks in
which a third of annual runoff occurs in a four month period. Stewart et al.
(2005) determined that less than 30% of mean annual flow in regional
transient basins occurs during the AMJJ period, suggesting that Mt.
46

Rainier transient streams better represent hybrid systems in which
snowmelt influences may be more influential that rainfall inputs. Thus,
snowmelt inputs are major components of annual flow in each watershed
regardless of hydrographic classification by providing critical flow
augmentation during the dry summer and fall months when other local
transient systems are well into recessional states.
Estimates of Mt. Rainier snowmelt proportions and quantities are
consistent with similar studies investigating snowmelt flows across
western North America (Stewart et al. 2004, 2005; Regonda et al. 2005;
Cayan et al. 2001; Dettinger and Cayan, 1995). However, studies
pertaining to Pacific Northwest snowmelt are more useful for corroboration
(Mote, 2008; Bach, 2002; Miles et al. 2000; Pelto, 1993). Bach’s 2002
study of snowshed contributions to the Nooksack River utilizes similar
methodologies to quantify snowmelt flows from Mt. Baker and serves as
the only fine grained comparison to this study. Hydrographic
characteristics of Mt. Baker and Mt. Rainier are very similar as both
systems originate on large glaciated volcanoes and possess similar
climates (Bach, 2002). Annual snowmelt flows in the Nooksack River
range from 27.9% to 63.9% percent of annual flow (Bach, 2002) which is
highly consistent with the range of annual snowmelt proportions at Mt.
Rainier. The high degree of coherence in estimates serves to further
validate the accuracy of AMJJ flow proportions calculated for Mt. Rainier.
The combined results of this study and Bach’s (2002) work on Mt. Baker
provide a firm baseline for estimating snowmelt dynamics in the Pacific
Northwest by establishing a working dataset for future water resource
planning.

5.2 Snowmelt Runoff Timing
Snowmelt timing dates (CT) calculated for Mt. Rainier corroborate
well with mean CT dates from similar snow fed systems observed across

large spatial domains (Stewart et al. 2004, 2005; Regonda et al. 2005;
Cayan et al. 2001; Dettinger and Cayan, 1995). Snowmelt flows
originating from Mt. Rainier generally begin to dominate the hydrographs
of all five rivers from early April to early May which is consistent with CT
dates for other Pacific Northwest rivers (Stewart et al. 2005, 2004; Bach,
2002, Miles et al. 2000; Hamlet and Lettenmaier, 1999). Similar to other
regions in the American West, Mt. Rainier CT dates vary annually on large
temporal scales, sometimes changing by periods measured in months
(Stewart et al. 2005; Regonda et al. 2005; Cayan et al. 2001). Highly
significant positive relationships (p=.008) established between aggregated
CT date and April 1st SWE volumes at Mt. Rainier illustrate fundamental
linkages between winter snow accumulation and snowmelt timing. Given
the importance of snowmelt timing to current reservoir operation
schedules, quantifications of CT date will provide helpful foundations for
water availability forecasting in the future.

5.3 Hydrologic Trend Analyses
Temporal trend analyses of AMJJ flow quantity, streamflow timing
(CT), and April 1st SWE provide the most valuable results of this study as
trends are of great interest to both water resource managers and climate
researchers. Overall, trends for all three parameters are negative;
declining trends were observed for both aggregated and individual basins
indicating that hydroclimatic variables at Mt. Rainier are shifting away from
observed long term conditions. Negative trends in AMJJ flow, CT, and
SWE parameters present profound implications to water resource sectors
and threaten to alter fundamental components of the current resource
operation regime. Furthermore, the observed declines in snowmelt
parameters serve to verify that regional trend signals are present in Mt.
Rainier watersheds, and suggest that climate change effects are already
beginning to influence local snow hydrology.

