Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005

Prepared in cooperation with Colorado Department of Public Health and Environment

Scientific Investigations Report 2007–5287

U.S. Department of the Interior U.S. Geological Survey Cover photograph. Agricultural area in foreground with Mancos Shale outcropping at the base of the Grand Mesa in the background, Whitewater Creek area (photograph by Cory A. Williams, U.S. Geological Survey). Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005

By Judith C. Thomas, Kenneth J. Leib, and John W. Mayo

Prepared in cooperation with Colorado Department of Public Health and Environment

Scientific Investigations Report 2007–5287

U.S. Department of the Interior U.S. Geological Survey U.S. Department of the Interior DIRK KEMPTHORNE, Secretary U.S. Geological Survey Mark D. Myers, Director

U.S. Geological Survey, Reston, Virginia: 2008

For product and ordering information: World Wide Web: http://www.usgs.gov/pubprod Telephone: 1-888-ASK-USGS

For more information on the USGS—the Federal source for science about the Earth, its natural and living resources, natural hazards, and the environment: World Wide Web: http://www.usgs.gov Telephone: 1-888-ASK-USGS

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Although this report is in the public domain, permission must be secured from the individual copyright owners to reproduce any copyrighted material contained within this report.

Suggested citation: Thomas, J.C., Leib, K.J., Mayo, J.W., 2008, Analysis of dissolved selenium loading for selected sites in the Lower Gunnison River Basin, Colorado, 1978–2005: Reston, VA, U.S. Geological Survey Scientific Investigations Report 2007–5287, 25 p. iii

Contents

Abstract...... 1 Introduction...... 1 Purpose and Scope...... 2 Acknowledgments...... 2 Description of Study Area...... 2 Summary of Selenium Issues in Study Area...... 4 Methods for Analysis of Selenium Loading...... 6 Site Selection...... 6 Estimating Selenium Loads, 85th Percentile Selenium Concentrations, and Load Reductions...... 8 Statistical Summary of Available Data...... 8 Load Estimation Using Annual Time-Weighted Means...... 8 Load Estimation Using Regression Analysis...... 9 85th Percentile Selenium Concentrations...... 9 Load Reductions...... 10 Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin...... 10 North Fork Gunnison River Basin and Red Rock Canyon at the Mouth...... 10 Uncompahgre River Basin...... 11 Gunnison River from North Fork Gunnison River to Grand Junction, Colorado...... 19 Applicability of Selenium Loading Analysis...... 21 Summary...... 22 References Cited...... 24

Figures

1–6. Maps showing: 1. Location of study area, sites, and 303(d) list segments within the Lower Gunnison River Basin, western Colorado...... 3 2. Select land-use types in the Lower Gunnison River Basin, western Colorado...... 4 3. Selected geologic units outcropping in the Lower Gunnison River Basin, western Colorado...... 5 4. Location of sites and land use in the North Fork Gunnison River Basin and Red Rock Canyon at the mouth (site 13), western Colorado...... 11 5. Location of sites and land use in the Uncompahgre River Basin, western Colorado...... 15 6. Location of sites and land use associated with the Gunnison River from North Fork . Gunnison River to Grand Junction, western Colorado...... 20 7. Graph showing relation between flow-duration curve (water years 1976–2005) and mean annual streamflow for two periods, 1997–2001 and 2001–05, at Gunnison River near Grand Junction, Colorado, western Colorado...... 22 iv

Tables

1. List of sites that represent the cumulative contribution of selenium concentration to 303(d) list segments in the Lower Gunnison River Basin, western Colorado, 1978–2005...... 7 2. Statistical summary of selenium concentration data for sites with limited data sets in the Lower Gunnison River Basin, western Colorado, 1978–2005...... 12 3. Selenium loading analyses for sites using average time-weighted mean (ATWM) method, Lower Gunnison River Basin, western Colorado, 1992–2005...... 14 4. Summary of regression model diagnostics at sites, Lower Gunnison River Basin, western Colorado, 2001–05...... 17 5. Selenium loading analyses for sites using regression analysis (LOADEST) method, Lower Gunnison River Basin, western Colorado, 2001–05 water years...... 18

Conversion Factors, Abbreviations, and Datums

Multiply By To obtain Length inch (in.) 2.54 centimeter (cm) foot (ft) 0.3048 meter (m) mile (mi) 1.609 kilometer (km) Flow rate cubic foot per second (ft3/s) 0.02832 cubic meter per second (m3/s) Mass pound, avoirdupois (lb) 0.4536 kilogram (kg) pound per day (lb/d) 0.4536 kilogram per day (kg/d) pound per year (lb/y) 0.4536 kilogram per year (kg/y)

Temperature in degrees Celsius (°C) may be converted to degrees Fahrenheit (°F) as follows: °F=(1.8×°C)+32 Temperature in degrees Fahrenheit (°F) may be converted to degrees Celsius (°C) as follows: °C=(°F-32)/1.8 Vertical coordinate information is referenced to the North American Vertical Datum of 1988 (NAVD 88). Horizontal coordinate information is referenced to the North American Datum of 1983 (NAD 83). Chemical concentrations are given in micrograms per liter (µg/L). Water year is the 12-month period from October 1 through September 30. The water year is designated by the year in which it ends; for example, the water year from October 1, 2004, through September 30, 2005, is called the 2005 water year.

Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005

By Judith C. Thomas, Kenneth J. Leib, and John W. Mayo

Abstract nium would be removed from streams by remediation. During a previous study of selenium loads in the Lower Gunnison Elevated selenium concentrations in streams are a water- River Basin, mean annual load reductions were estimated at quality concern in western Colorado. The U.S. Geologic Sur- the Gunnison River near Grand Junction for the 1997–2001 vey, in cooperation with the Colorado Department of Public study period. Mean annual load reductions estimated for this Health and Environment, summarized selenium loading in study period were less than those estimated for the 2001–05 the Lower Gunnison River Basin to support the development study period, emphasizing the importance of understanding of total maximum daily selenium loads at sites that represent that different study periods can result in different load reduc- the cumulative contribution to U.S. Environmental Protection tion estimates. Agency 303(d) list segments. Analysis of selenium loading included quantifying loads and determining the amount of load that would need to be reduced to bring the site into compli- Introduction ance, referred to as “the load reduction,” with the State chronic aquatic-life standard for dissolved selenium [85th percentile Selenium is a trace element that occurs naturally in the selenium concentration not to exceed 4.6 µg/L (micrograms environment. Various human-related activities such as min- per liter)], referred to as “the water-quality standard.” Stream- ing and agriculture can act to mobilize selenium in their flow and selenium concentration data for 54 historical water- waste products. Selenium readily dissolves in oxygenated quality/water-quantity monitoring sites were compiled from water and moves through the aquatic environment where it U.S. Geological Survey and Colorado Department of Public can bioaccumulate in organisms and potentially reach toxic Health and Environment data sources. Three methods were levels (Lemly, 2002). Because elevated selenium concentra- used for analysis of selenium concentration data to address tions are a water-quality concern, some stream segments in the the variable data density among sites. Mean annual selenium Lower Gunnison River Basin have been placed on the State loads were determined for only 10 of the 54 sites due to data 303(d) list as impaired for selenium (Colorado Department of availability limitations. Twenty-two sites had 85th percentile Public Health and Environment, 1998). Section 303(d) of the selenium concentrations that exceeded the water-quality stan- 1972 Clean Water Act requires that States identify impaired dard, 3 sites had 85th percentile selenium concentrations less streams. These impaired streams are segments of streams that than the State standard, and 29 sites could not be evaluated exceed the State chronic aquatic-life standard. To address with respect to 85th percentile selenium concentration (sample chronic aquatic-life standard exceedances for these impaired count less than 5). To bring selenium concentrations into com- streams, a priority ranking (the 303(d) list) is established, and pliance with the water-quality standard, more than 80 percent a total maximum daily load (TMDL) is determined. A TMDL of the mean annual selenium load would need to be reduced represents the sum of all contributing loading sources of a at Red Rock Canyon, Dry Cedar Creek, Cedar Creek, Lout- pollutant (both point and non-point sources) that are allowed zenhizer Arroyo, Sunflower Drain, and Whitewater Creek. to contribute to an impaired stream so that the chronic aquatic- More than 50 percent of the mean annual load would need to life standard is not exceeded (U.S. Environmental Protection be reduced at Dry Creek to bring the site into compliance with Agency, 2006). the water-quality standard. The Uncompahgre River, Gunnison In Colorado, the Water Quality Control Division (Divi- River at Delta, and Gunnison River near Grand Junction would sion) of the Colorado Department of Public Health and require 69, 34 and 53 percent, respectively, of the mean annual Environment is required to develop TMDLs for 303(d) list load to be reduced for water years 2001 through 2005 to meet segments (Colorado Department of Public Health and Envi- the water-quality standard. Mean annual load reductions can ronment, 2007a). The Water Quality Control Commission be further reduced by targeting the periods of time when sele- (Commission) through the Water Quality Control Division 2 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005 intends to prepare more than 90 percent of the TMDLs for mately 10 miles upstream from the confluence of the segments listed on the 1998 303(d) list by 2008 (Colorado Gunnison and Colorado Rivers in Grand Junction, Colo- Department of Public Health and Environment, 2007b). Owing rado, and is approximately 40 miles downstream from to the complexities associated with characterizing selenium Delta, Colorado. The Uncompahgre River joins the Gun- fate and transport, the U.S. Geological Survey (USGS), in nison River in Delta. From Delta, the study area extends cooperation with Colorado Department of Public Health and approximately 26 miles upstream along the Uncompahgre Environment conducted data analysis in support of the devel- River to Montrose, Colorado, and approximately 35 miles opment of selenium TMDLs in the Lower Gunnison River upstream along the Gunnison River and the North Fork Basin. Gunnison River to Paonia, Colorado. The land-surface elevation is 4,659 feet at Whitewater, 4,957 feet at Delta, 5,682 feet at Paonia, and 5,807 feet at Montrose (Geo- Purpose and Scope graphic Names Information System, 2007). The primary land use in the study area is irrigated This report summarizes selenium loading for selected agriculture (fig. 2); however, residential and urban land- sites in the Lower Gunnison River Basin for 1978–2005 to use development is increasing as population growth support the TMDL-development process. This report provides occurs in the study area. 2005 population estimates for a brief description of the study area and a summary of sele- major population centers in the study area were 15,479 nium issues, methods of data analysis, and results. Selenium in Montrose, 8,135 in Delta, 1,584 in Paonia, and 1,402 issues in the Lower Gunnison River Basin were summarized in Whitewater (U.S. Census Bureau, 2007). Areas near and include identification of primary land uses that cause sele- Kannah Creek, Whitewater Creek, and Callow Creek are nium loading. Results include the assessment of selenium con- experiencing increased demand for residential housing. centrations and loads, 85th percentile selenium concentrations, Changes in land use have resulted in a shift from open and quantification of the amount of selenium load (pounds range and irrigated agricultural land uses to residential annually) that would need to be reduced to bring the site into and urban land uses as well as the use of independent compliance with the State chronic aquatic-life standard for dis- septic drainage systems. These shifts in land use have the solved selenium. This assessment occurred at 54 sites that rep- potential to introduce new paths of selenium loading to resent the cumulative contribution of selenium concentration the main stem Gunnison River (Gunnison Basin Selenium to 303(d) list segments (table 1). The amount of selenium load Task Force, 2007). that would need to be reduced (herein referred to as a “load Climate in the study area is predominately reduction”) was determined for sites that exceeded the State semiarid, but some variation in climate occurs at higher chronic aquatic-life standard for dissolved selenium (85th per- elevations (in general, more precipitation and cooler centile selenium concentration not to exceed 4.6 µg/L (micro- temperatures occur at higher elevations). Average annual grams per liter)), referred to as the “water-quality standard” precipitation ranges from 8.9 inches at Grand Junction to (Colorado Department of Public Health and Environment, 9.6 inches at Montrose. Average annual snowfall ranges 2007c). Throughout the report, the use of the term “selenium” from 12.3 inches at Grand Junction to 25.9 inches at refers to dissolved selenium in micrograms per liter. Montrose. Average high temperatures are approximately 66.1 degrees Fahrenheit for Grand Junction and approxi- Acknowledgments mately 63.3 degrees Fahrenheit for Montrose; average lows are 40.5 and 34.6 degrees Fahrenheit, respectively The authors thank Colorado Department of Public Health (Western Regional Climate Center, 2007). and Environment for data contributed for this analysis. The The geology of the study area is dominated by two authors also thank David Nimick and Rob Runkel for techni- major structural features, the Uncompahgre Uplift and cal reviews, Lanna Combs for editorial review, and the Rolla, the north-dipping southwestern flank of the Piceance Missouri, Publishing Service Center for manuscript layout and structural basin (Brooks and Ackerman, 1985). The graphics. Uncompahgre Uplift forms the Uncompahgre Plateau and is an asymmetrical anticline that plunges northwest to southeast. Mancos Shale is the dominant bedrock Description of Study Area material outcropping in the study area (fig. 3) east of the Uncompahgre Uplift (Green, 1992). The Mancos Shale The study area is located in Delta, Mesa, and Montrose is of Late Cretaceous age and is composed of massive, Counties in west-central Colorado. The study is focused on fossiliferous marine shale with interbedded sandstone, selected streams within the Lower Gunnison River Basin. The siltstone, and devitrified volcanic ash layers. The Man- Lower Gunnison River Basin is locally defined as the Gun- cos is the lateral equivalent to the Niobrara Formation, nison River from below the Gunnison Tunnel to Whitewater, Cody Shale, and Pierre Shale in Colorado, Montana, Colorado, and includes the North Fork Gunnison River and the Nebraska, South Dakota, and Wyoming (Wright and But- Uncompahgre River Basins (fig. 1). Whitewater is approxi- ler, 1993). The other two dominant bedrock units in the

