Application of OpenAir and AgDRIFT Models to Estimate Organophosphate Pesticide Spray Drift: A Case Study in Macon County, Alabama

Abstract

Pesticides have been widely used in agriculture, resulting in significant pollution that affects both the environment and human health. This pollution is particularly prevalent in nearby agricultural areas, where sensitive resources are contaminated through spray drift exposure and surface runoff. Spray drift is a critical concern when it comes to environmental hazards. It poses health risks not only to farmers and pesticide applicators, but also to individuals living in nearby farm areas. To address this issue, developing reliable models and techniques for estimating spray drift and reducing its impact has become a crucial and efficient research topic. The current research has three primary objectives: firstly, to estimate the average pesticide application rates, trend analysis, and concentration distribution; secondly, to estimate the temporal variations of pesticide concentrations and identify the areas most likely to be affected by pesticide spray drift close to agricultural fields; and lastly, to develop a model for field spray drift and deposition integration between the OpenAir package for the R programming environment and the AgDRIFT atmospheric model. The drift model, along with precise supervised classifications, allowed for a more accurate estimation of potential drift in agricultural areas at a spatial resolution of 15 m. Additionally, multiple scenarios were conducted to evaluate the potential risks of pesticide drift outside of the target areas. This novel method effectively estimated organophosphate pesticide spray drift over two case studies in Macon County using a combination of OpenAir and AgDRIFT models and remotely sensed data. This method allowed for field simulations within completely defined exposure areas with little prior knowledge of pesticide quantities. This study concluded that 6% of total cropland is in danger of pesticide spray drift, with around 8% of crop areas exposed to potential strong drift on land use. Furthermore, 11% of cropped land is vulnerable to moderate drift, whereas around 75% of land use land cover is not vulnerable to pesticide drift. Through this research, an accurate and efficient approach has been developed to estimate spray drift and reduce its impact in agricultural areas, contributing to a safer and healthier environment for all.

1. Introduction

Pesticides can be unintentionally spread from treated farms to areas like residential neighborhoods, lakes, and sensitive crops through spray drift, vapor transfer, and streams [1]. According to [2], up to 50% of pesticides can move off-target through spray drift, and other studies have supported this finding [3,4,5]. Target spraying and drift spray are the two most common methods of pesticide application. Target spraying involves directing the spray from the tank to the specific location of the pests, while drift spray relies on wind to spread the spray over a wider area to reach the pests [6,7,8]. Drift spray is often considered more cost-effective as it can cover a larger area, reach intertwined orchards, and reduce the number of applicators needed. Pesticide drift occurs when pesticides unintentionally move through the air, away from their intended target site. This can happen either during or shortly after the application process. Fine droplets are more likely to drift than coarse droplets. A few factors can impact droplet sizes, such as the pesticide’s formulation, nozzle type, sprayer pressure, application method, and weather conditions [9,10]. The use of pesticides can be harmful to nearby agricultural communities when weather conditions cause them to drift away from their intended targets. High temperatures and low humidity can affect the rate at which droplets evaporate and make them more likely to be carried away by the wind. It is essential to obtain accurate information about the weather conditions to prevent pesticide drift. Field tests are typically used to assess exposure drift by gathering samples from the air cloud near the treatment site. However, these tests carry significant risks due to the unpredictable weather conditions during pesticide application. The spread of pesticide spray can have harmful effects on nearby crops, water sources, and residential areas. This is the main cause of pollution in farming regions and poses a significant risk to the health of farmers, applicators, and the environment. Exposure to these chemicals has also been linked to various health problems, such as cancer and reproductive issues in humans [4,11]. Spray drift models, utilized by regulatory organizations for forecasting pesticide dispersal, often use conservative assumptions, making it challenging to estimate environmental drift exposure or hazard [12].

It is widely acknowledged that people who work with pesticides, whether directly or indirectly, are at risk of experiencing varying degrees of toxic effects. This exposure usually occurs during the pesticide spraying process and can severely harm the health of farmers and applicators. Studies conducted by [13,14,15,16,17] have all confirmed this fact. In four Asian countries, a survey found that 3% of the global agricultural workforce suffers from pesticide poisoning every year. The survey also highlights the long-term impact of pesticides in urban farming. The authors in [18,19] suggested that exposure to organophosphate (OP) pesticides in children can increase oxidative stress and contribute to chronic diseases. This makes unintentional exposure to spray drift, which can occur through a ground boom, orchard air blasts, and aerial applications, a significant challenge for both the environment and society.

Reducing off-target drift is crucial to prevent the harmful effects of OP pesticides. The United States Environmental Protection Agency (EPA) has developed several models to assess pesticide risks, including the AgDRIFT model. This model is particularly useful as it can evaluate various spray drift scenarios, such as agricultural applications and the off-site deposition of liquid pesticides. According to the EPA in 2022, the AgDRIFT model has multiple applications, including the estimation of the downwind deposition of spray drift from aerial, ground boom, and orchard applications. The AgDRIFT model assesses pesticide spray drift based on distance downwind and the efficiency of the spray equipment. It was created to evaluate different spray drift conditions and measure potential exposure. The Spray Drift Task Force developed the AgDRIFT model, which can estimate potential drift for individual ponds distributed over a wide area using a flexible drift model and precise land cover data.

Over the past few years, a vast amount of data has been stored and made available online, including dispersion modeling results and other air pollution-related information. Due to the sheer size of these datasets, it has become essential to devise tools capable of managing and deciphering them [20]. One such tool is the OpenAir package for R programming, which provides data and features useful for air quality estimation and other related research. This study aims to integrate the capabilities of the OpenAir package with AgDRIFT models to achieve its main objectives.

