Mapping and Spatial Analysis to Expand Rural Broadband Access

Abstract

High-speed broadband internet access is a critically important issue for many aspects of daily life, yet populations in rural areas are often unserved or underserved with reliable internet connectivity. Expanding broadband internet coverage in rural areas may have significant economic potential, especially since it enables precision farming which in turn increases yields, particularly for row crops such as corn and soybeans. This paper introduces methods that utilize GIS spatial analysis and remote sensing to assist in efforts to expand rural broadband access using case study counties in Illinois. Specifically, the methods presented here: (1) quantify current cropland production as well as future potential production in currently unserved or underserved rural areas; and (2) automate mapping of vertical assets from Light Detection and Ranging (LiDAR) data that may be utilized as high points to expand broadband coverage. Collectively, these methodologies may be used for policy advocacy and to inform the decision-making process as future broadband expansion initiatives are considered in rural areas.

Keywords:

• Broadband
• agriculture
• GIS
• spatial analysis
• remote sensing

Introduction and background

Rural broadband challenge

Access to broadband or high-speed internet is a critically important societal issue with multiple economic, health, social, and quality-of-life benefits. The importance of broadband access was highlighted during the COVID-19 pandemic when reliable, high-speed internet was essential for online education, remote work, telemedicine, and other purposes which resulted in a significant increase in home broadband usage in the early months of the pandemic (Brake 2020). Populations living in rural areas are among those that benefit significantly from access to reliable broadband internet, yet rural areas often are overlooked by internet service providers which commonly use metrics that focus on population density when determining priority areas for broadband expansion to maximize profits. As a result, a “digital divide” exists as many rural communities across the country are categorized as “unserved” or “underserved” by broadband access according to maps and definitions provided by the Federal Communications Commission (FCC).¹

Access to reliable broadband is critical to agriculture which is an important foundation of the economy in many rural areas. High-speed internet enables download and upload of data necessary for precision or smart farming which minimizes costs of fertilizer and pesticide inputs, reduces labor, and increases yields. One study by the U.S. Department of Agriculture (USDA) estimated a nationwide annual added economic benefit of up to 12% in production value for row crops alone ($16.8 billion) if the full potential of broadband penetration in rural areas can be achieved for precision agriculture (USDA 2019). Another study estimated that doubling the number of rural households with adequate broadband speeds would increase annual yields of corn by 3.6% and soybeans by 3.8% (LoPiccalo 2021). In addition to increased yields of row crops, broadband access in rural areas can also enable or improve many other aspects of smart farming (Yaacoub and Alouini 2020), such as real-time weather monitoring, grain storage management, and livestock monitoring. As well as promoting economic growth through smart farming in rural areas, broadband internet may contribute to sustainable agriculture by reducing fertilizer and pesticide applications and conserving water usage. Despite these many benefits, broadband coverage on many American farms remains inadequate to realize this potential (Arnold 2021).

The Infrastructure Investment and Jobs Act enacted by federal legislation in November 2021 will provide an infusion of $65 billion toward expanding broadband coverage across the United States (National Telecommunications and Information Administration 2023). Funding for expanding broadband capacity from the Infrastructure Investment and Jobs Act will be administered through programs such as the Broadband Equity, Access, and Deployment (BEAD) Program which will distribute $42.45 billion across the states, territories, the District of Columbia (D.C.), and Puerto Rico with a focus on unserved and underserved areas (National Telecommunications and Information Administration 2023). The infusion of significant federal funding into broadband infrastructure for the nation provides much opportunity to close gaps in high-speed internet coverage, including in rural areas that previously have been overlooked by many privately operated broadband internet service providers due to the high cost of infrastructure to reach low numbers of households. Yet, there is need for metrics and analyses to determine return on investment and to inform broadband infrastructure that maximizes expansion of coverage.

Broadband access is fundamentally geographic in nature, and expanding coverage of high-speed internet service may benefit significantly from spatial analyses. First, accurate maps are needed to determine which areas are currently served, as well as those areas that are unserved or underserved, to understand the gaps in coverage. Broadband status maps provided by the Federal Communications Commission (FCC) and other sources through state-level maps² are a common source for understanding the current geographic extent of broadband coverage in an area. Such maps classify areas that are served, underserved, or unserved by broadband coverage using standard definitions developed by the FCC.³ Second, the return on investment for broadband (e.g., economic, environmental, social) may be estimated by quantifying future potential within an unserved or underserved geographic space if it were to be fully served. Third, broadband infrastructure requires significant financial investment, so identifying ways to expand fixed-wireless broadband that would maximize geographic coverage in a timely and cost-effective manner is an important priority. These geographic questions may be addressed in part by the GIS, remote sensing, and spatial analysis methodologies proposed in this paper.

