Citation
The influence of wind distribution on a basinwide snowmelt model

Material Information

Title:
The influence of wind distribution on a basinwide snowmelt model
Creator:
Harelson, Stephen James
Publication Date:
Language:
English
Physical Description:
ix, 84 leaves : illustrations ; 29 cm

Subjects

Subjects / Keywords:
Runoff -- Colorado ( lcsh )
Snow -- Colorado ( lcsh )
Heat -- Convection ( lcsh )
Heat -- Convection ( fast )
Runoff ( fast )
Snow ( fast )
Colorado ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 82-84).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Science, Civil Engineering.
General Note:
Department of Civil Engineering
Statement of Responsibility:
by Stephen James Harelson.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
34233660 ( OCLC )
ocm34233660
Classification:
LD1190.E53 1995m .H37 ( lcc )

Full Text
THE INFLUENCE OF WIND DISTRIBUTION
BASINWIDE SNOWMELT MODEL
by
Stephen James Harelson
B.S., Colorado State University, 1986
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Civil Engineering
1995
ON A


This thesis for the Master of Science
degree by
Stephen James Harelson
has been approved for the
Graduate School
by
Date


Harelson, Stephen James (M.S. Civil Engineering)
The Influence of Wind Distribution on a Basinwide
Snowmelt Model
Thesis directed by Associate Professor Lynn E. Johnson
ABSTRACT
The most important contributor to the water supply
of Colorado and similar regions is the runoff that
results from melting snow. Predicting the magnitude,
timing and duration of this runoff is an important part
of sizing water supply and flood control facilities, and
determining agricultural water allocations.
Snowmelt is governed by the laws of heat and mass
transfer. Energy is transferred to and from the snowpack
through six different mechanisms: Conduction, Convection,
Radiation, Latent Heat of Vaporization, Latent Heat of
Fusion and Precipitation Melt. This study focuses upon
the effect of convective heat transfer on snowmelt.
Existing basinwide snowmelt models use a lumped parameter
approach to model the influence of wind on convective
heat transfer. Though wind varies significantly in time
and space, convective heat transfer is usually
iii


represented as a linear function of temperature. The
linear coefficient can be changed periodically to
represent seasonal effects, but it remains a temporal and
spatial average. This type of approach is used because
few basins have meaningful records of wind speed and
distribution.
Geographic Information Systems (GIS) technology
offers a convenient tool to assist the hydrologist in
analyzing spatially varying phenomena such as wind. In
this study, wind and other meteorological data from the
University of Colorado's Long Term Environmental Research
Center on Niwot Ridge were used to model the streamflow
of North St. Vrain Creek, 15 kilometers north. The
influence of wind was modeled three ways. First, the
daily wind speed was varied linearly throughout the basin
by elevation. The linear variation was based on the
patterns at the Niwot Ridge site. Second, a basin
average wind speed was calculated using daily wind speed
data, and assumed to blow uniformly throughout the basin.
Third, a set of monthly average convection coefficients
was calculated using monthly average Niwot Ridge
meteorological conditions, and a traditional lumped
parameter approach was applied. These three methods were
iv


used with a model that held constant the influence of the
basin's hydrologic abstractions. It was found that daily
variation of wind has a significant effect on model
results, but that spatial variation does not.
This abstract accurately represents the content of the
candidate's thesis. I recommend it;
Signed
v


CONTENTS
ACKNOWLEDGEMENT...............................ix
CHAPTER
1. INTRODUCTION...............................1
Background................................1
Types of Models...........................7
Rationale............................... 8
Scope and Objective......................11
Thesis Format............................12
2. REVIEW OF THE LITERATURE..................13
3. COLLECTION AND DISTRIBUTION
OF CLIMATALOGICAL DATA.................. 19
Data Collection..........................19
Data Reduction...........................20
Data Distribution........................22
4. MODEL CONSTRUCTION........................27
Overview.................................27
PRMS Modeling............................31
Establishment of Hydrologic
Response Units...........................38
Establishment of PRMS Parameters.........48
Parameters Constant in
Space and Time......................49
Parameters Varying in Time,
Constant in Space...................50
Parameters Constant in Time,
Varying in Space....................51
vi


Parameters Varying in
Space and Time......................54
5. MODEL CALIBRATION AND RESULTS............56
6. CONCLUSIONS..............................62
Suggestions for Further Study...........65
APPENDIX. ....................................... 68
REFERENCES..........................................82
vii


FIGURES
Figure
1.1 Snowmelt Energy Summary...................6
1.2 Study Area...............................10
3.1 Precipitation Correction Curves..........25
3.2 Wind Adjustment Curves..................2 6
4.1 MMS Interface............................29
4.2 MMS Parameter Editing Screen.............30
4.3 PRMS Flowchart...........................33
4.4 Snowcomp.f Flowchart................... 34
4.5 St. Vrain Basin Elevation
Classifications.........................41
4.6 St. Vrain Basin Slope
Classifications.........................42
4.7 St. Vrain Basin Aspect
Classifications.........................43
4.8 St. Vrain Basin Watershed
Display.................................44
4.9 St. Vrain Basin HRUs.................... 45
5.1 Convection Coefficient Model.............59
5.2 Wind Varied Temporally Model.............60
5.3 Wind Varied Spatially and
Temporally Model........................61
viii


ACKNOWLEDGEMENTS
The author would like to acknowledge the assistance
of George Leavesley and Steve Markstrom at the U.S.G.S.
Water Resources Division in obtaining, compiling and
running the Modular Modeling System used in this project.
Bob Jarrett and Bob Ugland, also at the U.S.G.S. Water
Resources Division, provided streamflow data for North
St. Vrain Creek. Logistical support and/or data were
provided by the Niwot Ridge Long Term Ecological Research
Project (NSF DEB9211776) and the Mountain Research
Station (BIR 9115097). Rick Ingersoll at L.T.E.R. was
very helpful in providing this data. Thanks also to the
numerous investigators working on Niwot Ridge throughout
the past forty years, particularly J.W. Marr, Mark
Losleban, Jim Halfpenny and Skip Walker.
Thanks to Lynn Johnson for chairing the committee
and providing assistance, insight and encouragement along
the way. Thanks also to James Guo and William Hughes for
sitting on the committee. Special gratitude goes to
Daryl Lindeman, John Carroll and Jerry Lange of Carroll
and Lange, Inc., who supported this study both
financially and logistically. It's nice to work for a
company that truly values education.
ix


CHAPTER 1
INTRODUCTION
Background
The most important contributor to the water supply-
in Colorado and similar regions throughout the world is
the runoff that results from melting snow. Numerous
factors influence the quantity and timing of this runoff,
and many of these factors are difficult for hydrologists
to accurately quantify. The measurement of temperature,
wind speed, humidity, solar radiation and precipitation
is difficult throughout the remote, widespread areas of
most snowmelt basins in Colorado. Even more difficult to
measure are less tangible influences such as evaporation,
infiltration, convective heat and mass transfer, and the
influences of different types of vegetation and terrain.
The recent advent of Geographic Information Systems (GIS)
technology has provided the hydrologist with an effective
means of cataloguing the spatial distribution of factors
that influence snowmelt.
In order for snow to melt, heat energy must be
transferred to the snowpack. There are three major types
of heat transfer; radiation, convection, and conduction.
1


Radiation is the transfer of energy via electromagnetic
waves. It follows the Stefan-Boltzmann law, where heat
transfer is a function of the two transferring bodies'
temperatures raised to the fourth power. It can occur
between the sun and the snow, (short wave radiative heat
transfer) or between the snow and the surrounding
vegetation, clouds or outer space (long wave radiative
heat transfer). Long wave radiation can either heat or
cool the snow, while short wave radiation only heats the
snow.
Convective heat transfer, also known as sensitive
heat transfer, is the transfer of energy as a result of
the flow of fluid over the surface of a body. There are
two types of convection; free and forced. Free
convection results when the fluid is still, except for
the motion caused by density changes in the fluid as a
result of the heat transfer. An example of free
convection is still air being heated and rising over hot
pavement. As the heated air rises, cool air replaces it
near the pavement surface, and the cycle continues,
cooling the pavement. Free convection is most effective
when the density changes in the fluid result in an
unstable system. For this reason, free convection is
ineffective in the melting of snow, because as warm air
2


transfers its heat to the snowpack, the air is cooled and
its density increases, and tends to remain near the snow
surface. No uncooled "new" warm air can replace the
cooled air near the snow surface, so the system
stabilizes and the cycle ends.
Forced convection occurs when lateral fluid flow
(wind) transports a constant supply of uncooled warm air.
The effectiveness of forced convection is a function of
the lateral velocity of the fluid flow. Even modest
velocities can increase the effectiveness of forced
convection over free convection by a large amount.
The third method of heat transfer is conduction
between the ground surface and the snow. Ground
conduction is only a factor in thin snowpacks where
radiation can penetrate to the ground beneath and then be
conducted to the snowpack, or when snow falls on warm
ground in the fall or spring.
In addition to the three classic modes of heat
transfer; radiation, convection and conduction, energy
can be transferred to and from the snowpack via the mass
transfer mechanisms of precipitation, condensation and
o
evaporation. Liquid rain, always warmer than the 0 C
snowpack, will transfer energy to the snowpack.
Condensation heats the snow because as water vapor
3


