Adam L. Majeski

GIS in Water Resources

Term Project – Final Report

December 8, 2006

A Brief Introduction to My Masters Thesis

Lake Powell is created by Glen Canyon Dam, on the Colorado River. The drainage area above the lake encompasses ~108,335 mi², collecting enough water to make Lake Powell the 2nd largest reservoir in the U.S and the current depocenter for 868,231 acre-feet (as measured in 1986) of sediment (Ferrari, 1988). Since dam closure in 1963, the reservoir has experienced one transgression, filling to the top of its active storage (3700 ft) in 1980, and a minor regression from 1988 to 1993. The lake recovered by 1999, but this year also marked the beginning of the most significant flow deficit in the Colorado River basin in over 100 years (Webb et al, 2004). As a result, the lake dropped 150 ft over a 6 year period, and the delta sediments ~95 miles upstream of the dam, near the marina of Hite, were exposed for the first time. Since then, the Colorado River has been down cutting and reworking these sediments. Furthermore, delta sediments accumulating in side canyons and tributaries of the Colorado have been subject to exposure and reworking as well. Of particular interest are two tributaries, North Wash Creek and the Dirty Devil River, which join the Colorado at its delta near Hite. The Dirty Devil River is the only minor tributary of the lake to show significant sediment accumulation and it is one of the few to have its delta extend to the main channel storage area of the Colorado River. North Wash Creek enters ~3 km downstream of the Dirty Devil River and contrasts it greatly; its delta does not extend to the Colorado River and its sediment accumulation has been minor by comparison. Therefore, the current lowered reservoir state sets up a perfect natural laboratory to allow the response of these three different fluvial systems to the same drop in base-level to be compared and contrasted.

By surveying channel slopes, and cross sections based on those established by the Bureau of Reclamation in 1986, the present state of these systems can be compared to similar historical records and to pre-dam topography. This will also allow volumetric analysis of deposited delta sediment to be conducted in order to determine the amount mobilized and/or deposited. By observing photographs taken by several researchers, including John Dornwend, James Evans, William Vernieu, and myself, changes in delta morphology, and sediment deformation over time can be viewed in a qualitative manner. Finally, by observing delta stratigraphy and associated grain size distributions and bed material size distributions, inferences can be made on differences in incision rates, and the presence or absence of lateral slumping, mud cracking, and slope failure. Each piece of data will help to characterize each system and through comparison and contrast with the others, it will become more clear which responses are dependent and independent of the size and physical characteristics of the river system.

Where GIS Fits In

The driving theme behind my research is physical system characterization so comparisons can be made between these 3 systems. While much of my comparative data will come from field based methods, there is some that is more logically gathered in the office; drainage area delineation, drainage density, and interpolation of existing USGS gage records to streams where no recent ones exist. These three things will identify fundamental basin characteristics and the very different spatial scales of each system, which will help to frame and put into context the magnitudes of their responses to base level drop, to be investigated in my masters thesis.

Methods

Overview

To begin the investigation the necessary data sets were gathered; NHD and NHDplus for the United States Region 14 (The Upper Colorado Basin, from its headwaters to Lees Ferry, AZ), USGS gaging information point file, state boundaries, and major roads. From the NHD datasets, GIS was used to create feature datasets containing the HUC-6 basins of the Colorado River and their associated stream networks. From this large data set the HUC-10 watersheds of the Dirty Devil/Fremont River and North Wash creek and their associated stream networks was also isolated. The data was organized into personal geodatabases for the Colorado River and for the Dirty Devil/Fremont, North Wash features and everything was reprojected into the USA Contiguous Albers Equal Area Conic coordinate system using the NAD 83.

Drainage Basin Delineation

For the Colorado River above Lees Ferry, the HUC-6 basins were contained in 8 separate geodatabases for region 14. Each of these feature class layers were put into ArcGIS and the Colorado Drainage basin was identified by overlaying the NHDflowline layer and selecting the HUC-6’s it covered (the shape was also compared to a figure in the 1986 Bureau of Reclamation study). These were exported into a new feature class, duplicates from overlap were removed, and the data set was finalized. Drainage basin area was computed from the feature class’s attribute table category shape area with the summarize command.

For the Dirty Devil / Fremont Rivers, the HUC-8 subbasins were used as a template to select the necessary HUC-10 watersheds and the associated NHD flowlines. The HUC-8 for the Dirty Devil and Fremont Rivers were identified via the HUC-8 name column of the attribute table, and careful observation of the NHD flowlines in conjunction with previous research also dictated the inclusion of the Muddy Creek subbasin. For North Wash Creek a major highway feature class was added because the stream is known to trace Utah Highway 95 south from Hanksville. Using this spatial relationship and the identify tool, the HUC-10 for North Wash easily identified and a quick visual inspection of the small area showed that the entire drainage area had been captured. Again, drainage area was calculated using the summarize command on each feature class’ shape area attribute. See Figure 1.