Numerous studies have been conducted to evaluate the risks associated with spray drift, as its application has been observed to have negative effects. These studies include the research in [21,22,23,24,25]. The research on pesticide spray drift in the field is limited. Previous studies have used data from national reporting organizations, assuming that all farmers follow the instructions on pesticide products. However, due to the negative impact on both society and the environment, there has been increasing interest in investigating this phenomenon [26,27,28,29,30,31,32].

There are different methods to estimate spray drift, including advanced techniques like the 3D LiDAR algorithm and atmospheric models such as the AgDRIFT, AgDISP, PERFUM, SOFEA, RTDRIFT, and VALDRIFT [5,32,33,34,35,36,37]. Furthermore, various researchers have implemented the Spray Drift Task Force (SDTF) standard criteria to assess the likelihood of pesticide exposure drift [34,35,36,37]. Since the pesticide database is not available, some studies rely on comparing the distribution of pesticide use with maps that show how the land is being used and covered [33,37,38,39,40,41,42]. Other previous studies focused on the emission of pesticides into the air during sprayer application and the risk assessment of pesticide spray drift damage to crops [43,44,45,46].

Specifically, drift in spray operations is sensitive to factors like atmospheric conditions and equipment, making field testing difficult. Models have been developed to predict drift and deposition from multiple spray applications. These models are empirical and mechanistic, with the ideal model including mechanistic descriptions of processes like application mode. AgDRIFT, an extension of the original AGDISP Lagrangian model, is supported by comparisons with field data; these comparisons present a powerful argument for the applicability of AgDRIFT in predicting the downwind deposition and drift of both ground and aerially released spray material [26,29,42].

The OpenAir package model is widely used in atmospheric sciences worldwide and is frequently employed to assess air pollution dispersion modeling systems and trajectory clustering to examine the origins of pollutants. In addition, it has been effectively utilized to perform air quality monitoring analyses and estimate emission sources based on meteorological data. OpenAir was utilized by Carslaw and Ropkins (2012) to investigate relevant pollution exposure and events that occurred naturally [47]. Carslaw (2015, 2019) investigated the frequency with which the functions provided in this package collaborate to provide comprehensive details about the examined data, particularly when several variables and air pollutant drivers are involved [48,49]. Abdou (2014) and Munir (2016) used OpenAir to model the non-linear association of PM10 with metrological parameter trends [50,51]. Numerous studies have used OpenAir to investigate the variation in major air pollutants in different seasonal conditions in an urban environment, local source regions affecting fine particulate matter concentrations, and the accessible and repeatable analysis of urban air quality data, for instance, after conducting in Malaysia [52] and conducting in India [53]. Trajectory clustering, concentration-weighted trajectory, and other methods have been employed to investigate the origins of pollutants conducted in Chile [54] and Colombia [55].

After performing a study of the available literature review, it was discovered that there is no accurate pesticide usage database for Alabama and that the effects of pesticide use in Macon County, Alabama, have not been thoroughly assessed. Therefore, this study aims to investigate the use of the OpenAir and AgDRIFT models to estimate the spray drift of organophosphate pesticides in Macon County.

2. Materials and Methods

2.1. Study Area

This study looks at Macon County, Alabama, for several reasons. Firstly, it is a mostly rural area that relies heavily on agriculture and livestock farming. Secondly, it accounted for 1% of all pesticide use in the state of Alabama in 2017. Lastly, it ranks 16th in agricultural products and 17th in vegetables statewide [33]. Macon County is located in east-central Alabama, between latitudes 32°7′49″ and 32°35′7″ N and longitudes 85°30′48″ and 86°1′19″ W. It covers a total area of 392,830 acres, which includes water bodies, equivalent to 650 square miles. The county is bordered by Tallapoosa County to the north, Bullock County to the south, Russell County to the east, and Montgomery County to the west. Tuskegee is the county seat and largest city with a population of 9395, while smaller communities like Notasulga, Franklin, and Shorter have populations of 939, 598, and 385, respectively. Approximately 23% of Macon County’s land area, or 53,001.1 acres, is dedicated to cropland [56,57]. Cotton is the largest crop, covering 8909.6 acres, or 16.81% of the total cropland. Corn is the second-largest crop, covering 1941.3 acres, or 3.66% of the total cropland. This study will focus on the two largest fields of cotton and corn crops, located in Shorter to the west and Dicks Creek to the south, respectively (Figure 1).

2.2. Data

2.2.1. Cropland Data

The cropland data for Macon County were downloaded from the CropScape-Cropland Data Layer (CDL) dataset generated by the National Agricultural Statistical Service of the USDA (https://nassgeodata.gmu.edu/CropScape/ (accessed on 21 April 2023)). It is a high-resolution (30 m) satellite imagery with temporal coverage spanning more than 24 years (1997–2021) at annual intervals, with an extensive agricultural ground boom. It will be used to calculate the amount of pesticide usage and the total amount equals the crop area multiplied by 2.40 kg per acre, which is the standard amount of pesticide [56].

2.2.2. Pesticide Data

The agricultural organophosphate pesticide data for 2017 were obtained from the US Geological Survey (USGS) (https://earthjustice.org/organophosphates (accessed on 18 May 2023)), while the agricultural pesticide data for 2019 were obtained from [57,58]. The total volume of organophosphate pesticide used in this study was calculated by combining the use of seven organophosphate pesticide compounds, as shown in

Table 1. Agricultural organophosphate pesticide usage in Macon County, 2017, 2019.

Pesticides	ACEPHATE	CHLORPYR	DICROTOP	PHORATE	PHORATE	TERBUFOS	TRIBUFOS	Total Amount (Kg)
2017	24.70	50.60	94.30	410.50	170.30	150.00	2843.30	3743.70
2019	22.50	64.40	88.80	390.70	175.20	148.00	2450.60	3340.20

Sources: [57,58].

.