Objectives and guiding principles

This project addresses two primary challenges from a geographic approach to understand current limitations of fixed wireless broadband access in rural areas and to inform future investment of resources to expand broadband coverage. First, as priority areas are identified for broadband expansion, estimates of the return on investment for broadband in rural areas that are currently unserved or underserved will be beneficial to justify expansion of broadband coverage, especially in rural areas with low population densities but high agricultural productivity. Expanding on previous efforts to estimate the economic advantages of broadband expansion in rural areas, we develop a methodology using readily available GIS datasets to quantify crop yields in unserved or underserved rural areas in selected Illinois counties, and to forecast predicted gains with expanded broadband coverage. Resulting maps and statistics are proposed as an economic metric that may be considered to quantify the value-added impact of broadband expansion to the agricultural economy in rural areas.

Second, as broadband investment is allocated for rural areas, existing tall structures or vertical assets (e.g., grain silos) with adequate line-of-sight may be critical for expanding coverage in a timely and cost-effective manner (Denson 2021). Such vertical assets in unserved or underserved rural areas may serve as installation points for fixed wireless internet repeaters and other networking infrastructure to expand wireless signal propagation through “middle-mile” and “last-mile” solutions that connect end users (e.g., farm) to an existing broadband backbone network. Use of existing vertical assets has advantages due to the savings in cost and time that would not be possible with construction of new towers. However, despite the potential, detailed maps and databases of vertical assets in rural areas are not readily available. Using existing Light Detection and Ranging (LiDAR) data and aerial imagery, we develop a method for automatic extraction of existing vertical assets that may be considered as candidates for installation of middle-mile and last-mile fixed wireless infrastructure that connects with existing optical fiber networks to expand broadband penetration into rural areas.

The research leverages geographic information systems (GIS), satellite remote sensing, spatial analysis, and digital mapping to develop methodologies that may address these objectives using selected counties in Illinois as a study area. The project was guided by several important principles. First, both objectives are supported by methods that are repeatable and may be applied to other rural communities. Second, methodologies leverage the use of free or open-source data and software as much as possible as well as a toolkit of guided instructions. Third and most importantly, project methodologies were developed for actual implementation by rural communities for future broadband planning, so we also report here on how resulting maps from these analyses have already been incorporated into the rural broadband expansion efforts for five counties in Illinois.

Literature review

We connect our research to related efforts specifically in the areas of the digital divide and related examples of geographic gaps in services, the utility of maps and GIS methods to understand and address access to broadband internet, and the broader use of maps and GIS to inform public policy and for advocacy purposes.

The “digital divide” refers to the uneven access to the Internet and digital telecommunications and may be applied specifically to include disparities in broadband access in rural areas of the United States. More broadly, the term highlights the presence of geographic gaps or voids in access to a particular resource, service, or amenity, and has parallels to similar divides that have been examined for topics ranging from access to healthy foods (e.g., Hallett and McDermott 2011; McEntee and Agyeman 2010), public transportation (e.g., Kostelnick et al. 2020), and emergency response services (e.g., Lee et al. 2021). Studies have characterized the digital divide through analyses that vary by location, geographic scale, and type of digital telecommunication, Internet service, or speed/quality of service (e.g., Buys et al. 2009; Grubesic and Helderop 2022; Grubesic and Murray 2002; Manlove and Whitacre 2019). Key findings in several of these studies are the inequities that exist in broadband availability and speed between urban and rural areas (Riddlesden and Singleton 2014; Whitacre 2008), within a city (Reddick et al. 2020), and by socio-economic and demographic factors (Grubesic and Murray 2002; Oyana 2011; Prieger 2015; Rains 2008; Reddick et al. 2020; Riddlesden and Singleton 2014).

Numerous studies have demonstrated the utility of GIS analyses to visualize the current extent of the broadband digital divide, evaluate the quality of broadband coverage maps, estimate the overall impact of gaps in broadband access, and forecast solutions for closing the divide by expanding coverage. For example, Oyana (2011) used GIS in a supply and demand approach to compare broadband coverage maps with demand for service in southern Illinois counties to inform future expansion of broadband in the region. Grubesic (2010) incorporated spatial statistics to identify change over time in broadband providers and core broadband regions in the United States based on FCC Form 477 data for ZIP code areas, and to visualize results from a mathematical model that estimated efficiency for acquiring future broadband service. In another study that utilized GIS to understand gaps in broadband service, Dong and Statham (2020) used a Bayesian spatio-temporal statistical model to forecast broadband faults or disruptions to service.