condenses from the atmosphere, the latent heat of the
vapor (540 kcal/kg) is released and is at least partially
transferred to the snowpack. Similarly, as snow
sublimates or water evaporates into the atmosphere, the
latent heat required to cause the phase change is drawn
from the snowpack.
The heat balance for the snowpack is given by the
relation: (U.S. Army, 1956)
Hm = Hrs + Hr]_ + Hc + He + Hg + Hp + Hg
Where Hm is Heat of Melt
Hrs is short wave (solar) radiation
Hr]_ is long wave radiation
Hc is convective heat transfer
He is latent heat from condensation
Hg is conduction from ground
Hp is heat content of rain
Hg is the change in the heat content
of the snow.
The relative importance of each of these elements of
snowmelt depends upon temperature, humidity and solar
radiation intensity. The U.S. Army (1956) found that in
a Central Sierra location in California that radiation
melt was dominant in the spring and summer, providing
4


approximately two-thirds of the total melt. Convection
and condensation provided the remainder. In winter,
however, conduction and rain melt were the chief energy
components in snowmelt, with convection also
contributing. Radiation was shown to cause a net energy
loss to the snowpack in winter. Figure 1-1 shows this
seasonal variation.
Using the basic principles of heat and mass
transfer, the elements necessary to construct a snowmelt
model are readily identified. While the physics of even
multi-phase heat transfer are well understood under
controlled conditions, problems arise when trying to
apply these principles in the study of hydrology. It is
relatively simple to predict the heat and mass transfer
between the atmosphere and a small plane of ice in a
laboratory setting. Atmospheric conditions can be
measured and controlled and any variation can be noted
and accounted for. In the field of snow hydrology,
however, the areas are thousands of times larger and
often distributed over rough, remote terrain.
Atmospheric data is difficult to gather, and can vary
greatly throughout a watershed. Snow itself is an ever
changing material, whose characteristics depend upon its
age, temperature, and liquid water content. To apply the
5


classic heat transfer relations to the problem of
snowmelt, a relatively small number of data points must
be distributed, interpolated, and extrapolated over large
and rugged expanses of space. The collection and
cataloguing of data are as important as the design of the
snowmelt model that uses these data.
Snowmelt Energy Summary
Source-U.S. Army, (1956)
6


Types of Models
For many years, efforts have been made to
successfully model the phenomenon of melting snow and its
influence on hydrology. The simplest attempts were
variations of the "Degree-Day" method where melt was
correlated directly to temperature. Temperature was
considered the best indicator of the effectiveness of
heat transfer, and was often the only reliable and
regularly available meteorological variable available.
These models relied upon the transformation of large,
complicated systems into simple, lumped parameter models.
Accuracy depended on the uniformity of both the watershed
and the meteorological conditions affecting the runoff.
Other models have taken a more conceptual approach.
Much work has been done studying the melt patterns of
"point" sites, where the meteorological and snowpack
conditions can be closely measured and recorded. Classic
heat transfer analysis can be performed on these models,
because spatial and temporal variations do not exist.
Very accurate models have been developed, but the problem
lies in extrapolating the results on a basin wide scale.
Geographical information systems (GIS) technology
provides the opportunity for the hydrologist to combine
the lumped and conceptual types of modeling techniques.
7


Using GIS methods, the spatial distribution of
meteorological data and snowmelt processes can be
modeled. A basin can be divided into subbasins with
similar characteristics. Point meteorological data can
be distributed among these subbasins in a rational manner
within a GIS model. A model system that is a collection
of smaller submodels can then be created. Each of the
submodels more closely resembles a conceptual point
model, and does not rely on basin-wide averages of
geographic and weather data. This type of model provides
the ease of calculation of a lumped parameter model, with
the increased accuracy of a conceptual point model. The
GIS is a tool that both aids in cataloging parameters for
the subbasins and linking the submodels together.
Rationale
A significant challenge in the design of a snowmelt
model is the use of real world data in the classic heat
and mass transfer models. GIS technology provides a
convenient tool to manipulate and analyze geographic
influences on snowmelt, and is also useful in cataloguing
the spatial variation of the atmospheric influences in
snowmelt.
In many previous basinwide snowmelt models, the
effect of wind speed in convective heat transfer has been
8


simplified, because there is not often a record of wind
speed data. Most models have concluded that radiation is
the dominant mode of heat transfer in the melting of
snowpack. The effects of high wind speed, and
adiabatically warmed Chinook winds on both the melting
and evaporation of the snowpack, have not been thoroughly
studied.
The University of Colorado has operated a
climatological research center on the slopes of Niwot
Ridge in Boulder County, Colorado since the early 1950s.
The Niwot Ridge Long Term Environmental Research
(L.T.E.R.) Center has recorded temperature,
precipitation, solar radiation, wind and other
environmental factors at various elevations throughout
several periods since 1952. The records provide a rare
picture of the meteorological conditions of the alpine
environment on the eastern slope of the Colorado Rockies.
Approximately 15 kilometers north of Niwot Ridge lie
the headwaters of North St. Vrain Creek.(Figure 1-2.)
Near the town of Allenspark, the United States Geological
Survey operates a streamflow gage on this creek. This
gage has recorded the average daily flows of the creek
since 1986.
9


Longs Pk.
Study Area
10


The watershed^flowing into this gage contains an area of
85 square kilometers, and varies in elevation from 2525
meters at the gage site to 4344 meters above sea level,
at the summit of Longs Peak. The watershed lies on the
east side of the continental divide in the southerly
portion of Rocky Mountain National Park. It is
wilderness, covered by pine forests at its lower reaches
and alpine tundra at its upper elevations. No
transmountain water diversions take place within the
basin. The L.T.E.R. meteorological data from Niwot Ridge
provided a basis for the study of the phenomenon of
snowmelt in the North St. Vrain Creek basin.
Scope and Objective
It was desired to construct an energy balance
snowmelt model for North St. Vrain Creek using classical
heat transfer analysis. The streamflow gage record was
used to test and calibrate the model. The meteorological
parameters; temperature, wind speed, humidity, solar
radiation and precipitation were taken from the L.T.E.R.
records of Niwot Ridge. Methods to distribute these
point values throughout the watershed were developed and
compared. A mathematical model was constructed to
account for these factors, and the influence of the
11


factors compared. Few snowmelt watersheds have the
meteorological records that exist on Niwot Ridge, and it
was hoped to demonstrate the significance of wind speed
on the melting and evaporation of Colorado alpine
snowpack.
Thesis Format
Chapter 2 of the thesis provides a review of the
current and historical literature regarding snowmelt
hydrology and modeling. Chapters 3 is a discussion of
the methods used to collect and distribute the data used
throughout the watershed. Chapter 4 is a discussion of
the creation of the snowmelt model, and Chapter 5
describes the calibration and results of the model.
Conclusions and recommendations for future work are
contained in Chapter 6.


CHAPTER 2
REVIEW OF THE LITERATURE
Modern snowmelt hydrology is based on the handbook
"Snow Hydrology," published by the U.S. Army (1956) .
This work was a comprehensive collection of all existing
information on the subject of snow hydrology. This
handbook attempted to integrate all existing knowledge of
the deposition and distribution of snow, the evaluation
of the hydrologic water balance of snowy basins, the
physics of snowmelt, the storage and transmission of
water in the snowpack, and methods of estimating snowmelt
runoff. It resulted from 10 years of data collection and
analysis at three research watersheds in Montana,
California and Oregon. This book provides considerable
information on methods for both the collection and
analysis of snowmelt data and has been the basis for
nearly all other work in the field since 1956.
In 1986, the World Meteorological Organization (WMO,
1986) published the results of a comparison of snowmelt
models. In this work, eleven different models were
13


tested using the same set of data for six different river
basins. That study provided a set of guidelines for the
construction of snowmelt models, as well as a means for
real comparison of the various techniques in use.
Anderson (1976), constructed a model of the
temperature and runoff of a snowcover at a point using an
energy balance equation for snowcover and an equation for
energy transfer within the snowpack. The research
location was Danville, Vermont. Influences of varying
meteorological conditions and varying density of the snow
were evaluated. However, the influence of these effects
occurring simultaneously throughout an areal watershed
was not evaluated.
Price, Dunne and Colbeck (1976) developed an energy
balance model for a location in sub-arctic Quebec. They
sought to improve the assessment of turbulent convection
over the snowpack by using the Richardson Number as a
measurement of stability over the snowpack. They
developed their model based on seven test sites near
Schefferville, Quebec, Canada. Four of these sites were
located in forested areas, and three in tundra. The
sites, ranging in size from 1335 to 2810 square meters,
were all located at approximately 540 meters above sea
level. Using measurement data from May, 1973, they
14


