Accepted: April 25, 2025 | Published Online: April 28, 2025

Citation: Peterson BA, Hutchinson JMS, Olstad TA, et al. (2025) Developing and testing an automated GIS toolbox for tracking low-level overflights above United States’ National Park Service units. Int J Astronaut Aeronautical Eng 10:067

Copyright: © 2025 Peterson BA, et al. This This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Abstract

Low-level overflights are a significant and pervasive noise source in many international parks and protected areas. Understanding overflight travel patterns is important for managers to establish management strategies and to ensure flight compliance with specified rules. Overflight travel patterns can be determined by collecting and analyzing Automatic Dependent Surveillance-Broadcast (ADS-B) data. However, these datasets are often large, elaborate, and time-consuming to analyze, map, and interpret. Therefore, the purpose of this study was to develop and test an automated Geographic Information System (GIS) toolbox that can process ADS-B datasets. The toolbox was designed for use by the United States National Park Service (NPS) and intended to be intuitively used by NPS employees who have minimal GIS training. The toolbox was tested at Grand Canyon, Great Smoky Mountains, and Acadia National Parks and was assessed by GIS experts and validated by managers from these park units. The toolbox advances NPS’s ability to quickly and efficiently analyze ADS-B data, which can help inform the development of management strategies and soundscape management plans related to overflights.

Highlights

• This study developed and tested an automated GIS toolbox for processing ADS-B data.

• The toolbox is designed for use by the United States National Park Service but has broader utility.

• The toolbox is designed to support the development of low-level overflight management strategies for the protection of park acoustic resources.

Keywords

ADS-B, Arcgis Pro, GIS, Overflights, National Parks, Parks and Protected Areas

Introduction

Low-level overflights are a challenging management issue at many parks and protected areas (PPAs) around the world. In United States (U.S.) national parks, the most pervasive noise source is caused by overflights [1,2]. Issues surrounding low-level overflights also exist in places such as Aoraki Mount Cook and Westland Tai Poutini National Parks in New Zealand [3] and across PPAs in Spain [4]. Air tours, a type of low-level overflight, have been a focal point of management and research [5]. The primary natural resource and visitor experience concern of low-level overflights is noise. Noise refers to undesired or unwanted human-caused sounds that degrade park soundscapes [6]. While the acoustic environment has intrinsic value as a park resource, it is also a natural resource [7] that is especially important for both wildlife and quality visitor experiences in PPAs [1].

Anthropogenic noise challenges wildlife because it is often loud, erratic, and increases competition for acoustic space [8]. In a synthesis of two decades of research documenting the effects of noise on wildlife [9] found terrestrial wildlife responses to noise begin at levels as low as 40 dBA, which is approximately the magnitude of nighttime sound levels in typical residential areas [10]. Documented effects of noise on wildlife behavior include shifts in vocalizations, foraging behavior, mating behavior, vigilance, movement, and physiology. It is concluded that noise is detrimental to wildlife and natural ecosystems, and noise effects range from the individual to the community level [9].

The acoustic environment is also integral for the visitor experience at PPAs. Natural sounds have been shown to improve health, increase positive emotional affect, and lower stress and annoyance [11]. Research at Denali National Park and Preserve (U.S.), found 67% of participants reported sound-related motivations were ‘very’ or ‘extremely important’ for their visit [12]. This same study found aircraft noise was consistently one of the most unacceptable sounds evaluated. Research at Muir Woods National Monument (U.S.) found that visitors deem noise unacceptable when it surpasses approximately 37 dBA [6]. Even more concerning is that only 11.3% of U.S. national parks have low audibility of anthropogenic noise (i.e., < 25% of time) [11].

At PPAs where low-level overflights are causing high levels of acoustic disturbance, it is important for managers to establish management strategies and to ensure flight compliance where applicable [13]. There are two basic strategies for management actions related to low-level overflights. The first, where governmental authorities allow, is the management of overflight travel patterns. For example, in the U.S., the National Parks Air Tour Management Act (NPATMA) requires the U.S. Federal Aviation Administration (FAA) to work with the NPS to manage air tours over NPS units. Additionally, PPA managers can also work with local airports and aircraft owners on voluntary measures. Second, PPA managers can respond to noise from aircraft with on-the-ground management actions. For example, managers can schedule interpretative programing around times and locations where aircraft noise may be most pervasive. Both of these management strategies can be informed by reliable and accurate aircraft travel pattern data.

