Geospatial Impact Indexing of Agricultural Incidents: A Multi-Criteria Risk Assessment in the U.S. Midwest

This study introduces a multi-criteria geospatial framework utilizing Analytic Hierarchy Process and Geometric Fuzzy AHP weighting to reveal that while agricultural incident frequency in the U.S. Midwest is concentrated in northwestern Iowa, severity-weighted risk extends more broadly, thereby offering a superior evidence-based approach for targeting safety interventions and emergency medical planning compared to traditional frequency-only assessments.

The Gist

Imagine the American Midwest as a giant, bustling kitchen where millions of meals are prepared every year. But instead of just chopping vegetables, the "chefs" (farmers) are operating massive, powerful machines, handling heavy livestock, and working in remote corners of the kitchen. Sometimes, things go wrong.

This paper is like a new kind of safety report card for that kitchen. The authors, Ege Duran, Omer Mermer, and Ibrahim Demir, argue that the old way of keeping score was flawed. Here is the simple breakdown of their findings and methods:

The Problem: Counting vs. Understanding

The Old Way: Imagine a teacher who only counts how many times a student raises their hand to answer a question. They might say, "Wow, this student is very active!" But that doesn't tell you if the student is answering correctly, dangerously, or if they are in trouble.
The Paper's View: The researchers say that simply counting how many farm accidents happen (frequency) is like counting hand-raises. It tells you where the farmers are busy, but it doesn't tell you where the real danger lies. A busy farm might have many small, survivable bumps, while a quieter farm might have a few catastrophic events.

The Solution: A "Risk Score" Recipe

To fix this, the team created a new "Impact Index." Think of this as a complex recipe for a risk score. Instead of just counting accidents, they mixed in several ingredients to see how "dangerous" a specific location really is:

1. The Outcome (Fatality): Did someone die? This is the heaviest ingredient.
2. The Mechanism (Type of Accident): Was it a tractor rollover or a grain bin entrapment? These are like "explosive" ingredients that are inherently more dangerous than a simple slip-and-fall.
3. The Rescue Time (EMS Access): How long does it take to get to a hospital? If a farmer is hurt in a remote field with no cell service and a long drive to the hospital, the risk score goes up.
4. The "Lone Worker" Factor: Many farmers work alone. If they get hurt and can't call for help because of a "cell phone dead zone," the risk is higher.

The Two Methods: The Sharp Knife vs. The Soft Sponge

The researchers used two different mathematical tools to mix these ingredients, and they got slightly different pictures:

• Method A (AHP): Think of this as a sharp knife. It cuts through the data and highlights the absolute worst spots very clearly. It says, "Look right here! This is the most dangerous place!" It focuses heavily on the severity of the outcome.
• Method B (Geometric-Fuzzy AHP): Think of this as a soft sponge. It absorbs some of the sharp edges and uncertainty. It acknowledges that we don't know everything perfectly (like exactly how bad a specific accident was or how fast help would arrive). This method creates a smoother map, showing broader areas of risk rather than just pinpointing the absolute worst spots.

What They Found: The Map of Danger

When they applied this recipe to seven Midwestern states (Iowa, Minnesota, Kansas, etc.) using data from 2012 to 2023, they found some surprising things:

• Busy isn't always Deadly: The busiest farming areas (like Northwestern Iowa) had the most accidents overall. This makes sense because there are more farmers and machines there.
• Deadly isn't always Busy: However, the areas with the most fatalities were spread out differently. While Northwestern Iowa was still a hotspot, the "danger zones" stretched further into central Iowa and southern Minnesota.
• The "Silent" Killers: The most dangerous accidents weren't the common ones (like slipping on a road). They were the rare, high-energy ones: tractors flipping over, people getting stuck in grain bins, or ATV rollovers. These specific types of accidents were much more likely to result in death.
• Who is at Risk? The data showed that older male farmers are the most vulnerable group. As farmers get older, the chance that an accident will be fatal increases.

The Big Picture

The main takeaway is that you can't just look at a map of "where accidents happened" to know where to put safety resources. You have to look at a map of "where accidents are most likely to kill."

By combining the location of the accident with how hard it is to get help (hospitals and cell towers), the researchers created a new way to see the landscape. They found that the "danger zones" are not just where the farming is busiest, but where the farming is busiest plus where the machinery is most dangerous plus where help is furthest away.

This study doesn't offer a new machine or a new law. Instead, it offers a better flashlight. It helps safety officials and emergency planners see the true shape of the risk, so they can focus their efforts on the places where a farmer's life is most at stake, rather than just the places where they are most busy.

Technical Summary

Technical Summary: Geospatial Impact Indexing of Agricultural Incidents in the U.S. Midwest

Problem Statement
Traditional agricultural safety assessments frequently rely on raw incident counts, which emphasize exposure frequency but fail to capture the multidimensional nature of risk, particularly outcome severity and post-incident survivability. In the U.S. Midwest, a region of intensive agricultural production, incidents often occur in remote locations where "lone workers" face significant barriers to immediate discovery and assistance. Existing surveillance methods often suffer from fragmentation, underreporting, and a lack of geospatial resolution that fails to link incident locations with critical survival factors such as emergency medical service (EMS) accessibility and cellular communication infrastructure. Consequently, safety interventions may be misaligned with areas of highest actual risk, which are defined not just by how often incidents occur, but by the convergence of hazardous mechanisms, environmental isolation, and temporal constraints.

