2.5.1 Organized Landscapes
Have you ever asked yourself, why does something exist at a specific location? Why do certain people live in certain neighborhoods? Do cities or natural landscapes have order, or are they merely a random assortment of objects in space?
These questions are central to understanding geographic patterns. Consider the satellite image provided in Map 2.5.1 that shows the arid state of New Mexico. While this is one of the driest states in the US, you can see there is still a significant area covered by trees (look for the green patches – you can even zoom in to see individual trees). However, the trees existing in the natural landscapes of New Mexico are all clumped together in very specific regions of the state. The reason for this spatial pattern of clustering is because soil moisture, elevation, and temperature at these locations are amenable to the survival of trees, while the remaining uninhabited areas of the state have inhospitable conditions for trees. In this sense, locations closer together have similar physical and climatic characteristics, thus creating trees to exist close together in the landscape – essentially composing distinct forests.
Map 2.5.1
The same type of patterns can be observed with people. In the 1970s, the economist Thomas Shelling described the process by which people move to and from neighborhoods based on the preference of their neighbors. Using race as the primary example, Shelling hypothesized that people will move from their current location if the number of people dissimilar to them is beyond some threshold, and will attempt to move to a new location where more people are similar to them, thus leading to racially segregated neighborhoods. Shelling’s Theory of Segregation has since been used for helping to understand how age, ethnicity and other attributes also play a role in creating grouping similar people together in cities. This is evident looking at Map 2.5.2 that shows the median age of residents in the city of Seattle, Washington (you can pan around the map to look at other cities and see if similar patterns exist). You can clearly observe that locations within close proximity of one another share similar characteristics, as different neighborhoods are likely defined by young families, retirees, etc. as is the case with the trees in New Mexico example.
Map 2.5.2
2.5.2 Spatial Distribution of Objects
The New Mexico and segregation examples mentioned above represent two types of ways that we think about how landscapes are organized. For the New Mexico example, we are examining how a single type of object (trees) is distributed across space. In order to describe the distribution of objects, we use the terms clustered if objects are clumped together, dispersed if objects are spread out in some organized fashion across a landscape, and random if objects are neither clustered or dispersed, thus lacking any organization. Looking at landscapes and being able to observe and describe how objects are distributed across space a very important part of being spatially literate. Let’s take a look at another example and test your ability to think spatially.
Map 2.5.3 shows data on the location of grocery stores in the city of Pittsburg where people can purchase food that the US food guide considers part of a balanced diet (in other words, no convenience stores are included in the data set). How would you describe the spatial pattern of grocery stores? Are they clustered or evenly dispersed across the city? Or do they appear to be randomly located? While there exists several spatial statistics that we can use to answer this question (more about those in a later chapter), we can make some general observations based on what we see without evening having much knowledge about the city. The very first thing to notice is that grocery stores are not evenly dispersed across the city. If they were, we would be more likely to see them evenly spaced from each other. Instead, it appears that grocery store locations are clustered in specific areas, particularly in the south, west, and north parts of the city. While grocery stores in the east part of the city appear more dispersed, zooming in will show you that some stores are indeed very close together. Perhaps the most significant observation to make is not where grocery stores are located, but where they are not located. Observing that grocery stores are clustered means that there are some areas that do not have access to grocery stores, which could then lead us to ask, who does not have access to resources that can provide them with a balanced healthy diet?
Map 2.5.3
2.5.3 Spatial Distribution of Attributes
This brings us to the to the second way to think about how landscapes are organized. Both the New Mexico trees and the Pittsburg grocery stores are examples of how points or objects are distributed in space. But what if we are concerned with how objects are distributed with regards to some characteristic or attribute? Thinking back to Shelling’s Theory of Segregation and the median age of Seattle map, we are not merely looking at where people are located, but we are more interested in understanding where people with certain characteristics (e.g. age) are located with respect to one another.
In 1970, the acclaimed Geographer Waldo Tobler devised what is now known as the First Law of Geography that explains, “everything is related to everything else, but near things are more related to distant things”. This may appear intuitive to some people, but the First Law of Geography has profound impacts in how landscapes are organized. If Shelling’s Theory of Segregation explains why people move, Tobler’s First Law of Geography explains the type of patterns that we expect to observe as a result. When the First Law of Geography holds true and closer things are in fact more similar than distant things, we can state that the spatial distribution of attributes is clustered. Conversely, in the less common instance of when nearer things are more dissimilar than distant things, we can state that the spatial distribution of attributes is dispersed. Examples of dispersed attributes are certainly more rare than clustered attributes, but nonetheless do exist.
Returning to Pittsburg, looking at Map 2.5.4 we can observe the mean disposable income across the city. How would you explain the spatial distribution of people’s age in this city? It is very clear that census tracts in the two lowest disposable income categories are clustered more towards the center and east of the city, while census tracts with higher disposable incomes are clustered in the north, northwest, and southwest areas of the city. Does Tobler’s First Law of Geography hold true here? Just from observing the pattern we can confidently state that disposable income across the city looks to be organized in a clustered pattern – one that is certainly not random. In fact, one could argue that disposable income increases as you move outward from the city core, especially in the north, west and south parts of the city. There are several types of spatial statistics that can be used to test if these hypotheses hold true, but for now the spatial pattern is quite evident from a simple visualization of the data.
Map 2.5.4
2.5.4 Explaining Spatial Distributions
Continuing with the Pittsburg example, Thomas Shelling’s Theory of Segregation may provide some insight into the clustered pattern of disposable income (there are certainly many other factors that can help to explain this observation as well), but what explains the spatial distribution of grocery stores? Is there a reason why we see more grocery stores in some areas than in others? Often times the answer to this question is yes – objects (especially businesses) locate themselves in strategic locations in order to take advantage of available resources. In the case of grocery stores the resource is customers, where as in the New Mexico example the resources are water, nutrients and temperature that allow trees to survive. However, a more important question is focused around the areas in the landscape where things do not exist. Such is the case in Pittsburg where there is evidently large areas with no grocery stores. These areas have become referred to as food deserts, because they represent relatively vast areas with no or little access to nutritional food.
Let’s look at the location of grocery stores, mean disposable income, and retail spending potential in a single display in Map 2.5.5 or for a full view click HERE (this is called a web application – something you will learn how to create in a later chapter). We can now see the locations of grocery stores relative to two economic attributes of households across the city. In the center map we can observe the relationship between grocery store locations and income, while the map on the right shows us the relationship between locations and retail spending potential. There are two observations we can make immediately by looking at the map on the right. The first is that the vast area southeast of the city has few, and it is the area with a cluster of low retail spending potential. Second, the relatively few grocery stores that are located east of the city are located in or on the perimeter of census tracts where spending potential is above average. Therefore, from these observations we can hypothesize that food deserts do exist in Pittsburg, and areas with a lack of access to healthy food are those with the least amount of spending potential. While more research will certainly need to be accomplished to validate these observations, our visual analysis of spatial patterns and the relationship between objects and attributes has provided a very important glimpse into food deserts in Pittsburg.
Map 2.5.5