Air tour management plans are park-specific and often rely on several strategies to mitigate air tour noise, including lateral standoff requirements (horizontal distance rules), minimum altitude requirements (e.g., fly at altitudes higher than 2,000 ft above ground level (AGL), quiet technology requirements (e.g., EC130 helicopter), specific flight routes, temporal restrictions (e.g., no air tours between 8:00pm to 8:00am), restrictions during particular events, and a maximum number of overflights [14]. For example, Grand Canyon National Park implemented a Special Flight Rules Area (SFRA) that extends vertically to 17,999 feet above mean sea level (MSL) and includes several flight-free zones with maximum elevations of 14,500 feet MSL and 8,000 feet MSL (NPS, 2007). Additionally, within the SFRA, all commercial overflights are required to use quiet aircraft technology (QT) by 2027 [15].

Despite public policy requirements in the U.S. and other countries for managing low-level overflights above PPAs, in most cases there is a lack of information about travel patterns beyond general observations by staff [16]. This information deficit includes exact routes, altitudes, areas with the highest concentrations of flights, numbers of air tours, and flight metadata, such as aircraft tail number, aircraft type, engine type, and aircraft owner. However, recent technology has been shown to be effective in assisting park managers to gain a better understanding of low-level overflight travel patterns.

Automatic Dependent Surveillance-Broadcast (ADS-B) data can be passively collected and analyzed to understand overflight travel patterns above PPAs. These data are unencrypted, publicly accessible, and can be collected using a terrestrial data logger [1]. An aircraft’s ADS-B system derives its position from the satellite navigation system, using Global Positioning System (GPS), and then broadcasts a radio signal twice per second [17] that other aircraft receives and monitors [18]. ADS-B systems are A utomatic because they don’t require work from the pilot, D ependent because they depend on gathered GPS and other aircraft data, ­­ S urveillance because they assist air traffic controllers with tracking and monitoring aircraft, and B roadcast because aircraft broadcast their position using a radio signal. When there are no obstructions, such as mountains, ADS-B systems can broadcast their signal up to 288 miles (463 kilometers), which is the distance that radio wave propagation is limited by the curvature of the planet [19]. Broadcasted information includes aircraft position and velocity, as well as a unique identification code [20]. An aircraft’s unique identification code can be linked to data in the FAA Releasable Database [21], which provides additional metadata, including aircraft unique identification code, aircraft type, engine type, and aircraft owner.

ADS-B transmissions provide operators the ability to navigate crowded airspace continuously and precisely [22] and are effective in remote areas that lack radar coverage, such as areas within the Gulf of Mexico and Alaska [23]. Benefits include real-time precision, shared situational awareness, visualization of where other aircraft are located, exact locations of hazardous weather and terrain, and current flight information (e.g., temporary flight restrictions) [23]. As of 1 January 2020 in the U.S., the FAA requires all aircraft entering designated airspace to have ADS-B technology (14 Code of Federal Regulations [CFR] § 91.225 and 14 CFR § 91.227; see also [1] for details about designated airspace).

Beeco and Joyce (2019) tested the feasibility of collecting ADS-B data over PPAs by testing loggers at three U.S. NPS units and found that a terrestrial data logger with an expansive skyward exposure is effective at collecting large ADS-B datasets that can include millions of aircraft waypoints. When these data are cross-referenced with the FAA Releasable Database using aircraft unique identification codes, more than 15 identifying non-positional attributes can be added to the dataset. Traditional manual methods of processing, cleaning, and analyzing ADS-B datasets to assess travel patterns can be time-consuming, prone to procedural errors, and may not be feasible to perform on some computer systems if the dataset is too large. Automated data processing and analysis workflows can simplify analysis of these datasets and achieve more dependable results yielding an increased understanding of low-level overflight travel patterns. Therefore, the purpose of this study was to develop and test custom tools in a Geographic Information System (GIS) “toolbox” that can process large ADS-B datasets and provide useful insights for managing low-level overflights. Additionally, the tools can analyze all ADS-B data inputs, join data with the FAA Releasable Database, identify multiple flight events conducted by the same aircraft during a single day, and create a “cleaned” dataset that removes flights irrelevant to research questions (e.g., flights above certain altitudes, flights beyond certain boundaries, or operated by specific entities).

Materials and Methods

Toolbox Overview

Data were collected by ADS-B loggers tested by Beeco and Joyce (2019). These data were then used for toolbox development and evaluation. The GIS tools created were developed for Esri’s (Redlands, CA) ArcGIS Pro version 3.4.0 GIS application [24] as a custom “toolbox” to be executed at the desktop level by entry-level technicians. The custom toolbox is comprised of six total script-based geoprocessing tools written in Python 3.9.16 and incorporating features from three Python libraries or site packages, specifically ArcPy, Numpy, and Pandas. The code for each script-based tool was written to be compatible with the default coding environment for ArcGIS Pro versions 2.x and 3.x with no requirement for modification using the ArcGIS Pro or conda package managers. 

