Using LIDAR to Characterize Soil Roughness for Physical Weed Control Research
Daniel Priddy1 and Daniel Brainard2
1Graduate Research Assistant, and 2Professor, Michigan State University
Take-home points:
- Measurements of soil surface roughness are useful for answering many practical questions related to physical weed control including:
- How do soil surface conditions affect tool efficacy and optimal tool choice?
- How do adjustments in tool settings affect soil movement under various conditions?
- LIDAR scanners are becoming more affordable and precise, and can efficiently scan soils before and after cultivation events to provide insights into these questions.
Problem
Although it is common knowledge that soil surface roughness impacts the efficacy of tools, most published studies of tool efficacy (including our own) do not provide detailed information on what soil surface conditions were like at the time of cultivation. Moreover, most studies also do not adequately describe how the tools themselves move soils, and how that soil movement is influenced by tool settings. Without such knowledge it is difficult to understand the mechanisms responsible for results and to rationally calibrate tools based on that information.
Several previous studies have used LIDAR to characterize soil surface roughness successfully, but to our knowledge, this method has not been applied extensively to research on physical weed control.
Approach. To evaluate the potential for LIDAR to facilitate understanding of physical weed control, we ran scans before and after bed preparation and cultivation events including flextine cultivation, finger weeding and hilling in a series of field trials on sandy soil. For these evaluations, the LIDAR was mounted to a portable box to exclude light, with multiple scans taken both parallel and perpendicular to crop rows (Figure 1A).
Figure 1. Daniel Priddy taking LIDAR scans with Peruvian hat (A) and example scans (B).
The output from LIDAR scans was transformed to get the relative height of soils at different distances from crop rows (Figure 1B), and to calculate random roughness (standard deviation of the height of soil). Note in the example scans, the X axis is horizonal distance relative to the crop row, and the planter indentations can be seen at ~380 mm in either direction (between row crop spacing ~760 mm). The Y axis is relative height in mm, but stretched to better show surface characteristics. Note also the vertical lines on either side of the figure which are the edges of the scanner box.
Results. LIDAR scans were able to characterize differences in soil surface roughness associated with contrasting forms of soil disturbance including finger weeding vs hilling, and rolled vs unrolled beds (e.g. Figure 2). Note in this example, we were able to detect an interaction between rolling and compost application that would not have been obvious without these scans. (In this case, the effect of rolling on surface roughness was only significant where compost had not been applied historically)
Figure 2. Soil surface roughness determined from LIDAR scans in rolled and unrolled beds, with and without compost application.
Next Steps
We are currently evaluating other measurements of soil surface characteristics from our LIDAR data which we believe may be correlated with tool efficacy, including measures of the ‘clumping’ of soil (correlation in heights between adjacent points), which should improve understanding of (for example) the extent of soil fracturing resulting from cultivation with different tools in soils with different levels of crusting.