Methodology
This study constructs a unified, multi-state geospatial database covering seven Midwestern states (Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, and South Dakota) using agricultural incident records from 2012 to 2023 sourced from the Central States Center for Agricultural Safety and Health (CS-CASH). The methodology integrates longitudinal incident data with environmental and infrastructural layers, including USDA NASS Cropland Data Layer (CDL) for land use, HIFLD data for EMS and hospital locations, and FCC data for cellular tower proximity.

The core methodological innovation is the development of a Multi-Criteria Incident Impact Index using two weighting frameworks:

1. Analytic Hierarchy Process (AHP): A structured decision-making technique that decomposes the problem into a hierarchy of eight criteria: Fatality (FAT), Incident Type (TYPE), Temporal Relevance (REC), People Involved (PEOP), Age (AGE), EMS/Hospital Closeness (EM), Tower Closeness (TOW), and Farm-Type Fatality Ratio (FARM). Pairwise comparisons were used to derive crisp weights, prioritizing direct harm and intrinsic hazard.
2. Geometric Fuzzy AHP (Geometric-FAHP): An extension of AHP utilizing Buckley's fuzzy geometric mean method with Triangular Fuzzy Numbers (TFNs). This approach captures the uncertainty and vagueness inherent in human judgments regarding risk, converting linguistic variables into fuzzy sets to produce smoother, less volatile weight distributions.

Both frameworks normalize the eight parameters to a [0, 1] scale and aggregate them into a composite impact score. These scores are then applied to generate Kernel Density Estimation (KDE) surfaces using a quartic (biweight) kernel to visualize spatial clustering of high-impact incidents, distinguishing between frequency-driven patterns and severity-driven risk.

Key Contributions

• Unified Multi-State Database: The study creates a comprehensive safety profile for the seven-state Midwest region, overcoming the limitations of fragmented, single-state reporting by harmonizing 12 years of surveillance data.
• Granular Geospatial Integration: It explicitly overlays incident locations with survival infrastructure (EMS travel times and cellular tower distances), quantifying vulnerability specifically for lone workers in remote agricultural environments.
• Novel Impact Indexing: The research introduces a quantitative mechanism to assess agricultural incidents beyond simple fatality counts. By applying both AHP and Geometric-FAHP, the study provides a dual perspective: AHP highlights sharp, severity-driven hotspots, while Geometric-FAHP reveals broader, gradual vulnerability patterns by moderating extreme values through uncertainty modeling.

Results

• Divergent Spatial Patterns: Analysis reveals that while overall incident frequency is heavily concentrated in northwestern Iowa (reflecting intensive row-crop activity), fatal outcomes exhibit a broader spatial footprint extending into central/northern Iowa and central-southern Minnesota. Relying solely on frequency counts would significantly underrepresent localized high-consequence areas.
• Mechanism-Specific Severity: Fatal incidents are disproportionately associated with high-energy mechanisms such as tractors, grain bins, and skid-loaders (fatality rates ranging from 58% to 76.9%), whereas roadway and entanglement incidents are more frequent but more survivable.
• Demographic Vulnerability: Risk is heavily concentrated among male workers (85% of fatalities) and increases with age, with the 60–70 age group bearing the highest fatal burden.
• Weighting Comparison: The AHP framework produced higher mean impact scores and greater spatial dispersion, accentuating localized hotspots where catastrophic incidents dominate. In contrast, Geometric-FAHP yielded consistently lower mean scores and reduced dispersion, producing smoother spatial gradients that better capture regional vulnerability structures without over-emphasizing extreme outliers.
• Infrastructure Gaps: The spatial analysis highlights that fatal outcomes are not strictly proportional to incident frequency but are heavily influenced by the presence of high-severity mechanisms and the structural deficits in rural infrastructure (e.g., extended EMS response times and communication deserts).

Significance and Claims
The paper claims that its findings demonstrate that frequency-based mapping alone fails to capture the multi-dimensional nature of agricultural risk. By explicitly linking incident locations with survival infrastructure and utilizing multi-criteria decision analysis, the study provides an evidence-based framework for:

1. Targeting Safety Interventions: Moving beyond general exposure metrics to focus resources on areas where high-frequency activity coincides with high-severity mechanisms and poor survivability conditions.
2. Improving Rural EMS Planning: Identifying specific geographic gaps in emergency response and communication infrastructure that compromise the "golden hour" for trauma victims.
3. Addressing the Digital Divide: Highlighting the critical role of cellular connectivity in rural safety and the potential for communication deserts to turn survivable injuries into fatalities.

The authors position this research as a novel step in agricultural safety research, offering a robust, quantitative mechanism to distinguish between exposure-driven activity patterns and severity-driven risk concentrations, thereby supporting more effective policy and planning decisions in the U.S. Midwest.