The custom toolbox uses 49 unique ArcPy geoprocessing functions over six total tools which have a mean length of 194 lines of code. The longest (Tool # 2) and shortest (Tool # 3) scripts contain 239 and 114 lines of code, respectively. Of these 49 geoprocessing functions, 46 (94%) are available at the basic through advanced ArcGIS Pro license levels and do not require additional extensions. One geoprocessing function in Tool # 2 (Extract Values to Points) and two in Tool # 5 (Kernel Density, Band Collection Statistics) require the Spatial Analyst extension at any license level. 

At the time of development, only a limited number of geoprocessing tools in ArcGIS Pro (and functions in ArcPy) were able to use parallel processing. Where possible, this advantage was incorporated into the script-based tools. However, this involved only six implementations in the Buffer, Clip, and Kernel Density functions used in Tool # 2 and Tool # 5. Environment settings in the scripts of these tools specify a Parallel Processing Factor of 50%, enabling half of the cores available in a user’s computer workstation to be involved in the computations. Additional opportunities for tool performance boosts via parallel processing or eschewing ArcPy functions in favor of comparable approaches in Python libraries such as Geopandas, Fiona, or Shapely that are excluded from the ArcGIS Pro default coding environment were not considered.

The toolbox is also designed to analyze and output results for the 3-deminisonal aspect of these data. This was achieved by analyzing data using both above ground level (AGL) and mean sea level (MSL) altitude metrics. AGL altitude is a measurement of the distance between the ground surface and the aircraft and is useful for contextualizing how flights pass over variable terrain. MSL altitude refers to the altitude of an aircraft above sea level, regardless of the terrain below it, and is better suited for understanding aircraft patterns across larger spatial scales because the baseline (sea level) does not change.

Tool Descriptions

General descriptions for each of the six Python-based script tools in the ADS-B Overflight Analysis Toolbox follow. Detailed, Esri-style tool documentation is built into the toolbox tools and can also be accessed via a markdown file in the public GitHub repository or downloaded with the most recent software release.

Tool # 1 (Process Raw ADS-B Data Files) pre-processes raw ADS-B data collected from the logger to ensure proper data formatting and quality before constructing aircraft waypoint and flightline feature classes. It ingests and processes a single daily ADS-B file in native TSV format and returns a new daily output file in CSV format for later use. This tool can also be operated in “batch” mode within ArcGIS Pro to process several daily ADS-B TSV files in a single tool run. The tool performs basic structural checks on each file (e.g., presence of the required header row), formats fields to ensure proper data type is assigned, converts the original waypoint Unix timestamps to Python datetime objects in UTC, and calculates the time difference between sequential waypoints for each unique aircraft based on its International Civil Aviation Organization (ICAO) code included in the ADS-B broadcast. That time difference is used in conjunction with a user-defined tool parameter to identify unique flights by appending a zero-based index to the existing ICAO code to create a new flight id attribute and values (e.g., ICAO_0, ICAO_1, etc). A new flight during the same day by the same aircraft is indicated when two sequential waypoints have a time difference exceeding the Flight Duration Threshold (secs) which is set to a default value of 900 seconds (15 minutes). Identification of unique flights is important at U.S. national parks where some operators conduct several flights per day.

Additional quality control checks incorporated into Tool # 1 take advantage of the ValidFlags data field recorded by the data logger and its integrated uAvionix MavLink messaging protocol. The tool “unpacks” the binary data recorded for each waypoint in the valid flags field and removes any records having invalid latitude, longitude, and/or altitude values. The output of Tool # 1 is one or more cleaned daily CSV files each comprised of nine total attribute fields representing discrete waypoint records logged from aircraft ADS-B broadcasts (Table 1).

Tool # 2 (Create Waypoint and Flightline Feature Classes) ingests ADS-B data files in CSV format processed by Tool # 1 and produces aircraft waypoint and flightline geodatabase feature classes for data within a user-defined distance of a management unit polygon (e.g., PPA boundary) and below a user-defined altitude threshold. The tool converts the original altitude values broadcast by aircraft, in meters MSL, and writes them to a new field (alt_msl) in units of feet MSL. One of several required input parameters required by the tool includes a digital elevation model (DEM) for the area of interest which is used to calculate a new altitude above ground level (AGL) field (alt_agl) in units of feet. In addition to producing aircraft waypoints for each unique flight, flightlines are also created from waypoints having the same flight_id field value produced in Tool # 1. Additional attributes computed for the flightline feature class are length (Length Miles) and Sinuosity. Sinuosity is calculated as the ratio of the curvilinear length of the flightline and the Euclidean distance between the first and last waypoint comprising the flightline and is useful for identifying specific types of flights such as commercial airliners with straight line flight paths.