2.2.3. Satellite Imagery and GIS Dataset

The Landsat 8 images Operational Land Imager (OLI) will be used to create an accurate LULC map for Macon County with a spatial resolution of 15 m. GIS will be utilized to obtain better knowledge of how agricultural activities may be related to pesticide exposure in the environment. A GIS geospatial dataset will be created for the cotton and corn areas, pesticide use, the geographical extension of water bodies, and residential areas. Therefore, an accurate map for the distribution of pesticides in kilograms along the county will be developed.

2.2.4. Meteorology Data

For this study, the main source of meteorological data was gathered from the National Weather Service (NWS), under the direction of the NOAA Assistant Administrator for Weather Services. The NWS collects hourly information on surface weather, such as air temperature, pressure, humidity, wind direction and speed, and precipitation. All necessary climate variables and meteorological data can be accessed at https://www.weather.gov/wrh/Climate?wfo=bmx (accessed on 21 April 2023). These data will be utilized to obtain hourly wind speed and direction to run both AgDRIFT and OpenAir models.

2.2.5. Modeling and Its Boundary Conditions

OpenAir Model

This study will use an OpenAir model to monitor organophosphate pesticides and consider the current weather conditions. This model will help analyze the trend and dispersion of pesticide concentrations, examine their potential distribution, and measure the changes in pesticide concentrations over time. To achieve the research goals, the data will be analyzed by selecting appropriate functions in the OpenAir model, each with its analytical foundation and technique. OpenAir is a software package in R that is specifically built for air quality monitoring analyses. While its primary use is for air pollution measurement data, it is also useful in other areas of atmospheric science. The Natural Environment Research Council (NERC) initiated a model that uses statistical analysis to examine the correlations between pesticide concentrations and other relevant factors. The Environmental Research Group of King’s College London established this model as part of the NERC’s knowledge exchange initiative. The model can be downloaded from the official website of the OPENAIR Project at http://www.OPENAIRproject.org/ (accessed on 21 April 2023), according to [47].

AgDRIFT Model

This particular model is designed to estimate the amount of spray drift deposition and determine the quantity of spray drift that will be deposited downwind from aerial and ground boom applications. Teske et al. (2002) provide this model as part of the Cooperative Research and Development Agreement (CRADA) between the EPA’s Office of Research and Development and the Spray Drift Task Force (SDTF), a group of pesticide registrants who are developing a comprehensive database of off-target drift information in support of pesticide registration requirements [42]. This model requires specific meteorological and environmental boundary conditions, such as wind speed and direction, air temperature, relative humidity, and surface roughness. The Montgomery weather station was used as the closest station to the study area in this research. Wind data for Macon County were obtained from the Midwestern Regional Climate Center (MRCC) https://mrcc.purdue.edu/CLIMATE/ (accessed on 20 June 2023). These data are compiled and processed by the Climate Data Access Portal (Cli-DAP), which is maintained by the NOAA Regional Climate Centers (RCCs). Wind and pollution rose will be used to generate wind and pesticide particle trajectories at a specified spraying time in Macon County.

2.3. Methodology

In this study, the potential risks of pesticides drifting outside of the intended agricultural areas will be examined. Figure 2 provides a flowchart that outlines how this assessment will be conducted for ground boom, aerial, and mixed spraying methods. Additionally, the chart demonstrates how buffer zones can be utilized to protect sensitive habitats from unintended pesticide exposure. Our research will be conducted in Macon County using two case studies on a field-level scale. We will use the OpenAir package and the AgDRIFT model to estimate the potential drift exposure. Figure 2 shows that each of the five components is connected. Firstly, we will identify target fields and spray applications that have the potential to drift pesticides to non-intended areas using the precise location of the applied field and wind directions on the day of application. Then, we will use GIS techniques to generate outputs that will serve as the starting point for the OpenAir and AgDRIFT modeling. Finally, the AgDRIFT model will estimate the fraction of mass that moves off-target via spray drift for each drift event based on the pesticide application method and downwind distance from the applied field.

The preprocessing procedures of the CDL involved clipping the dataset’s spatial coverage for the Macon County boundaries and transforming the layers from the WGS 1984 Geographic Coordinate System (GCS) to the widely used Projected Coordinate System (PCS) of Universal Transverse Mercator (UTM) Zone 16 N (50,000,000 m square). Non-agricultural raster layers were masked to remove the vegetation raster layer because it included all agricultural and nonagricultural categories. Utilizing the statistical tools available in ArcGIS 10.8.2, the total acreage of diverse crop and rural–urban boundaries in the pesticide drift zone was measured and stored with those identical geographic sites in the GIS database. The USDA provided the Cropland Data Layer (CDL); these data were imported as a layer into ArcMap 10.8.2, and cotton and corn were extracted and isolated as separate layers.

The estimation of the impact of organophosphate pesticide spray drift on sensitive resources has relied on pesticide use data that is highly detailed in terms of both time and location. The goal was to determine the number of impacted acres that were exposed to spray drift. To account for data shortages in Macon County, the worst-case scenario values were used for the factors that most influence field-level spray drift. To calculate the total amount of pesticide applied to each crop, the crop area was multiplied by the average pesticide usage rate. The following equation was used for each crop to determine the potential pesticide exposure drift [32,33]:

$$X=\sum _{a}C\ast P$$

(1)

X = total potential pesticide applied for each crop.

${\displaystyle \sum _a}$ = sum of all crops.

C = total area of each crop in km².

P = quantity of pesticides applied in Kg for each crop.