calibrated their deterministic energy balance model with
moderate success. Again, this work was essentially a
point model with no attempt made to evaluate a large
scale watershed.
Bloschl, Kirnbauer and Gutknecht (1991) developed a
spatially distributed model for a small alpine catchment
in Tirol, Austria. Their catchment, 9.4 square
kilometers in size, was modeled using a digital elevation
model with a grid size of 25 meters. Areal snowcover
data was gathered from oblique aerial photographs. These
photographs were taken at intervals of one to two weeks,
and the resulting change in snowcover was used to
calibrate the snowmelt model. This verification method
was used, rather than runoff, because runoff is
influenced by hydraulic factors and infiltration and
other hydrologic abstractions. Meteorological data was
extrapolated throughout the basin based on simple
assumptions. Temperature measurements took place at two
stations, and the temperature at each grid element in the
Digital Elevation Model was assumed to vary linearly with
elevation between the two samples. Wind speed and
humidity were assumed constant throughout the basin, and
precipitation increased with elevation at the rate of 30%
15


per kilometer. Water equivalent was fit to local field
data based on elevation, aspect and local relief.
Ranzi and Rosso (1991) also developed a distributed
snowmelt model in the Alps. Their model, in the
Cordevole Valley of the Dolomite section of the Italian
Alps, stressed the importance of radiative heat transfer
and the spatial variation of snowpack albedo. The basin
studied had an area of 6.9 square kilometers and an
elevation of ranging from 1800 to 3152 meters above sea
level. The vegetation on the site was pasture and brush,
with negligble amounts of forest. The model included an
element to account for the convective effect of wind, but
did not model the spatial variability of the wind.
During the snowmelt season, the test site is
characterized by a low wind speed of 1 to 2 m/s. For
this reason, the turbulent mass and energy exchanges had
a small influence in the.output of the model, with the
dominant effect being short wave radiative transfer. An
important problem identified in the study was the effect
of grid size on the output of the model. A non-negligble
degree of sensitivity was identified in relation to the
size of D.E.M. grid spacing used in the model. Frankoski
(1994) also concluded that for small basins, of the order
of 20 square kilometers, spatial resolution does have a
16


significant effect on snowmelt runoff model output.
However for larger basins, the effect is insignificant.
Hottelet, Braun, Leibundgut and Rieg (1993)
developed a snowmelt runoff model for a 96 square
kilometer catchment in Switzerland. The most notable
element of their model was that it accounted for
subterranean transport of water through caverns. The
basin being studied was in a geological area where
limestone predominates. Because of this geology, there
are a great deal of caverns which complicate the
calculation of runoff. Using dye trace techniques, the
underground watershed was mapped, and combined with the
surface watershed and runoff hydrographs calculated.
The measurement of the areal variability of the snow
water equivalent is a difficult quantity to determine.
Much recent work has been done to better measure this
factor. Summerfeld, Musselman, Wooldridge and Conrad
(1991) used a simple degree-day model to estimate the
snow accumulation in small watershed in southern Wyoming.
The watershed had an average elevation of 3450 meters,
and an area of 2.87 square kilometers. The model,
developed by Martinec and Rango in 1981, used the areal
extent of the snowpack, temperature readings taken at 15
minute intervals, and an estimate of the density of the
17


snowpack, to calculate the yearly water equivalent. The
model was compared against core probe surveys and
precipitation gages and found to be as accurate as the
snow core probe gages.
Braun (1991) summarized problems with the estimation
of snow water equivalent and suggested the use of
remotely sensed data as a means to test physically based
models of snow water equivalent. Kumar, Haefner & Seidel
(1991) utilized Landsat data to map the snow cover in a
Himalayan basin, and Wankiewicz (1991) used NIMBUS 7
microwave observations to calculate snow-water
equivalents in the Canadian Rockies. Advanced Very High
Resolution Radiometer (AVHRR) data, obtained from the
National Oceanic and Atmospheric Administration
meteorological satellite, has been used by Xianzhang,
Qunzhu & Yongchao (1991) to map snowcover in China.
18


CHAPTER 3
COLLECTION AND DISTRIBUTION
OF CLIMATALOGICAL DATA
Data Collection
The collection of climatological data is one of the
chief problems in the design of a snowmelt model. Alpine
snowmelt basins of interest are usually located in remote
areas, and can cover vast areas of varied terrain. The
ruggedness of these sites do not allow for easy access
for the recording of data or maintenance of weather
instrumentation. Meteorological records of alpine
snowmelt basins are rare, and when they exist they are
typically represented by a relatively small number of
actual measurements, both spatially and temporally. The
conversion of these point meteorological records into
distributed values, useful throughout a large, varied
snowmelt basin is critical in the design of a snowmelt
model.
Niwot Ridge is an east-west flank of the Front Range
of the Rocky Mountains, located in Boulder County,
Colorado. The ridge is the home of the University of
Colorado's Long Term Environmental Research Center
19


(L.T.E.R.) This center has performed a wide variety of
meteorological measurements and research since the early
1950s. Measurement stations are located at elevations
ranging from 2195 meters to 3798 meters. Over the course
of over 40 years, both the instrumentation used and the
data recorded have changed many times. Though the data
from Niwot Ridge is not continuous or all inclusive, it
is a remarkable record of the alpine climate of the area.
Data Reduction
In reviewing a large amount of information, it was
necessary to decide which data was useful, and which
should be discarded. Since this model was to be
calibrated using USGS streamflow gage records which begin
in 1986, the Niwot Ridge Data collected since that time
was focused upon. All data available from the L.T.E.R.
was obtained, and catalogued in a spreadsheet. A bar
chart was created to show the available data from each
station for the period beginning in October, 1986 and
ending in September, 1993. There were 7 stations active
during this period, named A-l, B-l, C-l, D-l, Green Lake,
Saddle, and Arikaree. There were three types of Data
Collection Instrumentation in use at these stations; the
DP-211 Datapod, the DP-219 Datapod and the CR-21X
20


Datapod. The DP-211 data collectors recorded solar
radiation and temperature; the DP-219 recorded solar
radiation, temperature and evaporation; and the CR-21X
recorded solar radiation, temperature, humidity,
barometric pressure, wind speed, and wind direction.
These three types of recorders were installed at various
of the seven sites at various times during the period
between 1986 and 1993. In addition, precipitation gages
recorded both snow and rainfall at stations C-l and
Saddle. The precipitation gages used a system employing
ethylene glycol to melt the snow and prevent refreezing,
and an oil film to prevent evaporation. The gages were
read every 5 to 10 days, so the record is cumulative
rather than daily for these periods.
It was decided to base the model on temperature,
humidity, barometric pressure, solar radiation, wind
speed and precipitation. The precipitation records for
stations C-l and Saddle for this period do not begin
until January, 1988, so the starting point of the model
was moved to that date. In reviewing the temperature and
solar radiation data available, it was found that station
recorder A-l DP-219 and Arikaree DP-211 had the most
complete set of data. In addition, they were located at
very different elevations and allow for the accounting of
21


elevation effects on temperature. The CR-21X collector
at station C-l provided the most complete set of the
wind, humidity and barometric data.
Though the above collectors were the primary basis
of the model, other data collectors on the ridge were
used to replace faulty or missing data on the primary
data set. In each of the data sets, missing values were
replaced using data from the same type of collector at
another station. A correlation between the two sets was
identified, and the adjustment made on that basis. For
example, in cases where the Arikaree temperature was
missing, the A-l temperature was adjusted by the average
lapse rate between the two stations and inserted into the
Arikaree data set. The primary data set was selected
based on completeness, however, to minimize this type of
adjustment.
Data Distribution
The meteorological data was distributed spatially
according to elevation. To determine the precipitation
distribution for the basin, precipitation data from 4
gages on Niwot Ridge was used. Stations C-l, A-l and B-l
have a common data record from 1952 to 1964. Stations C-
1 and Saddle have a common data record from 1986 through
1993. Using the precipitation from gage C-l as a basis,
22


a monthly average index of each gage was calculated.
With the elevation of the gages, a monthly elevation-
precipitation curve was plotted. These curves are shown
in Figure 3-1.
The variation of wind speed was accounted for
similarly. The CR21-X datapods at stations C-l, D-l and
Saddle measured the wind speeds. Using the monthly
average scalar wind speeds at these three locations, the
relative wind speed versus elevation was plotted for each
month of the year. Because a portion of the basin was
located below the elevation of station C-l, the wind
speed at elevation 2700 meters was assumed to be 90
percent of station C-l. This is justified by the fact
that both elevations are in timber, and a linear
extrapolation between C-l and 2700 meters would have
made the wind speed at lower elevation unrealistically
slow. The monthly wind speed curves are shown in Figure
3-2.
Temperature was assumed to vary linearly with
elevation, with an average monthly lapse rate for both
the minimum and maximum daily temperature calculated
using temperature readings at stations A-l and Arikaree.
The geographic characteristics of the watershed were
obtained from a 3 arc-second resolution Digital Elevation
23