Further, Tool # 2 also performs two important table join operations between the output waypoint feature class and tables from the FAA Releasable Database (MASTER and ACFTREF). Using the aircraft ICAO code as the key field, joined fields from the MASTER table include aircraft N-NUMBER, TYPE AIRCRAFT, TYPE ENGINE, TYPE REGISTRANT, NAME (registrant’s name that appears on the FAA Application for Registration Form 8050-1), and MFR MODEL CODE (i.e., Aircraft Manufacturer Model Code). A single field called MODEL is joined from the ACFTREF table using MFR MODEL CODE as the key field. Attributes joined to the output waypoint feature class are also included in the corresponding flightline feature class. After the two table joins, it is possible to uncover several important details about a waypoint or flightline based solely on its ICAO code. For example, the aircraft transmitting ICAO code A895B0 was registered by the American Airlines corporation and is a fixed-wing multi-engine Embraer 145LR aircraft with turbo-fan engines and the tail number N652RS.

Tool # 3 (Merge Daily Waypoints and Flightlines) combines all daily waypoint and flightline feature classes produced by Tool # 2 into single point and line feature classes for later analyses. The tool considers any point or line feature class in the user-defined parameter input workspace as either a daily waypoint or flightline file for merging purposes. Waypoints are further filtered to identify and remove any duplicates which may be introduced when combining daily waypoint feature classes created from data recorded by two or more ADS - B data loggers within the study area.

Tool # 4 (Screen Suspected Non-Tourism Flights) creates a waypoint and flightline feature class containing features suspected of being unrelated to PPA tourism operations. It is included in the toolbox given this study’s overall goal of assisting the NPS in monitoring and managing air tours. Screening is based on parameters including the FAA Releasable Database attribute TYPE REGISTRANT, minimum and maximum Sinuosity values, minimum flight path length, and aircraft operator name. For the user-defined tool parameter Type Registrant Values , valid values are entered as one or more comma-separated numeric values from the TYPE REGISTRANT field in the MASTER file of the FAA Releasable Database (e.g., 1 = individual, 2 = partnership, 3 = corporation, 4 = co-owned, 5 = government, 7 = LLC, 8 = non-citizen corporation, and 9 = non-citizen co-owned). The default value is set to 5 for government flights. For the parameter Sinuosity Values, users enter a single or comma-separated numeric value for the minimum and maximum sinuosity values associated with flights to be excluded from further analysis. For reference, a sinuosity value of 1 is a straight line, while 0 represents a closed loop. The default minimum and maximum values are 0.10 and 0.99. The default value for minimum flight path length is 1 mile. Finally, for the parameter Aircraft Operator Name(s) , users can specify a comma-separated list of values for specific aircraft operator names such as those for commercial aircraft operators (e.g., AMERICAN AIRLINES INC, DELTA AIR LINES INC). Note that operator names must exactly match those appearing in the MASTER table of the FAA Releasable Database. The tool automatically removes these suspect waypoints and flightlines from the existing merged feature classes produced by Tool # 3 and creates new and "cleaned" merged waypoint and flightline feature classes. Any waypoints or flightlines meeting the screening criteria are also saved as separate “suspect” waypoints and flightline feature classes for further evaluation.

Tool # 5 (Summarize Waypoint Altitudes) allows users to analyze aircraft waypoints within a more restrictive buffer distance (e.g., NPATMA only applies to air tours within a half-mile of a park boundary) and then analyzes those waypoints across different AGL and MSL altitude intervals. Users are prompted by the tool to enter a minimum, maximum, and interval value for both AGL and MSL altitudes. The tool then determines how many total altitude classes for each altitude measure should be calculated by dividing the maximum altitude by the entered interval. Output includes two tables summarizing the frequency and percentage of total waypoints in each AGL and MSL altitude band and, to assist with visualizing airspace use, kernel density rasters for each AGL band. Kernel densities are only computed for AGL because this measure of altitude is commonly used to help define altitude standoff requirements when managing overflights. An optional correlation matrix can also be produced to illustrate the dependency between the kernel density rasters. This is implemented using the Band Collection Statistics function in ArcPy which generates a matrix similar to a typical correlation matrix [25].

