GIS-Based Erosion Modeling Using RUSLE

By Trevor Alsop

Submitted in partial fulfillment of CE 547 Spring 2009 course requirements

 

Introduction

Erosion and sediment mobilization is an important natural process in the development and maintenance of coastal habitats.  However, this natural process is being accelerated as the result of anthropogenic activities, to the detriment of receiving water bodies.  Typically associated with agriculture, soil erosion is increasingly the greater result of non-farming development; urban and suburban.  Erosion from U.S. farming accounts for nearly 3 billion tons of soil annually (NRCS 2003), while it is regarded that erosion due to construction can be 10 to 100 times the rate of loss as that of agriculture.  This is primarily the result of increased stormwater runoff resulting from the creation of more impervious surfaces, like pavement and rooftops, in conjunction with the stripping of naturally vegetated land in preparation for construction projects. 

The Southern Sandoval County Arroyo Flood Control Authority (SSCAFCA) is charged with constructing, operating, and maintaining flood control facilities to prevent flood damage to property.  Another goal of the Authority is to reduce sediment and erosion within the jurisdiction.  This goal is increasingly requiring more oversight, effort, and financial resources from the Authority as the land within the jurisdiction continues to develop at a rapid pace.  An abundance of sediment is available to move through flood control system because of the following reasons:

·         SSCAFCA area mostly undeveloped land

·         Fine soil types

·         Steep slopes

·         Monsoon storms resulting in intense runoff events 

·         Large inactive construction sites

Flood conveyance channels and detention basins receive large quantities of sediment each year, primarily the result of the summer rainy season with sufficiently energetic precipitation to cause erosion.  SSCAFCA is required to remove the deposited sediment at such time the facility no longer functions as designed because the deposited sediment has reduced the hydraulic capacity.  In recent years, SSCAFCA has allocated greater portions of the annual operation and maintenance budget to sediment removal and disposition. 

As a regulatory Authority, SSCAFCA can benefit from erosion modeling to formulate standards and guidelines, as well as prudent budgeting.    The Universal Soil Loss Equation (USLE), later revised (RUSLE), is a relatively simple empirical model developed by the U.S. Department of Agriculture that has remained one of the most practical methods for estimating soil erosion potential and the effects of different management practices for over 40 years (Kinnell, 2000).  RUSLE was chosen to model erosion due to its simplicity, wide acceptance and use, and manageable data requirements.  Also, RUSLE is used for construction site management at the federal level in National Pollutant Discharge Elimination System (NPDES) Phase II permitting (USEPA, 2000), for which SSCAFCA has involvement as a Municipal Separate Storm Sewer System (MS4) permit holder.  Geographic Information Systems (GIS) are a powerful tool for analyzing spatial data and integrate well with erosion models that when combined are helpful for evaluating the impact of land-use practices on soil loss. 

Methods and Procedures

Several data sources were utilized to generate the necessary information required for the RUSLE calculation.  To begin with, a study area in the northeast portion of the jurisdiction was selected as data was recently acquired for purposes of watershed planning:

      

Mosaic Orthoimagery                                                          TIN

 

The raw data was in the form of a triangulated irregular network (TIN) that was converted into a raster file format for analysis with the raster calculator functions.  The converted elevation data was projected into New Mexico State Plane coordinate system to match the orthoimagery and “filled” in ArcInfo.   The elevation raster was then translated into flow direction, accumulation, and slope grids in preparation for the specialized Length/Slope calculation required in RUSLE and for overall equation calculation.

RUSLE is represented by the following equation:

A = R K (LS) C P

Where:

A = average annual soil loss from sheet and rill erosion caused by rainfall and associated overland flow in tons/ac/yr, computed by selecting values for each factor and multiplying them together.

R = the factor for climate effect on erosivity

K = the factor for soil erodibility measured under a standard condition

LS = the combined effect for slope length and slope steepness

C = the factor for cover-management: the effect of vegetation, soil cover, soil biomass, and soil disturbing activities on erosion

P = the factor for support practices that mitigate erosion

The R-factor was provided by New Mexico Agronomy Technical Note No.28 Revised, as published by the local United States Department of Agriculture (USDA) Natural Resources Conservation Service (NRCS) office in Albuquerque.  An R-factor of R=25 was universally applied to the study area.