5.3.1 Spring Snowpack
The significantly negative trend in Paradise SWE indicates that
spring snowpack volumes have been steadily decreasing since records
began in 1940. SWE trend results for Mt. Rainier are consistent with
findings from several studies detailing Pacific Northwest snowpack (Mote
et al. 2008; Hamlet et al. 2005, Mote et al. 2003) and with studies
analyzing SWE trends from mountainous regions throughout western
North America (Mote et al. 2006; Hamlet et al. 2005; Mote et al. 2005;
Stewart et al. 2005; Regonda et al. 2005). Mote et al. (2008) find that
SWE declines from 1950-2006 range from 15%-35% in the Washington
Cascades, where studies encompassing larger domains find basin scale
decreases from 20%-80% during the latter half of the 20th century (Hamlet
et al. 2005, Mote et al. 2003, Regonda et al. 2005).
High spatial trend coherence (local and regional) indicates that Mt.
Rainier SWE trends are likely occurring in response to large scale climatic
variation, although it is unclear whether PDO influences or climate change
effects are significant drivers. Cayan et al. (2001) established that
warming trends from1948-2002 throughout the western US are regional in
extent and argued that a combination of PDO phase and greenhouse gas
forcing is driving the warming. Mote et al. (2003) found similar trends in
PNW temperature and related regional climatic variation to fluctuations in
spring snow water equivalencies. Results from Mote et al. (2003) indicate
that regional warming trends account for most of the declines in SWE
whereas precipitation had secondary effects. Furthermore, Mote et al.
(2003) conclude that temperature trends are occurring independently from
PDO phase state, and in a supplementary study (Mote et al. 2008) assert
that SWE declines in the Cascades are “largely unrelated to Pacific
climate variability and strongly congruent with trends expected from rising
greenhouse gases” (p.208). Therefore, it is likely that the observed

declining trends in Mt. Rainier spring snowpack can be partially attributed
to the effects of global and local climate change and will continue to
decline as anthropogenic warming accelerates.
Local water resource managers rely upon historic hydroclimatic
records for streamflow forecasting purposes in which April 1st SWE is the
central parameter used to estimate AMJJ streamflow potentials.
Continued declines in annual SWE will alter current runoff forecasting
methodologies by rendering contemporary historical approaches as
unreliable. Water managers will be required to implement new water
supply forecasting techniques that do not heavily integrate historical
variations in SWE. New methods must incorporate forecasted climate
change effects with increased field data in order to accurately establish
accurate water supply predictions. Failure to abandon historically
integrated approaches will result in the issuance of in-accurate streamflow
forecasts inflicting potentially large economic damage to agricultural and
municipal water sectors.
5.3.2 AMJJ Flows
Negative temporal trends were identified for the aggregate (19582006) and for the Cowlitz and Nisqually Rivers and a near stagnant trend
was identified for the Puyallup River. Declining trends in AMJJ flow
proportions are consistent with numerous studies analyzing snowmelt
systems across the western United States (Wahl, 1992; Aguado et al.
1992; Pupacko, 1993; Dettinger and Cayan, 1995; Stewart et al. 2004,
2005; Regonda et al. 2005) Of particular interest is the finding that
aggregated AMJJ flow quantities are not decreasing at a significant rate (p
= .154) whereas aggregated AMJJ flow proportions are significantly
decreasing (p = .03). The difference in decline magnitude between
aggregated AMJJ flow proportions and AMJJ flow quantity further
validates the findings of Mote et al. (2003, 2008) that temperature is