Tool # 6 (Summarize Waypoints by Time, Operator, and Type) is the final tool in the toolbox. Given an input waypoint feature class, the tool produces six output tables that summarize the frequency and percentage of total flights by (1) hour of the day, (2) day of week, (3) weekday vs. weekend, (4) month, (5) aircraft operator, and (6) aircraft type. UTC times for waypoints produced by Tool # 1 are converted to local times prior to summarization and each table summarizing temporal data does so based on TIME of the first waypoint in the sequence of each unique flight in the input file.

Toolbox Testing

Development of the GIS toolbox occurred over a one-year period and was advanced through multiple rounds of testing by the authors that focused on improving user experience, tool performance, output data quality, and the utility of output data summary products for PPA managers. Specific performance measures included execution times, computer resource utilization, and real-world use case scenarios. The toolbox was also used alongside manual ADS-B data processing methods in ArcGIS Pro to compare and validate results during concurrent research investigating low-level aircraft overflights at several U.S. national parks. In addition to internal assessment of results, data produced by the toolbox and its outputs were shared with NPS personnel involved with managing low-level overflights for further qualitative validation. These park-specific analyses served as use case studies for evaluating toolbox efficacy given variations across the parks with respect to the number of data loggers used to record ADS-B signals, the frequency and type of overflights, as well as important landscape characteristics. NPS reports featuring analysis of ADS-B data using the automated GIS toolbox discussed in this manuscript can be downloaded from the NPS’s Integrated Resource Management Applications (IRMA) Portal at https://irma.nps.gov/DataStore/Collection/Profile/9561 .

Use Case Study Areas

Grand Canyon National Park (GRCA)

GRCA was established as a national monument in 1908, a national park in 1919, and was designated a United Nations Educational, Scientific, and Cultural Organization (UNESCO) World Heritage Site in 1979 [26]. In 2023, GRCA received 4.7 million recreation visits [27]. The park, located in northern Arizona, encompasses 1,218,560 acres (493,134 hectares) and contains 278 miles (447 kilometers) of the Colorado River corridor [28]. Elevations range from 2,400 feet (732 meters) at Phantom Ranch [28] to 8,803 feet (2,683 meters) on top of Point Imperial [29] GRCA includes an established SFRA used to evaluate the ability to track overflights through designated flight corridors. ADS-B data were collected using one data logger located on top of an air quality station tower at The Abyss on the South Rim (36.05864°, - 112.18361°) at an elevation of 6,862 feet (2,092 meters).

Great Smoky Mountains National Park (GRSM)

GRSM was established in 1934 and was designated a UNESCO World Heritage Site in 1983 and an International Biosphere Reserve in 1988 [30,26]. It receives the highest visitation of any of the 63 national parks, with 13.3 million recreation visits in 2023 [27] and receives a high volume of overflights. Spanning the border of North Carolina and Tennessee, the mountainous park is comprised of 522,427 acres (211,419 hectares) of rolling hills, with elevations ranging between 875 feet (267 meters) at Abrams Creek to 6,643 feet (2,025 meters) on top of Clingmans Dome. ADS-B data were collected on a summit (4,150 feet/1,265 meters) at Cove Mountain (35.69667°, - 83.6097°) and in a valley (2,120 feet/1,265 meters) at Elkmont (35.66444°, -83.5903°).

Acadia National Park (ACAD)

ACAD was designated as a national park in 1919 [31]. In 2023, the park received 4 million recreation visits [27]. Located along the coast of Maine with 60 miles of coastline, the park encompasses 50,000 acres (20,234 hectares) of the highest rocky headlands along the Atlantic coastline [31]. The highest and lowest elevations within ACAD are Cadillac Mountain at 1,530 feet (466 meters) and the coastline at 0 feet (0 meters) [31]. ACAD was chosen as a use case study site because it experiences air tours and is topographically distinct (compared to GRCA and GRSM) as a set of islands. ADS-B data were collected by one data logger located at McFarland Hill Air Resource Station near middle of Mount Desert Island (44.377247°, - 68.260864°).

Results

In the following paragraphs, results are shared that show strengths and weaknesses of the toolbox that were highlighted by the case studies or went undetected during development. Strengths are: 1) processing and analyzing large datasets, 2) retrieval of flight metadata, 3) removal of duplicate flight data, 4) examination of lateral trends across altitude intervals, and 5) production of tabular outputs. Weaknesses are: 1) slow processing speed for Tool # 2 and 2) challenging for first time users. For each of the case studies, identical toolbox versions and input parameters were used, which were based on NPS recommendations (flight types removed from analysis: government flights, major airlines, straight-line flights, flights shorter than 1-mile, and survey flights). Case study summary statistics are shown in table 2. Lastly, results are displayed for the processing performance of the toolbox.

