GIS in Water Resources
Final Paper
Jeff Jensen
This paper contains information concerning the final project that was completed during fall of 2003 for CEE 5460 GIS in Water Resources. This part of the project deals mainly with the LS and SDR factors of the Modified Universal Soil Loss Equation (MULSE). However this project was completed in conjunction with the group of the Dirty Dodgers namely, Iosefa Matagi, Ryan Mcbride, Brad Taylor and myself. This project was completed as kind of a preliminary project to our final senior design project which will be completed in the Spring of 2004 at the end of our undergraduate careers.
The goal of this exercise was to complete in the most
general terms a sediment yield analysis for
There are some certain terms that should be noted when reviewing this material. An event refers to a flood of some time or a period of time that erosion occurs due to the presence of water.
Soil Loss Equations
There are several different methods for calculating soil loss including but not limited to Pacific Southwest Interagency Committee (PSIAC), the Universal Soil Loss Equation (ULSE), the Revised Universal Soil Loss Equation (RULSE), and the Modified Universal Soil Loss Equation (MULSE).
The
PSIAC method estimates total annual sediment yield,
not just sheet and rill erosion. The method is intended primarily for planning
purposes and results in a range of expected yield values. The procedure
considers nine factors that depend on surface geology, soils, climate, runoff,
topography, ground cover, land use, channel erosion, and upland erosion. The
procedure was developed for watersheds in the western
The
Universal Soil Loss Equation (USLE) can be used to estimate soil loss with
emphasis on sheet and rill erosion. It does not take sediment deposition into
account (Ouyang, 1997). USLE is perhaps the most widely used method for estimating
soil erosion. Although the equation was originally developed for small
agricultural areas, its use has been extended to basins with other land uses.
Average annual sheet and rill erosion is computed as the product of a rainfall
erosion index (R), a soil erodability factor (K), a slope length and steepness
factor (LS), a vegetative cover factor (C), and an erosion control practice
factor (P) (Forman, 2000)
The RUSLE method was developed to update and extend the USLE for non-agricultural applications and to incorporate additional data collected after development of the original USLE (Forman, 2000).
The MUSLE method modified the R in the USLE in order to predict
erosion from a single storm event whether it be rain or snow induced. The
modified R represents the cumulative effect of runoff volume and peak discharge
for a single storm event (Forman, 2000)
The Modified
Universal Soil Loss Equation (MULSE)
The MUSLE equation is as follows:
Ys=α(Q × q)β LS × K × CP × SDR
Where: Ys is the Sediment Yield for a particular event.
Q is the total volume of storm runoff in acre-ft.
q is the maximum flow in cfs.
LS is the Length-Slope factor.
K is the soil erodability factor.
CP is the Cover and Land Use factor.
SDR is the Sediment Delivery Ratio.
For a Rainfall event α = 95 and β = .56.
For a Snowmelt event α =120 and β = .3.
(Rahmeyer, 2003)
Data and Data Sources
The data required to perform this project using ArcGIS consisted of a variety of types of data. The first set of data that was obtained was the DEM (Digital Elevation Model) or the elevations for the watershed and there corresponding coordinates in order to create or represent the canyon in ArcGIS. This data was obtained from the website www.seamless.us.gov . The next bit of required data was a map of the water bodies and streams within the watershed that was obtained from the National Hydrography dataset at www.usgs.water.gov. Both these datasets where downloaded for free. Iosefa also downloaded a set of Land Cover data and Ryan downloaded time series values for stream flow from a USGS gauge located at the mouth of the canyon.
Data Manipulation and
Preprocessing
This step was performed by all of the members in our group at one time or another due to some amount of confusion and the difficulty of transferring ArcGIS files from one computer to another because of the data source requirements.
The first step in processing the data was to open ArcToolbox and project the data using the projection wizard function. The data was projected using a 15m grid cell size into Zone 12 and using a UTM coordinate system. The DEM was saved as the file ned and the National Hydrography Dataset was saved as nhd. ArcMap was then opened and the data was added to a new map.
Using the tools from the Spatial Analysis toolbar the data was then processed using the terrain preprocessing steps starting from the top. These include DEM reconditioning, Filling of sinks, Flow Direction, Stream Definition, Stream Segmentation, Catchment Grid Delineation, Catchment Polygon Processing, Drainage Line Processing, Adjoint Catchment Processing, and Drainage Point Processing.
Iosefa was able to delineate the watershed even further using TauDEM and make an outline for the entire watershed that was similar to a cookie cutter. This gave an updated watershed model that was then used as the base model for calculations in the MUSLE.
The way that we decided to use the ArcGIS program was to take the catchments that we had obtained from the watershed processing and use the areas taken from the ArcGIS to be able to calculate a sediment yield for the entire watershed.
Thus the catchment became the most important feature in ArcMap for use on this project.
Hydrology and the
Calculation Q, q
This part of the project was done by Ryan McBride and involved the review of flow records and the calculation of storm runoff using basic hydrology concepts.
The Q factor was obtained by assuming a rainfall event of 1 inch and a runoff duration of 6 hours by using the equation:
Q=CIA
Where: Q is the rainfall runoff in acre-ft.
C is a number based on the type of land use ranging from 1 to 0
I is the rainfall intensity in inches/hr
A is the area of the catchment in acres.
The variable little q was then found by converting Q to a flow rate over the time period of 6 hrs and multiplying in by a factor of 1.5 to account for a peaking in runoff.
For the snowmelt flood q was generated from a USGS gauging
station in the mouth of the canyon. A
maximum probable flood was assumed using the data from 1983 (the year that
For more information on the Calculation of Q and q see Ryan McBride’s webpage
Calculation of the LS
factor
The LS factor applies the effects of slope and length of slope to the MUSLE equation. The amount of sediment yield is directly proportional to the LS factor the longer the slope the more opportunity the water has to remove sediment from a particular area.
