Abstract:
Rainfall, discharge, traffic, and suspended sediment
were monitored for a period of 4.5 months at three
locations on a secondary haul road at Polk Inlet, Alaska to
determine the important processes and variables involved in
surface road erosion for this area. Three sites all less
than 500m² and within 5 kilometers of each other on the
same road were chosen to be instrumented for monitoring.
The proximity of the sites to each other resulted in the
sections all being nearly identical in age, topographical
location, aspect, elevation, and construction materials.
Also, the sites were subjected to the same traffic amounts
of approximately 3 to 4 loaded logging trucks per day plus
other light vehicles.
Maps were developed of the sites which helped
determine the source areas for each one. The gradients of
sites 2 and 3 were approximately 7%, and the gradient of
site 1 was 10%. Each study site was equipped with a flume,
pressure transducer, datalogger, and pumping sampler to
collect data on discharge and suspended sediment. Sites 1
and 3, had rain gages connected to the dataloggers which
recorded 5 minute rainfall intensities. Hourly suspended
sediment samples were collected at each site. An infrared
traffic counter was used to count the daily traffic amount.
An infiltration rate for the road was determined to be
0.9mm/hr using a simple water balance method and also by
determining the minimum amount of rainfall to initiate
runoff. The infiltration rate was used in development of
representative hydrographs for the three sites.
The runoff response of the sites were very similar
when normalized to an area of 280 m². The precipitation
catches for the two gages were very similar with
precipitation amounts of 893 mm for site 1 and 975 mm for
site 3 during 89 days of record. Several regression
analyses were completed for both hourly and storm data to
determine which variables and technique would be best for
estimating total sediment production. The method that
proved to be the best for determining hourly production was
to multiply the hourly sediment concentration by the
average hourly discharge to obtain a total estimated
sediment weight produced for that hour.
During multiple regression analysis, all three sites
and the combined model had rainfall as the most important
variable. The variable that averaged the number of axles
per day since the last runoff event was also found to be
significant in the combined model. Qualitative variables
were used to determine that timing of the events may have
an influence on the sediment production. The total storm
sediment production was determined by summing the total
hourly sediment weights for a given storm. The regression
analyses found rainfall to be the most significant of ten
variables for the total storm sediment production.
A comparison of all the different models coefficients
was developed. The multiple regression model with total
storm rainfall, a qualitative variable for gradient, and
axles per day was found to have the best coefficient of
determination of 0.66 for the combined data of all the
sites. The model for site 3 of axles per day and total
storm rainfall was found to have the highest coefficient of
determination, R² = 0.85. The simple linear regression
model of log of total sediment yield/km of road to total
storm precipitation was used to estimate the annual
sediment production from a kilometer of road at Polk Inlet.
The annual precipitation data was from a gage located about
16 kilometers northeast of Polk Inlet. The annual road
surface sediment erosion estimate is 8.1 tonnes/km of road.
A comparison of other studies shows this to be similar
to other locations in the United States and areas of New
Zealand. Several assumptions were made and the resulting
limitations are described for this estimate. Any use of
this estimate or equation for sites without very similar
characteristics would not be advised. Future studies are
in progress to expand the understanding of some of the
other variables not accounted for in this study.