Introduction 3

KANSAS

i e

i R

SAGUACHE

r s

v e

a r

t OKLAHOMA n

e s e

o NEBRASKA v

v i

s n

i MILES

R r a R

st e P a n

E ek k e 2 0 v e

i r r r n

107°

R C h A i u k

r o ve e io e t

i t u e G

R a h r a

l o O g

l S C S r

a w MINERAL

lo u

t il s

Colorado Springs P y

r W

ek Denver

C re

k ver r a

e 15

o B Pueblo

KILOMETERS

r C

h r

k e

r t e e e d w o k r re P 2 0 PITKIN C o la N ol e

r b d

e e n

k C

v a e

i r 1 0

R r r

G 15 e

l COLORADO Index map C v o

ta i i

e R

s t k R

i o y e

Study area r

r d c GUNNISON

C a r

a k

E e

e 1 0

t o

e v

Uncompahgre River h

W l

t i

n o 5 o

n s

i R

C A

r n

u n

e G

k o

v r

o s

HINSDALE

i i F

k k n

e r

e k

r R a

k e e L C r

l e C

a e n

r v

Co a C i

o u

p l

a p

So l i R

W G m

a Delta 0

Y 0

Grand es

eek k r r

r e lo C Junction e o

y r WYOMING D e

d C

v i

t e

d UTAH

i NEW MEXICO

u n eek v

a r

c i

e C

M r u e

t R

k u n

s C l

e B a e

E r g

s B C

i n

k l

a e k

t s

e n

r s

C y

ver r

G i r R

y n

d on ver C i d u r

C ar R

u im k Cimarron n

M r o t G C o s

107°30' s

1 e F

e tl n k # t t e

W e i s SAN JUAN

e L a H

r 2 E

C r

Tunnel

d i

Gunnison R r #

a 3 n

b o Paonia

r r a m i

b C

u k

r 4 H o

5 k

n C e

k re

r #

r I #

h o

t k

r F ee

o Cr h N w t o 8 i C #

S # # 9 7 Hotchkiss

k # e

e 11

r r e

10 e

C e

v Red Rock Canyon

i r x

u R

o k r e L

n r e

# a e 12 o r

is d

e C

n 13 #

n a C l

u l a

e D

e #

r G

Colona C

k t

Montrose

e n o OURAY

e a y

Cr r r o

g r #

Bi u 30 r

C A

k #

e e 14

r r 15

C 31

e z

c i

# 32

a h f 16

r en #

u # ts S ou er ek 33 # # L Riv re 19 re C mpahg g co # n 34 Un pri

35 S 108° 22 # 17 # 23 # 36 # 37 # 21 18 DELTA #

# k 24 # r ree 38 D y C # # # Delta 39 # 40

# k

# # e

28 e r # #

26 k

# e 20 C 41 e #

See figures 5 and 6 for detail of area r 43 42 e C i

y n

k e e f

# r 44

e k K

u s

e c

46 a

e e

d i b r u o M o

45 R

C H r

SAN MIGUEL h e v

a i r

n e

R k

n v

e i

e n R r

r l e

o C

s l

C u

i C

e i

r n

n w o s e

i M

u o

G C n k

n k

e a o

47 e

r S

w e t

r e

e # e

C r

t e

t C

i C

Callow Creek e

t o

h a t i

a o

W c

48 l

# s

a E

k c

F E

49 r MONTROSE

r D k r e e

C k

51 C

t e z

n r

k k e

# l

e e C # u

e e

g a

c r t

s i

v r

53 E

# m

Whitewater a

e t

50 h n

o D c

a k r

108°30'

k a d

e e

e o e

a e

u a

r t F

54 g

i C s

o C

t t

l L

b s

o o

a k

k s

a e

C l T r

E re E e

o C Grand K

F y

k r

Junction

e h D t

r r

C k N

t r

s u

g n i m o D

i g a B E k

h e

t e

r r

Study area 303(d) listed segment Sampling site and map index number

N k a k

e s e e EXPLANATION e

e r r

See figure 6 for detail of area

k C C

e t

k m n DOLORES r

r u e

o #

k s

C r m

F t

MESA e p

t n

e y i

v 17 h r

s t o

er i p

Location of study area, sites, and 303(d) list segments within the Lower Gunnison River Basin, western Colorado.

v u G e

C p

i R

e a

R u s l g W S s i B i s e D re r B o lo ol Do D tle Lit Base from U.S. Geological Survey digital data, 1:24,000 Mercator projection, Zone 13 Universal Transverse Horizontal coordinate information is referenced to the North American Data of 1983 (NAD 83) 39° 38° Figure 1. 38°30' 4 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005

108°30' 108° 107°30' 107°

B i g We Grand s C t E Boundary of r M PITKIN Junction e a ek u st d M C C d ol River study area y u r orado H dd ys ub C y tal b re River Callow Creek a e C r k stal R Whitewater d r ry e i C v r e N C e e k or k r 39° ek th For L ek i e t r t C l er e # Whitewat D # ## o MESA # reek lo nnah C re Ka E s R ek a i e i s v r hrac te t C nt C k e # x A ree R r u i k o er v e r unnison Riv er e North e G E r a L st C # S reek C la k e t e c # e e a R r f rk iv C r k o e st u e C r Ea DELTA S e F r o # C # t h a n t # l r a o C rr N # Paonia u r C Hotchkiss e e #G ## k u W ek n est Cre n # ## # is ## on # ## Rive # # k r ## # # # # e # # re ### z C # gue # min Delta Do O k e k h e ittl e Smi io e L e # th Fork r r ## C D # C r C

e e r

z t e n y # k e a k l e

a u c C e s r g E # k k r n e e C i e U r e e L G I e o r k C k # u o l ek m te r k n u B e n e n n E e r o scala C c t n

E e a C k o s t r e r is C D Fo t e s v m o e h C t r e u or n n n e g ek l e a d p h e r

i N r l o e W B o C k a a iz R k C B o k s e c i l h e v e e s r e w k r e e

g r e r # K E n e r A S r e k o e C C t e r o k r t r r l o a o o y C o ta k s e F C y p GUNNISON t R Cr e r n y u e e a i r C eek a r k e v v r l D e e Red Rock C r i esa C a k d r e M R rk c e e i i o s r e # Canyon t e F E b 38°30' ork C r n k n h F u t t C a o u r o s l S c is o a o l # e unn p S E t e R G i # r w r Ce u n s i da o i n r Creek C M r g C C r e e Montrose D k # o lo re s Ri ver Cree L ache k it begu r t Ta e le iv C R i m n a o C r r r r e o b a n o

m l i R l W a i i C C l Colona v l MONTROSE o e r k w r e e r e e e k C B r v r i e i C e g R k

B w o n l l o H u l o i s rsefly C e i re n ek C W N n r u a S e t a E e G

u n k

r a k k i M s e

t t r a e

i F o C g OURAY o r u F

r ek r e e C e k

ek l Cr C e

s k n R o C a i l w a r

v i l o EXPLANATION m L e r a C h D r a r e McK e r

e r nz e d k o ie e k Cre w n

SAGUACHE o R B Residential or urban land use P ig i G v

yp D e r r su y m C C re r ek e Irrigated agriculturale land use k D i HINSDALE sa 38° p reek p 303(d) listed segment n C o nso in e tm H en t C SAN MIGUEL # reSampling site ek SAN JUAN DOLORES MINERAL Base from U.S. Geological Survey digital data, 1:24,000 Land use from National Land Use Cover Data, 1983 Universal Transverse Mercator projection, Zone 13 0 5 10 15 20 MILES Horizontal coordinate information is referenced to the North American Data of 1983 (NAD 83) 0 5 10 15 20 KILOMETERS

Figure 2. Select land-use types in the Lower Gunnison River Basin, western Colorado. study area are the Dakota Sandstone of Early Cretaceous age observed at Kesterson Reservoir were an anomaly or an and various alluvial units of Quaternary age. indication of what could be expected to occur in other water bodies that receive irrigation drain water, the National Irriga- tion Water Quality Program (NIWQP) was formed. NIWQP’s Summary of Selenium Issues in Study Area objectives were to determine the extent and nature of irrigation–induced water-quality problems in the Western United Selenium has been the subject of a series of local water- States and to remediate those water-quality problems identi- quality studies throughout the Western United States since fied to be risks to human health or to DOI trust responsibilities 1983, when adverse biological effects were observed in (Seiler and others, 2003). aquatic bird populations at Kesterson Reservoir, a U.S. Depart- The NIWQP study was conducted as a five-phase ment of the Interior (DOI) National Wildlife Refuge in the approach that included site identification, reconnaissance western San Joaquin Valley (California). Increased incidents investigations, detailed studies, remediation planning, and of bird mortality, inherited birth defects, and reproductive finally remediation. Of the 26 study areas investigated as part failures were determined to be caused by selenium toxicity. of the NIWQP program, the Gunnison River ranked as the The selenium source was found to be water conveyed to the fourth most contaminated with respect to selenium (Seiler and area through constructed irrigation drainages (Ohlendorf and others, 2003). A study of selenium budgets for Lake Powell others, 1986, 1988). To determine if the biological effects Introduction 5

108°30' 108° 107°30' 107° Grand Junction PITKIN

Whitewater 39° # Piceance MESA #### Structural # Basin # DELTA # # # # # Paonia # Hotchkiss ## # ## # # ## ## # # # ### # # # # #### # Delta ##### # # # Boundary of study area # GUNNISON # 38°30' # ## EXPLANATION

Montrose Uncompahgre # Period Geologic units Uplift Quaternary Alluvium and gravel Andesitic volcanics Tertiary MONTROSE Colona Wasatch Formation Mancos Shale Cretaceous Dakota Sandstone OURAY Jurassic Morrison Formation Triassic Wingate Sandstone and

Chinle Formation SAGUACHE Precambrian Precambrian metamorphics

! NotHINSDALE identified 37° ! 303(d) listed segment SAN MIGUEL SAN JUAN # Sampling site DOLORES MINERAL Base from U.S. Geological Survey digital data, 1:24,000 Geology from Green (1992) Universal Transverse Mercator projection, Zone 13 0 5 10 15 20 MILES Horizontal coordinate information is referenced to the North American Data of 1983 (NAD 83) 0 5 10 15 20 KILOMETERS

Figure 3. Selected geologic units outcropping in the Lower Gunnison River Basin, western Colorado.

(located on the Colorado River downstream from Moab, Utah) nison River Basin on the 303(d) list. Task Force membership found that the Gunnison River Basin and the Grand Valley includes local governments, organizations, and citizens. The in Colorado produced 31 and 30 percent, respectively, of the Task Force and NIWQP (now inactive) worked together to total selenium load to Lake Powell (Engberg, 1999). Because identify and resolve selenium contamination issues while the Gunnison River Basin has been identified as a significant supporting the economic and lifestyle needs of the citizens contributor of selenium to streams, the Gunnison River Basin of the Gunnison River Basin (U.S. Environmental Protection has been the focus of many detailed studies with emphasis on Agency, 2007). Many focused studies have been conducted in the relations of climate, geology, geography, and land use to the Gunnison River Basin to aid in the understanding of sele- selenium mobilization. nium-related water-quality issues. USGS conducted several of Selenium is a priority pollutant and the U.S. Environ- these studies, including reconnaissance investigations of water mental Protection Agency has been working closely with quality, sediment, and biota; effects of selenium mobilization local entities to reduce selenium loading in the Gunnison on ground-water quality; quantification of the effectiveness of River Basin (U.S. Environmental Protection Agency, 2006). irrigation best management practices; and characterization of The Gunnison Basin Selenium Task Force (Task Force) was selenium loads from point and nonpoint sources (Butler and established by the CDPHE’s Water Quality Control Division others, 1991, 1996; Wright and Butler, 1993; Butler, 2001; in 1988 following the listing of several streams in the Gun- Butler and Leib, 2002). 6 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005