To determine the quantity of pesticides used in each case study, the total cropland was calculated and multiplied by the authorized proportion of pesticides per acre. This study considered the application methods for corn and cotton pixels, which were classified into ground, aerial, and mixed application, each with their corresponding potential for drift. To calculate the total deposition for the field, the drift for all pixels combined both residential areas and water bodies. AgDRIFT is a model that can evaluate various applications and predict the potential impact of changes in wind speed and direction on subsequent spray applications in the same area. By considering the location of the weather station, which provides data on wind speed, temperature, and relative humidity, as well as the direction of the wind and the number of applications, AgDRIFT calculates the percentage of drift and predicts the amount of spray drift that will settle downwind, as shown in

Table 2. Factors affecting field-level spray drift.

Factor	More Drift	Less Drift
Spray droplet size	smaller	larger
Release (or boom) height	higher	lower
Wind speed	higher	lower
Spray pressure	higher	lower
Nozzle size	smaller	larger
Nozzle orientation (aircraft)	forward	backward
Nozzle location (aircraft)	beyond ¾ wingspan	¾ or less wingspan
Nozzle type	smaller droplets	larger droplets
Air temperature	higher	lower
Air stability	inversion	lapse
Pesticide volatility	volatile	nonvolatile
Relative humidity	lower	higher

Modified from [5,9].

.

The study developed an effective and appropriate method to estimate the off-site agricultural pesticide spray drift caused by ground and aerial spraying methods. This method also evaluates the potential of buffer zones to protect populated areas, delicate crops, and water bodies from unintended exposure drift on both county and field levels. To make careful estimates for downwind deposition values, researchers use the Tier I technique. This is the first review of aerial and ground spraying methods based on fair averages of Tier I field data. According to the AgDRIFT model conditions illustrated in [9,22,34,42] when the wind speed is 4.5 m/s, the projected area of the pesticide drift zone in a non-target field with an area of 100 acres (0.40 km²) is 4.7% for the ground application and 47% for the aerial application of the pesticide. To determine the total area at risk from spray drift exposure, the total number of hectares planted with sensitive resources in the areas where the pesticide was administered is multiplied by either 47% for aerial applications or 4.7% for ground applications, as shown in Figure 3.

2.4. Study Hypotheses

This study utilized the OpenAir package to analyze the effects of organophosphate pesticides as an air pollutant. The study assumed that all cropland acreage for each crop was covered with pesticides, regardless of the number used. The primary risk from agricultural pesticide application drift was assumed to affect sensitive resources closest to the application, such as residential areas, water bodies, and sensitive crops. Residential areas and fields outside of the buffer zone were not at risk of spray drift. The AgDRIFT simulation model was used to determine the spray drift range based on wind speed and direction, assuming a range of (1000, 437.5, and 220 m) for an aerial, ground, and mixed application, respectively, at a wind speed of 4.5 m/s. The density of trees in the buffer zone was found to mitigate pesticide drifting into residential areas or canals and lakes close to the cultivated area. The study focused on organophosphate pesticides as the air pollutant analyzed by the OpenAir package. Due to the lack of spatial resolution in the pesticide database, the worst-case scenario was used for pesticide drift over Macon County.

3. Results and Discussion

3.1. Estimating Average Pesticide Application Rates

In the two case studies, the agricultural acreage of cotton and corn was precisely measured using geographic information systems. These systems were also used to determine the total annual rates of pesticide applications based on predicted crop areas and US Department of Agriculture estimates of allowable pesticide proportions for each crop. By estimating the quantity of pesticides applied to cotton and corn crops, the total pesticide application rates were calculated.

It was discovered that the application rates for cotton and corn ranged from 36 to 42 kg/ha. Surprisingly, the most heavily used crops accounted for less than 10% of the total agricultural area, while grasslands made up 62% of the inventory shown in Figure 4.

The data displayed in Figure 3 show the quantity of pesticides used on crops in Macon County according to the [58,59,60,61] USGS 2017 EPest-high estimations. The harvested farmland in this area had a rate of 199.61 pounds per square mile, and the total estimated pesticide usage for Macon County was 3721.6 kg, according to Table 1. The figure demonstrates that Tuskegee, Notasulga, Franklin, Woodland, and Shorter have a high pesticide usage rate, despite having a combined usage of over 32,000 kg of agricultural pesticides. However, it is important to note that this classification may not be well understood due to the uneven knowledge distribution of pesticides over time and space, especially in residential areas. Additionally, Figure 5 highlights the significant amount of vegetation that separates these cities from the issue.

Table 3. Cropland data layer statistics for Macon County.

Shorter Field
Crop	Pixel counts	Acreage
Corn	6.073	13,506
Cotton	15,667	34,843
Other crops	68,098	151,446
Total	89,838	199,795
Spring Union field
Crop	Pixel counts	Acreage
Corn	4578	10,181
Cotton	5514	12,263
Other crops	13,638	3033
Total	23,730	52,774

displays the cropland data layer statistics for both case studies. It is noticed that corn occupies 17% of the total cropped area in the first case study, while it covers 19% in the second one. Meanwhile, cotton covers 17% and 23% of the total cropped area in the first and second case studies, respectively.

Figure 6 displays the various land use and land cover (LULC) classifications in Macon County. These classifications have been divided into smaller units with a 15 m spatial resolution. To estimate the potential drift for each crop in the area, we utilized the AgDRIFT model along with reliable LULC data. The study area included 43 bodies of water that stayed stationary, covering 20,000 acres. Of these, 12 were within the spray drift range of cotton and corn. The overall study area encompassed 1400 km² and included approximately 108 houses. Supervised classification is used to analyze the spatial relationship between crop areas and LULC, focusing on wind direction and buffer distance. We were able to determine potential drift using the AgDRIFT model since we used precise data input from remote sensing techniques.