Note. *GRCA dataset had 975 days of data and 95 days of missing data; GRSM had 837 days of data and 144 days of missing data; and ACAD had 491 days of data and 77 days of missing data. Data were not collected every day because of ADS-B logger technological failures.

Strength: Processing and Analyzing Large Datasets

GRCA had the largest dataset (237 GB) and the toolbox successfully processed its 43,117,246 waypoints, which shows that the toolbox is successful at processing and analyzing large datasets. Figure 1 shows trends of flights adhering to the General Aviation Corridors (GAC) within the SFRA and it also shows flights above every area of the park. The flights that used the GAC were overflights flying at a low altitude. Flights above the Flight Free Zones were major airlines or other flights at higher altitudes. This result shows that the toolbox is capable of tracking flights through corridors and at higher altitudes.

Strength: Accessing and Analyzing Flight Metadata

The toolbox succeeded in connecting to the FAA Releasable database to retrieve flight metadata, such as aircraft type and owner. The addition of the flight metadata subsequently served as an opportunity for further analysis of the GRCA dataset. After reviewing map outputs, GRCA managers noticed that some of the rotorcraft flights, which have significant noise impacts, were entering into a Flight Free Zone in the SFRA (Figure 2). The path of these flights appeared to be going to the popular park destination Phantom Ranch, which offers overnight dormitories, cabins, and food [28]. Using tail numbers of aircraft for administrative flights, the ADS-B dataset was queried and it was confirmed that these were flights approved for entry into this portion of the park for administrative activities. This type of analysis is important for the NPS because many park units have contracts with companies for administrative uses (e.g., construction) and have air tour management plans/agreements for specific operators or aircraft types.

Strength: Removal of Duplicate Flight Data

At GRSM, two ADS-B loggers were deployed at adjacent locations with different elevations (Cove Mountain and Elkmont), which resulted in duplicate data for many flights. This served as an opportunity to test if the toolbox could detect and remove duplicate data. This case study confirmed that the toolbox can remove duplicate data and then produce useful outputs. Figure 3 shows the output of the Kernel Density tool applied to the cleaned dataset with flight waypoints in grey. The ability to remove duplicate data is an important feature of the toolbox because deploying multiple ADS-B loggers is often necessary at mountainous PPAs in which terrain shielding is likely. Loggers can be deployed around landscape obstructions (i.e., on opposite sides of a mountain range) to prevent the exclusion of signals blocked by terrain shielding. To understand if a location is suitable for a logger to effectively collect ADS-B signals, a viewshed analysis can be conducted.

Strength: Examination of Lateral Trends Across Altitude Intervals

Two primary means for managing overflight compliance are lateral and vertical offsets. The toolbox uses the Band Collection Statistics tool available in ArcGIS Pro to produce a correlation matrix that quantitatively compares lateral flight trends across altitude intervals. This information is helpful to understand which altitude intervals have similar lateral flight trends and how these patterns vary across altitudes. For testing, GRCA was highlighted because it has the greatest difference between its highest and lowest elevations. A 500-foot AGL altitude interval was used. Table 3 shows that the lateral flight patterns were highly correlated between 1,001 feet AGL and 2,500 feet AGL and between 4,001 and 5,000 feet AGL. In general, for each column of the correlation matrix, the correlation coefficient decreases as altitude increases. This three-dimensional finding shows that overflights change their lateral patterns as they traverse altitude intervals, which likely results as overflights enter and exit SFRA corridors.

Strength: Production of Tabular Outputs

The toolbox produces quantitative tabular outputs of overflight trends to support mapped data and to help contextualize low-level overflights. This section shows tables from ACAD. Tables were chosen that clearly show quantitative data trends for AGL altitudes, MSL altitudes, and flights per hour of the day. Table 4 displays the number and percentage of waypoints across AGL altitude intervals and table 5 displays the number and percentage of waypoints across MSL altitude intervals. Tables 4 and 5 were important to compare because overflights above the ocean, such as at ACAD, will have similar altitude values calculated for AGL and MSL. Thus, table 4 and table 5 display similar trends. The trends are not exact because ACAD receives some overflights above ground terrain, as not all ACAD’s overflights are above the ocean. Table 6 displays percentage of overflights by hours of the day. This information is useful for scheduling terrestrial programs (e.g., interpretive programs) around times when overflights are most common.