From these studies in the Gunnison River Basin, it was • Leroux Creek, Jay Creek, Big Gulch, and Short Draw, determined that the variation in the magnitude of selenium tributaries to the North Fork Gunnison River (segments concentration and load in streams is directly related to the COGUNNF05, COGUNNF06a, and COGUNNF06b; application of irrigation water. A USGS reconnaissance sites 1–12 in table 1), study conducted in 1988 and 1989 determined that the high- est selenium concentrations occurred at the Uncompahgre • Red Rock Canyon at the Mouth (COGUNLG04c; River at Delta (33 to 34 µg/L), whereas the lowest selenium site 13 in table 1), concentrations (2 µg/L) were observed farther upstream in • the Gunnison River main stem from the confluence the Uncompahgre River at Colona (Butler and others, 1991). with the Uncompahgre River to the confluence with Butler and others (1991) concluded that the observed increases the Colorado River (segment COGUNLG02; sites 40 in selenium concentration in the Uncompahgre River between and 52 in table 1), Colona and Delta were related to irrigation drainage. A 1998–2000 study of the Montrose Arroyo Basin determined • Gunnison River tributaries between Crystal Reservoir that replacing open-ditch irrigation laterals with pipes signifi- and the confluence with the Colorado River (segments cantly decreased the selenium and salt loads to the Montrose COGUNLG04a and COGUNLG07; sites 29–39, Arroyo. More than 90 percent of the decrease in selenium load 41–48, 50–51, and 53–54 in table 1), and was attributed to a decrease in ground-water selenium load (Butler, 2001). A detailed characterization study in 1999–2000 • Kannah Creek (segment COGUNLG04b; site 49 in on Cedar Creek and Loutzenhizer Arroyo determined that table 1). Montrose Arroyo was the largest source of selenium to Cedar A more detailed description of each segment can be Creek and that 41 percent of the total selenium load from found on the Water Quality Control Commission regulations Loutzenhizer Arroyo originated from its west tributary (Butler Web site (Colorado Department of Public Health and Environ- and Leib, 2002). ment, 2007a). Historical data for 54 stream water-quality/water-quantity monitoring sites were selected in consultation with Division staff who reviewed the site list and provided final approval. Methods for Analysis of Selenium Division staff approval involved modifying the site list by Loading either adding or removing sites based on Division needs. For example, Dominguez Creek, a tributary to the Gunnison River, was removed from the list and Red Rock Canyon at the Mouth was added to the list based on input from Division staff. Using Site Selection the Division-approved site list, USGS data were retrieved from NWIS (http://waterdata.usgs.gov/nwis), and CDPHE data A list of current and historical USGS sampling sites in were retrieved from the U.S. Environmental Protection Agen- the Lower Gunnison River Basin was retrieved from USGS cy’s (USEPA) STOrage and RETrieval (STORET) database National Water Information System (NWIS). The minimum (http://www.epa.gov/storet/). CDPHE selenium concentration criteria for data retrieval were location within the Lower Gun- data did not have accompanying streamflow data and were not nison River Basin and the presence of selenium concentration used to estimate mean daily loads. CDPHE selenium concen- data. The site list was refined to contain only sites that repre- tration data were used for sites with low sample counts to aid sent the cumulative contribution of selenium concentration to in determining 85th percentile selenium concentrations. For 303(d) list segments in the Lower Gunnison River Basin sites with low sample counts, CDPHE and USGS site aggrega- (table 1, fig. 1). Sites that represent cumulative contribution of tion occurred on the basis of site name and geographic loca- selenium concentration to a segment are typically sites that are tion. For six locations in the study area, two USGS sites were at or close to the mouth of a particular segment. The 303(d) near enough to one another to combine available data (table 1). list segments included in this report for selenium loading Although TMDL-development guidelines recommend using analysis are: the most recent 5 years of data (U.S. Environmental Protec- • the main stem of the Uncompahgre River near Mon- tion Agency, 2006), it was necessary in many cases to use all trose, Colorado, to the confluence with the Gunnison available selenium concentration data; therefore, USGS and River at Delta, Colorado (segments COGUNUN04b CDPHE data represent the period from 1978 through 2005. and COGUNUN04c; sites 26 and 28 in table 1), In this report, all annual values are expressed by water year where a water year is designated as the 12-month period from • Uncompahgre River tributaries between the South October 1 through September 30. Canal and the confluence with the Gunnison River (COGUNUN12; sites 14-25 and 27 in table 1), Methods for Analysis of Selenium Loading 7

Table 1. List of sites that represent the cumulative contribution of selenium concentration to 303(d) list segments in the Lower Gunnison River Basin, western Colorado, 1978–2005. [303(d) list segment, 303(d) list segment identifier (Colorado Department of Public Health and Environment, 2007a); USGS site ID, USGS site identification number; n, number of samples (selenium concentration and instantaneous discharge)] Site number Site name USGS site ID n 303(d) list seg- (fig. 1) ment North Fork Gunnison River Basin and Red Rock Canyon at the mouth 1 Hubbard Creek1 385532107310501 and 385532107310400 11 COGUNNF05 2 Terror Creek1 385414107334001 and 385414107334000 12 COGUNNF05 3 Minnesota Creek at Paonia 09134050 5 COGUNNF05 4 Roatcap Creek at Highway 133, near mouth 385144107371701 4 COGUNNF05 5 Reynolds Creek at Cty Road J75 385051107372701 4 COGUNNF06b 6 Bell Creek at county road and railroad tracks, near mouth 384922107402001 6 COGUNNF06b 7 Jay Creek at Highway 133, near mouth 384915107412101 4 COGUNNF05 8 Cottonwood Creek near Hotchkiss 09134200 7 COGUNNF06b 9 Short Draw West of county fairgrounds, at Hotchkiss 384747107430501 7 COGUNNF06a 10 Leroux Creek1 384732107434801 and 09135900 21 COGUNNF05 11 Alum Gulch at mouth 384610107455001 5 COGUNNF06b 12 Big Gulch at Highway 92 384756107490801 4 COGUNNF06b 13 Red Rock Canyon at mouth near Montrose 383537107471500 23 COGUNLG04c Uncompahgre River Basin 14 Dry Cedar Creek1 382711107520101 and 382716107520701 34 COGUNUN12 15 Cedar Creek near mouth 383041107544201 45 COGUNUN12 16 Mexican Gulch near mouth 383013107552501 1 COGUNUN12 17 Spring Creek1 09149400 and 383201107575301 14 COGUNUN12 18 Drain at Blossom Road, near Chipeta 383834108001701 1 COGUNUN12 19 Loutzenhizer Arroyo at N. River Road 383946107595301 87 COGUNUN12 20 Dry Creek at mouth, near Delta 384202108032001 17 COGUNUN12 21 Drain at B Road, near 1800, Ash Mesa 384043108012201 1 COGUNUN12 22 Drainage ditch near Highway 50, Overholt Wetland 384137108011401 1 COGUNUN12 23 Drainage ditch 2 at Overholt Wetland Area 384140108013601 1 COGUNUN12 24 Drain at D10 and Ash Mesa Roads 384152108024401 1 COGUNUN12 25 Drain at Ash Mesa Road, upper ditch 384150108023101 2 COGUNUN12 26 Uncompahgre River at Delta 09149500 124 COGUNUN04b 27 Ditch at 5th Street bridge at mouth, in Delta 384423108044801 1 COGUNUN12 28 Uncompahgre River at mouth 384523108052101 3 COGUNUN04c Gunnison River from North Fork Gunnison River to Grand Junction, Colorado 29 Sulphur Gulch at Highway 92 384752107502201 4 COGUNLG04a 30 Lawhead Gulch at Highway 92 384802107522201 4 COGUNLG04a 31 Unnamed drainage below Oasis Pond, at county road 384643107540301 6 COGUNLG04a 32 Currant Creek1 09137050 and 384651107561001 10 COGUNLG04a 33 Peach Valley Arroyo near mouth 384604107570701 10 COGUNLG04a 34 Alfalfa Run at Austin 384649107570501 7 COGUNLG04a 35 Sunflower Drain at Highway 92, near Read 384551107591901 106 COGUNLG04a 36 Tongue Creek at mouth 384635108010301 1 COGUNLG07 37 Unnamed drainage along 1825 Road, north of Delta 384643108020501 1 COGUNLG04a 38 Dry Gulch above confluence with Heartland ditch near Delta 384557108024300 2 COGUNLG04a 39 Drainage ditch at Confluence Park, at Delta 384502108042701 3 COGUNLG04a 40 Gunnison River at Delta 09144250 44 COGUNLG02 41 Drainage ditch near 1400 Road, at mouth 384457108055801 3 COGUNLG04a 42 East unnnamed drain at Highway 50, near Delta 384544108060001 4 COGUNLG04a 43 West unnamed drain at Highway 50, near Delta 384545108061601 4 COGUNLG04a 44 Cummings Gulch at mouth 384448108070301 13 COGUNLG04a 45 Seep Creek at G Road, near mouth 384408108091501 4 COGUNLG04a 46 Alkali Creek below Highway 50, near Delta 384510108111801 10 COGUNLG04a 47 Wells Gulch at Dominguez Road crossing 384813108184301 3 COGUNLG04a 48 Deer Creek below Windy Creek, near mouth 385104108213501 2 COGUNLG04a 49 Kannah Creek about .1 miles below Indian Creek 385600108250301 17 COGUNLG04b 50 East Creek at Highway 141 Bridge, near Whitewater 385824108274401 2 COGUNLG04a 51 Whitewater Creek .4 miles above mouth, at Whitewater 385839108264401 19 COGUNLG04a 52 Gunnison River near Grand Junction 09152500 202 COGUNLG02 53 Callow Creek at Whitewater 09152520 10 COGUNLG04a 54 Bangs Canyon at mouth, near Whitewater 385855108285501 1 COGUNLG04a 1 Combined two nearby sites into a single site. 8 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005

Estimating Selenium Loads, 85th Percentile percentile selenium concentrations for sites with five or more Selenium Concentrations, and Load Reductions samples, mean loads in pounds per day, and sample date range. Because CDPHE selenium concentration data did not Historical data for 54 stream water-quality/water-quantity have accompanying streamflow data, statistical summaries for monitoring sites with selenium sample results were compiled USGS data are presented separately from CDPHE data. Where from USGS and CDPHE data sources for this study. Three of CDPHE selenium concentration data were available, a recalcu- the sites were located at gaging stations (gaged) and therefore lated 85th percentile selenium concentration is determined that had continuous streamflow data. The remaining 51 sites were represents an 85th percentile selenium concentration based on not located at streamflow-gaging stations (ungaged); however, the combination of USGS and CDPHE selenium concentration instantaneous streamflow was measured at the time selenium data. concentration data were collected. The amount of available data and the period of data collection varied considerably Load Estimation Using Annual Time-Weighted from site to site. This resulted in a varying ability to estimate selenium loads and selenium load reductions. Means Three methods were used for loading analysis of sele- For ungaged sites with approximately 20 or more samples nium concentration data to address the variable data density that contained both selenium concentration and instantaneous among sites. For sites with low sample counts (typically streamflow data, an annual time-weighted mean (ATWM) was 20 samples or less), USGS and CDPHE data were combined determined that represents an annual selenium load in pounds. to provide statistical summaries. For ungaged sites that had To evaluate a site with respect to ATWM, selenium concentra- monthly samples for one or more water years (typically 20 or tion and instantaneous streamflow data were needed to propor- more total samples that contained both selenium concentration tionally represent water-quality conditions for 1 or more water and instantaneous streamflow data), a time-weighting tech- years. Proportional representation of a water year implies that nique was used for estimation of annual selenium loads using samples were collected during each season, and typically at USGS data. For gaged sites, a regression analysis was used least 1 sample was collected per month. A selenium load was to estimate daily mean selenium concentrations using USGS calculated for each matched pair of selenium concentration data from which annual selenium loads and 85th percentile and streamflow. Each calculated selenium load was assigned selenium concentrations were determined. Selenium load a weight on the basis of the number of days between samples. reductions were determined for sites where 85th percentile Weights were computed as the amount of time extending from selenium concentrations exceeded the water-quality standard one-half the time interval between a sample and the preced- and a mean annual selenium load had been determined. ing sample and one-half the time interval extending from the USGS water-quality data retrieved from NWIS were sample to the subsequent sample divided by the total time in collected and processed using standard USGS techniques and a year. Weights were computed using the following general procedures (U.S. Geological Survey, variously dated). USGS equation: streamflow data were collected using standard USGS stream- 1 1 gaging methods (Rantz and others, 1982). CDPHE water- [ /2 (Dc - Dp) - /2 (Dc - Ds)] Wc = X 100, (1) quality data retrieved from STORET were collected using 365.25 standard methods (Colorado Department of Public Health and where Environment, 1998). For the historical USGS data used in Wc is the weight for the sample, as a percentage; this report, two minimum reporting limits (MRL) were used, Dc is the date of the sample, in days; 1 and 2 µg/L. Concentrations found to be less than the MRL Dp is the date of the preceding sample, in days; are referred to as “censored data.” For censored data, one-half Ds is the date of the subsequent sample, in days; the value of the MRL was used for data analysis; however, and most selenium concentrations were substantially greater than 365.25 is the number of days in a year including leap the MRL. Additionally, it was not necessary to alter censored- years. data values for regression analysis as the method already has a The weight was multiplied by the corresponding load, and the means for dealing with censored data. weighted loads were summed to represent the mean annual load for a given water year (Crawford, 2004; Larson and oth- Statistical Summary of Available Data ers, 2004). Ideally, ATWM sample weights were approximately Statistical summaries were calculated for sites that did equal to one another; however, each water year was evaluated not have sufficient data to estimate an annual selenium load. to determine if the distribution of the selenium samples was Sites that did not have sufficient data typically had less than proportionally weighted. Proportional data typically implied 20 samples. Statistical summaries included sample counts samples collected monthly or more frequently, resulting in for matched pairs of selenium concentration and streamflow monthly weights that were approximately equal to the other data, median and range of selenium concentration data, 85th months of the year. In the event that a weight for one particu- Methods for Analysis of Selenium Loading 9