3.2. Pesticide Trend Analysis and Concentration Distribution

Figure 7 shows that the levels of pesticides vary depending on the time of application. This is particularly evident in Shorter and Spring Union, where the daily fluctuations correspond to the cyclic nature of pesticide amounts. The highest pesticide concentrations are found during application hours between 8:00 a.m. and 6:00 p.m., with an average of 37.800 µg m⁻³ and 15.300 µg m⁻³ in Shorter and Spring Union, respectively. On the other hand, the lowest pesticide concentrations are observed during application hours between 7:00 p.m. and 7:00 a.m., with an average of 3.200 µg m⁻³ and 1.350 µg m⁻³ in Shorter and Spring Union, respectively, during the analyzed period.

To reveal the distribution of pesticide concentration, bivariate polar plots were used, displaying wind speed and direction on a radial scale. This can help study the relationships between different parameters. As shown in Figure 8, high levels of organophosphate pesticide concentrations are mostly associated with low wind speed. The plot also shows that the southern and western areas of Shorter Town are the most affected by high concentrations of organophosphate pesticides, while the eastern and western areas of Spring Union town are considered the most affected by these concentrations.

3.3. Prevailing Weather and Potential Pesticide Dispersion

When it comes to understanding present-day weather data and its effects on pesticide concentrations and drift, the wind rose is an incredibly useful tool. By factoring in wind direction and speed, it can help calculate the potential exposure drift to nearby sensitive areas and water bodies. In this experiment, it is noted that wind speeds varied between 2 and 6 m/s. To estimate potential spray drift dependent on wind direction, we identified each non-target zone, including sensitive resources, in all eight directions (N, NE, E, SE, S, SW, W, and NW). We then calculated the total drift from all directions and divided it by the eight main and secondary wind directions to determine the average potential exposure drift. Using this method, we were able to estimate potential drift exposure to nearby residential areas, water bodies, and sensitive crops.

In Figure 9, the possible drift deposition from three different wind directions and the average drift deposition for all eight directions are shown. This pertains to the sensitive resources surrounding cotton and corn crop agriculture in the investigated regions.

The chart in Figure 9 displays the primary wind direction and highest wind speed during a pesticide application period. The average wind speed measured was 2.5 m/s, and calm conditions were observed 52.5% of the time, indicating a wind speed of 0 m/s. The most common wind direction was west and northwest, accounting for 14% of the time, followed by south wind at 13.3%. The maximum wind speed recorded was 4–6 m/s. This suggests that Shorter’s northern and northwestern regions may be at risk of pesticide exposure drift.

In Spring Union town, the average wind speed was 2.3 m/s, and calm conditions occurred 49.4% of the time, which means the wind speed was recorded as 0 m/s. The most frequent wind direction was south and southwest, with a frequency of 12.4%, followed by east wind with a frequency of 11.9%. The maximum wind speed was 4–6 m/s. Due to this, pesticide exposure drift may impact the north, west, and northwestern areas of Spring Union.

Moreover, the pollution rose is a useful tool to measure pollutant concentrations based on wind direction. It calculates the percentage of time the concentration is in a specific range. This approach is particularly informative for air pollution species, as demonstrated by [47]. In this study, it is used to measure pesticide gases released during a ground boom or aerial application.

Figure 10 displays the pollution rose plots, which were created through a process that involved investigating pesticide roses for cotton and corn crops. Simulations were conducted using both aerial and ground applications. The examination was conducted every four hours for the worst-case scenario, including 400, 800, 1200 h, and so on. The pollution was divided into four quadrants, namely, NE, NW, SE, and SW. Potential drift values were obtained by running the model for 72 h, considering meteorological data obtained from NWS.

In Figure 10A, it is evident that the east and south-westerly winds have a significant impact on the average concentration of organophosphate pesticides in the Shorter field. Half of the total concentration is attributed to three wind sectors in the east, south, and west. On the other hand, Figure 10B highlights the dominance of east and south winds in determining the average concentration of organophosphate pesticides in the Spring Union field. More than half of the total concentration is contributed by two wind sectors in the east and south sections.

3.4. Potential Pesticide Drift

In Shorter town, pesticides have the potential to drift toward nearby residential areas, water bodies, and sensitive crops from three directions: south, southwest, and east. The average values for these directions are 0.80, 1.06, and 1.10, respectively. Similarly, in Spring Union, pesticides may potentially drift toward residential areas, water bodies, and sensitive crops from four different directions: east, southeast, south, and southwest. The average values for these directions are 1.25, 1.32, and 1.16%, respectively.

Figure 11 depicts that about 4% of the total cropped area in Shorter is at risk of pesticide spray drift, while approximately 3% of cropland areas are exposed to potential strong drift on the main land use land cover. Additionally, 6% of the cropped area is exposed to moderate drift, and approximately 87% of the land use and land cover is not at risk of pesticide drift. Figure 11 provides a visual representation of these findings.

In contrast, Figure 12 illustrates that 6% of the total cropped area is at risk of pesticide spray drift in Spring Union, with about 8% of cropland areas exposed to potential strong drift on the main land use land cover. Moreover, 11% of the cropped area is exposed to moderate drift, while about 75% of the land use land cover is not at risk of pesticide drift.

It is important to note that the study’s results were based on the standard buffer zone distances outlined in a drift model guide, as well as previous research on calculating potential pesticide drift exposure. This includes studies by Wan (2015), Mayer (2019), and El Afandi (2023) [11,31,33], respectively.

3.5. Temporal Variations of Pesticide Concentrations

The polar Annulus function is used to consider the time-based changes in pesticide concentration based on wind direction. This function offers a way to visually represent trends, day of the week, and diurnal variations. By using an annulus instead of a circle, the issue of reading values near the origin is resolved.

In Figure 13, it is evident that organophosphate concentrations are dominated by southerly winds. There is minimal change in concentrations across the simulated period from Shorter and Spring Union fields, as indicated by the red coloring. The day of the week plot shows that pesticide concentrations are heightened for all wind directions during the application time. The diurnal plot in Figure 13 indicates that larger concentrations are observed from 8 a.m. to 6 p.m., with contributions from the west, north, and east displaying extremely low concentrations during the day over Shorter and Spring Union fields.

