Raster Analysis

Abstract

Soil erosion can lead to reductions in agricultural productivity as well as increases in surface water pollution. Modeling of soil erosion rates is conducted at various scales and applications to inform farming and other land management practices. The Revised Universal Soil Loss Equation model has been widely implemented in GIS applications of erosion processes. The Tillamook Bay area on the Oregon coast experiences frequent flooding and high sediment loading into the bay, which can result in water quality issues from high levels of bacteria. This project applies the RUSLE model to estimate annual soil loss in the Tillamook Bay area. Rasters representing the factors of rainfall erosivity, soil erodibility, topographic length-slope, land cover, and prevention measures were generated and used to calculate the average annual soil loss for the Tillamook bay watershed as well as the watersheds of the five rivers that drain into it. Soil loss values were high across the study area, likely due to high precipitation, steep slopes, and the maximum value set for the lack of prevention measures. The area of each river watershed was proportional to the total soil loss per watershed. Geospatial erosion modeling with the RUSLE method is approachable and versatile, but can be enhanced with more detailed information about the study area. In Tillamook Bay, further modeling of each individual river watershed could assist land managers in preventative practices to limit soil erosion.

Introduction

Soil erosion, while a natural process, can be affected by land use changes and human activities. It can lead to loss of productivity in agricultural lands, as well as an increase of pollutant loading into water bodies. This has implications for the sustainability of land management practices in terms of both food and water security. Numerous mathematical models have been developed to predict soil erosion rates in different areas of the world, with model selection based on varying spatial and temporal scales as well as the governing statistical or empirical equations used in the predictions. While many such models are highly complex or site specific, the use of GIS has led to efficient and widely applicable tools for estimating erosion (Mitasova et al. 2013).

One of the earliest models that has been implemented in GIS is the Universal Soil Loss Equation (USLE), which was developed by the U.S. Department of Agriculture and published in 1965. Field and plot studies were used to quantify a relationship that was not limited to a particular geography between average annual soil loss by overland flow with the factors of rainfall erosivity, soil erodibility, topographic length-slope, land cover, and prevention measures. This soil loss by overland flow is distinct from the measure of sediment yield, which would incorporate all erosion from channels, slopes, and other processes (Renard, 1997). After years of additional data collection as well as advances in computer technology, the Revised Universal Soil Loss Equation (RUSLE) was presented in 1992. The RUSLE model utilizes the same factors to estimate average annual soil loss, but modifications were made to how each factor is calculated as well as the availability of data it can draw from (Renard et al. 1991).

The Tillamook Bay area on the Oregon coast is characterized by high annual precipitation, steep slopes in the forested uplands that are used for timber production, and dairy farms and other agricultural practices in the lowlands around the city of Tillamook. High soil erosion rates can disrupt the local economy, which relies on timber, dairy farms, fishing, and coastal tourism. Five rivers, the Miami, Kilchis, Wilson, Trask, and Tillamook, drain into the bay (Figure 1). Frequent flooding and high sediment loading into Tillamook Bay can result in high levels of fecal coliform bacteria, which disrupts oyster harvesting as well as the recreational value of the water (Sullivan et al. 2005).

This study aims to develop a raster-based RUSLE model to estimate annual soil loss in the Tillamook Bay watershed to better inform local land management practices.

Methods

Preprocessing (Figure 2)

The RUSLE model estimates average annual soil loss, A, (metric tons/ha·yr) by multiplying five factors according to the following equation.

A = R * K * LS * C * P

• The Rainfall Erosivity factor, R, is a measure of how the kinetic energy of rainfall and runoff can erode soils (MJ·mm/ha·hr·yr).

• The Soil Erodibility factor, K, represents the physical properties of soils that govern their vulnerability to erosion (metric tons·hr/MJ·mm).

• The Topographic Length-Slope factor, LS, is a dimensionless estimate of the effect of slope steepness and length has on erosion.

• The Land Cover factor, C, is a dimensionless representation of how denser vegetative cover results in less soil erosion.

• The Prevention Measures factor, P, is a dimensionless factor that accounts for how agricultural practices affect erosion rates (Renard, 1997).