lar month was greater than that for the other months in the Not every regression model in this study used all of the year, that weight was evaluated to determine if it adequately explanatory variables listed above; each model was developed represented that portion of the water year. Factors such as the to best represent the site characteristics and the available data. amount of irrigation water that the site receives were used Where continuous specific conductance data were available, to evaluate the adequacy of a sample weight. The irrigation specific conduction was used to estimate selenium concentra- season occurs typically from April through October. The tions in place of streamflow in the general equation. Regres- remaining portion of the year (November through March) sion models were developed from a set of calibration data that is the nonirrigation season (no irrigation water is delivered). consisted of matched pairs of selenium concentration data Owing to the annual variability of the dates when irrigation and streamflow (or specific conduction) data. An estimation water is turned on or off, transition periods were identified that file containing continuous data (either mean daily streamflow represented intervals when selenium loads and concentrations or mean daily specific conductance) was used as input to the could be expected to be highly variable due to the presence regression model to predict mean daily selenium concentra- or absence of irrigation water. These transition periods were tions. defined as occurring from March through April (irrigation Model coefficients were developed using ordinary least- water on) and mid-October through the end of November squares (OLS) regression. OLS is a method for linear regres- (irrigation water off). For sites where water quality and water sion that estimates unknown quantities in a statistical model quantity were dominated by the use of irrigation water, empha- by minimizing the sum of the residuals (difference between sis was placed on the need for samples collected during these the predicted and observed values) squared (Helsel and Hirsch, transition periods. During periods when there was no use of 1992). Evaluation of a model’s significance included looking irrigation water, a single sample could be allowed to represent at the following diagnostics—the coefficient of determination more than 1 month as no appreciable changes were expected (r2), p-values, residual plots, and the standard error of esti- in selenium concentrations and loads during these periods. mate. Generally, a model exhibited poor correlation if the r2 value was less than 0.6 and the p-value was greater than 0.05. Residual plots (not included in this report) were evaluated to Load Estimation Using Regression Analysis confirm normality and constant variance throughout the range A multiple linear regression model was developed for of prediction (homoscedasticity) (Helsel and Hirsch, 1992). estimating mean daily selenium concentration using the FOR- Serial correlation also was evaluated using residual plots TRAN program LOAD ESTimator (LOADEST) for gaged (where a residual is an estimated value minus its correspond- sites. LOADEST provides users the ability to develop a regres- ing observed value) related to loading concentration estimates sion model for estimating constituent loads or concentrations and time. The presence of a serial correlation indicates that over a user-specified time period (Runkel and others, 2004). sampling bias or a trend may exist in the data set. All diagnos- Load estimates were derived in three steps—model develop- tics were considered collectively to determine the most appro- ment, calibration of the developed model, and estimates of priate regression model (Leib and others, 2003). Loads were loads using the model. LOADEST provides 11 predefined estimated using the calibrated regression equation and a time regression models and also allows for the creation of user- series of daily streamflow and/or specific conductance values. defined models. The 11 predefined models in LOADEST are a The primary load estimation method used within LOADEST series of models starting with the simplest model and increas- was Adjusted Maximum Likelihood Estimation (AMLE) ing in complexity by the addition of one or more explanatory (Runkel and others, 2004). variables. The regression model used in this analysis takes the following general form: 85th Percentile Selenium Concentrations

2 ln(C) = a0+ a1 lnQ+ a2 lnQ + a3 sin(2πdtime)+ a4 cos(2πdtime) + a5dtime, For all sites with five or more samples, 85th percentile (2) selenium concentrations were calculated. All 85th percentile where selenium concentrations were calculated in Microsoft Excel C is constituent concentration, in micrograms using the PERCENTILE function (Microsoft Corporation, per liter; 2003). For sites lacking sufficient data to calculate annual Q is the stream discharge, in cubic feet per selenium loads, all available data were used for calculation second; of 85th percentile selenium concentrations. For sites where dtime is decimal years from the beginning of the ATWM was used to estimate mean annual selenium loads, calibration period; the 85th percentile selenium concentration was computed sin(2πdtime)+cos(2πdtime) is used to describe from the selenium concentrations used to determine ATWM seasonality; (proportionally weighted samples). For sites where LOAD- and EST was used to estimate daily selenium concentrations, the a are model coefficients where n = 0, 1, 2,…., n. 85th percentile selenium concentration was computed from the n estimated daily selenium concentrations. 10 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005

Load Reductions reductions for sites whose 85th percentile selenium concentration exceeded the water-quality standard. Load reductions represent the amount of selenium load (pounds annually) that would need to be reduced to meet the water-quality standard. Load reductions were determined for North Fork Gunnison River Basin and Red Rock sites with mean annual selenium loads (ATWM or LOADEST Canyon at the Mouth methods). To determine if a load reduction would be needed at a site, the site was evaluated with respect to its 85th percentile Twelve sites within the North Fork Gunnison River Basin selenium concentration. If the 85th percentile selenium con- (sites 1–12, table 1) and Red Rock Canyon at the mouth centration exceeded the water-quality standard, the load reduc- (site 13, table 1) (fig. 4) were selected for analysis. For the tion was calculated using a simple mass-balance approach. Nork Fork Gunnison River Basin, sites 1 through 12 repre- This approach assumed that the selenium load reduction would sent the cumulative contribution of selenium concentration to take place with no change in streamflow volume and that only 303(d) list segments COGUNNF05, and COGUNNF06a and b selenium would be removed from the system. On the basis (table 1). All 12 sites lacked sufficient data to calculate mean of this assumption, a new selenium load was determined that annual selenium loads (table 2). Seven of the 12 sites had equaled the load that would result if the 85th percentile sele- CDPHE data in addition to USGS data (table 2). Sample dates nium concentration were to equal the water-quality standard of ranged from 1982 through 2000 for USGS selenium concen- 4.6 µg/L. tration data and from 1998 through 2005 for CDPHE selenium This method of calculating load reductions uses the ratio concentration data. Based on USGS data, 4 of the 12 sites of the water-quality standard to the 85th percentile selenium had 85th percentile selenium concentrations that exceeded the concentration, which inherently preserves the distribution of water-quality standard (4.6 µg/L). These sites were Bell Creek the concentrations through the year. The hypothetical load at (site 6), Cottonwood Creek (site 8), Short Draw (site 9), and compliance with the water-quality standard was subtracted Leroux Creek (site 10). from the mean annual load, and the resulting value represents The cumulative mean selenium load from the 12 sites in the amount of selenium load that would need to be reduced to the North Fork Gunnison River Basin was 2.3 pounds per day bring the site into compliance with the water-quality standard. (table 2). Assuming that this load is a reasonable representa- Load reduction is defined by the following equation: tion of the cumulative mean daily load from these sites, the resulting mean annual selenium load would be approximately

LR = LA – (STD [Se] / 85th [Se]) LA, (3) 840 pounds. In a previous report on selenium loading by Butler and Leib (2002), the mean annual selenium load for the where North Fork Gunnison River at the mouth was approximately L is mean annual load reduction, in pounds; R 1,300 and 1,400 pounds per year for 1999 and 2000, respec- L is mean annual load at a site, in pounds; A tively. The cumulative mean annual selenium load calculated STD [Se] is 4.6 µg/L, the State chronic aquatic-life for the selected sites (table 2) represents more than one-half standard for dissolved selenium; of the mean annual selenium load to the North Fork Gunnison and River. The source of the remaining mean annual selenium load 85th [Se] is the 85th percentile selenium concentration is not known, but sources may include unquantified naturally at a site, in micrograms per liter. occurring selenium load from ground water and surface water, deep percolation of irrigation water, or septic systems. Further quantification of the combination of selenium sources such Dissolved Selenium Loading for as agricultural land use, residential and urban development, and the occurrence of selenium parent material in this basin Selected Sites in the Lower Gunnison may aid in more completely understanding the source of the River Basin selenium observed in North Fork Gunnison River. Red Rock Canyon at the mouth near Montrose (site 13) This section of the report contains a summary of dis- represents the cumulative contribution of selenium concentra- solved selenium loading for the 54 selected sites in the Lower tion to 303(d) list segment COGUNLG04c (fig. 4, table 3). Gunnison River Basin, organized as follows: North Fork Red Rock Canyon data spans the 2001 to 2005 water-year Gunnison River Basin and Red Rock Canyon at the mouth, period. Water years 2004 and 2005 had sufficient data to Uncompahgre River Basin, and Gunnison River from North compute an ATWM. For the 2004 water year, the 85th per- Fork Gunnison River to Grand Junction. Each section contains centile selenium concentration was 72.1 µg/L and mean daily a map of the sites and reference to one or more of the three load was 0.571 pound (209 pounds annually). Based on 2004 tables that contain results of selenium loading analysis for the results, to bring this site into compliance with the water-qual- three methods used. This loading analysis summary contains ity standard, 93 percent (195 pounds) of the mean annual load summary statistics, 85th percentile selenium concentrations, would need to be reduced. For the 2005 water year, the mean daily and mean annual selenium loads, and annual load 85th percentile selenium concentration was 53.1 µg/L and the Dissolved Selenium Loading for Select Sites in the Lower Gunnison River Basin 11

108° 45' 30' 107°15' Wes t Mu

H dd E u y a b C b r s

a e t

rd e

C k M

39° r

e u

e d k d y

C r e e k k ite Cree e T rac k re e th C r n x DELTA r A u o r o r nison Rive r n Gu e C k # e L r J Ro . re a at 1 y c a C # e C p

r c Alfalka C

a e

f r

r e . C Run k Minnesota u 3 S ek # o e Creek a r k # l

C or C nt F 4 GUNNISON a Law h R r rr Sulfur rt # eno e u Hotchkiss o ld Paonia e Head s k C Gulch N C r 29 7## e 6 5B e 30 # el k # # #12 l C ## r Boundary of 10 8 e # 31 e 34 # # 9 k study area # # # C 32 Big ot 38°45' 35 33 11 ton Gulch Alum wo Peach od C Sunflower Valley Gulch reek Drain Arroyo Sm ith Fork EXPLANATION

k Residential or urban lande use e L I r o G r o C u u n n k t Agricultural land usel s n e is C E

o n r n t h e s i R e z i 303(d) listed segmente e v k r er MONTROSE W A r # 11 ro 13 k # Sampling site and map y Red Rock e o re Canyon C index number tal Crys Base from U.S. Geological Survey digital data, 1:24,000 Land use from National Land Use Cover Data, 1983 Universal Transverse Mercator projection, Zone 13 0 5 10 15 MILES Horizontal coordinate information is referenced to the North American Data of 1983 (NAD 83) 0 5 10 15 KILOMETERS