Model and a Digital Land Use Land Cover Map, both
obtained from the United States Geological Survey.
24


Precipitation Correction Factors
January through June
LL
d
o:
2000 2500 3000 3500 4000 4500
Bevation, Meters
-0- January
-A- February
-3 March
-0- April
May
June
PCF X Station C-1 Precip. = Elevation Adjusted Precipitation
Precipitation Correction Factors
July through Decerrber
2000 2500 3000 3500 4000 4500
--July
-A- August
-s- September
-0- October
-6- November
-#- December
Bevation, Meters
PCFX Station C-1 Precip. = Elevation Adjusted Precipitation
Figure 3-1
Precipitation Correction Curves
25


26


CHAPTER 4
MODEL CONSTRUCTION
Overview
The model consisted of two parts: the Geographical
Information System (GIS) portion where the spatial
analysis of the watershed was performed, and the
Hydrologic portion where the energy and water balance
analysis was performed. The GIS portion of the model
consisted of GRASS 4.1, a raster based GIS written by the
United States Army Construction Engineering Research
Laboratory. This GIS software was the tool used to
analyze the elevation, slope, aspect, and land use
characteristics of the site. The hydrologic analysis was
performed using a model written within the Modular
Modeling System, or MMS. MMS is a system that allows
linking of small hydrologic submodules into a whole
model. MMS also includes features that aid the
hydrologist in testing and calibrating models and in
sensitivity analysis and streamflow prediction. The
system has been developed jointly by the United States
Geological Survey, Water Resources Division; the United
27


States Bureau of Reclamation Upper Colorado Regional
Office, United States Department of Agriculture,
Agricultural Research Service; the TERRA Lab at Colorado
State University and the Center for Advanced Decision
Support for Water and Environmental Systems (CADSWES) at
the University of Colorado at Boulder. Figure 4-1 shows
the MMS interface, Figure 4-2 shows the MMS Parameter
Editor Screen
MMS was developed using modules first developed for
the Precipitation Runoff Modeling System or PRMS
(Leavesley, et al. 1983.) PRMS provides a library of
modules to model a large number of phenomena of interest
to hydrologists. Using MMS, these modules can be linked
together in different ways, modified, and linked with
other modules developed independently of PRMS. MMS
provides a programming protocol where modules can be
written and linked within existing models, allowing the
hydrologist to focus on the elements of the model within
their area of interest. This benefits the user in two
ways. First, it eases the programming task of model
development by allowing the easy reuse of non-modified
elements of the model. Secondly, it isolates the
modified elements of the model, allowing for easier
28


calibration and a better understanding of the influence
of the modifications.
Figure 4-1
MMS Interface
29


Figure 4-2
MMS Parameter Editing Screen
30


In this study, the PRMS model was modified to
calculate differently the convective heat transfer
involved in snowmelt. In PRMS, the convective heat
transfer is calculated using a monthly index. The energy
transferred between the air and the snowpack via
convection and condensation is a function of this monthly
index mutiplied by the air temperature. In this study,
the PRMS snowmelt module was modified to make the
convection-condensation energy transfer a function of
wind speed, humidity and barometric pressure. Using MMS,
it was possible to insert this change into PRMS and run
both PRMS and the Modified PRMS model on the same data
set, and review the influence of the modification.
PRMS Modeling
The Precipitation Runoff Modeling System is a
modular modeling system designed to evaluate the
influences of various hydrologic factors on storm runoff,
sediment yields, soil water relationships and ground
water recharge. The model's meteorological inputs are
air temperature, precipitation, and solar radiation. The
model contains modules that calculate the effect of
interception, infiltration, evaporation, snowmelt, and
groundwater recharge. Parameters that control these
phenomena can be input into the model in a structured
31


manner. The basis of this structure is the Hydrologic
Response Unit, or HRU.
Each HRU is a hydrologically similar area based on
slope, elevation, aspect, and land use or other factors.
If the model time step is significantly larger than the
basin's hydraulic travel time, it is not necessary that
an individual HRU be contiguous, only that it share
common hydrologic and hydraulic characteristics.
As the model is run, an energy and mass balance
analysis is performed on each HRU. All energy and water
that enters or leaves the HRU is accounted for, and these
influences on each of these HRUs is summed to provide the
watershed output. A flowchart of the PRMS process is
shown in Figure 4-3.
The portion of PRMS of greatest interest in this
study is the snowmelt module, known as "snowcomp.f" in
the MMS library. Flowcharts of both the original and
modified versions of this module are shown in Figure 4-4.
This module is based upon a snowpack system described by
Obled and Rosse (1977). It is an energy and mass balance
system where the snowpack is assumed to consist of a
surface layer and a lower layer. Energy is transferred
to and from the surface layer via convection, short and
long wave radiation, precipitation, sublimation and
32


long wave radiation, precipitation, sublimation and
evaporation. The surface layer of the snow also conducts
4
SOURCE: LEAVESLEY, ET AL. (1983)
Figure 4-3
PRMS Flowchart
33


A. UNMODIFIED PRMS
B. MODIFIED PRMS
Figure 4-4
Snowcomp.f Flowchart
34


energy to and from the lower layer. The snowpack is
treated as a reservoir, and as precipitation and melting
occur, the level of the reservoir rises and falls. PRMS
computes the snowpack water balance daily, and the energy
balance twice daily, for both day and night conditions.
"Snowcomp.f" computes the rate of convection-
condensation heat transfer based on a monthly index, CEC.
This monthly index is multiplied at each 12 hour time
step of the energy balance model by the average
temperature during this time step. The product of these
two factors, CECSUB, is the estimate of latent and
sensible heat flux for the time step, and has units of
calories:
CECSUB=CEC x Temperature
This approach to calculating latent and sensible
heat flux is a simplification. The complete equation by
the U.S. Army (1956) also includes terms for wind speed,
vapor pressure and atmospheric pressure. Snowcomp.f does
not directly account for these effects because they are
seldom measured in real watersheds, and they are
difficult to extrapolate in space and time. To account
for wind effects, snowcomp.f assumes HRUs within timber
receive one half of the magnitude of CECSUB, and to
account for vapor pressure, CECSUB is only calculated on
35


days of rainfall or when the ratio of observed to
potential shortwave radiation is less than or equal to
0.33.
In this study, CECSUB was calculated using the
relation from the U.S. Army (1956). The equation was
converted to SI units and now has the form:
CECSUB=0.3174 [ (Ta-Ts) (Pa/Ps)+ 15.46 (e.-e.) ] Vb
Where: Ta is Air Temperature, Degrees Celsius
Ts is Snow Temperature, 0 Celsius
Pa is Atmospheric Pressure, millibars
Ps is Standard Atmos. Pressure, 1013 mb
ea is vapor pressure of air, millibars
es is vapor pressure of snow, 6.11 mb
Vb is wind speed, meters per second
The vapor pressure of the air, ea, was calculated
using a method in Wallace and Hobbs (1977) First, the
saturated vapor pressure of the air as a function of
temperature was calculated using a relation that resulted
from a regression analysis of a National Weather Service
table from Ponce (1989):
0.0699T+1.81
esat=e
Using the saturated vapor pressure, the saturated
mixing ratio, wsat, was given by:
36


wsat=0.622 (esat/Pa)
The air mixing ratio was a product of the saturated
mixing ratio and the relative humidity:
w=RH(wsat)
Then, the vapor pressure of the air was calculated
by:
ea=w X P J (w+0.622)
In this study, the snowcomp.f module within the PRMS
library was modified using the 1956 U.S. Army equation
above. Modules to distribute the wind and atmospheric
pressure were written, and the model was run using daily
wind data from the Niwot Ridge L.T.E.R. data base. The
model was run under three conditions. First, the wind
was varied daily, and its velocity was assumed to vary
linearly, with elevation, throughout the watershed. The
wind velocity at each HRU was adjusted according to the
average elevation of each HRU, and this adjustment varied
monthly. Second, wind speed varied daily, but was assumed
to blow throughout the watershed at a uniform velocity.
The velocity used was the elevation adjusted velocity at
the watershed's average elevation. Third, the unmodified
PRMS model was run, using a monthly convection-
condensation coefficient, CEC, calculated using the 1956
37