driving snowpack decline. Aggregated annual (12 month) flows were not
found to be significantly decreasing, suggesting that declines in AMJJ flow
proportions are resultant of more precipitation falling as rain in the winter
limiting the amount of meltwater stored in high elevation snowsheds for
release during the AMJJ period (Bach, 2002). The observed imbalance in
streamflow decline provides further evidence that streamflow variations
are likely being driven by climate change effects.
The Nisqually and Cowlitz rivers are experiencing significant
declines in AMJJ streamflow whereas the Puyallup time series indicates a
trend neutral pattern. These findings suggest that AMJJ flows are
declining only in snowmelt dominated systems and transient rivers are not
experiencing significant alterations to historical variability. Data limitations
for the White and Carbon Rivers forbid conclusions to be made that
snowmelt dependence is exclusively controlling declines in AMJJ flows.
Here, the relationships between flow and SWE are utilized to hypothesize
that White River AMJJ flows are likely decreasing and Carbon River flows
are likely trend neutral. Snowmelt period runoff in the Cowlitz is strongly
correlated with SWE variation (R2= .69) indicating that AMJJ flows are
highly dictated by spring snowpack volumes suggesting that the overall
trend in White River snowmelt is decreasing at similar rates as those
identified for the Nisqually and Cowlitz systems. Alternatively, the
relatively low snowpack correlation observed for Carbon River snowmelt
(R2= .42) can be used to confidently infer that AMJJ flow trends are
trivially neutral in strength.
Decreases in annual AMJJ flow proportions and quantities have the
potential to drastically alter water resources operating regimes. Currently,
spring flows are utilized to re-fill reservoirs following winter flood control
drawdowns to provide water for the arid summer and early fall seasons
when inputs to reservoirs are at annual low volumes. The neutral trend in

overall discharge indicates that more runoff will occur during mid-winter
resulting in an increase of flood control releases. Increased flood control
releases will place even more dependence on snowmelt flows to meet
present reservoir demand curves therefore increasing the likelihood of
severe water scarcity and potential drought during the late summer and
fall.
The combination of projected climate change impacts and
population growth further confounds impending water scarcity conflicts
resulting from decreasing snowmelt flow. Shifts in energy demand are
likely to occur as a result of regional warming that will generate higher
demands for cooling energy during the summer (Hamlet et al. 2009).
Subsequently, heightened demands for hydropower will require utilities to
spill additional reservoir water during the summer further exacerbating
current tensions between the hydropower sector and instream flow rules
(Payne et al. 2004; Hamlet and Lettenmaier, 1999). Adaptation measures
will require the creation of new storage facilities or the expansion of
current reservoirs in an effort to capture more water during the winter
months while simultaneously complying with flood control guidelines.
5.3.3 Streamflow Timing
Trends in Mt. Rainier streamflow timing closely follow the observed
trends in SWE and AMJJ flow proportions for both the aggregated dataset
and individual rivers. For the aggregate period of 1958-2006 negative
trends in CT date were identified and are significant at the 90% confidence
level (p = .09); results which are highly consistent with similar studies
(Stewart et al. 2005, 2004; Regonda et al. 2005, Cayan et al. 2001;
Dettinger and Cayan, 1995). Similar to variations in AMJJ flows, CT date
displays significant positive coherence with SWE (p = .008), although
SWE is not an accurate explanation for annual CT variation (R2 = .122),

largely because CT is an expression of total annual flow and does not
isolate snowmelt inputs alone (Stewart et al. 2005).
CT dates are occurring significantly earlier in the year for the
Nisqually (p < .001) and Cowlitz (p = .03) Rivers, whereas CT dates for
the Puyallup remain trend neutral (p = .80). Due to similarities in
correlations with SWE and overall hydrographic regime, results from the
above rivers can be used to predict CT trends in the White and Carbon
Rivers. Utilizing this assumption, it is likely that the CT date for the White
River is occurring earlier in the year and remaining stagnant for the
Carbon River. Decreasing trends for the Nisqually and Cowlitz are very
consistent with findings from studies analyzing snow dominant rivers in
western North America (Stewart et al. 2005, 2004; Regonda et al. 2005,
Cayan et al. 2001; Dettinger and Cayan, 1995), and especially with
Stewart et al (2005) where the largest CT advances were identified in the
Pacific Northwest. High degrees of spatial coherence with snow
dominated rivers suggest that declines in CT are likely linked to regional
scale trends.
In terms of water resource management, streamflow timing is
equally important as flow quantity. Numerous management structures
dependant on snowmelt flows in the Pacific Northwest rely upon the timely
occurrence of the snowmelt peak in April to adequately meet allocation
goals (Payne et al. 2004; Hamlet and Lettenmaier, 1999). Shifts in annual
CT dates towards earlier center timing ultimately shift the hydrograph to
the left, where more flow occurs during the winter months, the spring
snowmelt peak occurs earlier, and summer and fall flows diminish.
Present water management schedules will need to be revised to
accommodate for the earlier influx of snowmelt.