Figure 4. Location of sites and land use in the North Fork Gunnison River Basin and Red Rock Canyon at the mouth (site 13), western Colorado. mean daily load was 0.710 pound (259 pounds annually). The rado, to the confluence with the Gunnison River at Delta mean annual load would need to be reduced by 92 percent to (fig. 5). bring this site into compliance with the water-quality standard. Data for 5 of the 15 sites were sufficient to determine The average amount of load that would need to be reduced at mean annual selenium loads (tables 3 and 4). For the remain- this site for water years 2004 and 2005 was 217 pounds or ing 10 sites that lacked sufficient data to calculate mean annual 93 percent of the mean annual load. This high percentage of selenium loads (table 2), only the Spring Creek site (site 17) load reduction represents almost the entire mean annual load had more than five samples, and its 85th percentile selenium at this site. concentration was 2.0 µg/L (table 2). Dry Cedar Creek (site 14), Cedar Creek (site 15), Loutzenhizer Arroyo (site 19), and Dry Creek (site 20) were ungaged and mean annual selenium Uncompahgre River Basin loads were estimated using the ATWM method (table 3). The Uncompahgre River at Delta (site 26) is gaged; therefore, Fifteen sites within the Uncompahgre River Basin were LOADEST was used to estimate mean annual selenium loads selected that represent the cumulative contribution of selenium for the most recent 5 years of data (2001 through 2005) concentration to 303(d) list segments COGUNUN12 (table (table 5). 1, sites 14-25 and 27), COGUNUN04b (table 1, site 26), and Dry Cedar Creek (site 14) data spans the 1991 through COGUNUN04c (table 1, site 28), which are segments on the 2001 water-year period. Water years 1995 and 1996 had suffi- main stem of the Uncompahgre River near Montrose, Colo- 12 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005 02/20/1992 06/05/2000 01/18/2000 04/30/1999 02/20/1992 02/20/1992 02/10/1992 02/29/1996 02/29/1996 06/29/1999 Date range 06/09/1982–03/13/2000 04/14/1982–03/13/2000 05/11/1999–03/13/2000 05/11/1999–03/13/2000 05/11/1999–03/13/2000 05/11/1999–03/13/2000 05/11/1999–03/13/2000 05/11/1999–07/25/2000 10/15/1998–03/14/2000 12/18/1990–12/05/2000 10/15/1998–03/14/2000 05/12/1999–03/14/2000 01/11/1978–08/25/2003 02/20/1992–11/22/1999 12/20/2000–07/11/2001 05/17/1999–03/14/2000 05/17/1999–03/14/2000 05/17/1999–03/14/2000 07/16/1991–03/28/2001 07/22/1991–03/15/2000 07/22/1991–03/15/2000 03/02/2002–07/23/2002 07/15/1993–09/10/1993 07/20/1993–09/10/1993 04/27/1999–03/20/2000 04/27/1999–11/16/1999 07/15/1991–03/20/2000 01/29/1992–09/10/1993 11/24/1995–04/04/2000 06/17/1999–03/21/2000 02/07/2000–03/21/2000 04/29/1999–03/12/2002 04/29/1999–09/08/1999 04/29/1999–05/23/2002 0.34 0.17 0.038 0.056 0.020 0.21 0.061 0.32 0.37 0.47 0.12 0.13 0.010 0.16 0.020 0.0030 0.19 0.014 0.021 0.12 0.35 0.0080 0.011 0.30 0.65 0.39 0.28 0.78 0.0053 1.4 0.76 0.42 0.082 0.12 0.92 0.55 0.042 0.021 0.0048 0.40 0.010 0.0034 0.0038 14.9 Mean load NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 1.4 7.0 9.6 3.1 2.0 19.1 15 14 34.2 27.9 17.5 14 42.0 15.2 117 85th percentile USGS 34 55 71 26 <1 2.0 8.0 38.5 11.9 32.8 9–11 28–37 <1–10 1.0–21 8.8–11 5.0–21 5.0–15 5.0–95 3.4–16 6.0–10 <1–1.5 <1–4.9 <1–8.0 <1–3.0 Range 3.0–7.0 1.6–3.2 7.0–9.0 5.0–8.0 2.0–5.0 2.0–5.0 1.3–1.5 10–43.6 <1–18.1 <0.7–<1 <0.7–<1 6.4–18.8 4.2–12.8 8.0–29.4 4.6–65.1 3.5–65.0 5.2–48.5 18.0–150 11.5–17.1 321–4580 NA NA NA NA NA NA NA NA NA NA NA NA 1.3 2.5 4.6 4.5 7.3 9.5 2.4 7.5 0.5 7.0 9.5 6.5 6.5 3.5 8.4 6.0 8.0 3.5 1.4 Selenium concentration (µg/L) 12.9 11.3 10 16.5 10 20.2 16.4 29 10 88.5 13.7 11.6 Median 2450 - 5/2 4/2 4/2 6/0 4/0 7/0 7/0 5/0 4/0 1/0 1/0 1/0 1/0 1/0 1/0 2/0 1/0 3/0 4/0 4/0 6/0 7/1 1/0 1/0 2/0 3/0 3/0 4/0 4/0 4/0 3/1 2/0 2/0 1/1 21/0 14/5 10/0 10/0 13/0 10/0 17/0 10/0 11/10 12/11 ber of censored values Number of values/num - Uncompahgre River Basin North Fork Gunnison River Basin ment COGUNNF05 COGUNNF05 COGUNNF05 COGUNNF05 COGUNNF05 COGUNNF05 COGUNLG07 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNNF06a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNNF06b COGUNNF06b COGUNNF06b COGUNNF06b COGUNNF06b COGUNLG04b COGUNUN04c 303(d) list seg Gunnison River from North Fork to Grand Junction, Colorado Site name 1 1 1 1 Hubbard Creek Creek Terror Minnesota Creek at Paonia 133, near mouth Roatcap Creek at Highway Creek at Cty Road J75 Reynolds Bell Creek at county road and railroad tracks, near mouth 133, near mouth Jay Creek at Highway Creek near Hotchkiss Cottonwood at Hotchkiss of county fairgrounds, West Short Draw Leroux Creek Alum Gulch at mouth 92 Big Gulch at Highway Gulch near mouth Mexican Spring Creek Drain at Blossom Road, near Chipeta Drain at B Road, near 1800, Ash Mesa Wetland 50, Overholt Drainage ditch near Highway Area Wetland Drainage ditch 2 at Overholt Drain at D10 and Ash Mesa Roads Drain at Ash Mesa Road, upper ditch Ditch at 5th Street bridge mouth, in Delta at mouth Uncompahgre River 92 Sulphur Gulch at Highway 92 Gulch at Highway Lawhead Oasis Pond, at county road Unnamed drainage below Currant Creek near mouth Arroyo Peach Valley Run at Austin Alfalfa Creek at mouth Tongue Unnamed drainage along 1825 Road, north of Delta confluence with Heartland ditch near Delta Dry Gulch above at Delta Drainage ditch at Confluence Park, Drainage ditch near 1400 Road, at mouth 50, near Delta East unnnamed drain at Highway 50, near Delta unnamed drain at Highway West Cummings Gulch at mouth Seep Creek at G Road, near mouth 50, near Delta Highway Alkali Creek below Gulch at Dominguez Road crossing Wells Creek, near mouth Windy Deer Creek below Indian Creek Kannah Creek about .1 miles below 141 Bridge, near Whitewater East Creek at Highway Creek at Whitewater Callow at mouth, near Whitewater Bangs Canyon Statistical summary of selenium concentration data for sites with limited sets in the Lower Gunnison River Basin, western Colorado, 1978–2005. 1 2 3 4 5 6 7 8 9 10 11 12 16 17 18 21 22 23 24 25 27 28 29 30 31 32 33 34 36 37 38 39 41 42 43 44 45 46 47 48 49 50 53 54 (fig. 1) Site number Table 2. Table CDPHE, Colorado Department of Public Health and 2007a); USGS, U.S. Geological Survey; (Colorado Department of Public Health and Environment, identifier 303(d) list segment [303(d) list segment, mean load; selenium load in pounds per day] computed; --, no data available; no value n, number of samples; NA, Environment; Dissolved Selenium Loading for Select Sites in the Lower Gunnison River Basin 13 ------04/01/1998 03/09/1999 03/10/1999 03/10/1999 03/09/1999 03/10/1999 04/01/1998–5/11/2005 07/24/2000–03/16/2005 07/24/2000–03/16/2005 03/31/1998–01/12/2005 03/31/1998–01/12/2005 03/31/1998–01/11/2005 04/01/1998–03/15/2005 03/09/1999–07/13/2005 Date range for CDPHE data - NA NA ------NA -- NA ------NA NA NA -- NA NA NA -- -- 2 3 8 21 centile 85th per 6 1 ------14 12 34 1.0 NA NA 1–26 1.1–2 CDPHE 24–37 (µg/L) 9.8–13 Range 1.0–3.0 2.2–9.0 NA NA ------NA ------NA NA NA -- NA NA -- -- 2 2.1 6.9 7.0 11 31 Selenium concentration Median ------6/6 6/6 5/0 5/0 3/0 1/0 5/0 1/0 1/0 1/0 1/0 2/0 1/0 22/0 ber of censored values Number of values/num ------11 (5) 3.8 (9) 3.1 (6) 12 (10) 12 (43) 14 (14) 41 (19) 2.0 (10) NA (17) NA (18) NA 116 (11) 33.6 (15) tration (µg/L) 85th percentile (n) Selenium concen USGS with CDPHE Uncompahgre River Basin - North Fork Gunnison River Basin ment COGUNNF05 COGUNNF05 COGUNNF05 COGUNNF05 COGUNNF05 COGUNNF05 COGUNLG07 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNNF06a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNLG04a COGUNNF06b COGUNNF06b COGUNNF06b COGUNNF06b COGUNNF06b COGUNLG04b COGUNUN04c 303(d) list seg Gunnison River from North Fork to Grand Junction, Colorado Site name 1 1 1 1 Hubbard Creek Creek Terror Minnesota Creek at Paonia 133, near mouth Roatcap Creek at Highway Creek at Cty Road J75 Reynolds Bell Creek at county road and railroad tracks, near mouth 133, near mouth Jay Creek at Highway Creek near Hotchkiss Cottonwood at Hotchkiss of county fairgrounds, West Short Draw Leroux Creek Alum Gulch at mouth 92 Big Gulch at Highway Gulch near mouth Mexican Spring Creek Drain at Blossom Road, near Chipeta Drain at B Road, near 1800, Ash Mesa Wetland 50, Overholt Drainage ditch near Highway Area Wetland Drainage ditch 2 at Overholt Drain at D10 and Ash Mesa Roads Drain at Ash Mesa Road, upper ditch Ditch at 5th Street bridge mouth, in Delta at mouth Uncompahgre River 92 Sulphur Gulch at Highway 92 Gulch at Highway Lawhead Oasis Pond, at county road Unnamed drainage below Currant Creek near mouth Arroyo Peach Valley Run at Austin Alfalfa Creek at mouth Tongue Unnamed drainage along 1825 Road, north of Delta confluence with Heartland ditch near Delta Dry Gulch above at Delta Drainage ditch at Confluence Park, Drainage ditch near 1400 Road, at mouth 50, near Delta East unnnamed drain at Highway 50, near Delta unnamed drain at Highway West Cummings Gulch at mouth Seep Creek at G Road, near mouth 50, near Delta Highway Alkali Creek below Gulch at Dominguez Road crossing Wells Creek, near mouth Windy Deer Creek below Indian Creek Kannah Creek about .1 miles below 141 Bridge, near Whitewater East Creek at Highway Creek at Whitewater Callow at mouth, near Whitewater Bangs Canyon Statistical summary of selenium concentration data for sites with limited sets in the Lower Gunnison River Basin, western Colorado, 1978–2005.—Continued 1 2 3 4 5 6 7 8 9 10 11 12 16 17 18 21 22 23 24 25 27 28 29 30 31 32 33 34 36 37 38 39 41 42 43 44 45 46 47 48 49 50 53 54 (fig. 1) Combined two nearby sites into a single site. Combined two 1 Site number Table 2. Table CDPHE, Colorado Department of Public Health and 2007a); USGS, U.S. Geological Survey; (Colorado Department of Public Health and Environment, identifier 303(d) list segment [303(d) list segment, mean load; selenium load in pounds per day] computed; --, no data available; no value n, number of samples; NA, Environment;

14 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005 93 92 93 92 93 93 89 86 88 88 84 87 98 97 98 98 53 97 88 96 94 93 reduction Percent load Load 195 238 217 213 287 250 653 844 112 2,170 1,860 1,650 1,690 1,300 1,730 5,110 4,360 5,740 5,070 1,420 1,110 1,130 reduction 209 259 234 232 309 270 964 121 load 2,440 2,160 1,870 1,910 1,540 1,980 5,240 4,490 5,870 5,200 1,230 1,470 1,160 1,200 Mean annual 0.571 0.710 0.641 0.636 0.843 0.740 6.69 5.92 5.12 5.22 4.20 5.43 3.37 4.02 2.64 3.18 3.28 0.332 load 14.4 12.3 16.1 14.2 Mean daily - tile 9.80 72.1 53.1 62.6 56.5 67.4 61.9 40.8 33.0 40.0 40.0 29.5 36.7 36.5 99.2 62.6 180 154 215 183 157 104 85th percen 1 1 1 1 1 2004 2005 1995 1996 1992 1993 1996 1997 2000 1996 1997 2002 1992 1996 2000 2001 2001 Average Average Average Average Average Water year Water n 6 6 9 5 9 8 6 9 6 Uncompahgre River basin 10 12 12 12 12 12 17 27 303(d) list segment Gunnison River from North Fork to Grand Junction, Colorado COGUNLG04c COGUNUN12 COGUNUN12 COGUNUN12 COGUNUN12 COGUNLG04a COGUNLG04a 2 Site name near Montrose Road River 92, near Read mouth, at Whitewater above Red Rock Canyon at mouth Red Rock Canyon Dry Cedar Creek Cedar Creek near mouth at N. Loutzenhizer Arroyo Dry Creek at mouth, near Delta Drain at Highway Sunflower Creek .4 miles Whitewater Selenium loading analyses for sites using average time-weighted mean (ATWM) method, Lower Gunnison River Basin, western Colorado, 1992–2005. Selenium loading analyses for sites using average time-weighted mean (ATWM) 13 14 15 19 20 35 51 (fig. 1) Average is the average of available water years. water of available is the average Average nearby sites into a single site. Combined two 1 2 Site number Table 3. Table 2007a); n, number of samples (selenium concentration and instantaneous discharge); (Colorado Department of Public Health and Environment, identifier 303(d) list segment [303(d) list segment, 85th percentile, percentile selenium concentration in µg/L; mean daily load, pounds per day; annual year; load reduction, Percent standards] percent of total load that needs to be reduced bring site into compliance with water-quality Dissolved Selenium Loading for Select Sites in the Lower Gunnison River Basin 15

Inset map

37# #36 ison Rive 42 28 40 # n r 43## 38 n u # # G 35 # 39 44# # # 41 26 U # n c # Delta o 27 m p a h g

r e

R iv er 24 Dry # 25 20 ## 23 Creek ##22

0 5 MILES

0 5 KILOMETERS

108°15' 108° 107°45'