U.S. Army CECSUB equation and monthly average
meteorological conditions. The three results were
compared to study the influence of wind distribution on
snowmelt.
Establishment of Hydrologic Response Units
An important part of running PRMS was the
establishment of the Hydrologic Response Units, or HRUs.
A GIS offers a convenient means of doing this. Using
methods similar to Frankoski (1994), the GRASS 4.1 GIS
was used to establish the HRUs for the North Fork St.
Vrain watershed. A Digital Elevation Model (DEM)
published by the United States Geological Survey was used
to develop the HRUs for the site. The "Greeley-W" USGS 1
degree quadrangle, a 1:250,000 scale 3 arc-second
resolution model, was used for this GIS analysis. The
North Fork St. Vrain Creek watershed was isolated from
this DEM using the "r.watershed" GRASS 4.1 command.
Next, the "r.slope.aspect" command was used to generate
maps of the slope and aspect of the watershed. The
original elevation map, and the derived slope and aspect
maps were then reclassified into three maps each with
three categories, according to the following:
38


Elevation(Meters)
Slope(Degrees)
Aspect(Degrees)
The above three maps, each with three
classifications, were crossed using the GRASS command
"r.cross," producing a map with 27 distinct
classifications (3 times 3 times 3). Because the PRMS
User's Guide, Leavesley, et al. (1983), suggests that
each HRU be no smaller than 4 or 5 percent of the basin
area, the eight smallest of these 27 classifications were
combined into classifications with the same elevation and
aspect, but differing slope. The final number of HRUs
was 19. A summary of the HRU's average characteristics
2529-3100 Zone 1
3101-3400 Zone 2
3401-4313 Zone 3
0-15 Zone 1
16-45 Zone 2
46-156 Zone 3
46-134
(North Facing) Zone 1
226-315
(South Facing) Zone 2
316-360, 0-45 & 135-225
(East-West Facing) Zone 3
39


is given in Table 1. Figures 4-5 through 4-7 show the
elevation , slope and aspect maps of the watershed.
Figure 4- 8 shows a view of the waterways of the basin,
generated by the GRASS "r.watershed" command, and Figure
4-9 shows the HRUs of the site.
HRU Elevation Aspect Slope Area
Meters Degrees Percent Sq. M.
1 2889 79 17.8 1670000
2 2833 288 17.3 1570000
3 2745 123 11.5 5300000
4 2868 96 56.1 4750000
5 2892 282 62.9 7280000
6 2894 232 53.4 5950000
7 2890 93 155.1 1350000
8 3222 173 16.2 3910000
9 3235 87 46.4 7940000
10 3260 231 52.7 12880000
11 3309 86 179.7 1220000
12 3253 259 166.4 2070000
13 3615 153 14.5 2270000
14 3570 85 50.9 4860000
15 3582 272 52.7 7040000
16 3591 202 57.0 6380000
17 3627 80 358.2 2550000
18 3703 268 203.2 5730000
19 3671 221 177.5 3140000
TABLE 1
HRU SUMMARY
40


I Zone 3
Figure 4-5
St. Vrain Basin Elevation Classifications
41


Figure 4-6
St.. Vraizx Basin Slope Classifications
42


fl Zone 3
Figure 4-7
St. Vrain Basin Aspect Classifications
43


Figure 4-8
St.Vrain Basin Watershed Display
44


Zone 1
Zone 2
Zone 3
Zone .4
Zone 5
Zone 6
Zone 7
Zone 8
Zone 9
Zone 10
Zone 11
Zone 12
Zone 13
Zone 14
Zone 15
Zone 16
Zone 17
Zone 18
Zone 19
Figure 4-9
St. Vrain Basin HRUs
45


The elevation, slope and aspect reclass operations
were performed using methods similar to Frankoski (1994).
Generally, the classification demarcations were selected
to obtain approximately 20 total HRUs. The elevation
reclassifications zones were chosen because they divided
the site approximately into thirds, by area.
Additionally, 3400 meters was chosen as a threshold
because it is approximately timberline elevation. The
aspect map was divided into north facing, south facing
and east-west facing. Frankoski divided her site into 3
elevation bands and 3 aspect bands, but she used 5 slope
bands. This resulted in 45 distinct classifications,
with an average size of just over 2 percent of the total
basin. Reducing these to twenty classifications, of
average size of 5 percent of the total basin, takes
considerable effort. It was decided to reduce the slope
reclassification to three bands, giving 27 (3 times 3
time 3) crossed categories, reducing the manual
recombination required to obtain 20 categories.
After the establishment of the HRUs, the land cover
characteristics of each had to be determined. The HRU
map was overlaid on a USGS Land Use and Land Cover Map.
Using r.stats, a summary of the vegetation types present
46


within each HRU was obtained. Using this summary/ the
predominant vegetation type of each HRU was determined,
and tabulated for use by the PRMS and Modified PRMS
models. This is shown in Table 2.
HRU_______Vegetation Type
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Pine Forest
Pine Forest
Pine Forest
Pine Forest
Pine Forest
Pine Forest
Pine Forest
Pine Forest
Pine Forest
Pine Forest
60% Tundra/40% Forest
40% Tundra/60% Forest
Bare Ground
Herbaceous Tundra
Herbaceous Tundra
Herbaceous Tundra
Herbaceous Tundra
Herbaceous Tundra
Herbaceous Tundra
TABLE 2
HRU VEGETATION
47


Establishment of PRMS Parameters
In hydrologic modeling, numerous parameters are used
within the model to account for evaporation and
sublimation, infiltration, interception, precipitation
distribution, temperature distribution, and solar
radiation effects. The modular modeling system includes
a spreadsheet system to provide a convenient means to
input and edit these parameters. The parameters take
four general forms: constant through space and time;
constant through space but variable through time;
variable through space but constant through time; and
variable through space and time. Calculating, cataloging
and calibrating these parameters is a formidable part of
running the PRMS and Modified PRMS models. It is beyond
the scope of this work to completely describe each of
these parameters, but a short description of each is
given here. Parameters of special interest to this study
are described more fully in the appendix, but for added
detail on the remainder, the reader is referred to the
PRMS manual (Leavesley, et al. 1983) and the MMS manual,
(CADSWES, 1994).
48


Parameters Constant in Space and Time
albset_rna albset_rnm albset_sna aUbset_snm Thresholds used to reset snow albedo, or reflectivity.
basin area Total area of the basin.
basin_tsta Index of temperature station.
den_init den_max Initial and maximum density of new fallen snow.
emiss_noppt Average emissivity of air.
free2o_cap Free water holding capacity of snowpack.
melt_force Date that model forces snowpack into spring snowmelt stage.
melt_look Date that model begins to look for spring snowmelt stage.
moyrsum Switch for monthly and yearly HRU sum.
pmo Print switch for HRU summary.
potet_sublim Fraction of the potential evapotranspiration sublimated.
print_freq Frequency of the output of the model. Month, Day or Year
49


radadj_intcp radad j_s lope radadj_sppt radadj_wppt radmax- Radiation adjustment factors for season, slope, and weather
settle_const Snowpack settlement time constant.
temp_units Defines whether Fahrenheit or Celsius degrees are used.
tmax_allsnow Temperature threshold controlling mixed snow/rain and all snow events.
Parameters Varying in Time. Constant in Space
adjm Amount of rain in a mixed rain- snow event
cecn_coef Convection-condensation heat transfer coefficient
dday_intcp- dday_slope- Parameters to adjust the amount radiation on an HRU.
epan_coef Evaporation pan coefficient.
jh_coef- Air temperature coefficient for evapotranspiration computations
tmax_allrain Temperature threshold above which a precipitation event is entirely rain.
50


tmax_index Determines precipitation adjustments due to radiation.
tmax_lapse tmin_lapse The rates at which the daily temperature decreases with increased elevation.
tstorm_mo Identifies whether frontal or convective storms predominate
Parameters Constant in Time. Varyina in Space
carea_max Maximum contributing area of an HRU to the surface runoff.
cov_type Vegetation cover type for HRU.
covden_sum covden_win Vegetation density in summer and winter for an HRU.
hru_area Area, in acres, of each HRU.
hru_dplcrv An index identifying a area- water equivalent depletion curve.
hru_elev- The elevation, in feet, of each HRU.
hru_gwres- An index identifying the groundwater reservoir for each HRU.
hru_percent A number representing the fraction of impervious area for each HRU.
51


hru psta Index identifying precipitation station for each HRU.
hru_rad.pl Index identifying raditation planes for each HRU.
hru_slope Average steepness of each HRU
hru_ssres An index identifying the subsurface reservoir for each HRU.
hru_tsta Index identifying temperature station for each HRU.
imperv_stor Maximum impervious storage, in inches, for each HRU
j h_coef f_hru Air temperature coefficient used for Jensen Haise evapotranspiration calculations.
pres Pressure adjustment factor (modified PRMS only
rad_tmcf Radiation transfer coefficient through vegetation canopy for HRU.
smidx_coef Coefficient used in area contributing algorithm.
smidx_exp Exponent used in area contributing algorithm.
snow_incp Amount of snow interception storage, in inches, of the major vegetation in the HRU.
52


