CEE 6440 GIS in Water Resources
Road
Surface Erosion Modeling and Spatial Analysis in a TMDL Watershed
Jeanny
Wang Miles
Term
Project –Final Report
December
3, 2010
Objective
The
objective of this project is to create and compare road erosion indices within
subwatersheds of the Bear Valley Watershed (BVW) in Idaho. Using readily available spatial data I will rank
the relative erosion risk from the roads system and compare them with predicted
sediment values modeled using the Geomorphic Road Analysis and Inventory
Package (GRAIP). It is hoped that this
sort of analysis may reduce the amount of field data needed to predict erosion
risk and may help land managers prioritize road treatment, drainage
improvements or road decommissioning opportunities that could reduce sediment
in downstream waterbodies.
Background
Fine
sediment erosion from unsealed forest roads negatively affect downstream
ecosystems. There are a limited number
of ways to measure the amount of sediment that reaches the stream channel or to
track the actual contribution of sediment erosion to the roads network. There are several models that attempt to
quantify road surface erosion; however, these depend on site specific data that
are often not available to calibrate or validate the model. The Geomorphic Road Assessment and Inventory
Package (GRAIP) is a process and a set of tools for analyzing the impacts of
roads on forested watersheds which combines a detailed road inventory with GIS
analysis tools to predict road sediment production and delivery, mass wasting
risk from gullies and landslides, and road hydrologic connectivity (USFS,
2010).
This model has been jointly developed by the USFS Rocky
Mountain Research Station (RMRS) and Utah State University (USU) to estimate
sediment erosion from roads. However the
process is very field intensive since it relies on an extensive GPS inventory
effort of the entire roads network, drainage features, pebble counts and an
assessment of connectivity to the stream channel. In this project I will determine if GIS
analysis of readily available image layers and other data sources can provide
road erosion metrics that will reduce or supplant the amount of field data
required to estimate road erosion hazard.
I will compare relative rankings from the indices to the GRAIP modeled
sediment estimates for the same subwatersheds.
While
devising this project I learned of a nationwide effort by the US Forest Service
to effort to classify the condition class of all 6th level HUCs on
Forest Service lands. This Watershed
Condition Framework (WCF) also looks to rank the condition class from the
impact of the roads and trails system (among other features) with regard to
four attributes: road density, stream
proximity, mass wasting potential, and maintenance (Pontyondy and Geier, 2010). The WCF methodology provided guidance for quantification
for the road density and stream proximity index. The other indices used in the analysis were
developed specifically for this project in an attempt to add quantifiable metrics
to be used in conjunction with this relative ranking methodology.
Although
using spatial data layers may not be as accurate or reliable as field
inventoried road features, the use of GIS derived road erosion indices may still
provide useful quantitative metrics for land managers to identify problem areas
or assign priority activities within watersheds.
CASE
STUDY - Bear Valley Watershed, Idaho (Boise National Forest)
The
Bear Valley Watershed (BVW) in Idaho provides an opportunity to analyze the
modeled sediment impacts of roads on water quality (Figure 1). BVW is comprised of seven 6th HUC subwatersheds,
several with streams that have been listed as 303(d) impaired for sediment
under the Clean Water Act. A TMDL (total
maximum daily load) study is under development for the BVW and significant work
has been done to map and model the sediment impacts of the roads throughout the
watershed. A significant portion of this
load is attributed to road related sediment delivery (Table 1). A large amount of GRAIP specific data has
been collected in the BVW by the RMRS and Boise National Forest staff. This unique data set could provide validation
of the model and provide useful application in the management decisions derived
from its results. At a minimum it provides
a relative means to compare the road erosion indices that I have developed for
each 6th HUC watershed in the BVW.
Figure 1. Bear
Valley Watershed Roads and Streams
Table
1. Bear Valley AUs recommended for Category 4b-Listing (Fly et al, 2010)
|
Assessment Unit
Name |
Stream Name |
Pollutant |
Length (mi) |
Beneficial Uses |
|
17060205SL012_02a
|
Upper
Bear Valley Creek and tributaries |
Sediment
|
28.86 |
Undesignated
|
|
17060205SL012_05
|
Bear
Valley Creek |
Sediment,
Temp. |
11.24 |
Undesignated
|
|
17060205SL013_03
|
Bearskin
Creek |
Sediment
|
1.83 |
Undesignated
|
|
17060205SL013_04
|
Elk
Creek |
Sediment
|
8.94 |
Undesignated
|
GRAIP
Within the GRAIP model, sediment production for each road
segment is calculated from slope, length, road surface condition and road-side
drain vegetation gathered by a GPS inventory and by overlaying the road path on
a Digital Elevation Model (DEM).