29# #12 10 # 42 30 r 28 40 N ve 37 36 34 #32 .F i 43 ## # 31# . G R G unnison ## # # # un # 38°45' ni # ## 38 35 33 son River 11 k # # # e 46 45 39 e 41 # r # 44 # Sm Fork C Delta DELTA ith 26 te 27 an cal 24 25 See inset Es # 23 map 20 #### 22 D

k r

e y

e #

r C

C r 21 e e t e # n k 19 la k a e c e r 18 L s # U E C o n k d u r o c t o o o s e F w m y n n r to k p h t a i D o ee z C r h e C g r r k e r A o e it e r r r n o

o C R y 13# k o M e u i e v r a e C e l r l d e i b w s u MONTROSE i o r R C 17# EXPLANATION 15 38°30' # Residential or urban land use 16# Mexican C Gulch S ed p ar Creek Agricultural land use r i ng C r 303(d) listed segment e Montrose e k #Dry 14 17# C Sampling site and map e d a index number r Cr eek Base from U.S. Geological Survey digital data, 1:24,000 Land use from National Land Use Cover Data, 1983 Universal Transverse Mercator projection, Zone 13 0 5 10 MILES Horizontal coordinate information is referenced to the North American Data of 1983 (NAD 83) 0 5 10 KILOMETERS

Figure 5. Location of sites and land use in the Uncompahgre River Basin, western Colorado. 16 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005 cient data to compute an ATWM. For the 1995 water year, the of the mean annual load would need to be reduced to bring this 85th percentile selenium concentration was 56.5 µg/L (table site into compliance with the water-quality standard. For the 3), and the mean daily load was 0.636 pound (232 pounds 1997 water year, the 85th percentile selenium concentration annually). For water year 1995, 92 percent (213 pounds) of the was 154 µg/L, and the mean daily load was 12.3 pounds mean annual load would need to be reduced to bring this site (4,490 pounds annually). Consistent with the 1996 water year, into compliance with the water-quality standard. Similarly for 97 percent of the mean annual load would need to be reduced the 1996 water year, the 85th percentile selenium concentra- to bring this site into compliance with the water-quality stan- tion was 67.4 µg/L, and the mean daily load was 0.843 pound dard. For the 2002 water year, the 85th percentile selenium (309 pounds annually). The mean annual load would need concentration was 215 µg/L, and the mean daily load was to be reduced by 93 percent (287 pounds) to bring this site 16.1 pounds (5,870 pounds annually). Ninety-eight percent of into compliance with the water-quality standard. The average the mean annual load would need to be reduced to bring this amount of load that would need to be reduced at this site for site into compliance with the water-quality standard. The aver- water years 1995 and 1996 was 250 pounds or 93 percent of age amount of load that would need to be reduced for this site the mean annual load. for water years 1996, 1997, and 2002 was 5,070 pounds, or Streamflow quantity and quality affect the results of 98 percent of the mean annual load. Streamflow at Loutzen- loading analysis at Dry Cedar Creek and similar sites like hizer Arroyo is almost entirely comprised of irrigation water, Cedar Creek and Loutzenhizer Arroyo. The percentage of the which is likely the reason for the high percentage of mean mean annual selenium load that needs to be reduced at a site annual load that needs to be reduced to bring the site into to bring it into compliance with the water-quality standard is compliance with the water-quality standard (Butler and Leib, affected by the amount of streamflow as well as the selenium 2002). concentration in the streamflow. Due to geology, land use, and Dry Creek (site 20) data spans the 1991 through 2001 climate in the Lower Gunnison River Basin, the concentration water-year period. Water year 1992 had sufficient data to com- of selenium in streamflow at some sites tends to be high. In pute an ATWM. For the 1992 water year, the 85th percentile the study area, streamflow with low selenium concentrations selenium concentration was 9.80 µg/L, and the mean daily tends to originate from snowmelt or storm-related streamflow load was 3.37 pounds (1,230 pounds annually). Fifty-three (Butler and others, 1991). Dry Cedar Creek does not receive percent of the mean annual load would need to be reduced to appreciable amounts of snowmelt or storm-related stream- bring this site into compliance with the water-quality standard. flow which, in combination with geology and land use, would The Uncompahgre River at Delta (site 26) is a gaged site; explain the high selenium concentrations observed at this site. therefore, mean annual selenium loads were calculated using Cedar Creek (site 15) data spans the 1991 through 2001 regression analysis developed with the FORTRAN program water-year period. Water years 1992, 1993, 1996, 1997, and LOADEST (Runkel and others, 2004) for water years 2001 2000 had sufficient data to compute an ATWM (table 3). through 2005. Daily selenium concentration was estimated Of the 5 water years analyzed, the greatest amount of load using the following regression model: that needed to be reduced to meet the water-quality standard occurred in the 1992 water year and the least amount of load a0 + a1 per + a2 lnQ + a3 lnQper, (4) that needed to be reduced occurred in 2000. For water year 1992, the 85th percentile selenium concentration was Explanatory variables used for this regression model include 40.8 µg/L, and the mean daily load of selenium was variables that describe selenium concentration dependence 6.69 pounds (2,440 pounds annually) (table 3). For 1992, on streamflow and using a dummy variable (per) to define the mean annual load would need to be reduced by 89 per- irrigation water use. The dummy variable was assigned a value cent to bring this site into compliance with the water-quality of either 1 or 0 to indicate the irrigation season. November standard. For water year 2000, the 85th percentile selenium through April is the non-irrigation season and was assigned concentration was 29.5 µg/L, the mean daily load was 4.20 a value of 1, whereas March through October are considered pounds (1,540 pounds annually). For 2000, the mean annual the irrigation season and were assigned a value of 0. Approxi- load would need to be reduced by 84 percent to bring this site mately 60 percent of the variability in the data was explained into compliance with the water-quality standard. The average (r2 = 0.589), the regression model was significant at the 5-per- amount of load that would need to be reduced for this site for cent level (p<0.05) (table 4), and generally showed constant water years 1992, 1993, 1996, 1997, and 2000 was variance throughout the range of prediction in residual plots. 1,730 pounds or 87 percent of the mean annual load. The low r2 value could be due in part to the effect of diversions Loutzenhizer Arroyo (site 19) data spans the 1991 by water users on streamflow conditions during the estimation through 2003 water-year period. Water years 1996, 1997, and period. 2002 had sufficient data to compute an ATWM. For the 1996 For water years 2001 through 2005, the average 85th water year, the 85th percentile selenium concentration was percentile selenium concentration was 14.8 µg/L (table 5), and 180 µg/L, which is nearly 40 times the water-quality standard the mean daily selenium load was 14.9 pounds (5,420 pounds of 4.6 µg/L. The mean daily load was 14.4 pounds annually). The average amount of load that would need to be (5,240 pounds annually) (table 3). Ninety-eight percent reduced for water years 2001 through 2005 was 3,730 pounds Dissolved Selenium Loading for Select Sites in the Lower Gunnison River Basin 17 6.04 6.65 9.51 12.9 Upper interval 3.88 5.88 7.45 11.0 Lower 95 percent confidence (µg/L) 4.87 6.26 8.43 11.9 concentration Mean selenium 0.49 0.55 0.19 0.53 (µg/L) of estimate Standard error 0.0134 0.0496 0.0181 0.0184 p-value 2 r 0.589 0.734 0.869 0.778 - 50/0 22/0 48/0 48/0 sored values number of cen Number of values/ - 4 5 6 7 ber from text Equation num - ment 303(d) list seg COGUNUN04b COGUNLG02 COGUNLG02 Site name Delta Delta Grand Junction Uncompahgre River at Uncompahgre River at Gunnison River near Gunnison River Summary of regression model diagnostics at sites, Lower Gunnison River Basin, western Colorado, 2001–05. 26 40 52 Site (fig. 1) , the coefficient of determination] , the coefficient number 2 Table 4. Table [r 18 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005 - 69 69 70 69 67 69 33 30 51 36 11 34 37 44 61 58 49 53 reduction Percent load 689 4,090 2,990 3,210 3,890 4,450 3,730 3,270 2,340 3,440 2,150 2,510 6,170 6,530 8,960 9,510 9,460 8,640 Load reduction 5,940 4,350 4,570 5,670 6,600 5,420 9,790 7,750 6,720 5,940 6,550 7,350 16,800 14,900 14,700 16,300 19,500 16,400 Mean annual load 16.3 11.9 12.5 15.5 18.1 14.9 26.8 21.2 18.4 16.3 17.9 20.1 45.9 40.9 40.1 44.7 53.3 45.0 Mean daily load 14.7 15.5 14.6 14.1 14.8 6.59 9.42 7.21 5.14 6.99 8.18 11.8 11.0 8.94 9.70 14.8 6.91 7.27 85th percentile 1 1 1 Uncompahgre River Basin 2002 2003 2004 2005 2002 2003 2004 2005 2002 2003 2004 2005 2001 2001 2001 Average Average Average Water year Water - ment Gunnison River from North Fork to Grand Junction, Colorado 303(d) list seg COGUNLG02 COGUNLG02 COGUNUN04b Site name Junction Uncompahgre River at Delta Uncompahgre River at Delta Gunnison River near Grand Gunnison River Selenium loading analyses for sites using regression analysis (LOADEST) method, Lower Gunnison River Basin, western Colorado, 2 001–05 water years. 26 40 52 (fig. 1) Average is computed with all the daily values available for 2001–05 water years. for 2001–05 water available is computed with all the daily values Average 1 Site number Table 5. Table 85th per 2007a); n, number of samples (selenium concentration and instantaneous discharge); (Colorado Department of Public Health and Environment, identifier 303(d) list segment [303(d) list segment, centile, 85th percentile selenium concentration in µg/L; mean daily load, pounds per day; annual year; load reduction, Percent percent of total that standards] needs to be reduced bring site into compliance with water-quality Dissolved Selenium Loading for Select Sites in the Lower Gunnison River Basin 19 or 69 percent of the mean annual load. The 85th percentile indicate the presence of more dilution water in 2000; however, selenium concentration for the 5 water years ranged from annual streamflow information was not available to confirm 14.1 to 15.5 µg/L (water years 2005 and 2003, respectively). this. The mean daily loads calculated for each year ranged from Whitewater Creek (site 51) data spans the 1999 through 11.9 to 18.1 pounds or 4,350 to 6,600 pounds annually (water 2002 water-year period. Only water year 2001 had sufficient years 2002 and 2005, respectively). data to compute an ATWM. The 85th percentile selenium concentration was 62.6 µg/L, and the mean daily load of selenium was 0.332 pound (121 pounds annually) (table 3). The average Gunnison River from North Fork Gunnison River amount of load that would need to be reduced for water year to Grand Junction, Colorado 2001 was 112 pounds or 93 percent of the mean annual load. The Gunnison River at Delta (site 40) is a gaged site; Twenty-six sites were selected along the Gunnison River therefore, mean annual selenium loads were calculated using from North Fork Gunnison River to Grand Junction, Colorado, LOADEST for water years 2001 through 2005. Daily sele- that represent the cumulative contribution to 303(d) list seg- nium concentration was estimated by the following regression ments COGUNLG02 (sites 40 and 52), COGUNLG04a and model: b (sites 29-35, 37-39, 41-51 and 53-54) , and COGUNLG07 (site 36) (table 1 and fig. 6). Of the 26 sites, 4 had sufficient 2 a0 + a1 lnQ + a2 lnQ + a3 sin(2πdtime) + a4 cos(2πdtime) + a5dtime, data to determine mean annual selenium loads (tables 3 and 5). (5) For the 22 sites that lacked sufficient data to calculate mean Explanatory variables used for the regression model included annual selenium loads, dates ranged from 1991 to 2002 for variables that describe selenium concentration’s dependence USGS data and from 1991 to 2005 for CDPHE data (table 2). on streamflow, seasonality, and time trends as described for Eight of the 22 sites had five or more samples and equation 2. Approximately 70 percent of the variability in the 85th percentile selenium concentrations that exceeded the data was explained (r2 = 0.734), the regression model was sig- water-quality standard (sites 31-34, 44, 46, 49, and 53, nificant at the 5-percent level (p<0.05) (table 4), and generally table 2). showed constant variance throughout the range of prediction Mean annual selenium loads were estimated for Sun- in residual plots. flower Drain (site 35) and Whitewater Creek (site 51) using For water years 2001 through 2005 at the Gunnison River ATWM method (table 3). Mean annual selenium loads were at Delta, the average 85th percentile selenium concentration estimated for Gunnison River at Delta (site 40) and Gunnison was 6.99 µg/L (table 5), and the mean daily load was River near Grand Junction (site 52) using LOADEST for the 20.1 pounds (7,350 pounds annually) (table 5). The average most recent 5 years of data (2001 through 2005) (table 5). amount of load that would need to be reduced for this site for Sunflower Drain (site 35) data spans the 1991 to 2001 water years 2001 through 2005 was 2,510 pounds or water-year period. Water years 1996, 2000, and 2001 had suf- 34 percent of the mean annual load. The 85th percentile ficient data to compute an ATWM (table 3). For the selenium concentration for the 5 water years ranged from 5.14 1996 water year, the 85th percentile selenium concentra- to 9.42 µg/L (water years 2005 and 2003, respectively). Mean tion was 157 µg/L, and the mean daily load was 4.02 pounds daily loads ranged from 16.3 to 26.8 pounds, or 5,940 to (1,470 pounds annually). For the 1996 water year, the mean 9,790 pounds annually (water years 2004 and 2001, respec- annual load would need to be reduced by 97 percent tively). Load reductions ranged from 689 pounds (or (1,420 pounds) to bring this site into compliance with the 11 percent of the mean annual load) to 3,440 pounds (or water-quality standard. For the 2000 water year, the 51 percent of the mean annual load) (water years 2005 and 85th percentile selenium concentration was 36.5 µg/L, and 2003, respectively). the mean daily load was 2.64 pounds (964 pounds annually). The Gunnison River near Grand Junction (site 52) is a For the 2000 water year, the mean annual load would need gaged site with continuous specific conductance data; there- to be reduced by 88 percent (844 pounds) to bring this site fore, mean annual selenium loads were calculated using into compliance with the water-quality standard. For the 2001 LOADEST for water years 2001 through 2005. Due to gaps water year, the 85th percentile selenium concentration was in the specific conductance data record, it was necessary to 104 µg/L, and the mean daily load was 3.18 pounds model selenium concentrations in two steps. Where specific (1,160 pounds annually). For 2001, the mean annual load conductance data were available, daily selenium concentration would need to be reduced by 96 percent (1,110 pounds) to was estimated by the following regression model: bring this site into compliance with the water-quality standard. The average amount of load that would need to be reduced for (6) a0 + a1ln(SC) + a3 sin(2πdtime) + a4 cos(2πdtime), this site for water years 1996, 2000, and 2001 was 1,130 pounds or 94 percent of the mean annual load. The Explanatory variables used for the regression model include 85th percentile selenium concentration in 2000 was appre- those that describe selenium concentration dependence on ciably less than that observed in 2001, whereas mean annual specific conductance and seasonality as described for equation selenium loads were approximately the same, which would 2. Approximately 90 percent of the variability in the data was 20 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005

