Decomposing Uncertainty in Spatial Accessibility Estimates of Urban Park Using Uncertainty Analysis and Global Sensitivity Analysis (UA-GSA)
This study proposes an uncertainty and global sensitivity analysis (UA-GSA) framework to evaluate how uncertainties in demand, supply, and travel impedance affect 2SFCA-based urban park accessibility estimates in Seoul, revealing that while spatial patterns remain consistent, the dominant sources of uncertainty vary significantly across space and time.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine trying to measure how easy it is for people to reach a park. For decades, planners have relied on a standard method that treats the city as a static snapshot: it counts how many people live in a neighborhood, how big the parks are, and how long it takes to walk there. This approach assumes that the number of people, the size of the parks, and the speed of walkers stay the same all day, every day. But cities are never still. People move to work in the morning, return home in the evening, and fill parks at lunch. When a model ignores these shifts, it presents a single, fixed number for accessibility that might look stable but is actually built on shifting ground. The question is not just whether a park is close, but how reliable that measurement is when the real world changes.
A team of researchers set out to test the stability of these measurements by looking at urban parks in Seoul, South Korea. They focused on four districts containing hundreds of small neighborhoods and over two hundred parks. Instead of accepting a single answer, they asked what would happen to the accessibility score if they changed the inputs. They considered three main factors that could vary: the number of people actually present in an area at a specific time, the amount of park space available after accounting for how crowded it already is, and the speed at which people walk. To simulate the real world, they did not just pick one number for each factor. Instead, they generated thousands of different scenarios, mixing and matching these variables to see how the final result would fluctuate. They ran these simulations for four distinct times: 8 AM, noon, 6 PM, and an average for the whole day.
The researchers found that while the general map of where parks are accessible looked similar across all times, the reasons for uncertainty changed dramatically depending on the hour and the location. In most neighborhoods, the biggest source of uncertainty was the number of people present. Because the population moves throughout the day, the "demand" side of the equation is the most volatile. However, this was not a uniform rule. In some specific areas, the availability of park space or the speed of walking became the dominant factor. For instance, in the morning, the amount of available park space mattered more in certain districts, while at noon, the speed at which people could walk became a more significant source of variation.
Crucially, the study revealed that these factors do not act alone. The influence of one variable often depends on the state of the others. When the researchers looked at how the variables interacted, the picture became more complex. A factor that seemed to be the main driver of uncertainty on its own might lose that dominance when its relationship with other factors was considered. This means that a planner cannot simply look at a single map and assume that adding more parks will fix a problem. If the uncertainty in a specific area is driven by fluctuating crowds rather than a lack of green space, adding more parks might not solve the issue. Conversely, if the uncertainty stems from how fast people can walk, the solution might lie in improving pedestrian infrastructure rather than expanding park size.
The study concludes that accessibility estimates should not be treated as fixed facts but as conditional outcomes that depend on how the inputs are represented. The researchers demonstrated that by breaking down the uncertainty, they could identify which specific inputs require the most attention in different parts of the city. In areas where the population density shifts wildly, understanding the timing of that movement is more important than the static count of residents. In other areas, the physical capacity of the park or the walking conditions might be the limiting factor. This approach offers a way to diagnose why an accessibility estimate might be unstable, moving beyond a simple "good" or "bad" rating to a nuanced understanding of what is driving the result.
Ultimately, the work suggests that planning decisions based on these measurements need to be more flexible. If a city wants to ensure fair access to green space, it must recognize that the "need" for a park changes throughout the day. A neighborhood that is underserved at 8 AM might be well-served at 6 PM, and the factors causing that difference are not always the same. By acknowledging that uncertainty is not just noise but a signal of how the city functions, planners can make more informed choices. The study does not provide a single new map of perfect accessibility, but rather a method to understand the reliability of any map and to identify which pieces of the puzzle are most likely to shift.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.