Source: APPLICATION INSIGHT, LLC submitted to NRP
RADAR-BASED SPRAY DRIFT DETECTION AND QUANTIFICATION SYSTEM
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
Reporting Frequency
Annual
Accession No.
1028596
Grant No.
2022-33530-37406
Cumulative Award Amt.
$175,000.00
Proposal No.
2022-01423
Multistate No.
(N/A)
Project Start Date
Jul 1, 2022
Project End Date
Sep 30, 2023
Grant Year
2022
Program Code
[8.13]- Plant Production and Protection-Engineering
Recipient Organization
APPLICATION INSIGHT, LLC
2519 WILSON AVE
LANSING,MI 489062737
Performing Department
(N/A)
Non Technical Summary
The problem: Most spray drift detection occurs as damage complaints long after a preventable event has occurred. Pesticide applicators, a critical link in the production of America's food, are tasked with cheap and efficient application of pesticides to protect our food and fiber. At the same time, they are tasked with stewarding our air, water, and land by making precise applications in an imprecise world filled with variables over which they have little control. It is a heavy burden and the tools to efficiently prove a responsible application free of drift damage don't exist. This leads to mistrust, expense, and litigation. Farms struggle to employ spray applicators, at least in part due to this heavy burden and the relatively low reward versus risk. Applicators need to be able to prove they did a good job and kept their spray where it belongs.Over 2.5 million acres of crops were reported to have been damaged by spray drift in 2017 (Parker 2017); many more acres, including crops, wildlands, and water courses were likely impacted and not reported. "Data on crop losses due to pesticides are difficult to obtain. Many losses are never reported to the state and federal agencies because the parties settle privately." (Pimentel et al., 1993). (Pimentel 2005) revised earlier estimates and put a value on total economic losses due to pesticide over-application on crops, including spray drift, as just under $1.4 billion dollars annually .No doubt Pimental's estimation would be low in today's dollars. In addition, with the market introduction three seasons ago of the Dicamba herbicide-resistant crop trait and the subsequent broadscale use of that herbicide, drift complaints have accelerated to the point they are a main-stream media topic. Dicamba is active on some plants in as low as parts per billion concentrations in the air (Kruger, Pers. com.) and lethal to others in parts per million. (Kittay, 2017).We often "see" drift in abstract, anthropocentric terms, i.e "I couldn't see it, so it must not have happened." However, it is impossible to make a spray without at least some small quantity of drifting particles. Good, responsible applications with spray drift below levels that would cause an effect are achievable with careful attention to weather, application speed, droplet size, and spray release height. The ability to detect and quantify spray drift operationally as it occurs, and before the drift spreads in sufficient quantity, will be a strong mitigation of risk for all involved. With present drift detection methodologies, real time detection is simply not possible, thorough it is desired by stakeholders. This is an unfilled and costly technology gap.
Animal Health Component
50%
Research Effort Categories
Basic
(N/A)
Applied
50%
Developmental
50%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
21615992020100%
Knowledge Area
216 - Integrated Pest Management Systems;

Subject Of Investigation
1599 - Grain crops, general/other;