The equation for the Calculation of LS is as follows:
LS = (l/72.6)m(430sin2θ +30sin θ+0.43)/6.613
Where: l is the length of the slope of the area.
θ is the slope of the area
.3 for slopes < 3%
m is slope dependant = .4 for slopes = 4%
.5 for slopes > 5%
(Rahmeyer, 2003)
All of this data was obtained using various tools in ArcGIS after the watershed processing had been done.
The first thing that was necessary to compute LS was the length of slope. On the tools toolbar in ArcMap there is a measuring tool that will measure horizontal distances in both strait lines and segmented lines. This tool was used to measure from the outlet of the watershed to the top of the watershed. This measuring is somewhat subjunctive on my part in theory the LS is based on a perfectly square area but that was not always and in fact never possible when working with catchments so I had to choose an appropriate length that I thought was representative of the entire catchment and used it to represent the average length of the catchment. This was done by simply looking at the watershed and dragging the cursor to a point that I felt was representative of the area.
These measurements where recorded and placed in a spreadsheet in order to be converted to first feet from meters and then length along the slope.
The next data requirement was to obtain the slope of the catchments. This was done using the 3-D analyst toolbar and the function Surface Analysis, Slope Function. This calculated the slope of every cell in the watershed using the ned layer as a reference and generated a slope map covering the entire watershed.
This is a mountainous canyon with steep sides so the slopes were relatively high throughout most of the watershed. The average slope for each catchment was then found using the Zonal Statistics tools also a part of the 3-D Analyst tools. The average values for the slopes were then exported to the spreadsheet containing the horizontal lengths for each catchment. The graph below comes from ArcMap and is a map of the mean slope of the catchments.
The length of the slope was then determined using the following equation that is based on simple geometry.
l = lh/
Where: l is the length of the slope.
lh is the horizontal length (found using the measuring tool).
θ is the average slope of the catchment.
Using all this information the LS factor was then found for each catchment using a spreadsheet function. A Table of values calculated for the LS Factor and also the minimum, maximum, mean and standard deviation can be found in appendix A.
Calculation of the K
factor
The K factor deals with the soils properties of the catchment area and represents a number of different factors including porosity and percentage of fines. Data for soils could not be found so it had to be collected using field work and lab analysis. Fifteen soils samples were taken throughout the watershed so there was not a soil sample for each of the catchments. Therefore the K values for the catchments with no soil samples were interpolated to match those around them.
For more information about the calculation of the K factor see Brad Taylor’s webpage.
Calculation of the CP
factor
The CP factor accounts for the basic cover and land use practices associated with the land area in question. The higher the CP factor the less ground cover and higher impact land use is present so the greater the potential for erosion.
Each catchment was assigned a CP value based on the type of land cover. The canyon in question has two to three distinct vegetation areas. On the south facing slope it is covered by small scrub-type trees and grass. The other side of the canyon is what amounts to an evergreen forest while at the far east end of the canyon as the main branch turns south it becomes more of a deciduous forest.
Land use is relatively high in the canyon as it is close to a populated area and there is a road that can is navigable by street vehicles that do not possess 4-wheel drive capabilities. There are also a number of campgrounds within the watershed itself as well as an US Forest Service station.
A number of assumptions were made in order to simplify the calculation of the CP factor mainly that the land cover and use that was in the majority in the catchment was taken as the CP for the entire catchment.
For more information about the calculation of the CP factor see Iosefa Matagi’s webpage.
SDR
The sediment delivery ratio is the factor that accounts for the amount of the sediment that is actually removed from the watershed in question.
Due to the fact that the attempt of this project was to calculate the amount of sediment that would leave the canyon due to a particular event the sediment delivery ratio was assumed to be equal to one. Or in other words all the sediment that was eroded or dislodged was removed from the watershed and deposited elsewhere. This assumption may or may not be true but was assumed for calculation purposes.
Final Data Processing
Once all these factors were accounted for the data was entered into spreadsheet and the sediment yield for each delineated catchment was found and then simply summed to come up with a total for the watershed. A total yield was found for both a snowmelt and rain induced floods. The spreadsheet can be found in Appendix ^(*)*)%&*).
Below is a map of both the sediment yield of a snowmelt flood and that of a rainfall event, the darker the area the higher the yield.
Conclusions
The preceding pictures although they may not have exactly correct figures as to the total amount of sediment removed from the canyon to however coincide with the MUSLE equation. The areas that are heavily forested and where runoff is low do to a lower K value have less yield than to the areas on the north slope that are more heavily used and have less vegetation. It should also be noted that sediment yield is largely lower in the top of the canyon due to the fact that slopes are generally less severe.
The difference between the two floods should also be noted the snowmelt flood yields have a progressively higher trend as the catchments move west implying that as the runoff from the snow moves down the canyon it accumulates and contributes to higher sediment yields. The rainfall yields are highest on the areas where the Area is the most significant, contributing to a high LS, and the also where the other factors such as the CP and the K have more effect on the yield.
Works Cited:
Forman, Selena M., Martin J. Teal, David T. Williams, Leo R. Kreymborg, and Craig
M. Burnett. Use of GIS, Geobased
Programs and Computer Models for
Watershed and Site Analysis-Part
2. Erosion Control. September/October
2000. http://www.forester.net/ec_0009_gis.html.
Ouyang, Da and Jon
Bartholic, Institute of Water Research,
Rahmeyer, William. Sedimentation Engineering Class Notes, Volume 2. CEE 5470/6470.
Spring 2003.
Appendix A
Spreadsheet for the Calculation of Sediment Yield