Sediment production is accumulated to drain points by adding the
sediment production from contributing road segments. These drain point sediment
loadings are then used in a DEM weighted flow accumulation function to
calculate sediment load inputs to streams based on inventoried delivery (Prasad,
2007). The GRAIP model is currently
being used by the US Forest Service (FS) in a number of National Forests to estimate
sediment production and delivery from forest roads, as well as sediment
production from landslides and gulleys.
GRAIP has also been used to monitor a number of the millions of dollars
spent on road decommissioning and drainage improvements through the USFS Legacy
Roads and Trails project. GRAIP data has
been collected in a number of watersheds that have 303(d) listed streams for
sediment and has been used as a basis in several TMDL[1]
watershed studies regulated by the EPA under the Clean Water Act.
The primary algorithm used by the GRAIP model in estimating
fine sediment production (E) from each road segment is calculated for each
inventoried road segment based on a base erosion rate and road properties (Luce
and Black, 1999) according to the following equation:
E = B x L x S x R x V
Where B = base erosion rate (kg/m),
L = road length contributing to the drain point (m),
S = slope of road in the travel direction,
R = road surfacing factor, and
V = vegetation cover factor for the flowpath (ditch
drainage).
For
this analysis, a base erosion rate of 210 kg/m/yr of road length per unit of
slope was assumed. This figure is based on BOISED model base rates for road sediment
production where maintained, heavy use roads are present (Reinig et al. 1991).
The base rates were developed from Zena Creek (in the nearby South Fork Salmon
River drainage) in erosion studies conducted by Megahan and Ketcheson, 1996).
Further work could determine if this rate is appropriate for this climate,
geology and road system. (Fly et al, 2010).
However, my project will not go into much detail regarding the other
specific variables that are used to calculate road erosion with the GRAIP
model. In order to streamline or
minimize required site specific and field intensive data collection procedures required
by GRAIP, I sought to develop proxies for the information required to estimate
sediment.
Methods
I
developed indices of road erosion hazard, and used GIS to produce spatial and
tabular representations of the information from available data layers, assigned
index ranking values, and compared these indices with sediment estimates
generated by the GRAIP model.
In order to streamline or minimize
required site specific and field intensive data collection procedures required
by GRAIP, I sought to develop proxies for the information required to estimate
sediment. I used available data layers
from several Forest Service databases, including information on stream
crossings and drainages (culverts and bridges in the INFRA[2]
database); the roads network (motorized vehicle maps, travel use plans,
authorized/unauthorized roads); and data layers of land types. For this watershed there were no records of Damage
Survey Reports (DSRs) of roads damaged by natural disasters or catastrophic
failures as reported by the Emergency Relief for Federally Owned Roads (ERFO)
Program.
The
data sources used in my analysis include:
·
Roads – provided by the Boise National
Forest (NF)
·
30m digital elevation model (DEM)
available at seamless.usgs.gov
·
Stream network provided by the Idaho
Department of Environmental Quality (DEQ)
·
TauDEM generated flow network (used in
the GRAIP model analysis)
·
GRAIP modeled sediment production and
delivery by road segment
·
STRMCROSS – locations of culverts,
bridges and other roads crossings from the USFS
·
LANDTYPES – landtypes and landtype
associations provided by the Boise NF
Indices of Road Failure
The following indices were
examined as indicators of road hazard and erosion risk:
1. Road
Density
2. Stream
Channel Proximity (within 300 foot buffer)
3. Predicted
Stream Crossings (risk of crossing failure)
4. Slope
Class (within 10 percent gradients)
5.
Slope Position (upper 20%, middle 40%, lower 40%)
6. Bedrock
Geology and Geomorphology
7.
Road Maintenance and Best Management Practices (BMPs)
Indices 5 and 7 were not used in
this analysis. Slope position was
considered insignificant since most of the roads within this watershed are
located in the valley bottoms (lower 20%).
Road Maintenance and BMPs, although important in reducing sediment
delivery to streams, were not easily quantifiable and were therefore eliminated
from this analysis as well.
For the remaining indices, I endeavored
to quantitatively address the appropriate questions:
A.
How and where the road system cause
surface erosion?