Based on Figure 14, the highest concentrations of organophosphate pesticides were found when the wind was blowing from the east on August 1st, 2nd, and 8th. On August 3rd and 8th, the concentrations were highest when the wind was blowing from the south, and on August 4th, 5th, 7th, 9th, and 10th, they were highest when the wind was blowing from the west over the Shorter field. Additionally, the highest concentrations were found when the wind was blowing from the east on August 1st, 2nd, 8th, and 10th. On August 6th, 7th, and 9th, the highest concentrations were found when the wind was blowing from the south, and on August 6th, 7th, and 8th, they were highest in the southwestern area of the Spring Union field.

3.6. Estimating Spray Drift Deposition Using the AgDRIFT Model

To determine the total potential drift exposure for a residential area, water bodies, and sensitive crops, the average of the AgDRIFT values was used for the main LULCs in eight directions.

Table 4. Potential exposure drift for all pesticide applications.

AgDRIFT Modeling Results
Crop	Potential Drift for Ground Application (M)	Potential Drift for Aerial Application (M)	Potential Drift for Mixed Application (M)
Cotton	237.5	450	200
Corn	237.5	500	225

and

Table 5. Average potential drift for eight directions in case studies under investigation.

Sensitive Resources	Total Acres	Average Potential Drift for Each Direction in Shorter Field								Drift Average
		N	NE	E	SE	S	SW	W	NW
Residential areas	214.50	0.00	0.00	0.00	0.00	2.41	2.65	1.30	0.00	0.80
Water bodies	1860.00	0.00	0.00	0.00	0.00	2.90	3.70	1.90	0.00	1.06
Sensitive crops	132.40	0.00	0.00	0.00	0.00	4.20	2.60	2.00	0.00	1.10
Sensitive Resources	Total Acres	Average Potential Drift for Each Direction In Spring Union Field								Drift Average
		N	NE	E	SE	S	SW	W	NW
Residential areas	103.50	0.00	0.00	2.14	3.40	2.90	1.50	0.00	0.00	1.25
Water bodies	2640.00	0.00	0.00	2.31	4.20	2.80	1.30	0.00	0.00	1.32
Sensitive crops	165.40	0.00	0.00	1.90	1.85	3.50	2.10	0.00	0.00	1.16

were considered, and assuming that wind speeds varied between 4 and 10 mph throughout the experiment, we calculated the average potential exposure drift to the nearby residential and sensitive crop areas as well as the nearby water bodies as the total drift from all directions, divided by the eight main and secondary wind directions. We identified each non-target zone, particularly the sensitive resources pixel, in all eight directions (N, NE, E, SE, S, SW, W, and NW), which enabled us to estimate the potential spray drift dependent on wind direction. We estimated the potential drift exposure onto the nearby residential area, water bodies, and sensitive crops using eight directions.

Illustrated in Table 5 are the potential drift depositions from three wind directions and the average drift deposition for all eight directions in the regions under investigation, concerning cotton and corn crop agriculture. These findings pertain to the sensitive resources close to the crops.

In-depth research was conducted on the potential for spray drift in the field. The AgDRIFT model was utilized to determine the rate of spray drift from various pesticide applications, considering factors such as spray droplet distribution, boom height, and wind speed (ranging from 4 to 10 mph). The parameters used in the AgDRIFT model simulations for ground and aerial applications in two case studies conducted in Macon County are listed in

Table 6. Estimated drift distance for ground and aerial application based on four scenarios.

Crops	High Boom; Fine Spray	Low Boom; Fine Spray	High Boom; Med/Coarse Spray	Low Boom; Med/Coarse Spray
Estimated Drift Distance for Ground Application
Cotton	300	150	125	100
Corn	260	130	100	75
Estimated Drift Distance for Aerial Application
Cotton	1000	750	600	400
Corn	800	700	500	400

.

The AgDRIFT model simulations for agricultural pesticide ground applications show that even with conservative conditions of low boom and medium/coarse spray, a buffer zone of over 300 m is necessary to dissipate spray drift. The estimated drift rates for a medium droplet distribution and boom height of 2 m were 3.6%, 2%, 0.6%, and 0.4% of ground application rates at distances of 25, 50, 75, and 100 m downwind (shown in

Table 7. Parameters downwind distance range for ground application.

Ground Application Method	Parameters Downwind Distance Range in Meters				Worst-Case Scenario
	25	50	75	100	Medium Droplet Distribution
Swath width (m)/number	12/2	12/4	12/6	12/8	Medium droplet distribution
Downwind drift	3.6	2	0.6	0.4	Medium droplet distribution
Drift rate (%)	17	9.6	5.8	2.3	Medium droplet distribution
Drift area (km2)	0.14	0.08	0.04	0.014	Medium droplet distribution

and Figure 15A). For a medium droplet distribution and a boom height of 3 m, the estimated drift rates at distances of 25, 50, 75, and 100 m downwind were 7%, 4%, 2%, and 0.7% of the ground application rates (shown in Figure 15D). Off-site spray deposition rates increased significantly from medium to fine to fine droplet distribution, but this increase was less noticeable as the boom height increased (shown in Figure 15B,C). Depending on the topography and weather, off-site drift at distances of 37.5 m downwind may reach 1% of the ground application rate.

To calculate the total acreage exposed to the risk of drift from ground applications in the study areas, potential drift rates for unintended areas (such as residential areas and water bodies) were calculated using knowledge of the field’s nature, buffer zones, field shape, and climatic data. The average cropping area in our case studies is less than 30 hectares. For a 30-hectare square field and a 100 m drift range for a ground application in an adjacent target field, the drift area of the non-target field is estimated to be 0.14 square kilometers or 35% of the total area.