108°30' 15' 108° 39° 53 Whitewater # W ek # # 52 hitewater Cre 54 ## 50 51 See inset Creek k nah e map 1 Kan e r C

t s 49 a # E

k ee r Cr Dee

k e 48 e MESA # r C

e ek h c re c fa C l r t Gu u an reek ls See inset S rr C el DELTA u ez W k u e C g 47 e map 2 in r m # G C o D i u l 30 ig n B a # 29 n k # is l 42 o A 28 34 n 40 39 37 # R 43 ## # 31# iver 44 38°45' ## # 36 # # # #### 38 35 33 32 46 45 # # 41 # Delta 27 26 24 23 25 20 # EXPLANATION #### 22 k e ek e Residential or urbanre land use r 21# C C 19 e U t D an d l n a o r c c # s o y Agricultural land use k o L o E e w C ut re n m se o r n C t p t e h e a # t o e i z 303(d) listed segment n k h a C 18 e ek g r Cre l a r A o c s e r l s r 18 # e E o K Sampling site and map y k R r o o i index number F v y e r MONTROSE r D

Base from U.S. Geological Survey digital data, 1:24,000 Land use from National Land Use Cover Data, 1983 Universal Transverse Mercator projection, Zone 13 0 5 10 15 MILES Horizontal coordinate information is referenced to the North American Data of 1983 (NAD 83) 0 5 10 15 KILOMETERS

Inset map 1 Inset map 2

e e 53 Whitewater ek r e C # r 54 C e # r c # a te f 52 a 37# r w 36 u # te # S 50# hi W on R

51 40 is iver G 42 28 # n

43# 38 n u # u #

n G 35

n # i

s # o 39 n 44 # # # R iv 41 U #26 k e e r K n e a c # Delta r nn o 27 C a m h t Cr s eek p # a a 49 h E g

r e

0 5 MILES R iv e 0 5 KILOMETERS r 24 25 Dry # 20 ## 23 Creek ##22

0 5 MILES

0 5 KILOMETERS

Figure 6. Location of sites and land use associated with the Gunnison River from North Fork Gunnison River to Grand Junction, western Colorado. Dissolved Selenium Loading for Select Sites in the Lower Gunnison River Basin 21 explained (r2 = 0.869), the regression model was significant at selenium concentration (see the “Methods for Analysis of the 5-percent level (p<0.05) (table 4), and generally showed Selenium Loading” section), which inherently preserves the constant variance throughout the range of prediction in distribution of the concentrations throughout the year. The residual plots. amount of mean annual load in tables 3 and 5 that need to Because specific conductance data were not available for be removed to bring a site into compliance (load reduction) the entire 2001-05 water-year period (94 percent of the period with the water-quality standard is calculated from a daily load had specific conductance data), streamflow was used to esti- which is then multiplied by 365.25 days. This would be a set mate selenium concentration where specific conductance data amount of selenium load that would need to be reduced every were not available. Where specific conductance data were not day over the entire year to achieve an 85th percentile selenium available, daily selenium concentration was estimated by the concentration that does not exceed the water-quality standard. following regression model: The water-quality standard is the 85th percentile selenium concentration that does not exceed 4.6 µg/L. The 85th

a0 + a1 lnQ + a2 sin(2πdtime) + a3 cos(2πdtime) + a4dtime, percentile selenium concentration is the concentration of (7) selenium below which 85 percent of the selenium concentra- Explanatory variables used for this regression model include tions occur. Therefore, load reduction does not need to occur those that describe selenium concentration dependence on during periods when selenium concentrations are less than the streamflow, seasonality, and time trends as described for equa- water-quality standard. Additionally, load reductions do not tion 2. Approximately 78 percent of the variability in the data need to occur during every instance when selenium concentra- was explained (r2 = 0.778), the regression model was signifi- tions exceed the water-quality standard. Only 85 percent of the cant at the 5-percent level (p<0.05) (table 4), and generally selenium concentrations need to be less than the water-quality showed constant variance throughout the range of prediction standard, which means 15 percent can exceed the standard. in residual plots. Specific conductance is an indication of the However, by reducing loads for a portion of the 15 percent amount of dissolved material (like selenium) that is in natural that exceeds the water-quality standard, the 85th percentile waters (Hem, 1985). Specific conductance cannot completely selenium concentration can be more effectively reduced. The describe the variability in selenium concentrations; however, number of days can be determined empirically by manipulat- specific conductance is a better estimator of selenium concen- ing the loading equation and solving for selenium concentra- tration than streamflow, as observed from the higher r2 value tion. By targeting load reductions, the resulting number of for equation 6 compared to equation 7. The selenium concen- days when load reductions would need to occur would be less tration data estimated using equation 6 were combined with than if load reductions occurred every day by some constant the estimated selenium concentration data from equation 7 to value as table 5 would indicate. create a complete daily record for selenium concentration over The sensitivity of when loads could be reduced was the time period. tested for the Gunnison River near Grand Junction (site 52). For water years 2001 through 2005 at the Gunnison River Multiple targeted load reduction scenarios were tested that near Grand Junction site, the average 85th percentile selenium included changing both the daily amount of selenium load concentration was 9.70 µg/L, and the mean daily load was reduced along with the number of days when load reductions 45.0 pounds (16,400 pounds annually) (table 5). The average were occurring within a water year. To determine the effect of amount of load that would need to be reduced for this site for these targeted load reduction scenarios, the load equation was water years 2001 through 2005 was 8,640 pounds or manipulated to solve for selenium concentration (where load 53 percent of the mean annual load. The 85th percentile = streamflow x selenium concentration x conversion factor). selenium concentration for the 5 water years ranged from Selenium concentrations were back calculated from the “new” 7.27 to 11.8 µg/L (water years 2001 and 2003, respectively). mean daily loads (post-load-reduction scenario) to calculate Mean daily load ranged from 40.1 to 53.3 pounds or 14,700 to the 85th percentile selenium concentration. As an additional 19,500 pounds annually (water years 2003 and 2005, respec- test of the sensitivity of load reductions, targeted load reduc- tively). Load reductions ranged from 6,170 pounds (or tion scenarios were tested for water years 2003 and 2005 for 37 percent of the mean annual load) to 9,510 pounds (or the Gunnison River near Grand Junction. Streamflow during 58 percent of the mean annual load) (water years 2001 and the 2003 water year was less than the 2005 water year. Mul- 2004, respectively). tiple load reduction scenarios were tested on these 2 years to demonstrate the relative effect that variable streamflow has on the amount and duration of load reductions. Applicability of Selenium Loading Analysis During 2003, the average annual load reduction needed to bring the Gunnison River near Grand Junction into compliance The average load reductions from table 5 can be reduced with the water-quality standard listed in table 5 was approxi- further by targeting the periods of time when selenium could mately 9,000 pounds (25 pounds daily for 365.25 days). By be removed from streams by remediation. The methods used targeting load reductions during specific periods of the water to calculate the load reductions provided in tables 3 and 5 use year when selenium concentrations exceeded the water-quality the ratio of the water-quality standard and the 85th percentile standard, the annual amount of load that would need to be 22 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005

25,000

22,500

20,000 d n o c

e 17,500 s

r e p

t 15,000 e e f

c i 12,500 b u c

w 10,000 o l f m

a 7,500

e Mean annual streamflow for r

S t period of study 1997–2001 5,000 Mean annual streamflow for period of study 2001–2005 2,500 2,500 1,500 0 0 10 20 30 40 50 60 70 80 90 100 Percent of time indicated discharge was equaled or exceeded Figure 7. Relation between flow-duration curve (water years 1976–2005) and mean annual streamflow for two periods, 1997–2001 and 2001–05, at Gunnison River near Grand Junction, Colorado, western Colorado. reduced would be approximately 8,000 pounds (32 pounds or exceeded (Searcy, 1959). Mean annual streamflows for the daily for 249 days). During 2005, the average annual load two study periods were plotted along the flow-duration curve reduction needed to bring the Gunnison River near to determine the percentage of time that each mean annual Grand Junction into compliance with the water-quality streamflow was equaled or exceeded. Streamflow during the standard listed in table 5 was approximately 9,500 pounds 1997–2001 study period was equaled or exceeded 35 percent (26 pounds daily for 365.25 days). By targeting load reduc- of the time, whereas streamflow observed during the 2001–05 tions during specific periods of the water year when selenium study period was equaled or exceeded 68 percent of the time, concentrations exceeded the water-quality standard, the annual indicating that streamflow during the 1997–2001 study period amount of load that would need to be reduced would be was appreciably higher (33 percent) than that observed during approximately 6,000 pounds (29 pounds daily for 205 days). the 2001–05 study period. Assuming that the source of sele- Therefore, using the Gunnison River near Grand Junction as nium in the study area has not increased from the 1997–2001 an example, if load reductions could be targeted, there is a study period, it can be assumed that the additional streamflow potential to greatly reduce the total amount of selenium load acted to dilute selenium concentrations during the 1997–2001 reduction required to bring a specific site into compliance with study period. This overall reduction in selenium concentra- the water-quality standard. tions would result in a reduced 85th percentile selenium con- To further illustrate the variability in load reductions, centration as compared to that observed for the 2001–05 study the results of this study were compared to a previous study period. This example illustrates that a different study period of load reductions at the Gunnison River near Grand Junc- can result in a different load reduction at the same site. tion. The previous study was conducted from 1997–2001 and used similar methods for determining load reductions (Bureau of Reclamation, 2006). The 1997–2001 study estimated the average load reduction at 5,000 lbs/year as compared to over Summary 8,000 lbs/year reported in this study. The difference between these two values can be explained in large part by variations in Elevated selenium concentrations in streams are a water- streamflow. Mean annual streamflows for the two study peri- quality concern in western Colorado. Segments of and tributar- ods were plotted on a flow-duration curve for Gunnison River ies to the North Fork Gunnison River, Red Rock Canyon at near Grand Junction, water years 1976–2005 (fig. 7). A flow- the mouth, Uncompahgre River, and the Gunnison River from duration curve is a cumulative frequency curve compiled by North Fork Gunnison River to Grand Junction, Colorado, have ranking all daily mean streamflows for a given time period (in been placed on the State 303(d) list as impaired with respect this case, 1976–2005) in order of their magnitude, then com- to dissolved selenium. In Colorado, the Water Quality Control puting the percentage of time a given streamflow is equaled Division (Division) of the Colorado Department of Public Summary 23