snowinf i l_ma Maximum snow infiltration, per day in inches.
soil2gw_max Maximum amount of soil water, in inches, routed to the groundwater reservoir each day.
soil_moist_i soil_ittoist_m s oil_rechr_i s oil_rechr_m Initial and maximum soil, moisture and recharge zone
soil type Index of soil type
srain_intcp Interception storage capacity
tmax_adj tmin_adj Adjustment of temperature based on slope and aspect
transp_beg Month of commencement of evapotranspiration computations
transp_end Month of end of evapotranspiration computations
trans_tmaxf Index, to control initiation of evapotranspiration computations
wrain_intcp Amount of winter rain intercepted by vegetation
gwflow_coef gwsink gwstor_init Groundwater routing and storage coefficients.
radpl_aspect radpl_lat radpl_slope Aspect,latitude and slope value for each of 10 radiation planes.
53


snarea_curve Snow area depletion curve
parameters.
ss2gw_exp
ss2gw_rate
ssr crwres
ssr_coef_lin Subsurface routing
ssr_coef_sq parameters
s s r_max_co e f
ssstor_init
tsta_elev Elevation of temperature
recording station.
Parameters Varying in Space and Time
rain_adj
snow_adj
Monthly precipitation
correction factor for each HRU
wind_adj
A monthly factor to adjust each
HRU's wind speed for elevation.
Using the above parameters, the model was run. As
is obvious, many of the parameters have little to do
directly with snowmelt, but are necessary to describe the
hydrologic routing of the basin. This is particularly
true with the subsurface and groundwater reservoir
parameters in the model. It was attempted to refine the
methods used in estimating these parameters during the
calibration stage of the project. The focus of the study
was the effect of wind on snowmelt, so parameters with
54


little direct effect on convective snowmelt received less
attention than those of greater importance.
55


CHAPTER 5
MODEL CALIBRATION AND RESULTS
The models were calibrated using the U.S.G.S.
streamflow gage records of North St. Vrain Creek at
Allenspark, Colorado. Since it was decided to measure
the influence of wind on the snowmelt, all three models
were calibrated identically, so the influence of wind
could be isolated. Three criteria were used in
calibration, in order of importance, yearly runoff
volume, yearly peak flow magnitude, and hydrograph shape.
Though many factors could be used to adjust the output of
the model, the primary ones used in this study were
precipitation distribution, interception influences and
evaporation influences.
The models were run from October 1, 1988, through
September 30, 1993. A significant problem in a model
such as this is setting the initial variable conditions.
The runoff from a watershed is dependent on many
antecedent conditions. The runoff on any given day is a
result of meteorological conditions of the day, month and
year prior.
MMS provides a reasonable method to set these
initial conditions. The output of a prior run of the
model can be used as initial conditions for a later run.
56


In this case, the model was run using null initial
conditions, for the duration of the data record. The
model variable states from September 30, 1993 were saved
in an initialization file, and this file was used to
initialize the same variables on October 1, 1988 on a
second run. In the first run, the six year duration
allows the model to rid itself of the influence of the
null initial conditions. The variables on September 30,
1993 are assumed to have reached a steady state.
Recycling them back into the model gives a more realistic
picture of the true situation in the early stages of the
model.
Once the model was properly initialized, it was
attempted to balance the predicted yearly runoff totals
with the recorded. Initial runs of the model provided an
average 75 percent excess in volume, and a 110 percent
excess in peak flows.. The vegetation interception,
infiltration, and evaporation coefficients were increased
in an attempt to account for this excess water. Finally,
the precipitation correction factors were all reduced by
15 percent. This provided a more reasonable fit for the
data.
The modified model was run twice, once to
demonstrate the effect of spatially varying the wind
57


speed, and once to demonstrate the effect of assuming the
daily average wind speed throughout the watershed. The
non-modified form of the model was run, using a monthly
convection-condensation coefficient to account for
sensitive heat transfer. The following figures show
observed versus predicted flows for the three model runs
for water years 1990 and 1991.
58


Water Year 1990
Vobs=14612 cfs-yr Vmod=14742 cfs-yr i
Qobs=270 cfs Qmod-279 cfs RMS=40 |
Observed and Predicted Runoff
Convection Coefficient Model
Vobs=18039 cfs-yr Vmod=18870 cfs-yr j
Qobs=415 cfs Qmod=428 cfs RMS=48 I
Predicted Runoff
Observed Runoff
Figure 5-1
Convection Coefficient Model
59


Water Year 1990
Vobs=14612 cfs-yr Vmod=15089 cfs-yr
Qobs=270 cfs Qmod=259 cfs RMS=42
Figure 5-2
Wind Varied Temporally Model
60


Water Year 1990
Vobs=14612 cfs-yr Vmod=15193 cfs-yr |
Qobs=270 cfs Qmod=271 cfs RMS=431
Observed and Predicted Runoff
Wind Varied Spatially and Temporally
Water Year 1991
Vobs=18039 cfs-yr Vmod=19314 cfs-yr
Qobs=415cfs Qmod=438 cfs RMS=49
Predicted Runoff :
>
Observed Runoff ;
Figure 5-3
Wind Varied Spatially and Temporally Model
61


CHAPTER 6
CONCLUSIONS
In this study, three situations were modeled. The
effect of wind and condensation was first modeled using a
monthly convection-condensation coefficient, calculated
using average monthly wind speed, humidity and
temperature data. This monthly coefficient was
multiplied by a daily temperature index to calculate the
sensitive heat transfer on the watershed, and this heat
transfer was translated to snowmelt. Second, the model
was modified to account for daily wind speed in the
watershed. A spatially uniform wind was assumed to blow
throughout the watershed, providing heat transfer as a
function of this daily wind speed, temperature and
humidity. Finally, the wind model was adjusted to vary
the wind speed according to elevation. Higher elevations
in the Niwot Ridge area have greater wind speeds. Thus,
the model assumed that HRUs with higher elevations had
higher wind velocities, and the heat transfer in each HRU
was calculated using this elevation adjusted wind speed.
The wind adjusted model runs provided very similar
results. Yearly volume totals are within 1 percent, and
62


the majority of daily totals are identical. The yearly
volume totals for the monthly convective coefficient
model are about 10 percent below both of the wind
adjusted models.
Two major conclusions were drawn from this study.
The first was that spatially distributed wind velocity
has very little influence on snowmelt, in most instances.
In the alpine environment of the North St. Vrain Creek
watershed, temperatures seldom rose significantly above
freezing. If the air temperature was below freezing, no
matter what the wind velocity, the snow did. not melt
through convection. Though the wind at the upper end of
this basin was modeled to blow 4 to 5 times faster than
at the stream gage, it was seldom warm enough to melt
snow, and when it did, the snow at the lower levels had
long been melted.
The second major conclusion was that time varied
wind data did make a difference in snowmelt calculations.
The fact that the wind adjusted models showed a 10%
increase in runoff over the convective coefficient model
demonstrates this. Averaging wind, humidity and
temperature, and calculating the convective heat transfer
coefficient, CEC, underestimated runoff. This problem
could be solved several ways; either by adding a
63


correction factor to the calculated CEC to increase the
magnitude of the convective runoff, or by using daily
wind velocities, averaged over the basin, to calculate
the sensitive heat transfer In basins with constant wind
speed, the difference between the lumped approach and the
daily approach is small, but in areas of widely varying
wind speed, the daily wind can have significant effect.
As this study was developed within the Modular
Modeling System, several observations regarding the
system are warranted. MMS provided an effective means to
construct linked modular models, and an easily learned
programming protocol to construct and modify the modules.
The library of infiltration, evaporation, and groundwater
modules allowed the construction of a complex model
without the need to independently develop these modules.
The tasks of input/output and execution programming were
handled by the MMS shell, allowing the author's limited
programming skills to be focused on the problem of
snowmelt. Modular systems such as. MMS will likely play
an important role in the study of hydrology and other
phenomena that result from a combination of a large
number of natural processes.
64


Suggestions for Further Study
The primary conclusion in this study was that
snowmelt in alpine areas is not appreciably affected by
spatial wind distribution. Temperatures in the.alpine
basin studied were often too cold to result in
significant convective heat transfer to the snowpack,
regardless of the wind speed. Regions at lower
elevations and with different precipitation, wind and
temperature patterns might possibly have a different
response than the St. Vrain Basin. The effect of polar
latitudes, where solar radiation influences are more
extreme, would be interesting. A similar study to this
one on a site on the eastern slope of the Rockies in
Alberta, Canada would be interesting to compare, as would
a study in an equatorial region.
In comparing the PRMS and Modified PRMS models on an
annualized basis, the differences are notable, but not
huge. Using the models in forecasting spring runoff
floods could provide a more dramatic difference between
the two approaches.
There are many basic areas in the field of snowmelt
that warrant further study. Nearly all of the conceptual
65


work on snowmelt has been done on point models, and these1
point models have been extrapolated to basins. More
study needs to be done identifying phenomena that cannot
be modeled using a flat-plane point model. The effect of
wind on evaporation, and the spatial variation of
evaporation and sublimation are two areas that do not
seem to be well understood. For example, the phenomenon
of fresh snow being blown off ridge tops, and then
refalling or sublimating, is very localized, and
dependent on wind speed, duration and temperature. A GIS
offers a convenient tool for identifying the locations
where such phenomena occur. The influence of this
phenomenon on windward versus leeward ridges could
provide some interesting insights in the study of water
resources and avalanches.
The effect of infiltration in high alpine areas, is
another problem that has not been adequately addressed.
In the North St. Vrain Creek watershed, upper reaches of
the snowmelt basin consist of granite talus on cracked
and faulted bedrock. A proven infiltration model of this
highly irregular material does not exist. Similarly,
this type of irregular material, though frozen, continues
to have interstitial spaces through which water flows.
66


