This document discusses using passive open-path Fourier transform infrared (OP-FTIR) spectroscopy to monitor pesticide drift. The researchers conducted greenhouse and outdoor experiments to detect and quantify pesticide clouds using OP-FTIR. In the greenhouse experiments, the OP-FTIR was able to successfully detect water clouds 94.8% of the time with a 5% false positive rate. Outdoor experiments detected pesticide clouds created by common sprayers. Future work will incorporate radiative transfer modeling to estimate cloud parameters like concentration from the OP-FTIR measurements.
Monitoring Pesticide Drift Using Passive FTIR Measurements
1. MONITORING PESTICIDE DRIFT USING PASSIVE MEASUREMENTS OF OPEN-PATH FTIR
Oz Kira, Yael Dubowski and Raphael Linker
Faculty of Civil and Environmental Engineering, Technion, Haifa, Israel
1 Introduction
Pesticide drift, originated from agricultural
use, may be a threat for the health of close by
population and ecological systems. Airborne
pesticides spend significant part of their
lifetime as condensed aerosols, which are
more difficult to measure and estimate than
gaseous compounds.
In the present research we are using OpenPath
Fourier
Transform
Infra-Red
spectrophotometer (OP-FTIR) for real time
monitoring of pesticide drift. Infrared
instruments in general have shown significant
potential for detecting and identifying
airborne compounds. Yet, the use of OP-FTIR
has been restricted mainly to measuring
gaseous components, and its use for
investigating toxic airborne particulates or
condensed-phase pollutants has been
reported only in very few studies.
2 Radiative transfer – remote sensing principles
Passive remote sensing of aerosols in the atmosphere in general and of pesticide drift in particular is based on radiative transfer.
The LOS, consists of the ambient air, the background, and the pesticides cloud. Each medium emits radiation according to its
optical properties and temperature, and interacts with incoming radiation. Remote sensing by spectral instruments is unable of
measuring aerosols size distribution and concentration directly, and in order to estimate these, a suitable model is needed. BenDavid et al. 2003 developed a radiative transfer model to assist the detection and estimation of biological aerosols.
𝟏
𝟐
𝑴 𝝀 = 𝑴 𝟏 𝝀 + 𝟏 − 𝒆𝒙𝒑 −𝜶 𝝀 𝝆
𝟑
𝑩 𝝀, 𝑻 𝒄 𝒕 𝟏 𝝀 + 𝑴 𝟐 𝝀 𝒆𝒙𝒑 −𝜶 𝝀 𝝆 𝒕 𝟏 𝝀 + 𝜺 𝒃 𝝀 𝑩 𝝀, 𝑻 𝒃 𝒕 𝟐 𝝀 𝒆𝒙𝒑 −𝜶 𝝀 𝝆 𝒕 𝟏 𝝀
The atmospheric radiance between the aerosol cloud and the sensor (1).
The radiance of the aerosol cloud (2).
The atmospheric radiance between the aerosol cloud and the background (3).
The radiance of the background surface (4).
α(λ) is the mass extinction coefficient
ρ is the mass column density
𝒆𝒙𝒑 −𝜶 𝝀 𝝆 is the cloud’s transmission
𝜺 𝒃 𝝀 is the background emissivity
3
𝑩 𝝀, 𝑻 𝒃 is the blackbody radiation
𝑴 𝝀 is the radiance of the ambient atmosphere
𝒕 𝝀 is the atmospheric transmission
Figure 1: The geometry of the LOS between the
sensor and the background (Harig., 2004).
3 Remote sensing of aerosol - challenges
The calculation of scattered radiation from molecules and extra fine particles in the UVVis-IR range is rather simple and straight forward; however this is not the case with
larger particles which are calculated using the complicated Mie theory. The signal
depends on optical parameters (e.g. refractive index) and quantitative parameters (e.g.
size distribution and concentration). One of these parameters and one of the major
challenges is to isolate each of the parameters' influence.
Figure 2: Signal dependence on the droplets’ diameter.
A result of radiative transfer modeling using Modtran 5.
𝟒
Figure 3: Signal dependence on the droplets’ concentration.
A result of radiative transfer modeling using Modtran 5.
4 Goals
The main goals of this study are the detection and quantification of a pesticide cloud in
a contained environment and in the field. First of all experiments where conducted to
detect the cloud’s presence in the line of sight. After the successful detection of the
cloud, estimation of parameters such as concentration and size distribution will be
attempted.
Figure 4: The OP-FTIR
measurement setup at the Matityahu orchard.
Figure 5: The greenhouse
measurement setup at the Technion.
5 Results – Greenhouse measurements
6 Results – Outdoor measurements
The greenhouse measurements were intended to study the limits of detection using the
OP-FTIR. The temperature of the polyethylene sheet is very similar to the background
(soil and vegetation), which causes a lower signal extinction due to the cloud’s presence
in the line of sight.
Detection of a water cloud using Neural Quantification of the flow rate using Neural
Networks: The first goal was to detect a Networks: A preliminary attempt was made
cloud in the line of sight. An experiment of to quantify the signal using vendor data of
440 measurements using several types of flow rates and the same 440 measurements.
nozzles with different size distributions and The RMSE obtained was 9 μm.
flow rates was carried out.
Outdoor measurement were intended to study the actual abilities of the OP-FTIR
during a representative spraying of pesticides.
True
positive:
94.8%
True
negative:
82.9%
Water cloud signals as a function of nozzles
type and number: An outdoor experiment
yielded good results in which a clear
distinction between nozzle type and
quantity was observed.
Total true
detection
: 90.5%
Figure 4: The results of the cloud’s detection experiment.
Measurements of a standard pesticide cloud
created by a common sprayer: An initial
attempt was made to detect a cloud in the
field in real conditions. There were four
spraying events at different distances.
1st event
2nd event
Figure 5: The results of the cloud’s quantification experiment.
Figure 6: Measurements of a cloud produced by
5,3, and 1 nozzles with d50 of 119 μm .
7 Conclusions
The preliminary experiments show promising results. The OP-FTIR was able to detect a cloud in
all of the experiment, even in the limited radiative conditions in the greenhouse. The
quantitative part of this study is much more difficult due to the large number of parameters
influencing on the total signal. Future plans of this study include the incorporation of radiative
transfer modeling using Modtran 5 in order to estimate the cloud’s quantitative parameters.
Additionally, chemometric methods, such as the mentioned neural network, will be used to
quantify these parameters. The use of the OP-FTIR could contribute, in the future, to
calibrating and validating models of Aeolian transport and aerosolized pollutants.
3rd event
4th event
Figure 7: The results of the field experiment.
Supported by:
Center for Security Science
and Technology (CSST)