All data sets used in this analysis were projected into the NAD 1983 (2011) Oregon Statewide Lambert coordinate system in meters. Hydrologic Unit Code 10-digit watersheds for the five rivers that drain into Tillamook Bay were downloaded from the National Hydrography Dataset and were merged to create the overall area of analysis (Figure 1). Mean annual precipitation values in millimeters from the National Centers for Environmental Information were taken from the eight nearest monitoring stations and were interpolated across the study area using inverse distance weighting to create a 10-meter raster. Soil erodibility values were found from the National Resources Conservation Service, resampled into a 10-meter raster, and were converted into metric units (Figure 4). A 10-meter digital elevation model (DEM) was acquired from the U.S. Geological Survey and was extracted to the study area. Land cover classes across the study area were found from the Multi-Resolution Land Characteristics Consortium. In the absence of detailed knowledge about local farming practices, a maximum value of one was set for the P-factor for the entire study area. This assumes that no erosion prevention measures are in place.

Methods 

LS-factor Calculation (Figure 3)

Sinks present in the DEM were filled in so as to designate the D8 flow directions to the steepest downslope cell (Figure 5A). From this, a raster of the flow accumulated into each cell was calculated. Outlets were manually placed near where each of the five rivers drain into the Tillamook Bay area and were snapped to the flow accumulation raster in order to delineate the five upstream watersheds as rasters for later calculations. A raster of the calculated slope for each cell was also derived from the DEM (Figure 5B). Finally, the LS-factor was calculated from the flow accumulation and slope rasters by the following equation from Moore & Wilson (1992).

LS = (A F * cell size / 22.13) 0.4  * (sinβ * 0.01745 / 0.0896) 1.3 

The variable AF  is the flow accumulation, the cell size is 10 meters, and β is the slope in degrees (Figure 5C). The exponent values were chosen as the typical values found by Moore & Wilson (1992).

Methods

Annual Soil Loss Calculation (Figure 6)

The relationship between precipitation and rainfall erosivity can vary with climate. Rainfall Erosivity, R, was calculated with the equation below from Renard & Freimund (1994) for areas where annual precipitation, P, is greater than 850 millimeters (Figures 7A & 7B).

R = 587.8 - 1.219*P + 0.004105*P 2 

Corresponding C-factors for each land cover class from the National Land Cover Database were derived from the literature by Woznicki et al. (2020) and are presented in Table 1. The land cover raster (Figure 8A) was reclassified based on these values (Figure 8B). Finally, the average annual soil loss for the Tillamook Bay watershed was calculated by multiplying each of the five factors together (Figure 10).

Methods

Annual Soil Loss by Watershed (Figure 9)

The watershed raster that was created during the calculation of the LS-factor was used to generate rasters of each of the five watersheds upstream of Tillamook Bay. Each of these was used to extract the annual soil loss values for each watershed. The pixel type of each raster was converted in order for attribute tables to be built for summary statistics. For visualization purposes only, symbology values for the annual soil loss per watershed were scaled by the maximum value found in each watershed (Figure 11). The percentage of the total area as well as the percentage of the total annual soil loss were also calculated for each watershed (Table 2).

Discussion

The range of values for the annual soil loss in Figure 10 is pronounced. However, approximately 99% of the cells have a value less than 5,000 metric tons/ha·yr with a comparatively small number of outlier cells above this value. One component that could have led to a high estimate of erosion rates is the use of the maximum P-factor across the study area. Further refinement of the model could include local knowledge of preventative farming practices such as contouring, cover cropping, or strip cropping. Overall, the calculated soil erosion rates seem to follow the expected trend of higher values in the steep uplands to the east and lower values in the flatter areas towards the coast.

The Oregon coast receives very high amounts of precipitation in a typical year, with some areas receiving 100 in/yr. This led to high rainfall erosivity values across the entire bay area. However, taking into account the intensity of rainfall, instead of the annual average, could potentially provide more accuracy. The soil erodibility values also seem to follow the expected trend, with soil being more susceptible to erosion in the uplands and less susceptible in the floodplains towards the bay. The calculated LS-factor seems to align with the physical landscape, in that steep slopes in the uplands leads to channelization of water flow that conveys downslope to the coast. The land cover factor intuitively reverses the general trend, in that the forested uplands are less susceptible to erosion than the pastures and agricultural lands around the city of Tillamook near the coast.