According to

Table 8. Parameters of downwind distance range for aerial application.

Aerial	Parameters Downwind Distance Range m				Worst-Case Scenario
	50	100	150	300	Medium Droplet Distribution
Swath width (m)/number	15/3	15/6	15/10	15/20	Medium droplet distribution
Downwind drift	7.2	5.2	3.3	1.4	Medium droplet distribution
Drift rate (%)	52	48.3	46.5	33.5	Medium droplet distribution
Drift area (km2)	0.47	0.40	0.35	0.25	Medium droplet distribution

and Figure 16D, the estimated deposition rates for a medium droplet distribution during aerial application are 11%, 6%, 4%, and 0.5% at distances of 50, 100, 150, and 300 m downwind. Off-site deposition may reach 1% of the aerial application rate at distances of 300 m downwind, depending on the topography and weather conditions. The estimated drift zone for a non-target field of 30 hectares during aerial application is 0.47 km or 11%. Based on the AgDRIFT model for a medium spray droplet distribution, a drift range of over 500 m represents the worst-case scenario.

In contrast, the non-target field’s drift area is calculated to be 0.35 km², which makes up 46.5% of the entire area. This is based on a 45 ha (0.45 km²) square field and a drift range of over 300 m during aerial application in a nearby target field, as illustrated in Figure 17.

As per the regulations set by the US Environmental Protection Agency, two simulations were conducted using the AgDRIFT model to estimate spray drift under actual meteorological conditions. The simulations took place when the air temperature at a height of two meters fluctuated between 26 and 36 °C, the relative humidity ranged from 75 to 88%, the wind direction did not deviate by more than 36 °C, and the average wind speed varied from 4 to 10 mph.

Table 9. Agriculture pesticide drift events associated with the ground and aerial application of the two case studies over Macon County.

Shorter Field
Crop	Method	Treated Area (ha)	Wind Direction	Applied Rate Rp (kg/ha)	Drift Fraction Fd	Deposition Area (ha)	Swath Width (m)	Drift Mass (kg)	Average Drift Applied (%)
Cotton	Ground Aerial	26.8 34.2	NE NW	2.30 2.60	0.222 0.350	3.61 4.80	1–8 1–20	0.7 1.3	3.97 5.62
Corn	Ground Aerial	28.3 36.5	NE NW	2.30 2.60	0.222 0.350	2.82 5.60	1–8 1–20	0.7 1.3	3.97 5.50
Spring Union Field
Crop	Method	Treated Area (ha)	Wind Direction	Applied Rate Rp (kg/ha)	Drift Fraction Fd	Deposition Area (ha)	Swath Width (m)	Drift Mass (kg)	Drift/ Total Applied (%)
Cotton	Ground Aerial	26.8 34.2	SE SW	2.30 2.60	0.222 0.350	3.61 4.80	1–8 1–20	0.7 1.3	3.97 5.62
Corn	Ground Aerial	28.3 36.5	SE SW	2.30 2.60	0.222 0.350	2.82 5.60	1–8 1–20	0.7 1.3	3.97 5.50

displays the findings of the simulation regarding agriculture pesticide drift events in Macon County, specifically concerning ground and aerial application. Both of the areas studied exhibited equivalent results. The treated area affected by pesticide drift ranged from 26.8 to 36.5 hectares, with a pesticide application rate of 2.30–2.30 kg/ha. The drift fraction deposition ranged from 0.222 to 0.350, and the deposition area ranged from 2.82 to 5.60 hectares. For ground application, the swath width ranged from 1 to 4 m, while for aerial application, the swath width ranged from 1 to 8 m. The drift mass reached 0.7 kg/ha, while the average drift applied ranged from 3.97 to 5.50%.

Regarding the potential risk of off-site spray drift from agricultural pesticide applications, the study estimated a risk of 4320 hectares or 1.3% of the total crop acreage planted in sensitive crops. This was based on exposure from near-field spray drift, which was accurately and precisely estimated. Additionally, the potential risk was estimated to be 3440 hectares or 1.24% of all water bodies from off-site spray drift from agricultural pesticide applications. Finally, 9.37% of homes in the vicinity of the two case studies were at risk from off-site spray drift from agricultural pesticide applications. Overall, the study concludes that for ground application, the estimated average pesticide drift in the two case studies was over 100 m from the intended target, while the estimated average pesticide drift was over 300 m from the intended target for aerial application.

Understanding sensitive resources’ exposure to organophosphate pesticide drift is important for evaluating the adverse health effects of the exposure and designing intervention strategies to prevent the exposure. The lack of exposure data hinders such investigations. To address this challenge, this study developed a method that takes advantage of the OpenAir package, AgDRIFT model, land use information derived from remote sensing data, and the county-level Cropland Data Layer (CDL) dataset.

The developed method provided more comprehensive data regarding organophosphate pesticide exposure drift than traditional exposure models and indicated significant potential in addressing pesticide-related environmental issues. This promising method is used to estimate organophosphate pesticide spray drift over Macon County, Alabama. This study is among the first endeavors to examine county-level high-resolution data of pesticide exposure drift. The estimations for drift, pesticide concentration, and deposition provided in this study were generated to reflect off-target movement predicted based on the US EPA report. Based on [10,33], a range of application sizes was created for each of the usage scenarios.