Field Of Science
2020 - Engineering;
Goals / Objectives
To measure the change in radar reflectivity of spray laden air versus non-spray laden air. Reflectivity measurements, combined with measurements of temperature, relative humidity, air speed and direction, account for the psychrometric effects of evaporation on the spray's volatile component and allow calculation of air volume. This information will allow us to resolve the quantity of spray drift as it passes a plane at the downwind edge of a field.Our key innovation is the concept that the total change in reflected intensity will correlate to the change in aqueous content in the air. In many radar applications, these measurements would mostly be interpreted as noise and be filtered out. While spray drift clouds are extremely dispersed compared to atmospheric aqueous clouds, which have been able to be assessed for water content for nearly 70 years, we expect that as the quantity of spray drift reaches levels that would cause concern, we will be able to detect it. Atlas reported detection to 0.1 ml/meter3 (100 parts per billion (ppb)). Bertoldo reported similar detection sensitivities with rain drops. Typical spray drift would be in the order of magnitude of 0.2 ml/m3. Dicamba drift damage can occur down to several ppb of active ingredient, but it is always sprayed at less than 1.25% concentration (OSU extension 2016) so carrier liquid would be present to facilitate detection. Therefore, radar is expected to be appropriate technology to detect relevant levels of drift in most situations.There are several means to use radar to detect the spray-laden air. 77 gigahertz is a well established frequency for which aqueous water particles exhibit strong reflectivity(Gourova et. al 2017). Bertoldo, et.al. 2018, describe the use of 77 GHz radar as a near-field rain gauge. We will also evaluate 24 GHz (K-Band) radar systems due to high electromagnetic absorption from water (H2O) (Bertie & Lan 1996).The high attenuation of 24GHz is generally unsuitable for weather applications, but can benefit the detection of drift in several ways. First, the phenomenon of radio absorption from water will limit the distance of transmission, keeping the operational distance localized to the drift measurement region. Second, the limited transmission range should reduce the chances of interference from other devices operating on the same frequencies. Third, the radio absorption of water may be observed as reflected signal power loss when observing a known target, thus providing the necessary information to detect the presence of drifting material (Chen 1975). This means that there could be two methods to validate radar sensors when measuring water droplets and vapor, one by direct reflected signal and one by reflected signal degradation to a known, highly reflective target compared to conditions prior to the presence of water. It may be possible to combine the use of 2 or more of these phenomena to create a more robust output, though that would not occur until Phase II.We propose to leverage smaller footprint radars to identify water particles and weather detection algorithms to design a system able to detect field scale drift. (Acconeer Corp, Infineon Technologies, Lufft Corp, others). Devices proposed for testing are already used in industry (Pisa et. al 2016) and their licensing requirements are mostly public ISM bands (USFCC, 2020).
Project Methods
The overarching objectives of Phase I are to validate that we can find and quantify dispersed clouds of spray drift with low energy radar devices in a laboratory environment and gain insight as to the variability inherent to real-world use. Spray drift for the purposes of this series of experiments is defined as suspended aqueous particles in the 12 to 150 micron diameter range and of varying particle and mass ratio densities in relation to the air. While smaller droplets may exist in spray drift clouds, their contribution to spray drift as a pollution problem is negligible. This will be done through a series of experiments designed to answer the following key questions related to the technical and commercial feasibility:Does the equipment detect water droplets consistent with spray drift? What mechanism(s) work for the detection given the hardware parameters?Upon successful detection of water, do these radar system options allow for the quantification of the drift itself (density of the cloud, size of the cloud, etc.)? What are the limits of each radar system measurement capabilities?What degree of variability exists in field conditions from environmental factors? Does the measured variability prevent the ability to distinguish and quantify a passing spray cloud?Technical Objective 1: Initial validation of the core detection process.Acceptance Criterion: For each candidate technology, there are three tests to determine detection of simulated spray drift simply at a presence or absence level. As detailed below, they are direct reflection, single-point signal degradation, and multi-point signal degradation. The combination of a candidate technology and test method will define a "use case". For each baseline, experiment, and post baseline, 1000 samples will be collected. 80% of the experiment samples must show differences versus the averaged baseline measurements to be considered a success. Use cases will be eliminated for future consideration if no difference is detected. Successful use cases will advance to Objective 2.Technical Objective 2: Quantification of the radar system sensitivity and validation of the ability to quantify the detected droplet cloud.Acceptance Criterion: Using the same protocol as Objective 1, each remaining use case will be evaluated in a broader and more precise range of variables to quantify the sensitivity of the device. Success for a given use case will be defined at a basic level as the ability to generate a trend on one or more variables that is useful for defining a spray drift presence threshold. Higher level success will be defined by trends that are of sufficient quality to allow the creation of a quantitative model or define geometric boundaries.