Drainage Density

To compute drainage density, the NHD flowlines were added for the Upper Colorado Basin. In the case of the Dirty Devil / Fremont and North Wash, the flowline file was clipped by using the select by location tool and the drainage area polygons as the template. It is important to note that the NHD flowlines were used and NOT the NHDplus flowlines, because the NHD are of higher resolution (Figure 2). The summary command was used to sum the length of the drainage lines and this value was entered into an excel spread sheet, along with the drainage area. Drainage Density = Total Stream Length / Drainage Area.

North Wash Gage Interpolation

North Wash Creek only contains gaging records from 1950 to 1970 so interpolation from nearby gages with longer records on the Dirty Devil / Fremont River (DDF) was used to approximate current mean annual flow (MAF). The USGS gage feature class was added to the map and those lying along the Dirty Devil / Fremont and North Wash were selected by location and exported into their own feature class. Then, the gages on the DDF were further thinned to those containing at least 15 years of record. Each of these gages was ten selected with the identify tool and the active hyperlink displayed in the information box was followed to download the gage records (Figure 3). These were then placed into excel for analysis. Records were broken into intervals of the first and last 10 or 20 years and averaged. These averages were compared with the t-test to determine if they were significantly different, or if MAF was consistant for any given 10 to 20 year period. If MAF on the DDF showed to be invariant with time, then it could be assumed that the MAF on NW is as well. Therefore, the 1950-1970 MAF value would still hold true today.

As another method to calculate MAF on North Wash, the drainage area above each gage selected above was gathered from the info file. The MAF value was divided by this drainage area to compute a cfs per mi^2 vale, which was then multiplied by the drainage area of North Wash.

Results

Drainage Basin Delineation

The Upper Colorado Drainage Basin above Lees Ferry, AZ spans 5 states (Wyoming, Utah, Colorado New Mexico and Arizona) and covers a total area of 113,390.55 sq. mi. (Figure 4). It is comprised of the10 HUC-6 basins listed in the table 1 and also displayed on figure 4.

	Table 1
HUC_6	HU_6_Name	Area (km^2)	Area (mi^2)
140100	Colorado Headwaters	25496.89	9844.40
140200	Gunnison	20822.10	8039.46
140300	Upper Colorado-Dolores	21664.64	8364.76
140401	Upper Green	43799.45	16911.06
140402	Great Divide Closed Basin	9947.52	3840.76
140500	White-Yampa	34332.46	13255.84
140600	Lower Green	37701.43	14556.60
140700	Upper Colorado-Dirty Devil	35395.27	13666.19
140801	Upper San Juan	37583.40	14511.03
140802	Lower San Juan	26937.01	10400.44
	Total Area	293680.17	113390.55

To compute the drainage area above Glen Canyon Dam, it was necessary to remove any downstream drainage area. A majority of this excess area belongs to the Paria River and this area was removed by selecting its HUC-8 subbasin (Figure 5). Several other minor tributaries also enter below the dam and their NHDplus catchments were overlaid and selected, roughly removing all of their excess drainage area as well. The end result was a drainage area of 111,726.54 sq. mi., which is larger than the108,335 sq. mi. drainage area reported by the Bureau of Reclamation website on Glen Canyon Dam (Table 2). To reconcile this discrepancy flowline network tracing was implemented, with considerable difficulty, to isolate the NHDplus flowlines above Glen Canyon Dam (and therefore isolate the drainage area above the dam because each flowline is attributed with its contributing area via linking the flowlineattributeflow table to the flowline feature class). Upon tracing it became apparent that the network was not entirely connected, and that the effort to connect each segment by hand would be far too time consuming and also require referencing DEM’s to properly connect many of the segments (Figure 6). So the effort would not be in vain however, the connected and traced network was saved as a feature class and used to select by location the associated catchments. Then, those catchments not selected because of unconnected flowlines were roughly selected by hand. The total sum of the these two areas was very near 111,700 sq. mi., confirming that calculating the area above GCD with GIS will yield a higher value than that reported by the Bureau of Reclamation.

Figure 5 – Removing the Paria River drainage area to calculate the drainage area above Glen Canyon Dam.