Other studies have incorporated GIS analyses as a spatial decision support system to estimate the impacts of future broadband growth. For example, Lehtnonen (2020) estimated the impact of broadband expansion on rural population development in Finland through a regression analysis of population and broadband availability using GIS databases. Sawada et al. (2005/2006) developed a GIS-based method to determine broadband market potential for siting wireless internet service provider (WISP) communication towers as well as estimated numbers of households served by the expanded coverage. Grubesic and Murray (2002) conducted a spatial analysis of potential broadband service areas for Franklin County, Ohio, provided by wireless hubs with fixed locations through a simple buffer analysis that defined coverage. Their analysis provided summaries of future expansion of coverage for several demographic variables and was followed by a more complex analysis that incorporated the maximal covering location problem to maximize broadband coverage for the fixed locations based on total households, income, and education.

Studies also have highlighted challenges associated with broadband maps more broadly for accurately depicting broadband coverage, which remains a persistent challenge given the inherent uncertainties prevalent in national broadband maps (Grubesic and Helderop 2022). Early studies based on an analysis of the National Broadband Map and related maps derived from FCC Form 477 (e.g., Grubesic 2012a; Grubesic 2012b) documented challenges such as inconsistencies in data quality and completeness, bulkiness of data for analysis and visualization, and overestimates of coverage resulting from provider data spatially aggregated by ZIP Code or Census geography units which under previous FCC definitions only required the presence of one household and one provider to be designated entirely as served by broadband. Recent studies (e.g., Grubesic and Helderop 2022; Whitacre and Biedny 2022) have noted improvements to newer iterations of national broadband maps through the broadband serviceable location fabric which factors in agricultural and other nonresidential structures for broadband coverage estimates and requires more detailed information on provider service areas (Whitacre and Biedny 2022).

From a broader perspective, GIS and maps serve an indispensable role in public policy for decision-making, resource allocation, and monitoring return on government investment in an efficient and transparent manner (Thomas and Humenik-Sappington 2009). Broadband accessibility serves an important example of how GIS may address the geographic issues associated with an important societal issue with much government involvement. Notably, many of the aforementioned studies (e.g., Grubesic and Helderop 2022; Oyana 2011; Riddlesden and Singleton 2014; Sawada et al. 2006) have acknowledged the implications of their GIS-based analyses toward guiding efficient and equitable policy for broadband expansion. It also is important to highlight the role of maps and GIS for advocacy purposes at the grassroots level to self-empower groups on the other side of the digital divide (McMahon, Smith, and Whiteduck 2017). As priority areas for broadband expansion are considered as a result of the Infrastructure Investment and Jobs Act, maps and GIS analyses arguably will become even more important for influencing broadband policy.

Quantifying agricultural production in areas unserved and underserved by broadband

Overview

Precision agriculture provides several advantages to crop productivity, including the potential to increase yields (LoPiccalo 2021). Access to high-speed broadband internet is central to networked farms that utilize technology to collect and analyze data to support smart, data-driven agricultural practices (Arnold 2021). Increased farm yields facilitated by smart farming that is enabled by broadband access may in turn boost the overall economic growth in a predominately rural county (Spell and Low 2021), and therefore may be viewed as a positive return on investment for broadband infrastructure that expands high-speed internet coverage. However, broadband internet service providers often focus on areas with high population densities when priority areas are identified for expansion in order to maximize profits, which results in many rural areas with little or no broadband coverage despite this economic potential due to the high-cost-to-deployment to reach fewer customers. Federal funding such as the BEAD Program will help to alleviate the high-cost-to-deployment by prioritizing funding for costly infrastructure in unserved areas that will encourage providers to expand into rural areas.

Row crops such as corn and soybeans are an integral part of the Illinois economy. In the 2022 growing season, Illinois farmers harvested 10.6 million acres of corn for an estimated 2.268 billion total bushels, and 10.75 million acres of soybeans for an estimated 677 million bushels (USDA NASS 2023). We argue that an important consideration as broadband priorities are determined should be the additional economic gain that might result in rural areas due to additional crop production resulting from precision farming that requires fast and reliable internet speed for data upload and download in farm operations.

We introduce a geographic and economic method of analysis to estimate corn and soybean production on areas unserved or underserved by fixed wireless broadband internet access based on the estimates by LoPiccalo (2021), and then present results applying the methodology to a case study of fifteen Illinois counties for the 2021 growing season. Corn and soybean production in areas with both unserved and underserved broadband availability were estimated, along with the economic value of production and an estimate of potential gain in production if broadband availability is expanded in the future to support additional precision farming in these rural areas. We focus only on expanded crop yields and not on estimating additional gains to the agricultural economy that may accompany broadband expansion in rural areas given the wide range of short and long-term impacts that have been reported in other studies (Bai, Wang, and Jayakar 2022).