Most models assume frozen ground does not support
infiltration.
The properties of snow can be very different
depending on age, meteorological conditions when formed,
climate where formed, and meteorological conditions since
formation. As snow ages and becomes ripe for melting, it
becomes more homogeneous with ripe snow the world over.
However, new fallen snow is very different by region, and
even by storm, and methods for accounting for these
differences could be developed.
67


APPENDIX
Further Discussion of Parameter Properties
Parameters Constant in Space and Time
albset_ma-Proportion of rain in a rain-snow mix
precipitation event above which albedo is not set.
Applied during the snowpack accumulation stage.
Snowpack albedo is the reflectivity of the snow and
is used in radiative heat transfer computations.
New fallen snow is more reflective than old snow.
Snowcomp.f sets an initial albedo at the time snow
falls, and the albedo decays with time as the snow
crystals age, coalesce and become dirty. This
parameter sets a threshold in a rain-snow mixed
event upon which the snowpack albedo is not reset to
the initial value.
albset__ram-Proportion of rain in a rain-snow mix
precipitation event above which albedo is not set.
Applied during the snowpack melt stage.
albset_sna-Minimum snowfall, in water equivalent, needed
to reset snow albedo during snowpack accumulation
stage.
albset_snm-Minimum snowfall, in water equivalent, needed
to reset snow albedo during snowpack melt stage.
68


basin area-Total area of the basin, in acres.
basin_tsta-Index of temperature recording station. In
this study, one station was used within the model.
However, a lapse rate was calculated using a second
station, so effectively 2 stations were used.
den_init-Initial density of new fallen snow.
den_max-Aver age maximum density of new fallen snow.
emiss_noppt-Average emissivity of air on days without
precipitation. Used in radiation transfer
calculations.
£ree2o_cap-Free water holding capacity of snowpack
expressed as a fraction of total snowpack water
equivalent.
melt_force-Date that model forces snowpack into spring
snowmelt stage. Varies with region depending on
length of time that permanent snowpack exists. The
melt phase is defined as when the snowpack is at 0
Celsius for 5 consecutive days, or when the melt
force date is passed. Some studies refer to the
melt stage as a "ripe" snowpack. The heat transfer
characteristics of melting snow are different from
the characteristics of accumulating snow, and this
parameter helps define the state of the snowpack to
the model.
69


melt_look-Date that model begins to look for spring
snowmelt stage.
moyrsum-Switch for monthly and yearly HRU sum.
pmo-Switch to print month form HRU summary.
potet_sublim-Fraction of the potential evapotranspiration
that is sublimated from the snow surface.
print_freq-Controls the frequency of the output of the
model. Can be adjusted for daily, monthly or
yearly.
radadj_intcp-Parameter to adjust the amount of radiation
delivered to a sloped surface.
radadj_slope-Parameter to adjust the amount of radiation
delivered to a sloped surface.
radadj_sppt-Adjustment for the maximum solar radiation
delivered on a summer day with greater than 0.02
inches precipitation.
radadj_wpptAdjustment for the maximum solar radiation
delivered on a winter day with greater than 0.02
inches precipitation.
radmax-Maximum amount of solar radiation that can reach
the ground due to haze, dust smog, etc.
settle_const-Snowpack settlement time constant.
temp_units-Defines whether Fahrenheit or Celsius degrees
are used. Celsius was used in this model.
70


tmax_allsnow-Temperature threshold below which
precipitation is assumed to be entirely snow. Above
this threshold, a mixed rain-snow event can occur.
Parameters Varying in Time. Constant in Space
These parameters are constant for all HRUs, but are
variable on a monthly basis.
adjm-Amount of rain in a mixed rain-snow event
cecn_coef-Also known as CEC, Convection-condensation heat
transfer coefficient. It is only used in the
P.R.M.S. model. In the modified P.R.M.S. model, it
is the element replaced by the 1956 U.S. Army CECSUB
relation. In this study, the P.R.M.S. model was run
to compare the effect of wind distribution on
snowmelt. The 1956 CECSUB equation was used to
calculate the heat transferred into the snowpack for
each month using average recorded values of wind,
temperature, humidity and atmospheric pressure
adjusted for the average elevation of the watershed.
These values were divided by the average monthly
temperature to obtain CEC. This CEC value was used
within the PRMS model
dday_intcp-Parameter to adjust the amount of radiation
delivered to a HRU.
71


dday_slope-Parameter to adjust the amount of radiation
delivered to a HRU.
epan_coef-Evaporation pan coefficient. Called by the
model, but not used.
jh_coef-Monthly air temperature coefficient used in
Jensen-Haise potential evapotranspiration
computations. Given by the relations:
jh_coef=l/[C1+(13.OxCH)]
Cl=68.0-[3.6 X ELEV/1000]
CH=50/(e2-el)
Where: ELEV is the mean elevation of the watershed in
feet.
e2 is the saturation vapor pressure of the
maximum mean air temperature, in
millibars.
el is the saturation vapor pressure for the
minimum mean air temperature, in
millibars.
tmax_allrain-Temperature threshold above which a
precipitation event is entirely rain. Below this
threshold, a mixed rain-snow event can occur.
tmax_index-Index temperature used to determine
precipitation adjustments due to solar radiation.
72


tmax_lapse-The rate at which the daily maximum
temperature decreases with increased elevation.
Calculated using monthly averages of daily lapse
rates of Niwot Ridge L.T.E.R. data from stations A-l
(Elevation=2195 Meters) and Arikaree (Elevation=3798
Meters)
tmin__lapse- The rate at which the daily minimum
temperature decreases with increased elevation.
Calculated using monthly averages of daily lapse
rates of Niwot Ridge L.T.E.R. data from stations A-l
(Elevation=2195 Meters) and Arikaree (Elevation=3798
Meters)
tstorm_mo-An index to identify the predominant storm type
that occurs during the month. Summer months are
convective storms, while winter months are frontal
storms.
Parameters Constant in Time, Varying in Space
carea_max-Maximum contributing area of an HRU to the
surface runoff.
cov_type-Vegetation cover type for HRU. 0 denotes bare
soil, 1 denotes grass, 2 denotes shrubs, and 3
denotes trees.
73


covden_sum-Vegetation density in summer for the
predominant vegetation in the HRU.
covden_win-Vegetation density in winter for the
predominant vegetation in the HRU.
hru_area-The area, in acres, of each HRU.
hru_dplcrv-An index identifying a area-water equivalent
depletion curve for the basins snowpack. This study
used one.
hru_elev-The elevation above sea level, in feet, of each
HRU.
hru crwres-An index identifying the groundwater reservoir
for each HRU.
hru_percent-A number representing the fraction of
impervious area for each HRU. Held at zero for all
HRUs in this study.
hru psta-Index identifying precipitation station for each
HRU. Two used in this study.
hru_radpl-Index identifying precipitation station for
each HRU. Ten used in this study, one to cover flat
planes, and nine others to cover north, south, and
east-west facing slopes of small, medium and large
steepness.
hru_slope-Average steepness of each HRU in percent.
74


hru_ssres-An index identifying the subsurface reservoir
for each HRU. One used in this study.
hru_tsta-Index identifying temperature station for each
HRU. One used in this study, however, lapse rates
calculated using a second.
imperv_stor-Maximum impervious storage, in inches, for
each HRU. Since site was assumed to have no
impervious area, this was held at zero.
jh_coeff_hru-Air temperature coefficient used for Jensen
Haise evapotranspiration calculations. Given by:
JH=27.5-0.25 (e^e,)-El/1000
\
Where: e2 is saturated vapor pressure-mean
maximum air temperature.
e2 is saturated vapor pressure-mean
minimum air temperature.
El is mean elevation, in feet, of HRU.
pres-Pressure adjustment factor (modified PRMS only),
used to. adjust barometric readings taken at station
C-l. Multiplicative adjustment.
rad_tmcf-Radiation transfer coefficient through
vegetation canopy for HRU.
smidx_coef-Coefficient used in area contributing
algorithm.
smidx_exp-Exponent used in area contributing algorithm.
75