Table 2
Total Drainage Area Below GCD =	4309760147.3100	m^2	4309.760147	km^2	1664.007698	mi^2
Upper Colorado River Drainage Basin Area =			293680.1706	km^2	113390.5479	mi^2
Drainage Basin Area Above GCD =			289370.4105	km^2	111726.5402	mi^2

The Dirty Devil / Fremont drainage area is comprised of 22 HUC-10 watersheds and it covers an area of 4374.05 sq. mi. (Figure 7 and Table 3). North Wash creek drains an area of 142.85 sq. mi. and is represented by 1 HUC-10 watershed (Figure 7 and Table 3).

Table 3
Dirty Devil Drainage Area
HUC_10	HU_10_Name	Shape_Area (m^2)	Shape_Area (km^2)		Shape Area (mi^2)
1407000408	Lower Dirty Devil River	326137972.7	326.1379727		125.9225754
1407000202	Headwaters Muddy Creek	419029537.7	419.0295377		161.7882092
1407000204	Salt Wash-Muddy Creek	451449431.8	451.4494318		174.3056003
1407000201	Ivie Creek	663385884.6	663.3858846		256.1347223
1407000203	Willow Spring Wash	338727335.8	338.7273358		130.7833557
1407000207	Wild Horse Creek	304816189.9	304.8161899		117.690189
1407000206	Red Canyon-Muddy Creek	352701457.3	352.7014573		136.1787942
1407000403	Upper Dirty Devil River	417976216.1	417.9762161		161.3815194
1407000205	Salt Wash	902300916.1	902.3009161		348.3803318
1407000301	Headwaters Fremont River	1004475775	1004.475775		387.8302654
1407000208	Outlet Muddy Creek	584417662.3	584.4176623		225.6449212
1407000303	Deep Creek-Fremont River	896583306	896.583306		346.1727502
1407000402	Robbers Roost Canyon	215183048.1	215.1830481		83.08263944
1407000306	Town Wash-Fremont River	400267456.6	400.2674566		154.5441292
1407000401	Dry Valley Wash	229381149.3	229.3811493		88.56455698
1407000406	Middle Dirty Devil River	370808952.3	370.8089523		143.1701371
1407000305	Sweetwater Creek-Fremont River	858902217.3	858.9022173		331.6240005
1407000405	French Spring Fork-Happy Canyon	234091961.8	234.0919618		90.38341185
1407000304	Sandy Creek-Fremont River	992904500.3	992.9045003		383.3625712
1407000407	Poison Spring Canyon	247287953.8	247.2879538		95.47841286
1407000404	Beaver Canyon-Granite Creek	208031956.1	208.0319561		80.32158738
1407000302	Pine Creek-Fremont River	909872998.3	909.8729983		351.3039291
		Total Drainage Area =	11328.73	km^2	4374.048609	mi^2
North Wash Drainage Area
	HU_10_Name	Shape_Area (m^2)	Shape Area (km^2)		Shape Area (mi^2)
	North Wash	369981699	369.981699		142.8507328
		Total Drainage Area =	369.981699	km^2	142.8507328	mi^2

As can be seen from the drainage basin delineation for each system, they represent 3 quite different spatial scales, all of which are responding to the base level drop of Lake Powell. By understanding the magnitude of this difference, each system’s response can be more fully understood and responses dependent or independent of basin size will be easier to identify.

Drainage Density

Drainage Density = Total Stream Length / Total Basin Area. Table 4 reports the results of drainage density analysis for the Upper Colorado, Dirty Devil / Fremont, and North Wash drainage areas.

Table 4

	Colorado River Drainage Basin Area =	293680.1706	km^2	113390.5479	mi^2
From summary statistics of the NDH Flowlines ==>	Total Stream Length in Drainage Basin Area =	463652.849	km	288100.5234	mi
	Drainage Density =	1.578767977		2.540780768
	Dirty Devil / Fremont River Drainage Basin Area =	11328.73388	km^2	4374.048609	mi^2
From summary statistics of the NDH Flowlines ==>	Total Stream Length in Drainage Basin Area =	17549.426	km	10904.70775	mi
	Drainage Density =	1.549107445		2.493046769
Table 3 - continued
	North Wash Creek Drainage Basin Area =	369.981699	km^2	142.8507328	mi^2
From summary statistics of the NDH Flowlines ==>	Total Stream Length in Drainage Basin Area =	481.66807	km	299.2946628	mi
	Drainage Density =	1.301869988		2.09515665

The topology of a basin depends in large part on the interaction between hillslope and fluvial processes. Where a hillslope transitions from a straight or convex form to a concave one is generally recognized as representing a transition in dominant process. One of the most fundamental ways to express this transition/interaction, and also one of the most fundamental basin characteristics, is drainage density (Tucker and Bras, 1998). By comparing the values for the Upper Colorado, Dirty Devil / Fremont, and North Wash it is apparent that even though the spatial scales of the basins are quite different, the interaction between hillslope and fluvial processes is comparable. Therefore it is reasonable to assume that any difference in the response of the systems to the base level drop of Lake Powell is not on account of differences in the variables that influence drainage density, such as vegetation, precipitation, soil/rock types, relief, and etcetera.