R-Factor Map of New Mexico

 

A default P-factor of P=1 was used assuming no support practices are applied in the study area to attenuate soil erosion. 


The K-factors were obtained for the study area from the USDA Soil Survey Geographic (SSURGO) database, in which attribute tables were joined to tease out the K factors from the voluminous data contained within the database. 

K-Factor Grid

 

The LS factor was calculated in the raster calculator with the following equation (Wischmeier and Smith, 1978, updated by Moore and Burch, 1986):

LS = (flowaccum grid × cell size∕22.13)0.4 × (sin(slope grid × 0.01745∕0.0896)1.4 × 1.4

LS Factor Grid

 

The C-factor may be the most important factor because its range of possible variation affects computed soil loss more than any other and it is the factor most easily changed through soil management to control erosion (Pierce et al., 1986; Foster, 1982).  Published C-factors are sparse and may not be current, therefore  I made estimates of the C-factor ranging from 0 to 1 according to four classifications I developed specific to this project: urban (0.01, most attenuation), light urban (0.05), undisturbed (0.1), and highly disturbed (1.0, no attenuation).  This approach is very rudimentary and can be expanded upon with a host of subfactors that the NRCS employs to better reflect real-life conditions.  Finally, I created polygons roughly encompassing areas according to the classifications and then created a raster grid. 

C-Factor Grid

 

Results and Conclusion

With all the factors accounted for, the final step was performed in the raster calculator as the product of all the factors applied to each grid cell.  The resulting raster grid shown below indicates areas of heightened potential for erosion corresponding with areas of highly disturbed soils, as to be expected.  These areas should be targeted for management practices seeking to reduce erosion to the maximum extent practicable.  As far as quantified estimates for sediment loss from this area, the model does not appear to give valid results as implemented.  Extracting quantified values from the estimate grid appear to be too large.  The model is being reviewed for correct factor generation, step-wise calculations, and interpretation of results.  In light of the numerical uncertainty, I classified the results qualitatively as indicated below.

Erosion Estimate Grid

 

Future Work

To be of actual benefit to the Authority as a regulatory and budgeting tool, the following future work is identified:

·         The RUSLE equation only estimates the amount of sediment lost from sheet and rill erosion from a landscape profile, which is not the actual sediment yield leaving the sight, nor is it capable of calculating sediment transport within a channel.  The total sediment at the end of a channel reach or deposited in a detention basin is instead the erosion estimate multiplied by a sediment delivery ratio (SDR), which is a power function specific to the watershed.  Development of an appropriate SDR in conjunction with the local NRCS conservationist/agronomist is recommended.

·         Utilize custom ESRI scripts and extensions to automate tasks and improve tool efficiency: promising tools produced by groups from Purdue University and Central Washington State

·         Factor and Subfactor refinement: interface with local NRCS conservationist/agronomist

·         Single storm event modeling vs. annualized estimate: MUSLE vs. RUSLE     (Williams and Berndt, 1977)

 

 

References

Foster, G. 1982. Modeling the soil erosion process. Pp. 297-382 Hydrologic Modeling of Small Watersheds.

Kinnel, P.I.A., 2000, AGNPS-UM: Applying the USLE-M Within the Agricultural Nonpoint Source Pollution Model. Envinronmental Modeling and Software 15:331-341.

Moore, I., and G. Burch. 1986. Physical Basis of the Length-Slope Factor in the Universal Soil Loss Equation. Soil Science Society of America Journal. 50:1294-1298

Natural Resource Conservation Service, 2003. Natural Resources Inventory.

Pierce , F.J., W.E. Larson, and R.H. Dowdy.1986. Field Estimates of C Factors: How good are they and how do they affect calculations of erosion? In Soil Conservation: Assessing the National Resources Inventory, Vol. 2. Washington D.C.: National Academy Press.

U.S. Environmental Protection Agency (USEPA, 2000). Stormwater Phase II Final Rule: Small Construction Site Program Overview. EPA 833-F-00-013. Washington, D.C.

Williams, J.R., and Berndt, H.D., 1977. Sediment yield prediction based on watershed hydrology. Transactions of the ASAE 20 (6), 1100-1104.

Wischmeier, W.H., and D.D. Smith. 1978. Prediction Rainfall Erosion Losses: A Guide To Conservation Planning. USDA Agricultural Handbook No. 537. U.S. Department of Agriculture.