snarea_thresh-Threshold water equivalent below which the
snow covered area curve is applied.
snow_incp-Amount of snow interception storage, in inches,
of the major vegetation in the HRU.
snowinfil_ma-Maximum snow infiltration, per day in
inches.
soil2gw_jmax-Maximum amount of soil water, in inches,
routed to the groundwater reservoir each day.
soil_moist_i-Initial soil moisture, inches.
soil_moist_m-Maximum soil moisture, inches.
soil_rechr_i-Initial soil recharge zone, must be less
than initial soil moisture.
soil_rechr_m-Maximum soil recharge zone, must be less
than maximum soil moisture,
soil type-index of soil type. 1 indicates sand, 2
indicates loam, and 3 indicates clay. No soil
surveys have been performed on this basin, so
forested areas were assumed to be loam, and areas
above timberline were assumed to be sand.
srain_intcp-Interception storage capacity for the major
vegetation type of the HRU, during summer rains.
tmax_adj-Adjustment of maximum temperature based on slope
and aspect. In this study, southerly facing slopes
76


were assumed to be warmed an additional 4 degrees,
and east west slopes by 2 degrees.
tmin_adj-Adjustment of minimum temperature based on slope
and aspect. In this study, southerly facing slopes
were assumed to be warmed an additional 4 degrees,
and east west slopes by 2 degrees.
transp_beg-Month of commencement of evapotranspiration
computations. Assumed to be April.
transp_end-Month of end of evapotranspiration
computations. Assumed to be October.
trans_tmaxf-Cumulative Temperature index, to control
initiation of evapotranspiration computations. In
the month of commencement, the daily temperatures
are summed until they reach this threshold, allowing
evapotranspiration to commence.
wrain_intcp-Amount of winter rain intercepted by
vegetation, inches.
gwflow_coef-Groundwater routing coefficient.
gwsink_coef-Groundwater storage coefficient.
gwstor_init-Initial groundwater storage.
radpl_aspect-Aspect value for each of 10 radiation
planes. Held at north, south, and east-west for
each HRUS of low, moderate and steep slope.
77


radpl_lat-Latitude value for each of 10 radiation planes.
Held at 40 degrees north.
radpl_slope-Slope value for each of 10 radiation planes.
snarea_curve-Ordinal values for area-water equivalent
snow area depletion curve.
ss2gw_exp-Subsurface routing exponent controlling the
manner which subsurface reservoir water is
transported to the groundwater reservoir.
ss2gw_rate-Subsurface routing coefficient controlling the
manner which subsurface reservoir water is
transported to the groundwater reservoir,
ssr crwres-Index of the groundwater reservoir receiving
water from the subsurface reservoir.
ssr_coef_lin-Subsurface routing coefficient controlling
the manner which subsurface reservoir water is
transported to streamflow.
ssr_coef_sq-Subsurface routing coefficient controlling
the manner which subsurface reservoir water is
transported to streamflow.
ssr_max_coef-Coefficient controlling the manner which
subsurface reservoir water is transported to the
groundwater reservoir. Recommended value of 1.00.
ssstor_init-Initial storage in inches for each subsurface
reservoir.
78


tsta_elev-Elevation of temperature recording station.
79


Parameters Varying in Space and Time
rain_adj-Monthly precipitation correction factor for each
HRU. This factor, multiplied by a precipitation
data point, gives an elevation adjusted
precipitation for each HRU. Since precipitation
patterns differ according to time of year, these
factors were calculated for each month, for each of
the 19 HRUs. Lower elevation HRUs were tied to the
precipitation station C-l (3018 meters) record,
while upper were tied to the precipitation station
Saddle (3525 meters) record. Used in both modified
and unmodified PRMS study. The curves shown in
chapter 3 are the basis for the adjustment.
snow_adj-Since the precipitation gages did not
differentiate between snow and rain, the factors
used for rain_adj were used also for snow_adj.
wind_adj-A monthly factor to adjust each HRU's wind speed
for elevation. The data record of station C-l was
most complete, so the model was indexed to this
station. Using wind speed-elevation correction
curve plotted with Niwot Ridge data at stations C-l,
D-l and Saddle in chapter 3, an elevation correction
factor for each HRU for each month was calculated
80


Used only in the modified PRMS model. For one run,
the adjustment was held constant, to give a daily
average velocity of wind, at the average elevation
of the watershed. In another run, the wind speed
was varied with each HRU's elevation.
81


REFERENCES
Anderson, E.A. (1976) A point energy and mass balance
model of snow cover. In: NOAA Tech Report NWS no 19.
U.S. Dept, of Commerce, Silver Spring, Maryland,
USA.
Bloschl, G., Kirnbauer, R. & Gutknecht, D. (1991) A
spatially distributed snowmelt model for application
in alpine terrain. In: Snow, Hydrology and Forests
in High Alpine Areas (Proceedings of the Vienna
Symposium, August 1991.) I.A.H.S. Publ. no. 205.
Braun, L.N. (1991) Modeling of the snow-water equivalent
in the mountain environment. In: Snow. Hydrology and
Forests in High Alpine Areas (Proceedings of the
Vienna Symposium, August 1991.) I.A.H.S. Publ. no.
205.
Center for Advanced Decision Support for Water and
Environmental Systems, "CADSWES". (1994) Modular
Modeling System-Users Manual, University of
Colorado, Department of Civil, Environmental and
Architectural Engineering, Boulder, Colorado, U.S.A.
Frankoski, L (1994) Effect of spatial resolution on
hydrologic model results. Master'S Thesis,
University of Colorado, Boulder, Colorado, U.S.A.
Hottellet, Ch., Braun, L.N., Leibundgut, Ch., & Rieg, A.
(1993) Simulation of snowpack and discharge in an
alpine karst basin. In: Snow and Glacier Hydrology
(Proceedings of the Kathmandu Symposium, November,
1992.) I.A.H.S. Publ. no. 205
Kumar, V.S., Haefner, H. & Seidel, K. (1991) Satellite
snow cover mapping and snowmelt runoff modeling in
Beas basin. In: Snow, Hydrology and Forests in High
Alpine Areas (Proceedings of the Vienna Symposium,
August 1991.) I.A.H.S. Publ. no. 205.
Leavesley, G.H.,(1989) Problems of Snowmelt Runoff
Modelling for a Variety of Physiographic and
Climactic Conditions. In: Hydrological Sciences
Journal,V 34, No. 6 December, 1989
82


Leavesley, G.H., Lichty, R.W., Troutman, B.M. & Saindon,
L.G. (1983) Precipitation Runoff Modeling System:
Users Manual. Water-Resources Investigations 83-
4238 United States Geological Survey, Denver,
Colorado, U.S.A.
Obled, Charles, & Rosse, B.B., (1977) Mathematical Models
of a Melting Snowpack at an Index Plot. In: Journal
of Hydrology, no. 32
Ponce, V.M.,(1989) Engineering Hydrology, Prentice Hall,
Englewood Cliffs, New Jersey, U.S.A.
Price, A.J., Dunne, T. & Colbeck, S.C. (1976) Energy
balance and runoff from a subarctic snowpack. In:
CRREL Report 76-27. Corps of Engineers, U.S. Army,
Cold Regions Research and Engineering Laboratory,
Hanover, New Hampshire, U.S.A.
Ranzi, R. & Rosso, R. (1991) A physically based approach
to modeling distributed snowmelt in a small alpine
catchment. In: Snow. Hydrology- and Forests in High
Alpine Areas (Proceedings of the Vienna Symposium,
August 1991.) I.A.H.S. Publ. no. 205.
Summerfeld, R.A., Musselman, R.C., Wooldridge, G.L. &
Conrad, M.A.(1991) The performance of a simple
degree-day estimate of snow accumulation to an
alpine watershed. In: Snowr Hydrology and Forests in
High Alpine Areas (Proceedings of the Vienna
Symposium, August 1991.) I.A.H.S. Publ. no. 205.
U.S. Army (1956) Snow Hydrology. U.S. Army Corps Of
Engineers, Portland, Oregon, U.S.A.
Wallace, J.M., Hobbs, P.V. (1977) Atmospheric Science-An
Introductory Survey. Academic Press, Orlando,
Florida, U.S.A.
Wankiewicz, A. (1991) Mountain snowpack observations by
microwave satellite. In: Snow. Hydrology and Forests
in High Alpine Areas (Proceedings of the Vienna
Symposium, August 1991.) I.A.H.S. Publ. no. 205.
World Meteorological Organization (1986) Intercomparison
of models of snowmelt runoff. Operational Hydrology
Report 23, World Meteorological Organization,
Geneva, Switzerland.
83


Xianzhang, C., Qunzhu, Z., & Yongchao, L. (1991)
Satellite snow cover monitoring and snowmelt runoff
prediction in the high alpine area of northwestern
China. In: Snow. Hydrology and Forests in High
Alpine Areas (Proceedings of the Vienna Symposium,
August 1991.) I.A.H.S. Publ. no. 205.
84