Reservoir operations are the most sensitive water resource sector
to shifts in CT as regional refill schedules are fine tuned to snowmelt pulse
timing for hydropower and flood control optimization. As a result, reservoir
retention and release schedules dictate water availability beyond
impoundments which greatly affects the ability of downstream sectors to
meet reliability goals. Earlier CT dates have profound implications for
current and future instream flow compliance in each WRIA primarily due to
further reductions in summer and fall flows. If current trends continue,
existing tensions between instream flow rules and reservoir optimization
will be exacerbated necessitating policy reform and further litigation.
Stewart et al. (2004) applied a “business as usual” climate change
model to historical streamflow timing records to assess potential shifts in
CT if current greenhouse gas emissions continue to escalate. Results
indicate substantial advances in CT dates from 30-40 days by 2099, with
the greatest advances occurring in the Pacific Northwest. Observed CT
trends at Mt. Rainier are certainly consistent with the aforementioned
forecast and may represent the initial stages of climate change induced
CT shifts. Thus, the implications of climate change to Mt. Rainier CT dates
are drastic, especially when coupled with expected declines in SWE (Mote
et al. 2003, 2008) and AMJJ fractional flows (Stewart et al. 204, 2005;
Regonda et al. 2005, Bach, 2002).
Mt. Rainier Hydroclimate and the PDO
The PDO has been shown to be an accurate proxy for water
forecasting in the Pacific Northwest (Mantua et al. 1997; Hamlet and
Lettenmaier, 1999; Miles et al. 2000; Mote et al. 2003, Payne et al. 2004).
Significant differences between positive and negative PDO phase state
and hydroclimatic variables at Mt. Rainier confirm that oscillations of the
PDO influence variations in SWE and AMJJ fractional flows. The stark
differences observed in hydrology during differing phase periods indicate

that Mt. Rainier streamflow is highly responsive to large scale climatic
variation and suggest that the impacts of climate change on Mt. Rainier
hydroclimate will be profound. PDO indices will continue to be a valuable
tool for water forecasting although uncertainties remain as to the effects of
climate change on PDO behavior (Mote et al. 2003, 2008). In the event
that climate change effects intensify current PDO regimes its
consideration in planning purposes will increase in value by providing a
reliable baseline for water availability purposes.

6. Conclusions
The purpose of this study was to investigate the role of snowmelt
parameters as a source of hydrologic variation in Mt. Rainier watersheds
with the overarching goal of providing water managers with hydrologic
data for use in future water resources forecasting applications. Detailed
fine-grained data will be required to develop effective adaptive strategies
for water resource sectors facing impending strains from population
growth and climate change.
In order to provide useful results, quantifications and temporal trend
analyses were carried out for three hydroclimatic parameters: AMJJ
fractional flows, snowmelt pulse timing (CT), and snow water
equivalencies (SWE). The combined analyses of the three parameters
yielded useful results that synthesize sources of variability to illustrate the
major dynamics influencing the snowmelt hydrology of Mt. Rainier Rivers.
Furthermore, the influence of the Pacific Decadal Oscillation on Mt.
Rainier streamflow and snowpack variation was investigated to identify
linkages between local hydrography and global scale climate; a key
indicator of how Mt. Rainier rivers may respond to large scale climatic
change. It is apparent from the findings of this study that significant
hydrologic change is occurring in Mt. Rainier watersheds, and that drastic
revisions to current water resource management regimes will be
necessary for adaptation if observed trends continue.
Key findings can be summarized in five ways: (1) Annual AMJJ
flows account for essentially half of the annual runoff for the aggregate
and snow-dominant systems where transient systems display distinct
snowmelt pulses (2) AMJJ flow quantities and proportions are decreasing
for the aggregate, Nisqually, and Cowlitz Rivers (3) SWE has been
significantly decreasing since at least 1940 (4) Runoff timing (CT) is
shifting to earlier in the year for the aggregate and snow dominated basins;
56