Methods

Fifteen counties in Illinois (Bond, Christian, Clinton, Edgar, Hancock, Henry, Iroquois, Kankakee, LaSalle, Macoupin, McLean, Ogle, Schuyler, Washington, and Wayne) were chosen for the analysis, representing different geographic regions in the state (Figure 1). The highest corn and soybean producing counties in each general region of the state were chosen to maximize focus on agricultural lands, with additional counties added due to the presence of strategic project partners.

Figure 1. Counties included in the analysis.

Fixed wireless broadband availability estimates are available from the Illinois Broadband map.⁴ Broadband availability layers for the analysis were current as of November 2021 and derived by Connected Nation through a combination of FCC Form 477 data along with finer granular data to improve detail in broadband coverage (Figure 2). Unserved areas are defined as those areas with less than 25 Mbps download and 3 Mbps upload or no service available, while underserved areas are defined as those areas with a minimum 25 Mbps download and 3 Mbps upload, but less than 100 Mbps download and 20 Mbps upload.

Figure 2. Areas unserved, underserved, and served by fixed wireless broadband in Illinois as of November 2021. Source: Connected Nation, Connect Illinois.

The Cropland Data Layer (CDL) for the 2021 growing season, available from the U.S. Department of Agriculture (USDA), was utilized in the analysis to delineate areas of corn and soybean cropland in each county for the selected year to correspond with the broadband availability maps for the same time period (Figure 3). The Cropland Data Layer (CDL) is a crop-specific land cover dataset at a native spatial resolution of 30 meters that is derived annually by the USDA from Landsat 8 and other satellite imagery sources (USDA 2022). Corn and soybeans are the predominate row crops throughout Illinois which are included as separate categories in the CDL. Accuracy of crop type classifications in the CDL is generally 85 – 95% (USDA 2022).

Figure 3. Corn and soybean cropland in Illinois, 2021. Source: USDA NASS, Cropland Data Layer, 2022.

First, geographic overlay identified only areas in each county that were unserved or underserved and also were classified as corn or soybeans in the 2021 growing season according to the CDL (Figure 4). Next, total number of acres were calculated in the GIS using an equal-area map projection for each of the four combinations of unserved or underserved broadband internet access and corn or soybean classification in the CDL (Figure 5). Average corn and soybean yields in bushels per acre for each county in 2021, available from the USDA National Agricultural Statistics Service (USDA NASS 2023), were used to estimate total yields for all acres of corn and soybeans by multiplying the number of acres of corn and soybeans, respectively, by the average yields for each county (Figure 5).

Figure 4. Geographic overlay to identify areas of corn and soybean in underserved areas for the 2021 growing season in McLean County, Illinois. In this example, the county boundary is first identified (A), followed by underserved areas in the county (B). From the Cropland Data Layer (C), areas where corn or soybeans intersect underserved areas are identified (D).

Figure 5. Calculations of current corn and soybean production using underserved areas for the 2021 growing season using McLean County, Illinois, as an example. First, total acres of corn and soybean production displayed on the map are calculated. Next, current production is estimated by multiplying total acres by average yield per acre in bushels based on the county average for the growing season.

Additional potential production on acres in unserved or underserved areas was estimated with assumed additional average yields of 3.6% for corn acres and 3.8% for soybean acres following the findings of LoPiccalo (2021). The estimates assume a doubling of the number of farm households with minimum 25 Mbps download and 3 Mbps upload speeds which would expand broadband infrastructure to support precision agriculture. Additional production estimates are based on the assumption that farmers would maximize use of precision agriculture if broadband availability was improved in current unserved and underserved areas (Figure 6). Finally, state-wide estimates for the average price per bushel for both corn and soybeans were used to assess overall economic value by multiplying the estimated additional production in bushels by the average price per bushel. The average price for soybeans in Illinois in 2021 was $10.69/bushel, while the average price for corn was $5.40/bushel.

Figure 6. Future corn and soybean production estimates in underserved areas for the 2021 growing season using McLean County, Illinois, as an example. First, additional future production is estimated for corn and soybeans by assuming a 3.6% increase in corn production and 3.8% increase in soybean production following LoPiccalo (2021). Next, economic value of the additional production is estimated by multiplying the additional production by the average price per bushel of corn ($5.40/bushel) and soybeans ($10.69/bushel) to estimate total additional production value. For McLean County, the estimated additional production is approximately $5.7 million in corn and $3.9 million in soybeans for the 2021 growing season.

Results

Results from the analysis estimate that nearly $110 million in total could have been added in 2021 to the agricultural economies of the fifteen counties in the study area from additional corn ($68.3 million) and soybean ($41.5 million) production alone if full broadband coverage were to be realized (Table 1). The analysis only estimates additional crop yields, and not any economic gain associated with decreases in expected agricultural labor or cost savings due to reduced use of fertilizer and pesticide inputs, and for this reason likely underestimates the actual economic impact of broadband to the agricultural economy of each county. It is important to note as well that the estimates are for one growing season only, and similar gains could be expected for each subsequent growing season.

