One of the key factors that drives crop yields is the availability of key nutrients in the soil. Erosion plays a large factor in the absolute amount of those nutrients available. Unfortunately, most farmers underestimate how much erosion is actually occurring in their fields. In some areas (like the Upstate of South Carolina, where we are based), hilly topography means that erosion and nutrient loss are not uniform. Variations in slope will result in areas with higher erosion potential and more nutrient loss due to erosion every year. By using drones and a bit of detailed analysis, we can:
Identify the areas with the highest erosion potential and focus our soil testing efforts there.
Estimate nutrient losses to predict additional fertilization needs.
Vary our rates of fertilization to apply more to the areas that need it the most.
Adjust crop cover and management practices to reduce erosion in those areas.
Before we can discuss the details of how to complete this analysis, we need to understand the model we will use. The RUSLE model is a widely accepted tool for estimating soil erosion. It incorporates multiple factors: rainfall erosivity (R), soil erodibility (K), topographic factors (LS), vegetation cover (C), and conservation practices (P).
Rainfall erosivity measures the amount and energy of rainfall events in an area. It can be estimated based on historical data and varies by location.
Soil erodibility describes the physical characteristics of a type of soil (clay, sand, loam, etc.) and how easily it erodes due to rain.
Topographic factors describe the steepness of the terrain. Water will flow more quickly down a steeper slope, carrying more energy and removing more soil.
cover and conservation practices are aspects of the model under the direct control of the farmer and include things like the type of crop grown and what type of tillage is used. Each of these factors plays a vital role in determining the overall potential for soil erosion in a given area.
The RUSLE calculator is available here: https://fargo.nserl.purdue.edu/rusle2_dataweb/RUSLE2_Index.htm
This guide will discuss the processing and analysis steps WE do after we collect images with a drone in the field. The flight planning and image collection steps will be discussed in detail in a later article. This is a general guide to help farmers and drone service providers who want to explore ways that drones can integrate into farming operations.
Fly a drone and collect images:
Identify Key Factors for RUSLE:
While in the field, it is important to identify the key information we need for the RUSLE calculations.
The first step is to identify the crop and the base management strategy the farmer is using. This can be a quick discussion to determine the crop type, planting density, and what sort of tillage they use. An example option from the RUSLE tool is corn, no-till, 150 bushel/acre.
The next step is to identify the soil type by observing or sampling areas throughout the field. If the client has a recent soil test, we can use the organic matter content to refine the results. An example of soil would be clay with a low-to-medium organic matter content. Soil maps are also available online. This link is for a soil type data set from ArcGIS https://www.arcgis.com/home/item.html?id=ac1bc7c30bd4455e85f01fc51055e586
Lastly, determine any additional supporting practices used by observing the field or discussing them with the farmer. Examples include counturing on slopes, vegitative buffers like fescue buffer strips in the middle of a slope, or terracing.
Create a Point Cloud: A point cloud is a 3D model that represents the world as a cloud of points based on data from LiDAR or images. We use Pix4Dmatic to create our point cloud and Pix4Dsurvey to turn that into a usable digital elevation model (DEM) for analysis. To create the point cloud, we follow the following steps:
Open Pix4Dmatic, select your images from the file explorer, and drag them into the application.
Name the project and click Start.
Click on the processing tab. Here we select the checkboxes for calibration, dense point cloud, and orthomosaic. The mesh output is useful, but we don't need it for this analysis. Processing time for a point cloud and orthomosaic scales with the number of images, so it can take a while for larger projects.
Save the file before you exit.
Classify Point Cloud and Create DEM: If we used the DEM output from Pix4Dmatic, it would include trees and buildings as part of the terrain. We don't want our erosion values to include the slope of trees, so we need to classify our points before we create our digital elevation model. We use Pix4D Survey for this.
On the file tab for Pix4Dmatic, we see an option to open Pix4Dsurvey. If we click this, our project will launch in the Pix4D survey.
Click the process tab and simply run through the steps from top to bottom.
The distant outlier filter filters out points that are far away from any other point and deletes them.
Terrain filter: classifies points into terrain and non-terrestrial. This separates objects like buildings and trees from the ground surface.
Grid of points: Creates a regular grid that measures the elevation of different points. Think of this as a surveyor marking out locations and elevations across the property.
TIN: This step uses that terrain grid of points to make your digital elevation model. This is our key output from this part of the process. We did the previous steps to eliminate trees and buildings to get this model of the surface of the ground.
We then click on the Export tab, make sure TIN is selected, and choose GeoTIFF as our file type (the default is LandXML, so we have to change this in the drop-down menu).
Analyze in ARCGIS: We're now going to use our outputs from the previous two steps to do our RUSLE analysis in ArcGIS Pro.
Find your Orthomosaic and TIN files in the folders they were exported to. The default is a folder in documents called Pix4Dmatic and Pix4Dsurvey, respectively.
Drag them individually into the ArcGIS window and let them process.
We should see those files in the contents pane on the left, and our map should have zoomed to show those layers.
Run a RUSLE script in ARCGIS. We downloaded that script here: https://www.arcgis.com/home/item.html?id=d9cf2bdcb64e47d39df8410cb6814d20
To use this script, we save the file to our desktop, then go to the insert tab, click on the toolbox, click add toolbox, and then select the file from our desktop, "LSP Soil Erosion Tools V1.tbx."
The tool will now be in our toolbox. If you can't find it, just type RUSLE in the search bar.
The inputs for this tool are R, K, C, and P.
R- rainfall value can be looked up on this site at address https://lew.epa.gov/
K: Use the soil tool in the RUSLE2 app to find the K value for each soil type or look up the soil erosivity for your location from a national data set like the one mentioned earlier in the artlicle.
For C and P, we generally set a constant value of.1 because we are not trying to get accurate values on this step.
Click Run, and then identify areas with a high soil loss value.
Quantify Erosion in Pix4Dsurvey: Measure slopes and distances in high-risk areas.
From here, I like to go back to Pix4D Survey and take measurements in those high-risk areas. I do this by drawing a polyline down the slope. The measurements will show the length and slope of that line. This is the key input for our RUSLE calculator.
Take several measurements in those areas, including measurements across the entire slope and measurements of the steepest points of the slope. Our example shows terracing down the hill. The faces of those terraces will erode significantly and deposit soil into the flat portions
Open the RUSLE2 calculator and input values: Calculate the estimated soil loss using observed conditions.
Use The length and slope of your measurements
Choose the appropriate location, soil type, and management (crop and tillage practices).
You can use the rule of thumb (soil loss in tons/acre)*(nutrient in lbs/1000 tons) to estimate pounds of nutrient lost per year. You need a previous soil test to determine the nutrient density of the soil.
Soil Fertility and Nutrient Management Strategies
The primary reason we want to do this sort of analysis is to identify areas that will need more attention. Fertilization decisions should be made based on soil testing and consultation with agronomists and soil scientists, but an erosion and nutrient loss analysis can guide those efforts. We use this data to identify zones to do additional soil testing throughout the year to monitor nutri
ent losses. In a multiyear crop monitoring plan, we will see correlations between erosion zones and our plant health analyses later in the season. These can drive management decisions like sidedressing or foliar fertilization. We can also use our spraying drones to apply additional fertilization in those areas if needed.