B.
How and where the road system cause
mass wasting?
C.
How and where does the road system
modify surface or subsurface hydrology?
D.
How and where is the road system
hydrologically connected to stream system?
E.
How and where does road system create
potential for pollutants to enter waters?
GIS
Tools
I used
many different GIS tools to overlay, extract, buffer, intersect, query, and
export the spatial information needed for each of the attributes of the
following road erosion indices. The
primary methods used include:
1. Road
Density - used “clip” to divide
watershed into subwatersheds; used “intersect” tool with a line output to
obtain length of roads with each subwatershed; index in units of miles road / square miles in each subwatershed.
2. Stream
Channel Proximity – used “buffer” tool on DEQ flow network with a total of 600
feet total buffer width; used “dissolve” to create continuous shapefile of
stream buffer; used “identity” and “frequency” to overlay roads layer onto
buffer width for each subwatershed; index in units of mi road within buffer /
total miles roads in percent (%) within each subwatershed.
3. Predicted
Stream Crossings – used “intersect” tool on road layer and DEQ stream layer
with a point output to determine number of road-stream crossings in each
watershed; used “clip” to subdivide the DEQ stream layer into the seven
subwatersheds, and tabulated summary statistics on the total length of streams
within each subwatershed; index in units of number of crossings per stream
mile. (Process also repeated for the
TauDEM generated flowlines, and compared with locations of GRAIP inventoried
stream crossings and for the STRMCROSS layer.)
4. Slope
Class – created shapefile DrainSlope
that links slope with drainpoint to stream connectivity by first converting
values in bvdemSD8 to points using
“Extract Value to Points” in Spatial Analyst; created five slope classes
between 0 to 30% slope; created slope histogram of drainpoints within each
class. Also created a scatter plot of
one type of drainpoint (BroadBasedDips) with hydrologic connectivity (2) with
stream.
5. Bedrock
Geology and Geomorphology – used “intersect” tool to determine miles of road
within each landtype association code with each subwatershed; units in miles
road in each type / total miles of road within each subwatershed.
Relative weighting and ranking of indices
I ranked the index obtained for each of
the above attributes in several ways.
One is based on a whole number value of good (1) /fair (2) /poor (3)
metric which was derived from the WFC guidance documents (Pontyondy and Geier,
2010). The second indices multiply out
the values derived for each attribute.
The “123” index multiplies out the values of the first three indices, and
the “1to4” multiplies out the values of the first four indices. The categorization into the good/fair/poor
categories are based on the following rubric.
1. Road
Density - based on WFC
guidance (Potyondy and Geier, 2010) index evaluated as Good (1: road density is
< 1 mi/mi2); Fair (2: road density is between 1-2.5 mi/mi2);
and Poor (3: road density is > 2.5 mi/mi2).
2. Stream
Channel Proximity – based on WFC guidance (Potyondy and Geier, 2010) index evaluated
as Good (1: <10% roads in buffer); Fair (2: 10-20% roads in buffer); and
Poor (3: >20% roads of within
buffer).
3. Predicted
Stream Crossings – index for the intersect of the DEQ stream and roads
layers evaluated in terms of a simply distributed designation of Good (1: < 0.2 crossings/stream mile); Fair (2: 0.2 – 0.4
crossings/stream mile); Poor (3: > 0.4 crossings/stream mile).
4. Mass
Wasting – % of roads network in “unstable” landuse categories
(C1: fluvial lands with moderate
potential for mass wasting); distributed designation of Good (1: <10%), Fair
(10-20%), and Poor (<20%) percent of the roads network lies in the C1 landuse
type.
I devised the “123” index and the “1to4”
index as a metric that would be more “precise” relative ranking of the
subwatersheds, rather than the whole number assignments used in with the
adapted WFC methodology. The values for
each road erosion index and two index classifications are tabulated by
subwatershed.
RESULTS
1.