transient basins are trend neutral confirming that SWE is a main driver of
CT date in snow-dominant basins (5) AMJJ runoff and SWE are sensitive
to PDO phase state suggesting that Mt. Rainier hydroclimate is highly
responsive to large scale climatic variability.
Increasing water scarcity is the starkest implication illustrated by
the results of this study. The combination of decreasing SWE and AMJJ
flows compounded with earlier snowmelt timing will ultimately lead to
reductions in summer and fall water supplies and increase the frequency
of droughts. Currently, water is over allocated in each basin and tensions
already exist surrounding the allocation of present day late season supply.
Thus, intensified declines in streamflows coupled with population growth
will exacerbate current water allocation conflicts and require intensive
policy reform.
Importantly, changes will have to be made to current reservoir refill
schedules to alleviate growing tensions between water use interests.
Revisions must include optimization changes for hydropower as likely
increases in flood frequency during the winter will necessitate the release
of power generation storage for flood control evacuations leading to
scarcer and more expensive hydropower during the summer. Projected
increases in temperature are likely to cause a shift in the regional
electricity demand curve resulting from heightened summer cooling
requirements (Hamlet at al. 2009). Expected increases in summer
hydropower production range from 9-11% (Hamlet et al. 2009) which will
place further strains on reservoir rule curves especially for the large
hydropower facilities located on the Nisqually and Cowlitz Rivers. Thus,
further increases in both climate change intensity and population growth
will inevitably cause summer hydropower demand to increase concurrent
with further decreasing AMJJ flows and earlier CT dates, and may force

utilities to purchase power from the grid or explore alternate energy
sources.
Additional reductions in hydropower allocations will be
necessitated in order to maintain instream flow requirements for wildlife
and senior water rights, irrigation, and recreation levels. The impending
tradeoffs between hydropower production and downstream uses elucidate
the multi-faceted nature of the implications likely to occur in response to
decreases in snowmelt flows; ramifications extend to both physical and
social arenas. Upholding the environmental quality of each river will come
at high societal costs to power producers and consumers, irrigators, and
water right holders as water rationing to fulfill each sector’s goals will need
to become commonplace. Consequently, disagreements are sure to arise
leading to lengthy and expensive policy reform and related environmental
litigation between water resource sectors, environmental groups,
government regulatory bodies and tribes. Therefore, environmental
restoration and instream flow rule making efforts are likely to proceed at
slower rates further complicating already sensitive political issues.
PDO phase state was shown to significantly affect Mt. Rainier
hydroclimate indicating that large scale climatic processes are significant
drivers of hydrologic variation. These results provide ample reason to
believe that SWE, CT, and AMJJ flows will be altered as a result of
anthropogenic climate change. To date, PDO signals are not strong
enough to completely dictate the declines in regional spring snowpack
(Mote et al. 2008) suggesting that climate change effects are, at least in
part, responsible for the observed declines in Mt. Rainier SWE, AMJJ
flows, and earlier CT dates. Continued global warming will result in
intensified negative trends for all three parameters, especially in snow
dominant basins. Therefore it is likely that the Nisqually, Cowlitz, and
White Rivers will experience the most drastic declines in AMJJ flows.