Table 1. Sum of estimated additional production on unserved and underserved areas in selected Illinois counties for corn and soybeans in the 2021 growing season.

County	Corn	Soybeans	Total
Bond	$2,603,756	$1,770,881	$4,374,637
Christian	$852,325	$601,216	$1,453,541
Clinton	$3,157,735	$1,728,910	$4,886,645
Edgar	$6,344,316	$4,312,338	$10,656,654
Hancock	$4,983,905	$3,296,469	$8,280,374
Henry	$359,883	$177,650	$537,533
Iroquois	$11,527,355	$6,584,935	$18,112,290
Kankakee	$4,316,864	$2,083,794	$6,400,658
LaSalle	$7,992,687	$5,153,209	$13,145,896
Macoupin	$6,038,996	$4,310,802	$10,349,798
McLean	$7,291,381	$5,035,828	$12,327,209
Ogle	$6,969,762	$2,386,371	$9,356,133
Schuyler	$1,283,122	$875,981	$2,159,103
Washington	$2,795,873	$1,699,859	$4,495,732
Wayne	$1,772,628	$1,453,890	$3,226,518
Total	$68,290,588	$41,472,133	$109,762,721

In addition to these statistics, a map layout was created for each county to facilitate visualization of the overall economic impact on the county, especially for rural broadband advocacy efforts by county stakeholders (Figure 7). Each map layout includes two maps displaying current corn and soybean production in both unserved and underserved areas, along with a table of statistical calculations for the county. The purpose of the map layout is to provide visuals to demonstrate the geographic extent of corn and soybean fields that are currently unserved or underserved by broadband coverage through maps, while also providing statistics in tabular form to indicate the overall economic impact in dollars.

Figure 7. Map layout for Hancock County, Illinois, with calculated additional corn and soybean production. Source: USDA NASS, Cropland Data Layer, 2022. Esri. “World Topographic Map” [basemap]. 1:14,000,000. February 1,2024. https://basemaps.arcgis.com/arcgis/rest/services/World_Basemap_v2/VectorTileServer (March 20, 2024).

Discussion

Estimates from the analysis indicate a substantial amount of agricultural gain with improved broadband availability for rural areas that are currently unserved or underserved. The estimates provided here only consider additional yields from corn and soybean production, and do not quantify gains for other agricultural industries (e.g., dairying, livestock) that also may benefit from broadband availability. The estimates also do not estimate broader economic gains, such as increased overall Gross Domestic Product (GDP) or employment growth which also accompany broadband expansion in rural counties (Spell and Low 2021). Likewise, the estimates do not account for labor savings and time efficiency in operations for farmers, who may otherwise need to transmit data manually in precision agriculture operations in the absence of reliable internet. More broadly, the estimates do not take into account the non-economic benefits of broadband, such as environmental gains that may result through increased precision agriculture due to reduced use of fertilizers and pesticides (Arnold 2021). Finally, it is important to emphasize that these results are a projected gain for a single growing season, and longer-term economic impact could be calculated by repeating the analysis for additional years.

The results should be interpreted with care in the context of a number of important assumptions and limitations. The analysis is based on “best available” maps of broadband availability, and any geographic overestimates or underestimates in broadband unserved or underserved areas will impact the estimated gains. The analysis assumes that farmers would utilize precision agriculture fully if broadband were available at the farm, and farmers would invest in the necessary equipment to utilize broadband coverage to maximize production with precision farming. Although it is difficult to predict the number of farmers that would adopt precision farming, a survey by the United Soybean Board found that nearly 87% of farmers are considering or planning to use additional data in their operations, while 60% of farmers believe that current broadband service for their farm is inadequate (Arnold 2021).

Ultimately, agricultural gains accompanying broadband are challenging to predict given the number of variables, so we would advocate for a future study to compare the projected yields presented here with actual yields if broadband coverage were to improve in the study area counties. Given the significant economic potential in rural areas that is often overlooked and has evaded quantification, we propose this methodology as an important economic metric that should be considered to provide a general picture of the value-added impact of broadband expansion in rural areas.