For ground boom and aerial simulation inputs, there are no variations between the results of this study and the US EPA analysis. The American Society of Agricultural and Biological Engineers (ASAE) class fine to medium/coarse droplet spectra were employed in both low and high boom applications. The US EPA, on the other hand, reported the 90th percentile estimate. The AgDRIFT user’s manual has the same format regardless of model version, and the Tier I default parameters are the same across AgDRIFT 2.0.05 and AgDRIFT 2.1.1. Tier I AgDRIFT [42,61]

In terms of ground applications, the averaging of the results of the field experiments conducted on the case studies was carried out at different wind speeds and according to the predominant wind direction at the time the model was run. The label states that 10 mph is the maximum wind speed that can be used to operate the model for agricultural pesticide ground applications, with the average wind speed in all experiments falling between 4 and 10 mph. Since pesticide labels do not specify droplet size or release height for ground applications, the AgDRIFT model was run with all four scenarios (low boom, medium/coarse, high boom and fine spray, high boom, and medium/coarse spray, low boom, and fine spray) to determine potential buffer distances. All drop size descriptions are based on standard parameters, with high and low denoting release heights of 2 and 4 feet and boom heights of 4 feet, respectively.

Previous studies have utilized statistical models that consider meteorological conditions and pesticide toxicity to calculate pesticide concentrations and impact [4,5,8,11,14,31,33,58,59,60,61,62,63,64]. These studies’ findings have been consistent in documenting evidence that pesticides sprayed in fields are drifting off-target and that the quantity of off-target drift is dependent upon various variables such as wind speed, direction, and other weather conditions, as well as the application type applied. A few previous studies examined how levels of pesticide residues were affected by residents’ proximity to treated farms. Data from 60 farming families in a rural area of Central Washington State were utilized by [48] to demonstrate that homes closer to treated orchards had higher dust concentrations than those farther away.

Notably, our study comprised 38 homes, 12 water bodies, and 6 acres of sensitive crops within 125 m of treated cropland in Shorter fields, and 17 homes, 10 water bodies, and 11 acres of sensitive crops within 125 m of treated cropland in Union Spring field, whereas Lu’s study included 45 (75% of the sample) homes within this distance. According to earlier studies, the amount of pesticide drift decreases dramatically as the field proximity increases. For instance, [11,33,59,63] used passive dosimeters to assess drift from 10 pesticides; spray was detected up to 150 m off-target for the ground application.

Previous models for spray drift deposition at the ground surface and airborne levels of pesticides close to application regions were created to predict the environmental destiny of these chemicals. Because residential exposure was not considered when these models were being developed, there were knowledge gaps in forecasting sensitive resource exposure drift, particularly at greater distances (>100 m for the ground application and >400 m for the aerial application). Experimental research on spray drift at greater distances (100 m to >400 m) was conducted on two sizable agricultural fields in Shorter and Spring Union in order to fill in these knowledge gaps.

The results of this study agree with the previous studies’ findings, as these studies provided a broad estimation of pesticide drift that can highlight potential areas of high pesticide exposure drift. Highlighting geographic areas of substantial risk can assist in determining areas of potential concern. The present study developed an application of the OpenAir and AgDRIFT models to estimate organophosphate pesticide spray drift over Macon County, Alabama, particularly in regions with limited pesticide data. Despite the study’s limitations, it provided an effective and reliable estimation of pesticide exposure drift in order to address the gap in pesticide use data conducted in most US states as well as generating high spatial resolution pesticide data influenced by minimal inconsistencies.

4. Conclusions

This study presented a novel method for estimating high-resolution information about organophosphate pesticide exposure at the field level. This proposed method revealed more extensive data concerning pesticide exposure than standard exposure models and demonstrated tremendous promise in examining pesticide-related environment and health concerns. The findings of the current study in Macon County, Alabama, may apply to other states with similar agricultural production systems, except for the lack of data on pesticide use. A probabilistic risk assessment model is necessary to simulate agricultural pesticide applications over time and space, including spray drift events and subsequent effects on unexpected sensitive crops, in states where complete pesticide use statistics are unavailable.

To reduce off-target air pollution on sensitive crops and rural residential areas and to increase production, an awareness of the risks of pesticide spray drift is essential. In Macon County, no-spray zones and approved techniques to reduce drift exposure are frequently necessary to reduce the risks and challenges of spray drift. In unfavorable wind conditions, agricultural pesticides’ spray drift to unintended land use and land cover (LULC) poses a significant risk to sensitive crops, residential areas, and bodies of water. To generate accurate supervised classifications based on spectral signatures for all relevant CDL and LULC and to derive maps for land use with a 15 m spatial resolution, remote generation was relied on. Geographic information systems (GISs) and modeling were integrated to enable field-level analysis and research of areas using high spatial resolution suitable for field-level satellite images such as Landsat.

The satellite image land cover classification provided information on the location of cropland fields, aquatic habitats, and their proximity to one another. The classification used spectral and spatial classification methods. The present study adopted an integrated data approach for measuring the accurate geographical distribution of crops in Macon County on a large scale (field level). The accuracy of spatial correlations established from satellite images enabled modeling of the probable drift of agricultural pesticides from their targets in Macon County. Predicting the current weather accurately is essential for mitigating pesticide drift, as the consequences will be more common the farther pesticides drift from their intended targets.

AgDRIFT was used to calculate the potential agricultural pesticide drift per 15 square meters of cotton and corn fields in two distinct areas: one in the north and one in the south of Macon County. Using information from the drift model about droplet size, application kind, and wind speed, the number of variations in potential drift deposition was also tracked. The efficiency of the model in estimating drift was determined, particularly in investigations that do not require a great deal of detail regarding the amount of pesticide applied.

The study found that 4320 ha, or 1.3% of the total cropland, was planted in sensitive crops that could be at risk from off-site spray drift from agricultural pesticide applications. In two case studies in Macon County, it was determined that the total crop area potentially at risk from near-field spray drift from ground and aerial applications was approximately 550 ha or 0.7% of the total area planted in sensitive crops. The type of application, LULC data, pesticide label data, and meteorological conditions all influence the possible drift distance for numerous studies. The authors speculate that the results obtained based on regulatory assumptions are specific to the current study.