Road
Density as an indicator of road hazard
Figure 2. Bear Valley Watershed –Roads by Subwatershed
Table 1. Bear Valley Watershed – Open Road density by
Subwatershed
|
HU_6_NAME |
Road Miles |
Square Mile |
Density (mi/mi2) |
|
|
170602050101 |
Upper Bear Valley |
49.00 |
26.28 |
1.86 |
|
170602050202 |
Bearskin |
26.33 |
17.58 |
1.50 |
|
170602050203 |
Lower Elk |
18.27 |
20.79 |
0.88 |
|
170602050102 |
Cache |
34.37 |
39.97 |
0.86 |
|
170602050103 |
Wyoming |
11.95 |
25.74 |
0.46 |
|
170602050104 |
Fir Creek |
5.51 |
20.18 |
0.27 |
|
170602050201 |
Upper Elk |
2.85 |
40.81 |
0.07 |
|
TOTAL |
Bear Valley Watershed |
148.27 |
191.34 |
0.77 |
Overall,
Bear Valley Watershed has good to fair ratings based on open road density for
its subwatershed, with 5 of 7 watershed have a good rating, with 2 watersheds
(Upper Bear Valley and Bearskin with a fair rating).
1. Stream Channel Proximity as an
indicator of road hazard
Figure 3. Stream Proximity: Road miles within 300 ft of stream by
watershed
Table 2. Stream
Proximity: Roadlength within 300 ft of
stream by watershed
|
Subwatershed |
Meters |
Miles in Buffer |
Total Miles |
% Roads in Buffer |
|
Fir Creek |
6930.6 |
4.3 |
5.1 |
79% |
|
Bearskin |
23080.3 |
14.3 |
14.9 |
56% |
|
Wyoming |
9385.1 |
5.8 |
6.3 |
51% |
|
Cache |
23313.6 |
14.5 |
14.9 |
43% |
|
Upper Bear Valley |
27127.9 |
16.9 |
17.2 |
35% |
|
Lower Elk |
8597.2 |
5.3 |
5.6 |
30% |
|
Upper Elk |
956.7 |
0.6 |
0.9 |
28% |
By this rating, all the
subwatersheds rank as poor condition class, since the roads in this watershed
mostly parallel the stream network within a 300 foot buffer on each side of
stream. Upper Elk has the fewest
percentage (28%) of its roads in the stream corridor while Fir Creek and
Bearskin have the highest at 79% and 56%, respectively.
2. Predicted
Stream Crossings as an indicator of
road hazard (risk of crossing failure)
Figure
4. Generated Stream Crossings Variables
by Subwatershed
Stream
crossing events were generated in several ways and vary based on the streamflow
network used. StreamCross yielded 73 intersected values of the roads clip[3]
and the DEQ streams layer. This includes
only 1st and 2nd order tributaries within each watershed
and a total of 253 miles of streams over the Bear Valley Watershed. Using a TauDEM model generated flowline by
over-estimates of the actual stream network.
However, inventoried stream crossings following the GRAIP procedure
would also yield a large number of stream crossings, due to the definition of the
road and stream drain points for the data dictionary. The GRAIP inventory also yields a high number
of stream crossings (191) and the TauDEM
flowlines yield a comparable number of stream crossing intersects (180 for StreamNetCross). STRM_CRO is a shapefile of stream crossing
locations provided by the Boise NF and was presented for comparison.
Stream
density (stream miles per square mile) is fairly uniform across all subwatersheds
with values of 1.2-1.6 mi/mi2. However, since stream crossings are a major
route of sediment to the stream, the greater the number of stream crossings on
the stream reach, the greater the hazard of sediment input to the stream. I created an index of number of stream
crossings per total stream miles within subwatershed in number per stream mile
(Table 3). Ordering the watershed by
most stream crossings per stream mile, Upper Bear Valley (0.6) and Bearskin
(0.55) again are the highest risk watersheds for sediment input from stream
crossings.
Table 3. Stream Density and
Road Crossings per Stream Mile
|
HU_6_NAME |
Stream Density |
Stream Crossings |
Stream Miles |
# Crossing / Stream Mile |
rank |
|
Upper Bear Valley |
1.2 |
19.0 |
31.5 |
0.60 |
3 |
|
Bearskin |
1.4 |
14.0 |
25.5 |
0.55 |
3 |
|
Cache |
1.2 |
13.0 |
49.1 |
0.26 |
2 |
|
Wyoming |
1.5 |
7.0 |
38.6 |
0.18 |
1 |
|
Fir Creek |
1.2 |
4.0 |
23.4 |
0.17 |
1 |
|
Lower Elk |
1.6 |
3.0 |
33.0 |
0.09 |
1 |
|
Upper Elk |
1.3 |
1.0 |
51.7 |
0.02 |
1 |
|
TOTAL (or average) |
1.3 |
61 |
252.8 |
0.24 |
1.7 |
For
purposes of comparison, I created a stream crossing index of Good (1), Fair (2)
and Poor (3) for less than 0.2 , between 0.2-0.4 and greater than 0.4 crossings
per stream mile, respectively.