Health and Environment is required to develop total maximum their mean annual load to be reduced in order to bring these daily loads of selenium for the 303(d) list segments. sites into compliance with the water-quality standard on the The U.S. Geological Survey, in cooperation with Colo- basis of available data. Cedar Creek and Loutzenhizer Arroyo, rado Department of Public Health and Environment, sum- like Dry Cedar Creek, do not receive appreciable snowmelt- marized selenium loading in the Lower Gunnison River related streamflow and are more influenced by the application Basin to support the total maximum daily loads development of irrigation water. process. This report provides a brief description of the study Four of the 26 sites along the Gunnison River from area including a summary of selenium issues, methods of data North Fork Gunnison River to Grand Junction, Colorado, had analysis, and results. Selenium issues in the study area were sufficient data to determine mean annual selenium loads. For summarized and include identification of primary land uses Sunflower Drain and Whitewater Creek, on average more that cause selenium loading. Results include the assessment of than 90 percent of their mean annual loads would need to be available selenium concentrations and loads, 85th percentiles reduced to bring these sites into compliance with the water- of selenium concentration, and quantification of the amount quality standard based on available data. For the Gunnison of the selenium load (pounds annually) that would need to River at Delta, 34 percent of the mean annual load on average be reduced to bring the site into compliance with the State needed to be reduced to bring the site into compliance with the chronic aquatic-life standard for dissolved selenium, herein water-quality standard from 2001 through 2005. For Gunnison referred to as “a load reduction.” Selenium load reductions River near Grand Junction, 53 percent of the mean annual load were determined for sites where selenium concentrations on average needed to be reduced to bring the site into compli- exceeded the State chronic aquatic-life standard for dissolved ance with the water-quality standard from 2001 through 2005. selenium (85th percentile selenium concentration not to The method used to calculate the load reductions pro- exceed 4.6 µg/L), referred to as the “water-quality standard.” vided in this report used the ratio of the water-quality standard Data for 54 historical monitoring sites with selenium and the 85th percentile selenium concentration, which inher- sample results were compiled from U.S. Geological Survey ently preserves the distribution of the concentrations through- and Colorado Department of Public Health and Enivronment out the year. The average load reductions needed to bring these data sources for this study. Three of the sites were located sites into compliance with the water-quality standard could be at continuous streamflow-gaging stations (gaged) and the reduced further by targeting the periods of time when selenium remaining 51 sites were not located at streamflow-gaging sta- is removed from streams through remediation. The sensitivity tions (ungaged) but did have streamflow data associated with of when loads could be reduced was tested on the Gunnison the selenium concentration data. This resulted in a varying River near Grand Junction. ability to estimate mean annual selenium loads. Three meth- Multiple targeted load reduction scenarios were tested ods were used for analysis of selenium concentration data to that included changing both the daily amount of selenium load address variable data density among sites. For sites with low reduced along with the number of days when load reductions sample counts (typically 20 samples or less), statistical sum- were occurring within a water year. As an additional test of the maries were determined. For ungaged sites with approximately sensitivity of load reductions, targeted load reduction scenarios 20 or more samples, a time-weighting technique was used for were tested on water years 2003 and 2005 for the Gunnison estimation of mean annual selenium loads. For gaged sites, a River near Grand Junction. Streamflow during the 2003 water regression analysis was used to estimate daily mean selenium year was lower than the 2005 water year. Multiple load reduc- concentration and mean annual selenium loads. The 85th tion scenarios were tested for the 2003 and 2005 water years percentile selenium concentrations were determined for all to demonstrate the relative effect that variable streamflow has sites with five or more samples. In the event of a water-quality on the amount and duration of load reductions. standard exceedance, the amount of the selenium load (pounds During 2003, the average annual load reduction for the annually) that would need to be reduced to bring the site into Gunnison River near Grand Junction was approximately compliance with the water-quality standard was determined. 9,000 pounds (25 pounds for 365.25 days). By targeting load The load reduction calculation used a simple mass-balance reductions during specific periods of the water year when approach that assumed no change in streamflow volume and selenium concentrations exceeded the water-quality standard, that only selenium would be removed from the system. the annual amount of load that would need to be reduced None of the sites in the North Fork Gunnison River Basin would be approximately 8,000 pounds (32 pounds for 249 had sufficient data to determine a mean annual selenium load. days). During 2005, the average annual load reduction for Red Rock Canyon at the mouth, on average, would need the Gunnison River near Grand Junction was approximately 92 percent of the mean annual load reduced to bring this site 9,500 pounds (26 pounds for 365.25 days). By targeting load into compliance with the water-quality standard. reductions during specific periods of the water year when sele- Five of the 15 sites in the Uncompahgre River Basin (Dry nium concentrations exceeded the water-quality standard, the Cedar Creek, Cedar Creek, Dry Creek, Loutzenhizer Arroyo, annual amount of load that would need to be reduced would be and the Uncompahgre River at Delta) had sufficient data to approximately 6,000 pounds (29 pounds for 205 days). determine mean annual selenium loads. All five sites except Load reductions were previously studied from 1997- Dry Creek (53 percent) would require more than 70 percent of 2001 at the Gunnison River near Grand Junction. Estimated 24 Analysis of Dissolved Selenium Loading for Selected Sites in the Lower Gunnison River Basin, Colorado, 1978–2005 load reduction was 5,000 pounds per year as compared to Colorado Department of Public Health and Environment, more than 8,000 pounds per year estimated in this report for 1998, The basic standards and methodologies for surface the 2001–05 study period. Streamflow during the 1997–2001 water: Denver, Regulation 31, Water Quality Control Com- study period was appreciably higher (33 percent) than that mission, 140 p. observed during the 2001–05 study period. Higher streamflow during the 1997–2001 study period was likely the reason for Colorado Department of Public Health and Environment, the smaller load reduction (more dilution water). This example 2007a, Colorado 303(d) lists: Information available on Web, illustrates that a different study period can result in a different accessed August 29, 2007, at http://www.cdphe.state.co.us/ load reduction at the same site. wq/Assessment/TMDL/303dlists.html. Colorado Department of Public Health and Environment, 2007b, Colorado’s 1998 303(d) List and Related Water Quality Management Lists Text: Information available on References Cited Web, accessed May 29, 2007, at http://www.cdphe.state. co.us/op/wqcc/SpecialTopics/303(d)/303dtext.pdf

Brooks, T., and Ackerman, D.J., 1985, Reconnaissance of Colorado Department of Public Health and Environment, ground-water resources in the Lower Gunnison River Basin, 2007c, Water quality control commission regulations: Infor- southwestern Colorado: U.S. Geological Survey Water- mation available on Web, accessed May 29, 2007, at http:// Resources Investigations Report 84–4185, 35 p. www.cdphe.State.co.us/regulations/wqccregs/index.html

Bureau of Reclamation, 2006, Evaluation of selenium Crawford, C.G., 2004, Sampling strategies for estimating remediation concepts for the lower Gunnison and lower acute and chronic exposures of pesticides in streams: Jour- Uncompahgre Rivers, Colorado: National Irrigation Water nal of American Water Resources Association, 489 p. Quality Program, Reclamation Technical Assistance to States Program, Gunnison Basin Selenium Task Force, 26 p. Engberg, R.A., 1999, Selenium budgets for Lake Powell and the upper Colorado River Basin: Journal of the American Butler, D.L., 2001, Effects of piping irrigation laterals on Water Resources Association, v. 35, no. 4, p. 771–786. selenium and salt loads, Montrose Arroyo Basin, western Colorado: U.S. Geological Survey Water-Resources Investi- Geographic Names Information System, 2007, Query form for gations Report 01–4204, 14 p. the United States and its territories: Information available on the Web, accessed January 29, 2007, at http://geonames. Butler, D.L., Krueger, R.P., Osmundson, B.C., Thompson, usgs.gov/pls/gnispublic A.L., and McCall, S.K., 1991, Reconnaissance investiga- tion of water quality, bottom sediment, and biota associated Green, G.N., 1992, The digital geologic map of Colorado with irrigation drainage in the Gunnison and Uncompahgre in ARC/INFO format: U.S. Geological Survey Open-File River Basins and at Sweitzer Lake, west-central Colorado, Report 92–0507, 12 diskettes. 1988–89: U.S. Geological Survey Water-Resources Investi- Gunnison Basin Selenium Task Force, 2007, Current proj- gations Report 91–4103, 99 p. ects—effects of changes in land use on selenium loading Butler, D.L., and Leib, K.J., 2002, Characterization of in the Whitewater, CO, Area: Information available on the selenium in the Lower Gunnison River Basin, Colorado, Web, accessed February 14, 2007 at http://www.seleni- 1988–2000: U.S. Geological Survey Water-Resources umtaskforce.org/projects1.html. Investigations Report 02–4151, 26 p. Helsel, D.R., and Hirsch, R.M., 1992, Statistical methods in Butler, D.L., Wright, W.G., Stewart, K.C., Osmundson, B.C., water resources: New York, Elsevier, 522 p. Krueger, R.P., and Crabtree, D.W., 1996, Detailed study of Hem, J. D., 1985, Study and interpretation of the chemical selenium and other constituents in water, bottom sediment, characteristics of natural water: U.S. Geological Survey soil, alfalfa, and biota associated with irrigation drainage Water Supply Paper 2254, 264 p. in the Uncompahgre Project Area and in the Grand Valley, west-central Colorado 1991–93: U.S. Geological Survey Homer, C., Huang, C., Yang, L., Wylie, B., and Coan, M., Water-Resources Investigations Report 96–4138, 136 p. 2004, Development of a 2001 National Landcover Database for the United States: Photogrammetric Engineering and Colorado’s Decision Support Systems, 2007, GIS Layer Data Remote Sensing, v. 70, no. 7, July 2004, p. 829–840. - Division 4 - Gunnison: Information available on the Web, accessed January 23, 2007 at http://cdss.state.co.us/DNN/ Gunnison/tabid/56/Default.aspx. References Cited 25

Larson, S.J., Crawford, C.G., and Gilliom, R. J., 2004, U.S. Geological Survey, variously dated, National field manual Development and application of watershed regressions for for the collection of water-quality data: U.S. Geological pesticides (WARP) for estimating atrazine concentration Survey Techniques of Water-Resources Investigations, book distributions in streams: U.S. Geological Survey Water- 9, chaps. A1-A9; Available online at http://pubs.water.usgs. Resources Investigations Report 03–4047, 15 p. gov/twri9A. Leib, K.J., Mast, M.A., and Wright, W.G., 2003, Using water- Western Regional Climate Center, 2007, Colorado climate quality profiles to characterize seasonal water quality and summaries: Information available on Web, accessed April 4, loading in the upper Animas River Basin, southwestern 2007, at http://www.wrcc.dri.edu/summary/climsmco.html Colorado: U.S. Geological Survey Water-Resources Investi- gations Report 02–4230, 43 p. Wright, W.G., and Butler, D.L., 1993, Distribution and mobilization of dissolved selenium in ground water of the Lemly, A.D., 2002, Selenium assessment in aquatic ecosys- irrigated Grand and Uncompahgre Valleys, western Colo- tems—a guide for hazard evaluation and water quality rado, in Allen, R.G., and Neale, C.M.U., eds., Management criteria: New York, Springer-Verlag, 3 p. of irrigation and drainage systems—integrated perspectives: American Society of Civil Engineers, Proceedings of the Microsoft Corporation, 2003, Microsoft Excel: Redmond, 1993 National Conference on Irrigation and Drainage Engi- Wash., copyright © Microsoft Corporation 1985–2003. neering: Park City, Utah, July 21-23, 1993, p. 770–777. Ohlendorf, H.M., Hoffman, D.J., Saiki, M.K., and Aldrich, T.W., 1986, Embryonic mortality and abnormalities of aquatic birds—apparent impact of selenium from irrigation drainwater: Science of the Total Environment, v. 52, p. 49–63. Ohlendorf, H.M., Kilness, A.W., Simmons, J.L., Stroud, R.K., Hoffman, D.J., and Moore, J.F., 1988, Selenium toxicosis in wild aquatic birds: Journal of Toxicology and Environmen- tal Health, v. 24, p. 67–92. Rantz, S.E., and others, 1982, Measurement and computation of streamflow: U.S. Geological Survey Water-Supply Paper 2175, 2 v., 631 p. Runkel, R.L., Crawford, C.G., and Cohn, T.A., 2004, Load estimator (LOADEST)—a FORTRAN program for estimating constituent loads in streams and rivers: U.S. Geological Survey Techniques and Methods, book 4, chap. A5, 69 p. Searcy, J.K., 1959, Flow-duration curves, manual of hydrol- ogy-part 2, Low-flow techniques: U.S. Geological Survey Water-Supply Paper 1542-A, 33 p. Seiler, R.L., Skorupa, J.P., Naftz, D.L., and Nolan, B.T., 2003, Irrigation-induced contamination of water, sediment, and biota in the western United States-synthesis of data from the National Irrigation Water Quality Program: U.S. Geological Survey Professional Paper 1655, 123 p. U.S. Census Bureau, 2007, Population finder: Information available on Web, accessed May 29, 2007, at http://www. census.gov/

U.S. Environmental Protection Agency, 2006, Overview of current total maximum daily load - TMDL - program and regulations: Information available on Web, accessed January 18, 2007, at http://www.epa.gov/owow/tmdl/overviewfs.html U.S. Environmental Protection Agency, 2007, Non point source news — notes January 2007, Issue #80, 32 p.

Publishing support provided by: Rolla Publishing Service Center

For more information concerning this publication, contact: Director, USGS Colorado Water Science Center Denver Federal Center, MS 415 Bldg 53, Room E-2313 Lakewood, CO 80225

Or visit the Colorado Water Science Center Web site at: http://co.water.usgs.gov

Thomas and others—Analysis of Dissolved Selenium Loading for Selected Sites—Scientific Investigations Report 2007–5287 Printed on recycled paper