Mapping vertical assets

Overview

Expanding broadband coverage to farmers and residents in rural areas is a complex challenge that will involve many different technologies and creative strategies to ensure every rural residence has adequate service. Although fiber-optic cable connected directly to each farm and rural residence is an ideal solution for overall download and upload speed performance, it is challenging to implement in many rural communities because of overall cost, time to deployment, and land ownership issues (e.g., right-of-way). An alternative solution is to connect tall structures, or vertical assets, to an existing broadband backbone from which a signal can propagate wireless coverage to multiple farms with internet repeaters that utilize fiber and fixed wireless as a hybrid solution (Figure 8). In the context of broadband access, the term “middle-mile” refers to the segment of a telecommunications network linking a core network to the local area network, or the last-mile connectivity that physically reaches the end-user’s premises for delivering telecommunication services to customers. Such an approach is particularly beneficial in environments with flat terrain and relatively few trees, where line-of-sight for wireless signals may be beneficial for spanning middle-mile and last-mile connections to allow tractors and farm implements to download and upload data needed to support precision agriculture (Figure 9).

Figure 8. Broadband miles along a network, and the role of vertical assets for connecting farms. Figure adapted from Platte Institute, A Blueprint for Better Broadband in Nebraska. Source: https://platteinstitute.org/a-blueprint-for-better-broadband-in-Nebraska/.

Figure 9. A Hypothetical solution for expanding rural broadband coverage with existing vertical assets. (A) Candidate vertical assets in a rural area; (B) Thiessen polygons for the vertical assets, or the most efficient areas of coverage for each respective vertical asset that provide complete wireless coverage of the area with no overlap; and (C) fixed wireless broadband coverage provided by the vertical assets based on internet repeaters with various geographic ranges as depicted by the circles. Note the hypothetical solution above does not account for factors such as terrain, tree cover, and other signal obstructions.

A significant expense associated with line-of-sight wireless networking is construction of towers upon which equipment may be mounted. The cost as measured in both money and time can be reduced significantly if existing structures, such as grain silos, water towers, radio towers, buildings, and other vertical assets, are used instead of erecting new structures. The taller the vertical asset, the higher likelihood of an unobstructed line-of-sight which increases the geographic coverage of broadband service. Key to the use of existing vertical assets are accurate maps to decipher both the location of suitable vertical assets along with the respective height of each asset for incorporation into line-of-sight analyses. However, for many rural communities, such vertical asset maps do not exist already. The Federal Aviation Administration (FAA) maintains a publicly available list of tall structures mainly for aviation purposes called the Digital Obstacle File (DOF) (Figure 10). A limitation of the DOF for mapping vertical assets for broadband purposes is that it only includes structures that are over 200 feet tall, except in the close vicinity of an airport in which case shorter structures are also included, but does not include the wide array of other vertical assets in rural areas.⁵

Figure 10. Vertical assets in the FAA’s Digital Obstacle File (DOF) for McLean County, Illinois, are displayed in blue. Areas with high clusters of vertical assets are wind farms with wind turbines or are in close proximity to the Central Illinois Regional Airport. Additional vertical assets greater than 50 feet in height that were identified from the proposed methodology are displayed in red. Source: FAA, 2019. Esri. “World Topographic Map” [basemap]. 1:289,817. February 1, 2024. https://basemaps.arcgis.com/arcgis/rest/services/World_Basemap_v2/VectorTileServer (March 20, 2024).

Given the gap in readily available maps of vertical assets in rural areas, we present a methodology for mapping existing tall structures that might serve as candidate locations for wireless internet stations. The method employs visible/near-infrared aerial imagery and Light Detection and Ranging (LiDAR) data, both of which are collected regularly by state governments in partnership with federal agencies and are often freely available throughout the United States. The methodology may be adapted by communities who wish to map and identify vertical assets in their own communities as part of the broadband planning process.

Methods

Five counties in Illinois (Edgar, Hancock, McLean, Ogle, and Schuyler) were chosen as a study area to develop and refine the vertical assets mapping methodology (Figure 11). The selected counties also are representative of the varied terrain characteristics, tree density, and hydrography coverage across the state. Additionally, all five counties simultaneously participated in a Broadband Breakthrough workshop series sponsored by the Benton Institute for Broadband & Society from January through May 2023 which provided a firsthand opportunity to observe how the completed vertical assets map could be integrated into the ongoing rural broadband planning initiatives for each county.

Figure 11. Counties included in the analysis.