3. Slope Position as an
indicator of road hazard (upper 20%, middle 40%, lower 40%)
As the terrain where most of the
roads in this watershed occur is generally flat, this indicator was not seen to
be of significance and was not considered in this analysis.
4. Slope Class as an
indicator of road hazard (slope class within 10 percent gradients)
Slope at specific drain points was
considered to have some influence on road erosion delivery to the stream. Slope values for GRAIP collected drainpoints
were sorted into 10% classes and charted as the slope class against frequency
and cumulative distribution (Figure 5).
54% of road drainpoints had slopes of less than 10% and 3% of road
drainpoints were found on slopes just under 50% (Table 4). Hydrologic connection from one type of road
drainage feature (Broad Based Dips) was charted against slope with little
correlation (Figure 6). More work can be
done to determine the relationship between slope and delivery of sediment to
the stream network.
Table 3. Bear Valley Drain Slope Frequency and
Cumulative Distribution
|
Slope Class |
Frequency |
Cumulative % |
|
0.1 |
249 |
53.6% |
|
0.2 |
94 |
73.8% |
|
0.3 |
6 |
87.5% |
|
0.4 |
44 |
97.0% |
|
0.5 |
14 |
100.0% |
Figure 6. Slope at Hydrologically Connected Broad
Based Dips
5. Bedrock Geology and Geomorphology as an indicators of road
hazard
Using
LandTypes layer acquired from the Boise National Forest and intersected
roadlines over each of the LandTypes (Wendt & Bliss, 1976). Roadlines were categorized into the miles of
roads that intersected each LandTypeAssociation (Figure 5, Table 4).
Figure 7. LandType Associations Along Roads Signifying
Mass Wasting Potential
Table 4. Percent Roads Network within each LandType
Association - by subwatershed
|
|
C1 |
C2 |
D1 |
D2 |
F1 |
F5 |
G1 |
G2 |
|
Bearskin |
24.7% |
12.9% |
10.5% |
24.7% |
1.1% |
|
|
1.0% |
|
|
Fir Creek |
19.4% |
|
2.0% |
|
|
4.5% |
|
74.1% |
|
|
Upper Bear Valley |
19.0% |
49.9% |
0.7% |
20.0% |
|
|
|
2.7% |
|
|
Cache |
17.3% |
19.6% |
47.1% |
|
|
|
0.2% |
2.6% |
|
|
Wyoming |
6.8% |
3.5% |
62.9% |
12.4% |
|
0.4% |
|
|
|
|
Lower Elk |
1.2% |
|
0.3% |
|
|
2.1% |
|
26.8% |
|
|
Upper Elk |
|
31.5% |
18.8% |
15.1% |
|
|
|
|
|
The
main categories of LandTypeAssoc where
Bear Valley roads occur include:
C1:
Fluvial lands which have a moderate to high mass wasting potential due to pockets of heavier than normal
soils formed in remnant lacustrine deposits (Cryic uplands and hillslopes)
C2: Other fluvial lands adjacent to
canyons with live streams, with shallow, parallel 1st and 2nd order drainages,
and with slopes dominantly greater than 60 percent
D1: Volcanic lands which show evidence
of recent geologic activity by the presence of volcanic flows and cinder cones (Morainal
and outwash lands)
D2: Other volcanic flow lands
(basaltic) which have been faulted and tilted from 30 to 50 percent, with scarp
slopes up to 70 percent (alluvial lands from 5,000 to 7,000)
F6: Mass wasting lands
The percent of the roads network
intersecting “unstable” LandTypes or LandTypeAssoc would be an indication of
the potential for erosion risk from mass wasting. Although this has not been confirmed with a
soil stability specialist, as an initial evaluation I characterized C1 and F6
as “unstable.” Soils in the F6 category
do not occur in this watershed.
Therefore the roads miles that occurred within the C1 category were used
for this analysis. In order of the greatest
to least percent of the roads network overlaying “unstable” or the C1 category,
the subwatershed order is: Bearskin
(25%), Upper Bear Valley and Fir Creek (19%), Cache (17%), Wyoming (7%), Lower
Elk (1%), and Upper Elk (0%).
Comparison
of Road Erosion Indices to GRAIP modeled sediment production or delivery.