Weakness: Slow Processing Speed for Tool # 2

Tool # 2 has the longest script with 239 lines of code. It performs several functions including ingesting CSV files (produced by Tool # 1), generating buffers around park units, clipping spatial data to buffer boundaries, retrieving metadata from the FAA Releasable Database, calculating waypoint AGL values using a DEM, calculating flightline sinuosity values, producing waypoints feature classes, and producing flightline feature classes. These operations streamline data processing and enhance spatial analysis capabilities. During toolbox development, it was challenging to anticipate how long Tool # 2 processing would take for large datasets. For GRCA the CSV files had a size of 141 GB, for GRSM the CSV files had a size of 74.6 GB, and for ACAD the CSV files had a size of 4.3 GB. For GRCA and GRSM this resulted in Tool # 2 conducting extended processing. Although Tool # 2 was successful with processing these large datasets, it became apparent that Tool #2 is a bottleneck in terms of processing time. For larger datasets, it is recommended for Tool # 2 to conduct processing in batches. These outputs (daily waypoint feature classes and flightline feature classes) can then be merged using Tool # 3.

Weakness: Challenging for First Time Users

Despite the intent to design the toolbox for NPS employees who have minimal GIS training, some new users struggled with file organization, inputs, and outputs. Subsequently, user instructions and guidance were developed as video tutorials and as written instructions. The video tutorials show users how to get started, how to use each tool, input parameters, and save outputs. Similarly, written instructions describe the same procedures.

Processing Performance

The performance metrics for each tool were evaluated by processing five daily ADS - B data files logged at GRSM in September and October of 2019. Tests were performed on a Dell XPS 8900 workstation with the Windows 10 64-bit operating system, an Intel Core i7-6700K CPU with 4 cores operating at 4 GHz, and 32 GB of RAM. The size of the daily ADS-B files ranged from a minimum and maximum of 137,314 and 637,114 rows, respectively. Given these differences in the number of rows (i.e., waypoints), total processing time for each tool varied but the average time is shown in figure 4A. For each tool, the measured CPU frequency, or number of cycles executed per second, was at the system maximum of 4 billion cycles per second. Additional key performance characteristics are presented in figure - 4B,figure - 4C,figure - 4D,figure - 4E,figure - 4F. It is important to note that the maximum RAM usage was 14.15 GB during operation of Tool 5. Also, Tool 5 is the only tool that produces raster output and does so at a rate of 285.3 seconds per 10 million cells.

Discussion

This study developed and tested an automated GIS toolbox that can process large, elaborate ADS - B datasets and join metadata from the FAA Releasable Database. Three park-specific analyses served as use cases for evaluating toolbox efficacy and were conducted at GRCA, GRSM, and ACAD. The ADS-B toolbox advances methodological processes by automating and standardizing procedures as compared to relying on manual techniques.

Benefits to the U.S. NPS

This study was specifically focused on improving the ability of the U.S. NPS to track aircraft above NPS units and was customized to their needs. However, the toolbox could be used by any PPA interested in understanding the airspace utilization above them. The ability to process large ADS-B datasets is an important step for understanding and managing low-level overflights at PPAs, including informing management strategies and ensuring operator compliance with management plans where applicable [5]. Additionally, it was important to develop an intuitive toolbox that can be used by park staff who might have minimal GIS training or understanding of programming methods.

Successful aircraft tracking requires processing large, elaborate datasets [5]. Prior to developing the ADS-B toolbox, the research team worked extensively for three years with the NPS to manually process and analyze ADS-B data of manageable size. Development of the toolbox took approximately a year, leveraging prior experiences and utilizing feedback from NPS staff.

The toolbox consists of several data cleaning processes allowing the NPS to focus on non-governmental and lower-level flights. Eliminating flights with specific parameters, such as government flights, like fire departments, civil air patrol, or commercial airliner flights, from further processing and analysis was possible by connecting to the FAA Releasable Database. Major airlines are cleaned by using a list function in Python. However, this list function requires that all major airlines are listed, which could be prone to human error due to omission or inconsistent naming. To overcome this, the toolbox also identifies straight-line flights by calculating sinuosity (Tool #4). Straight line flightpaths are characteristic of major airlines flying commuter routes over PPAs. The last cleaning technique to exclude flights that are less than one mile in length accounts for erroneous ADS-B data collection. Further cleaning procedures can be applied, such as excluding flights above an altitude threshold. For example, if managers want to focus on low-level overflights with the greatest impact on the acoustic environment, they might want to input a threshold of 5,000 ft AGL. Also, a smaller buffer can be applied around the PPA, such as a 0.5 - mile buffer. Both a 5,000 ft AGL altitude requirement and a 0.5 - mile buffer is of interest to national parks because NPATMA applies to overflights that are below 5,000 ft AGL and within a half mile of a park boundary [16].