When comparing the river watersheds in Figure 11 and Table 2, the percent of the total area of each watershed is relatively proportional to the percent of total soil loss. The larger Wilson and Trask watersheds have a wider range of soil loss values from the uplands down through the developed areas. Sullivan et al. (2005) also found that the Wilson and Trask watersheds had the highest concentrations of total suspended solids as well as total phosphorus from samples collected at the downstream end of each of the five rivers. The less developed and smaller Kilchis, Trask, and Miami watersheds have a narrower range and overall less total soil loss. In order to better trace the fate and transport of sediments and pollutants into the bay, it would be important to better understand how much soil loss each river watershed contributes. Further analysis could include modeling each river watershed individually, so as to better ascertain how the five factors vary as well as their significance in each watershed.

Conclusion

GIS-based erosion modeling seems to be adaptable at the watershed scale across a range of agricultural and environmental applications. For the RUSLE model, several methods were available to calculate each factor, depending on data availability. However, localized knowledge of the study area as well as ground truthed data could likely provide further predictive power to the model. In Tillamook Bay, RUSLE modeling could aid in highlighting areas that need to enhance erosion control practices such as replanting areas after timber harvesting as well as limiting overgrazing of farmlands.

References

Mitasova, H., Barton, C. M., Ullah, I., Hofierka, J., & Harmon, R. (2013). GIS-based soil erosion modeling. In Treatise on Geomorphology (Vol. 3, pp. 228–258). https://doi.org/10.1016/B978-0-12-374739-6.00052-X

Moore, I. D., & Wilson, J. (1992). Length-slope factors for the revised universal soil loss equation: Simplified method of estimation. Journal of Soil & Water Conservation, 47, 423–428.

Renard, K. G. (1997). Predicting Soil Erosion by Water: A Guide to Conservation Planning with the Revised Universal Soil Loss Equation (RUSLE). U.S. Department of Agriculture, Agricultural Research Service.

Renard, K. G., Foster, G. R., Weesies, G. A., & Porter, J. P. (1991). RUSLE: Revised universal soil loss equation. Journal of Soil and Water Conservation, 46(1), 30–33.

Renard, K. G., & Freimund, J. R. (1994). Using monthly precipitation data to estimate the R-factor in the revised USLE. Journal of Hydrology, 157(1), 287–306. https://doi.org/10.1016/0022-1694(94)90110-4

Sullivan, T. J., Snyder, K. U., Gilbert, E., Bischoff, J. M., Wustenberg, M., Moore, J., & Moore, D. (2005). Assessment of Water Quality in Association with Land use in the Tillamook Bay Watershed, Oregon, USA. Water, Air, and Soil Pollution, 161(1), 3–23. https://doi.org/10.1007/s11270-005-2443-7

Woznicki, S. A., Cada, P., Wickham, J., Schmidt, M., Baynes, J., Mehaffey, M., & Neale, A. (2020). Sediment retention by natural landscapes in the conterminous United States. Science of The Total Environment, 745, 140972. https://doi.org/10.1016/j.scitotenv.2020.140972

Data Sources

Dewitz, J. (2023). National Land Cover Database 2021 (CONUS) [Data set]. Multi-Resolution Land Characteristics Consortium.                        https://www.mrlc.gov/data/nlcd-2021-land-cover-conus

National Centers for Environmental Information. (n.d.). U.S. Climate Normals Data (2006 – 2020) [Data set]. National Oceanic and Atmospheric Administration. https://www.ncei.noaa.gov/maps/normals/

Natural Resources Conservation Service. (2023). USA SSURGO – Erodibility Factor [Image service]. Esri inc. https://landscape11.arcgis.com/arcgis/rest/services/USA_Soils_Erodibility_Factor/ImageServer

United States Geological Survey. (2023). National Hydrography Dataset Best Resolution (NHD) – Oregon [Data set]. https://www.sciencebase.gov/catalog/item/61f8b8e9d34e622189c32935

United States Geological Survey. (2024). USGS 1/3 Arc Second n46w124 20240124 [Data set]. National Map Viewer. https://www.sciencebase.gov/catalog/item/65b1f900d34e36a39045208c