Table 5 provides a summary of the
quantifiable road erosion indices as well as each modified WCF ranking, average
ranking, the “123” index and the “1to4” index. The “123” index was obtained by
multiplying the actual value of the first three road erosion attributes in each
subwatershed, and the “1to4” index was obtained by multiplying out the values
of the first four road erosion attributes.
Table 5. Summary of Road Erosion Indices by
Subwatershed
|
HU_6_NAME |
1-Road Density |
Rank |
2-Stream Proximity |
Rank |
3-Cross / Str Mi |
Rank |
4-Mass
Waste |
Rank |
WCF Rank |
123 Index |
1to4 Index |
|
Bearskin |
1.50 |
2 |
54% |
3 |
0.55 |
3 |
24.7% |
3 |
2.75 |
44.86% |
11.1% |
|
Upper Bear Valley |
1.86 |
2 |
34% |
3 |
0.60 |
3 |
19.0% |
2 |
2.5 |
38.72% |
7.3% |
|
Lower Elk |
0.88 |
1 |
29% |
3 |
0.26 |
2 |
1.2% |
1 |
1.75 |
6.80% |
0.1% |
|
Cache |
0.86 |
1 |
42% |
3 |
0.18 |
1 |
17.3% |
2 |
1.75 |
6.57% |
1.1% |
|
Wyoming |
0.46 |
1 |
49% |
3 |
0.17 |
1 |
6.8% |
1 |
1.5 |
3.87% |
0.3% |
|
Fir Creek |
0.27 |
1 |
78% |
3 |
0.09 |
1 |
19.4% |
2 |
1.75 |
1.94% |
0.4% |
|
Upper Elk |
0.07 |
1 |
21% |
3 |
0.02 |
1 |
0.0% |
1 |
1.5 |
0.03% |
0.0% |
|
TOTAL/ave |
0.77 |
1.3 |
44% |
3.0 |
0.24 |
1.7 |
13% |
1.7 |
1.93 |
14.7% |
2.9% |
GRAIP
Model Outputs
Estimates of sediment accumulation and
sediment delivery from by the GRAIP model process are shown in Figures 8 and 9,
respectively. Tables 6 and 7 show the
total amounts of sediment delivered and as a fraction of the sediment
accumulated through the model.
Figure 8.
Sediment Accumulation
Table 6.
Sediment Delivered as Fraction of Sediment Produced
|
|
sedDel |
sedProd |
% |
|
Fir Creek |
58,874 |
288,089 |
20.4% |
|
Upper Bear Valley |
388,658 |
3,072,915 |
12.6% |
|
Bearskin |
140,574 |
1,252,226 |
11.2% |
|
Wyoming |
10,500 |
129,630 |
8.1% |
|
Cache |
94,441 |
1,259,637 |
7.5% |
|
Lower Elk |
48,336 |
1,095,730 |
4.4% |
|
Upper Elk |
3,150 |
154,031 |
2.0% |
Table 6. Sediment Delivered
in each watershed (kg/yr)
|
|
sedDel |
|
Upper Bear Valley |
388,658 |
|
Bearskin |
140,574 |
|
Cache |
94,441 |
|
Fir Creek |
58,874 |
|
Lower Elk |
48,336 |
|
Wyoming |
10,500 |
|
Upper Elk |
3,150 |
Figure 9.
Sediment Delivery at Drainpoints
Table
8 provides a summary of the road erosion indices, WCF ranking, “123” index and
GRAIP modeled sediment delivered (sedDel) in kg/yr and as a % of the total
sediment delivered in the Bear Valley Watershed. The maximum or worst values (in terms of road
erosion risk or sediment delivery) obtained in the summary indices are noted in
bold.