Progress 07/01/22 to 09/30/23

Outputs
Target Audience: Nothing Reported Changes/Problems:Several radar-on-chip development boards were purchased and evaluated for the opportunity to detect drift-sized spray droplets. In short, none of the evaluated radar-on-chip systems met this criteria. While the TI mmWave board was successful in detecting larger droplets from a garden hose (as replicated by the TI mmWave video documentation), the board was not suited to detect smaller droplets suitable for the target application. The Acconeer board appeared to sense drift size droplets but the significant warm-up time and delayed response prevented this board from being suitable for a drift detector application as well. The remaining boards either performed poorly or were limited in capabilities that would qualify them for further evaluation. When we proposed this project, there was a chance failure could occur. These boards are designed primarily for object and presence applications with typical uses in industrial, commercial, and automotive applications. These applications involve sensing the position and motion of larger solid objects while rejecting signal noise from dust, spray, and other small objects. Our intention was to determine if these boards had other mechanisms to detect spray droplets outside of their intended use. In the case of the TI mmWave board, one could argue this to be true. The noise profile data from the development kit clearly described changes to signal when large droplets from a garden hose were sprayed in its sensing region. Unfortunately, smaller droplets were not sensed either because the hardware was not sensitive enough or because the inherent noise of these systems overshadowed the behaviors we were attempting to capture. Similarly, the Acconeer board seemed to have some level of sensitivity to small droplets, but in many cases it was unclear if the observed changes were from directly detecting the droplets. The long duration of time required to achieve signal equilibrium to measure any sort of change also induced complications. In short, the Acconneer board was not responsive enough for this application. The OmniPreSense and Infineon BGT boards were not suitable for this application as they only provided velocity information. Again, these boards were selected based on criteria that suggested they might be able to sense droplets and that we might need to tweak operating parameters or look at analysis techniques to extract the behaviors we wanted to sense. Unfortunately, both of these devices were more "locked down" than expected only providing limited information, primarily for larger object and surface detection only. The Infineon Position2Go board was selected due to its similar features to the TI mmWave board. Our expectation was that this board would be as equally feature rich as the TI board. Unfortunately, the Infineon Positon2Go board was both limited and yielded highly noisy data, even when using it for its primary purpose. It is unclear if this board was damaged or if the board is an inherently poor detector. Overall, this project was deemed a failure. While radar has the ability to detect water droplets, it's clear that several of the commercially available radar-on-chip development boards have design intent to reduce noise from liquid droplets ranging in diameters common from rain to fog in an effort to emphasize information about larger rigid objects. This is particularly helpful in automotive applications. Radar-on-chip technology remains quite new. If this project were revisited several years from today, the odds of success are likely to improve. The project goal of constructing a radar-based drift detector still holds merit. If the project was expanded beyond radar-on-chip technology, a custom-built radar system is likely to have a greater chance at sensing spray drift as discovered in prior research. While developing such a system would come at great cost, it could help shape the specifications necessary to develop a radar-on-chip integrated circuit solution, ideal for reduced costs and commercialization. What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? Several radar-on-chip development boards were purchased and evaluated for the opportunity to detect drift-sized spray droplets. In short, none of the evaluated radar-on-chip systems met this criteria. While the TI mmWave board was successful in detecting larger droplets from a garden hose (as replicated by the TI mmWave video documentation), the board was not suited to detect smaller droplets suitable for the target application. The Acconeer board appeared to sense drift size droplets but the significant warm-up time and delayed response prevented this board from being suitable for a drift detector application as well. The remaining boards either performed poorly or were limited in capabilities that would qualify them for further evaluation. When we proposed this project, there was a chance failure could occur. These boards are designed primarily for object and presence applications with typical uses in industrial, commercial, and automotive applications. These applications involve sensing the position and motion of larger solid objects while rejecting signal noise from dust, spray, and other small objects. Our intention was to determine if these boards had other mechanisms to detect spray droplets outside of their intended use. In the case of the TI mmWave board, one could argue this to be true. The noise profile data from the development kit clearly described changes to signal when large droplets from a garden hose were sprayed in its sensing region. Unfortunately, smaller droplets were not sensed either because the hardware was not sensitive enough or because the inherent noise of these systems overshadowed the behaviors we were attempting to capture. Similarly, the Acconeer board seemed to have some level of sensitivity to small droplets, but in many cases it was unclear if the observed changes were from directly detecting the droplets. The long duration of time required to achieve signal equilibrium to measure any sort of change also induced complications. In short, the Acconneer board was not responsive enough for this application. The OmniPreSense and Infineon BGT boards were not suitable for this application as they only provided velocity information. Again, these boards were selected based on criteria that suggested they might be able to sense droplets and that we might need to tweak operating parameters or look at analysis techniques to extract the behaviors we wanted to sense. Unfortunately, both of these devices were more "locked down" than expected only providing limited information, primarily for larger object and surface detection only. The Infineon Position2Go board was selected due to its similar features to the TI mmWave board. Our expectation was that this board would be as equally feature rich as the TI board. Unfortunately, the Infineon Positon2Go board was both limited and yielded highly noisy data, even when using it for its primary purpose. It is unclear if this board was damaged or if the board is an inherently poor detector. Overall, this project was deemed a failure. While radar has the ability to detect water droplets, it's clear that several of the commercially available radar-on-chip development boards have design intent to reduce noise from liquid droplets ranging in diameters common from rain to fog in an effort to emphasize information about larger rigid objects. This is particularly helpful in automotive applications. Radar-on-chip technology remains quite new. If this project were revisited several years from today, the odds of success are likely to improve. The project goal of constructing a radar-based drift detector still holds merit. If the project was expanded beyond radar-on-chip technology, a custom-built radar system is likely to have a greater chance at sensing spray drift as discovered in prior research. While developing such a system would come at great cost, it could help shape the specifications necessary to develop a radar-on-chip integrated circuit solution, ideal for reduced costs and commercialization.