Understanding aircraft travel patterns is important for several reasons, including assisting with the development of air tour management strategies and assessing air tour compliance with plans in place. Also, it supports the NPS Natural Sounds and Night Skies Division’s mission to protect and restore natural acoustic environments [32]. At national parks, a common strategy for managing air tours, where allowable, is to assign specific routes and flight altitudes. Using the toolbox, ADS-B data can be processed and analyzed to isolate specific tail numbers, check their travel patterns, and produce figures and tables for sharing the information internally and with operators. This information is useful for designing overflight lateral requirements, altitude standoff requirements, specific flight corridors, temporal restrictions, and maximum number of flights.

Benefits for Conservation

For conservation more broadly, the identification of anthropogenic noise and its sources can help assist in the management of the acoustic environment. Limiting or reducing noise is important for wildlife and providing opportunities for visitors to experience natural sounds. Broadly, ADS-B can inform park managers about the numbers and distributions of aircraft travel patterns, giving great insight into the spatial and temporal trends of probable noise disturbances from aircraft.

ADS-B data can also be linked with acoustic data using their respective timestamps. Using acoustic recording data, it is possible to identify when aircraft noise begins and ends, which can then be used to highlight the audible transit of an aircraft’s flight path that disrupted the acoustic environment, and to map the geographic scope of the noise impact [14]. Audible transits can be examined for altitude, horizontal velocity, vertical velocity, heading, and aircraft metadata. This integration of otherwise disparate datasets offers significant potential for extending understanding of how different aircraft types, engine configurations, and flight parameters disrupt the natural acoustic environment in PPAs.

Future research should aim to connect ADS-B data with acoustic data to model low-level overflights’ noise impact in three-dimensional (3D) space for understanding detection distances. A 3D model can be used to assist with formulating specific lateral and altitude standoff requirements. This type of research is currently underway [14] but is in its infancy, limited by complexities of implementing acoustic propagation models at the landscape scale and the potential for spatiotemporal misalignment of the acoustic and position records. Future studies should advance the ability to model 3D spatial impacts of large datasets.

Additionally, future studies may want to develop and add a seventh tool to the toolbox that analyzes space-time relationships of overflights. Furthermore, future research should examine how to quantitatively optimize ADS-B data logger deployment location. One of the most important considerations when deploying ADS-B loggers is terrain shielding. For this study, the locations chosen for the loggers all had expansive skyward exposure, yet terrain shielding was an issue at all three parks. If ADS-B data are collected again at GRCA, research should aim to deploy multiple loggers. At GRSM, data for the east side of the park indicated fewer overflights than elsewhere in the park, which is likely to be inaccurate, as there is an airport nearby.

The toolbox calculates several other variables that have not been used for research purposes, such as horizontal velocity, vertical velocity, and heading. These variables could be used to understand how air tours approach attraction sites and how they relate to noise impacts. This information could supplement air tour management plans.

Limitations

This study has some limitations. First, the toolbox was developed to ingest data collected by NPS designed ADS-B logging units. Additionally, some of the toolbox analyses are dependent upon the FAA Releasable Database. Perhaps not all PPAs have access to such resources. Second, the research team has not identified an automated method for cleaning survey flights, such as those used by park managers to survey vegetation or wildlife. This is a minor concern, as the number of survey flights is typically limited and the irregular flight paths can be visually detected. There’s a possibility that a GIS-based fractal analysis could successfully identify survey flights, but it might be more efficient to continue to use visual inspection methods rather than applying a lengthy effort towards developing another tool. Third, Tool # 2 is currently a bottleneck in terms of processing time; parallel processing cannot be applied to functions beyond Buffer and Clip at the time of publishing this manuscript. Also, Tool # 3 requires the use of parallel processing. Some of these limitations are self-imposed given the desire to produce tools for use within the ArcGIS ecosystem by entry-level GIS technicians. Others will likely be resolved as multi-core processing methods continue to proliferate in consumer GIS applications.

Conclusions

The development of the ADS-B toolbox provides an important advancement for tracking aircraft at PPAs. This product will be useful for future management needs regarding flight compliance and assisting with the development of air tour and soundscape management strategies. The toolbox is designed for future use by park managers who have minimal GIS training and need assistance processing these large and elaborate ADS-B datasets. Its design serves to simplify processing, analysis, mapping, and interpretation of the data. Managing low-level overflights is challenging, but as advancements are made, the ability for data-driven decision-making will increase, helping conserve PPAs for future generations.

Funding Statement: Funding provided by the United States National Park Service (Agreement P21AC10586).

Disclosure Statement: The authors have no disclosures or competing interests to declare.

---