Table 8. Comparison of Road Erosion Indices with
GRAIP Predicted Sediment Delivery
|
HUC_6_NAME |
1. Density |
2. Stream Proximity |
3. Crossings / Stream Mile |
WCF Rank |
123 Index |
1to4 Index |
sedDel (kg/yr) |
% Total sedDel |
|
Bearskin |
1.50 |
54% |
0.55 |
2.75 |
44.86% |
11.1% |
140,574 |
18.9% |
|
Upper Bear Valley |
1.86 |
34% |
0.60 |
2.5 |
38.72% |
7.3% |
388,658 |
52.2% |
|
Lower Elk |
0.88 |
29% |
0.26 |
1.75 |
6.80% |
0.1% |
48,336 |
6.5% |
|
Cache |
0.86 |
42% |
0.18 |
1.75 |
6.57% |
1.1% |
94,441 |
12.7% |
|
Wyoming |
0.46 |
49% |
0.17 |
1.5 |
3.87% |
0.3% |
10,500 |
1.4% |
|
Fir Creek |
0.27 |
78% |
0.09 |
1.75 |
1.94% |
0.4% |
58,874 |
7.9% |
|
Upper Elk |
0.07 |
21% |
0.02 |
1.5 |
0.03% |
0.0% |
3,150 |
0.4% |
|
Total / Ave |
0.77 |
44% |
0.24 |
1.93 |
14.7% |
2.9% |
744,532 |
100% |
Discussion
The
road density index provides little applicability for sediment erosion risk
ranking in this watershed with relatively low road densities overall. Stream proximity and stream crossings index would
provide a better indicator since increased road mileage close to the stream and
increased number of stream crossings would signify greater opportunities for
road erosion to enter the stream. The
mass wasting index could be improved by using a landtype maps with a finer
level specification of geologic or geomorphic features confirmed to be
unstable. Slope was not well correlated
to erosion risk potential in this watershed and therefore was not used as a
quantitative factor within the ranking indices.
The summary indices of WCF ranking, “123” and “1to4” index have the same
relative rankings with Bearskin and Upper Bear Valley posing the greatest
erosion risk.
The GRAIP
estimated fine sediment erosion from the BVW roads network could be as high as
745,000 kg per year under current model assumptions. The highest amounts of sediment delivered to
the stream are in the Upper Bear Valley and Bearskin watershed. The other five subwatersheds do not
consistently show the same relative ranking with the modeled sediment
delivered, and need more analysis to be useful.
Therefore there is only marginal consistency in the relative ranking of road
erosion indices with respect to the GRAIP model outputs.
Conclusion
Spatial
data layers can provide a useful metric of relative road erosion risk in a
watershed and subwatershed. Analysis of
available data layers is useful to indicate the locations of certain features (roads
network, stream crossings, drainage features) and to derive slopes and provide
relative indicators of road erosion risk.
These indices have a minor correlation with estimates of sediment
production from the GRAIP model. Spatial
layers may also supplement future modeling efforts to estimate or monitor road
erosion using the GRAIP model. These or
other sorts of risk erosion indices can be developed to provide a quantitative
metric for land managers to identify problem areas and assign priority
activities within watersheds.
References:
Fly, Chase; Grover-Wier, Kari; Thornton, John;
Black, Tom; and Charlie Luce, 2010. Bear Valley Road Inventory (GRAIP) Report -
In Support of the Bear Valley Category 4b Demonstration, RMRS, February, 54 pp.
Wendt, George and Timothy Bliss, 1976. Key to Landtype Associations and Landtypes on
the Boise National Forest, Boise National Forest, February 5, 1976
Luce,
C. and T. A. Black, 1999. "Sediment Production from Forest Roads in
Western Oregon," Water Resources Research, 35(8): 2561-2570.
Megahan,
W.F., and G.L. Ketcheson, 1996.
Predicting Downslope Travel of Granitic Sediments from Forest Roads in
Idaho. Water Resources Bulletin, Vol.
32, No. 2, pp. 371-382.
Prasad, Ajay, 2007. A Tool to Analyze
Environmental Impacts of Roads on Forest Watersheds, MS Thesis, Civil and
Environmental Engineering, Utah State University.
Pontyondy,
John and Theodore Geier, 2010. Forest Service Watershed Condition
Classification Technical Guide , November 6, 2010; 72 pp.
Reinig,
L., Beveridge, R.L., Potyondy, J.P., and F.M. Hernandez. 1991. BOISED User’s
Guide and Program Documentation. USDA Forest Service, Boise National Forest. V.
3.01.
US
Forest Service, 2010. Forest Service Watershed Condition Framework –
Implementation Guide, October 26, 2010, 31 pp.
US Forest Service - Rocky
Mountain Research Station (USFS) 2010. GRAIP - Geomorphic Road Analysis and
Inventory Package, GRAIP - Geomorphic Road Analysis and Inventory Package;
available at: http://www.fs.fed.us/GRAIP/index.shtml
Washington
Forest Practices Board, (1995), "Standard Methodology for Conducting
Watershed Analysis," Version 3.0, Washington State Department of Natural
Resources, Olympia, Washington, p.B1-B52, November 1995.