Climate change impacts to Mt. Rainier hydrology will require local
water managers to implement revisions to fundamental forecasting and
planning methodologies in order to meet the future demands of a rapidly
growing population. Importantly, managers will be required to abandon
historically integrated approaches to water forecasting as such techniques
will no longer accurately predict AMJJ flow and CT date due to hydrologic
uncertainty introduced by a warming climate. Importantly, managers must
begin to incorporate climate change into resource planning processes
(Whitely, 2006) to assess both the physical and social implications of
warming to water availability. A key step in that process will involve the
application of hydrologic data, similar to the findings presented here, to
hydrologic modeling to assess effects at basin scales. Such models
integrated with watershed planning efforts will greatly improve the abilities
of water managers to prepare for adaptations.
Data and methodological limitations combined with the conclusions
generated from this study provide opportunities for improved future
research. Data limitations existed as the largest barrier to acquiring
accurate AMJJ flow estimates and CT dates for the White and Carbon
Rivers. Inconsistent records for the respective basins prevented high
accuracy results from being attained thus further investigation into
historical hydrologic conditions is required to improve upon the results
presented here. Interpolation may be utilized to accurately estimate
historical flows for the upper reaches of the White and Carbon Rivers
providing more robust quantifications for each respective basin and the
aggregate. Improved monitoring of snowpack volumes would greatly
enhance the understanding of climate and streamflow relationships,
however implementation would be costly. Questions still remain
surrounding aggregated temporal trend discrepancies between AMJJ

quantities and proportions where more detailed meteorological analyses
are needed to fully understand the causes of differing trend magnitudes.
Overall, the findings of this study provide water managers in the
South Sound region with an accurate hydrologic dataset to assist in future
watershed planning applications. Importantly, the detection of declining
streamflow and advanced runoff peaks provides confirmation that regional
snowmelt resources are decreasing and actions must begin to take place
to prepare for future perturbations in snowmelt flows. Additionally, this
study provides a framework for future basin scale studies across the
Northwest and makes important contributions to the greater discourse
pertaining to climate change associated effects on snowmelt hydrology.

7. References
Aguado, E., D. R. Cayan, L. G. Riddle, and M. Roos. 1992. Climatic
fluctuations and the timing of West Coast streamflow. J. Climate,
5:1468–1483.
Bach, A. 2002. Snowshed Contributions to the Nooksack River Watershed,
North Cascades Range, Washington. The Geographical Review
92(2):192-212.
Cascade Water Alliance. 2010. Water Planning. Retrieved from:
http://www.cascadewater.org/supply_overview.php
Cayan, D. R., S. A. Kammerdiener, M. D. Dettinger, J. M. Caprio, and D.
H. Peterson. 2001. Changes in the onset of spring in the western
United States. Bull. Amer. Meteor. Soc., 82:399–415.
Cayan, D. R. 1996. Interannual climate variability and snowpack in the
western United States. J. Climate, 9:928–948.
Dettinger, M. D. and D. R. Cayan. 1995. Large-scale atmospheric forcing
of recent trends toward early snowmelt runoff in California. J. Climate,
8:606–623.
Fountain, A.G., and W. V. Tangborn. 1985. The Effects of Glaciers on
Streamflow Variations. Water Resources Research 21(4): 579-586.
Hamlet, A.F., S.Y. Lee, K.E.B. Mickelson, and M.M. Elsner. 2009. Effects
of projected climate change on energy supply and demand in the
Pacific Northwest and Washington State. Chapter 4 in The Washington