North Wash Gage Interpolation

When the 4 selected gage records of mean annual flow on the Dirty Devil / Fremont were broken into 10 or 20 year intervals, the t-test showed that MAF did not significantly vary through time. Since North Wash is a directly adjacent basin, it is assumed that its MAF has not varied through time either, and that the 1950-1970 record of 1.20 cfs still holds true today (Table 4). çFollow the hyperlink to the excel file containing the gage records, broken down, and the results of the t-test as run using XLStat.

North Wash MAF was also estimated by dividing each gage’s MAF by the USGS drainage area above the gage. This flow per unit area for each gage was then applied to North Wash’s drainage area to generate MAF. Table 5 shows the results.

Table 5
North Wash's MAF = 1.191 cfs from the 1950-1970 gage record
	Most Upstream				3rd
SEVEN MILE CREEK NEAR FISH LAKE, UT				FREMONT RIVER NEAR CAINEVILLE, UT
	Mean Annual Flow	Drainage Area (mi^2) Above the Gage	cfs / mi^2		Mean Annual Flow	Drainage Area (mi^2) Above the Gage	cfs / mi^2
	15.157	24	0.631542		74.281	1208	0.061491
	Applied to North Wash's Drainage Area yields a MAF =	90.20309625	cfs		Applied to North Wash's Drainage Area yields a MAF =	8.782744396	cfs
	2nd				Most Downstream
PINE CREEK NEAR BICKNELL, UTAH				DIRTY DEVIL R AB POISON SP WSH NR HANKSVILLE UT
	Mean Annual Flow	Drainage Area (mi^2) Above the Gage	cfs / mi^2		Mean Annual Flow	Drainage Area (mi^2) Above the Gage	cfs / mi^2
	3.986	104	0.038327		97.174	4159	0.023365
	Applied to North Wash's Drainage Area yields a MAF =	5.474234423	cfs		Applied to North Wash's Drainage Area yields a MAF =	3.337187406	cfs

It can be seen that the most upstream (and farthest from NW) gage provided the worst approximation of the North Wash MAF and the most downstream (and closest to NW) provided the best. As shown in table 4, the gage at Poison Spider Wash also has the longest flow record (from 1949-1993 and 2002-2005), adding robustness to the MAF calculated for NW from the contribution per unit area value. The difference between 1.191 cfs and 3.337 seems small at only 2.15 cfs, but it represents an increase of 35% which is not trivial. However, MAF is not generally regarded as the flow responsible for geomorphic work so it is not necessary at this point to reach an accurate present day value of MAF on North Wash. However, by showing that it is more than an order of magnitude smaller than that of the Dirty Devil and over 3 orders of magnitude smaller than that of the Colorado it again provides a frame of reference in which system responses to base level drop can be compared.

Conclusion

GIS is an extremely powerful tool for organizing and analyzing data sets pertaining to water resource issues. This utility is further enhanced by the initiative of many different groups collaborating to produce comprehensive datasets like NHD and NHDplus, among others. However, this term project stressed the fact that some data errors exist and that adaptations and processing are still necessary (although greatly reduced) to analyze the data. It also stressed that computing power is still a limiting factor on working with a highly detailed representation of a large physical area. At the end of the day though, fundamental and important basin characteristics were calculated in an accurate and easily displayed format which can be packaged and distributed; the very essence of geographic information systems.

References:

Data

NHD

http://nhd.usgs.gov/

NHDplus

http://www.horizon-systems.com/nhdplus/

USGS Gage Records

http://waterdata.usgs.gov/ut/nwis/rt

Publications

Ferrari, Ronald L. 1986. 1986 Lake Powell Survey. Bureau of Reclamation, Denver Office, December 1988.

Webb et al. 2004. Climatic Fluctuations, Drought, and Flow in the Colorado River Basin. USGS Fact Sheet 2004-3062 Version 2, August 2004.

Tucker, Gregory E. and Bras, Rafael L., 1998. Hillslope Processes, Drainage Density, and Landscape Morphology. Water Resources Research, Vol. 34, NO 10, pp. 2751-2764