The methodology identifies structures taller than 20 feet above ground level that are not trees and are at least 50 meters from a tree (Figure 12). The assumption here is that vertical assets that are shorter than 20 feet or near another tall object would not have the necessary line of sight to serve as effective candidates for broadband infrastructure, although these thresholds could be adjusted to fit the characteristics of a specific location. Light Detection and Ranging (LiDAR), which is an active sensor that uses light beams to create highly detailed and accurate terrain and surface maps, is used to extract vertical heights of tall objects by subtracting the ground elevation from the surface height (Figure 13). All LiDAR data utilized for the study area counties are freely available through the Illinois Height Modernization Project and were captured with an aggregate nominal pulse spacing of <0.71 meters and a vertical RMSE of <0.10 meters.⁶ LiDAR data for each county were converted to raster Digital Elevation Models (DEMs) and sampled to a spatial resolution of 4 meters. The Normalized Difference Vegetation Index (NDVI) (Rouse et al. 1974), a reliable estimate of photosynthesis, was used to identify and remove trees from the tall objects extracted from the LiDAR data as these are not deemed suitable as vertical asset candidates (Figure 14). Healthy vegetation such as trees absorb most incoming red lightwaves for photosynthesis and reflect most incoming near-infrared light. NDVI was calculated from near-infrared imagery available from the National Agriculture Imagery Program (NAIP) from the USDA and acquired from the Illinois Geospatial Data Clearinghouse with a spatial resolution of 0.6 m (USDA Farm Service Agency 2019).⁷ NDVI was calculated from near-infrared imagery using the formula NDVI = (NIR – R)/(NIR + R), where NIR is near-infrared light and R is red light (Rouse et al. 1974). Resulting NDVI values were scaled from 0 (no vegetation) to 1 (high vegetation). Through trial and error, an NDVI threshold of 0.1 was selected as the threshold to identify the deciduous trees that are commonly found in Illinois, but this value could be adjusted by geographic region. Since some trees have artificially depressed NDVI values caused by the shadow of neighboring trees, any tall object within 50 meters of a tree was removed. The centroid of each vertical asset was calculated and includes the mean and maximum heights of the asset stored as attribute values (Figure 15). Data clean-up was necessary as a final step to remove centroids that are not viable vertical assets, which included erroneous centroids especially near forest edges where NDVI values were impacted by shadows.

Figure 12. Overview of the steps and thresholds employed in the methodology to extract vertical assets.

Figure 13. Subtraction of LiDAR surface height (A) and ground elevation (B) to find tall objects (C).

Figure 14. NDVI to identify trees (in red) for removal from the tall objects that were identified in the previous step.

Figure 15. Examples of two vertical assets mapped as centroids or points with the mean and maximum height displayed for the silos.

Results

Vertical assets were mapped using the methods described above for Edgar, Hancock, McLean, Ogle, and Schuyler Counties as case study counties to develop and refine the methodology. A web map was created with the vertical assets for each county to facilitate ease in identifying vertical assets on high-resolution imagery and other basemaps for consideration in broadband planning efforts (Figure 16). The web map also includes an interactive tool to filter vertical assets based on a specific height range, and includes a catalog with example vertical assets found in the county, a picture of each asset type, and height ranges that may be used to identify a specific type of asset through the filter tool available in the web map.

Figure 16. Screenshot from a web map for Hancock County with vertical assets (red dots) filtered to show only vertical assets 50 feet or taller (available at https://isu-geomap.maps.arcgis.com/apps/webappviewer/index.html?id=c5fbe05e5af34005a7ed0b0b6cdae979).

Discussion

The methodology provides a cost-effective and accurate method for mapping vertical assets that may be considered as candidate locations for middle-mile and last-mile infrastructure to expand broadband coverage in unserved or unserved areas through a hybrid approach that relies on both fiber and fixed wireless technologies. The methodology relies on freely available datasets commonly available for the entire United States, and may be implemented in other rural areas that seek to leverage existing vertical assets to expand broadband coverage.

As a next step, line-of-sight analyses may be conducted with the vertical assets identified from the method described here to account for factors such as terrain, tree cover, and other obstructions to determine optimal selection of candidate vertical assets as well as specific technology solutions for fixed wireless to maximize range of coverage and minimize costs for a defined geographic area. Although beyond the scope of the work presented here, we refer readers to the works of Guttman (2006) and Sawada et al. (2006) as examples of this approach.

The methodology is subject to limitations that should be considered when mapping vertical assets. First, LiDAR interactions with water may be unpredictable, and underwater features may be detected and appear erroneously in the resulting vertical assets. Second, shadows in the imagery will lower NDVI values, resulting in some trees that are not removed from the vertical assets, which is particularly an issue along the edges of forested areas and requires manual post-processing to remove. Third, a threshold of 50 meters was used to minimize the effects of shadowing from trees and other features, but this distance could be customized to a specific location to improve overall results. Finally, any vertical assets constructed after LiDAR data were captured will not appear on the resulting vertical assets map, which may be a disadvantage for counties without current LiDAR data acquisition.

Outreach, implementation, and testing

An important step in the development of the methodologies described previously is to determine their effectiveness for informing the broadband planning process through implementation in actual real-world settings. Here we offer preliminary results from initial outreach, implementation, and testing efforts.