Climate Change Impacts Assessment: Evaluating Washington's Future
in a Changing Climate, Climate Impacts Group, University of
Washington, Seattle, Washington.
Hamlet, A. F. and Lettenmaier, D. P. 1999. Effects of Climate Change on
Hydrology and Water Resources in the Columbia River Basin. J. Amer.
W. Res. Assn. 35 (6), 1597–1623.
Hamlet, A.F., Mote, P.W., Clark, M.C, Lettenmaier, D.P. 2005. Effects of
Temperature and Precipitation Variability on Snowpack Trends in the
Western United States. Amer. Meteor. Soc. November, 2005. p. 45454561.
Miles, E. L., Snover, A. K., Hamlet, A. F., Callahan, B., and Fluharty, D.
2000. Pacific Northwest Regional Assessment: The Impacts of Climate
Variability and Climate Change on the Water Resources of the
Columbia River Basin. J. Amer. Water Res. Assn. 36 (2), 399–420.
Mantua, N. J., Hare, S. R., Zhang, Y.,Wallace, J. M., and Francis, R. C.
1997. A Pacific Interdecadal Climate Oscillation with Impacts on
Salmon Production. Bull. Amer. Meteorol. Soc. 78, 1069.
Mote, P. W. 2003. Trends in snow water equivalent in the Pacific
Northwest and their climatic causes. Geophys. Res. Lett., 30, 1601.
Mote, P. W. 2006. Climate-driven variability and trends in mountain
snowpack in western North America. J. Climate, 19, 6209–6220.
Mote, P.W., Hamlet, A. F., Clark, M. P., and Lettenmaier, D. P.: 2005.
Declining mountain snowpack in western North America. Bull. Am.
Meteorol. Soc. 86, 39–49.

Mote. P.W., Hamlet A.F. Salathe. E. 2008. Has spring snowpack declined
in the Washington Cascades? Hydrologic and Earth Systems Science,
12, 193-206.
Natural Resources Conservation Service.2010. SNOTEL. Paradise, WA
1940-2006. Data available at:
http://www.wcc.nrcs.usda.gov/nwcc/site?sitenum=679&state=wa
Payne, J.T., A.W. Wood, A.F. Hamlet, R.N. Palmer, and D.P. Lettenmaier.
2004. Mitigating the effects of climate change on the water resources
of the Columbia River basin. Climatic Change 62(1-3):233-256.
Pelto, M.S.1993. Current behaviors of glaciers in the North Cascades and
effect on regional water supplies. Washington Geology 21(2):3-10.
Puget Sound Energy. 2009. Electron Hydroelectric Project. Retrieved from:
http://www.pse.com/SiteCollectionDocuments/mediaKit/060_Electron.pdf

Pupacko, A. 1993. Variations in Northern Sierra Nevada streamflow:
Implications of climate change. Water Resour. Bull., 29:283–290.
Regonda, S., B. Rajagopalan, M. Clark, and J. Pitlick. 2005. Seasonal
cycle shifts in hydroclimatology over the western United States. J.
Climate, 18, 372–384.
Stewart, I. T., D. R. Cayan, and M. D. Dettinger. 2005. Changes toward
earlier streamflow timing across western North America. J. Climate, 18,
1136–1155.

Stewart, I. T., D. R. Cayan, and M. D. Dettinger. 2004. Changes in
snowmelt runoff timing in western North America under a “business as
usual” climate change scenario. Climatic Change, 62, 217–232.
Tacoma Public Utilities. 2010. Hydropower. Retrieved from:
http://www.mytpu.org/tacomapower/power-system/hydro-power/Default.htm

United States Geological Survey.2010. Surface Water Data Network.
Retrieved from:
http://waterdata.usgs.gov/nwis
Washington Department of Natural Resources. 2009. Geologic Map of
Washington. Retrieved from:
http://www.dnr.wa.gov/Publications/ger_geol_map_washington_pagesize.pdf

Washington Department of Ecology. 2010. Personal communication: Dave
Nazy. January 26th 2010.
Washington Department of Ecology. 2008. Report to the LegislatureProgress on Watershed Planning and Setting Instream Flows.
Publication # 08-06-027
Washington Office of Financial Management. 2007. Historical and
projected population for growth management and other purposes.
Retrieved from: http://www.ofm.wa.gov/pop/gma/gmamedproj.xls
Watershed Professionals Network. 2002. Nisqually River Level 1
Assessment (WRIA 11) Summary Report. Retrieved from:
http://www.ecy.wa.gov/programs/eap/wrias/assessments/wria11/summary.pdf

Whitely Binder, L.C. 2006. Climate change and watershed planning in
Washington State. Journal of the American Water Resources
Association 42(4):915-926.

Implications for Future Water Resources Management · The Evergreen State College Institutional Repository · Evergreen Digital Collections (2024)

References