Publications


    Progress 07/01/23 to 09/30/23

    Outputs
    Target Audience: Nothing Reported Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals?Additional testing to boards and with different droplet generating devices will be continued as per the technical objectives.

    Impacts
    What was accomplished under these goals? The following radar-on-chip development boards were purchased and evaluated based on their core specifications and the expectation they would have a chance at detecting spray drift: OmniPreSense OPS 243 Infineon BGT24LTR11 Infineon Position2Go TI mmWave IWR6843ISK Acconeer XE132 The OmniPreSense OPS 243 and Infineon BGT24LTR11 boards implemented doppler radar functionality. Ideally, these could track the motion of the drift particles, rather than the presence of the mist itself. Preliminary evaluation determined these boards were not sensitive enough for this task. The boards were capable of tracking larger solid objects however with spray droplets, the results resulted in negligible average flux (undetected). Unfortunately, there were few tuning parameters and no other relevant data sets to extract from the boards. These boards received no further evaluation as they were deemed incapable of measuring spray drift. The Infineon Position2Go board resulted in very inconsistent tracking for both amplitude and position, even when the target was large and solid, like a human. The noise completely obscured any readings when testing with drift droplets. As a result, this board received no further evaluation as it was deemed incapable of measuring spray drift. The TI mmWave board was very successful in sensing droplets from a garden hose sprinkler but these droplets are outside of the scope of spray drift. This validated basic functionality of the radar (as noted in early project research) but the droplets are large, meaning this is not useful in the spray drift detection application.

    Publications


      Progress 07/01/22 to 06/30/23

      Outputs
      Target Audience: Nothing Reported Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals?There will be continued testing of different boards and different droplet sizes as per the technical objectives.

      Impacts
      What was accomplished under these goals? The following radar-on-chip development boards were purchased and evaluated based on their core specifications and the expectation they would have a chance at detecting spray drift: OmniPreSense OPS 243 Infineon BGT24LTR11 Infineon Position2Go TI mmWave IWR6843ISK Acconeer XE132 The OmniPreSense OPS 243 and Infineon BGT24LTR11 boards implemented doppler radar functionality. Ideally, these could track the motion of the drift particles, rather than the presence of the mist itself. Preliminary evaluation determined these boards were not sensitive enough for this task. The boards were capable of tracking larger solid objects however with spray droplets, the results resulted in negligible average flux (undetected). Unfortunately, there were few tuning parameters and no other relevant data sets to extract from the boards. These boards received no further evaluation as they were deemed incapable of measuring spray drift. The Infineon Position2Go board resulted in very inconsistent tracking for both amplitude and position, even when the target was large and solid, like a human. The noise completely obscured any readings when testing with drift droplets. As a result, this board received no further evaluation as it was deemed incapable of measuring spray drift. The TI mmWave board was very successful in sensing droplets from a garden hose sprinkler but these droplets are outside of the scope of spray drift. This validated basic functionality of the radar (as noted in early project research) but the droplets are large, meaning this is not useful in the spray drift detection application.

      Publications