In fall 2022, the project team was invited to partner with the Benton Institute for Broadband & Society to field test the mapping methods presented here as part of Broadband Breakthrough, a 14-week series of workshops with participating rural counties that aims to expand broadband capacity. The purpose of the workshop series was to bring together broadband planning teams in cohort counties selected for the workshop with experts across a range of specialty areas, including mapping, to develop rural broadband expansion plans customized for each county.

Five counties in Illinois (Edgar, Hancock, McLean, Ogle, and Schuyler) were selected to participate in the Broadband Breakthrough workshops beginning in January 2023 and concluding in May. A key component of the workshop series was the inclusion of maps to quantify agricultural production in unserved and underserved areas, and maps of existing vertical assets as part of the planning process. To support this goal, the project team first participated in a workshop session in which the mapping methodologies and results were presented along with guidance for how the maps might be utilized to inform the broadband planning process. Next, county workshop participants were provided with maps that quantify agricultural production in unserved and underserved areas for their county (See Figure 7), as well as an interactive web map of existing vertical assets for the county (See Figure 16). Once provided, these resources were then available to the county team for consideration in the development of their respective county broadband plans.

At the conclusion of the workshop, each county drafted a broadband plan which they presented to workshop participants before implementation. Notably, each of the five counties included maps developed during the workshop as important components of their plans, and several emphasized the importance of these maps for both broadband advocacy and planning purposes. One county used their vertical assets map to create a survey for vertical assets located on private lands to inquire if a land owner would be willing to host broadband infrastructure. The maps displaying quantification of agriculture likewise served a key role in advocacy and planning efforts for the county planning teams. For example, some counties mentioned the value of the maps for convincing skeptical decision-makers about the return on investment for expanded broadband coverage to the rural economy. Counties also found the maps beneficial for farmers to visualize potential yield gains if they incorporated smart farming practices.

Conclusions and future directions

The project described in this paper has developed new methods for investigating a range of important questions related to current broadband availability, estimated impacts of increased broadband coverage, and approaches for expanding broadband availability in rural areas. In particular, we emphasize the important role of investigating these questions from a geographic perspective, in which case maps, GIS, remote sensing, and spatial analysis may assist in providing guidance to “what if” questions and proposing potential solutions. We also highlight the importance of these methods to guide the planning process in an economically feasible manner to inform investment of financial resources for maximum gain. Feedback from counties that participated in broadband advocacy efforts through the Broadband Breakthrough workshop series was very positive and demonstrated much potential for these maps in the planning process. We advocate for additional implementation of these methods in planning efforts to address the broadband challenge in rural areas.

Acknowledgements

This work was supported by the Illinois Innovation Network under a Sustaining Illinois Seed Grant and completed in partnership with the Illinois Soybean Association. The authors wish to acknowledge the following individuals and organizations for their collaborations and contributions to the work presented in this paper:

• Todd Main, Scott Gaffner, Stephen Sostaric, Brad Daugherty, Illinois Soybean Association

• Rex Schaeffer, Craig McLauchlan, Research and Sponsored Programs, Illinois State University

• Jake Brasen, Luke Brasen, Ryan Hogan, Eric Richmond, Sara Schelinski, Illinois State University

• Kyle Harfst, Brian Rogers, Illinois Innovation Network

• Mike Wever, Mary Gay, Southwest Illinois Leadership Council

• Shubhika Agarwal, Illinois Broadband Lab

• Adrianne Furniss, Benton Institute for Broadband & Society

• Bill Coleman, Community Technology Advisors

• Nancy Esarey Ouedraogo, University of Illinois Extension

• Bill Bodine, Illinois Farm Bureau

• Bond County (IL), Edgar County (IL), Hancock County (IL), McLean County (IL), Ogle County (IL), and Schuyler County (IL).

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Standard FCC definitions of less than 25 Mbps downstream or 3 Mpbs upstream are used in this paper to define “unserved” areas while “underserved” areas are defined as minimum 25 Mbps downstream and 3 Mpbs upstream, but less than 100 Mbps downstream and 20 Mbps upstream.

2 For example, see https://broadband.uillinois.edu/ for the Illinois broadband map.

3 Current FCC definitions are unserved = 25 Mbps download/3 Mbps upload or less, underserved = greater than 25 Mbps download/3 Mpbs upload but less than 100 Mbps/20 Mbps, and served = greater than 100 Mbps/20 Mbps.

4 See https://broadband.uillinois.edu/.

5 For more information about the FAA’s Digital Obstacle File (DOF), see https://www.faa.gov/air_traffic/flight_info/aeronav/obst_data/doffaqs/.

6 See https://clearinghouse.isgs.illinois.edu/data/elevation/illinois-height-modernization-ilhmp.

7 See https://clearinghouse.isgs.illinois.edu/data/imagery/2019-illinois-naip-digital-orthophoto-quadrangle-data.