1. Introduction:

Urban expansion is a set of interrelated, complex processes that – in multiple ways – can put unprecedented stress on medical facilities (Frumkin, 2002; Polo et. all, 2015). The Center for Disease Control sites urban sprawl as a determinant in the quality of environmental health as it tends to increase the dependence on automobiles and road networks for transportation provoking accidents, compromising air quality and facilitating sedentary lifestyles that puts additional pressures on health facilities (Frumkin, 2002). In addition to this, travel impedances associated with an increased population can impede emergency response teams from accessing residents who are in need of immediate care (Trowbridge et. al,  2009).

Lane County is expected to experience a large population increase over the next two decades (Population Forecast for Lane County, 2008, p. 17). Thus is remains a topic of concern to city planners, engineers and public health officials in the county to investigate possible forms of urban expansion and to scrutinize spatial characteristics of the medical facility system in order to accommodate a growing population.

Location allocation analysis provides a methodological framework for investigating the stresses placed on medical facilities due to a growing population, while spatio-temporal modeling can be used to predict the spatial distribution of land use change over time (Polo et. al, 2015; PeuQuet, 2005; Aguda, 2013). In particular, Polo et. al (2015) employs a location allocation analysis under the terms that such a model “provide[s] a framework for investigating the use of health care services and for generating alternatives either to suggest an efficient service or to improve an existing one” (p. 3), while Aguda (2004) sites temporal and spatial modeling as method that is successful in describing the spatial and temporal patterns of urban development (p. 94).

Inspired by the literature, an investigation that employs spacio-temporal modeling in GIS is devised to highlight areas in the medical facility system that may be compromised in the face of an increasing population and urban expansion. Desired outcomes of this study include: the identification of medical facilities that are expected to be under stress in regards to urban expansion, as well as an informed selection of an area to place a new medical facility that will justly accommodate the expected population.

2 Methods:

2.1 Site

2.1.1 Spatial Extent:

The study focuses on the predicted population increase of Lane County, Oregon – an administrative, geographical body spanning the southern Willamette Valley, Central Oregon Coast and Cascades (Kline et. al, 2004).

While the majority of the incoming population is expected to settle near larger metropolitan areas of Lane County (Population Forecast for Lane County, 2008, p. 17), the spatial extent of the analysis encompasses the entire county. Benefits of investigating this particular extent  include: illustrating interactions between small and large metropolitan areas that are also subject to growth, and successfully addressing land use change with respect to agricultural and forested areas in remote regions of the county.

2.1.2 Temporal Extent:

A temporal extent of 15 years is addressed in the analysis. Adequately reviewed, detailed approximations of population growth have been published for the next two decades (Hubbard, 2015; Kline et. al, 2004; Weitz et. al,1998), providing ample reliable predictions for this temporal extent. Time is represented in the following model as discrete timesteps accumulating the growth over five years in the spatial and non-spatial representation of each timestep. This enables a simplistic and effective approach to modeling with time (Peudet, 93).

2.2 Data

Three datasets were gathered to employ the analysis: a set of Lane County hospitals, street network data, and Lane County Zoning Data.

The hospital and street network data remained unaltered throughout the analysis, and played an integral part in the Location Allocation analysis.

The land use data was initially a vector zoning shapefile. The shapefile was rasterized into 10 by 10 acre cells, and grossly simplified on the basis of producing a simple binary dataset engage in a urban expansion model.

This analysis capitalizes on the idea that new growth will be spatially appended to existing residential and urban land use types, and will result in contiguous urban forms (Weitz et. al 1998) that invade forested and agricultural zones (Kline, 2004, p. 33). This allows us to derive two classifications of land use types that participate in the urban expansion model, urban and rural. Medium and high-density residential areas were determined to be attractors or urban growth, while very low-density residential areas were classified as rural, likening them to farmland or forest (Kline, 2004). Additionally, rural commercial zones were considered attractions of urban growth (Population Forecast for Lane County, 2008, p. 21). Undefined land use types, industrial areas, and other land use types were classified as null values. 

2.3 Model

The methodological approach following the posed inquiry is driven by two central tasks: (1) to integrate non-spatial population growth statistics with a spatio-temporal model to infer the spatial arrangement of various land-use types and population growth over time, and (2) to incorporate the modeled land use and population change in a location-allocation analysis to determine increased stresses on Lane County’s medical facilities.

2.3.1 Land Use Change and Population Model

An iterative algorithm is devised in order to describe the land use change and population growth throughout the specified spatial and temporal extents, where a simplified zoning dataset containing binary values and a constant linear growth rate are accepted as parameters.

Once the urban and rural areas are extracted from the land use dataset, a euclidean distance is calculated from the center of each raster cell to determine cells that are potential candidates for urbanization. A value of 2087 feet is then received as the second parameter, and cells lying within 2087 feet of an existing urban cell are valued as candidates for urbanization. This is done through a raster overlay, between the proposed urban cells and the previously defined ‘rural’ areas, where the output only includes areas that were nominated by a euclidean distance calculation, and whose previous land use type was a forest or farmland. The resulting raster is then incorporated into the existing land use raster dataset in preparation for the following iteration of the algorithm.

2.3.2 Location Allocation Analysis

A broad generalization is made that land use change is proportional to population growth by assuming a constant population density for each new urban. These urban cells could thus be used as unweighted points of demand in a location analysis. The urban cells were converted to points to act as spatial weights in determining usage by hospital.

A median problem is addressed in our analysis as the objective of the location analysis is to draw a relationship between the estimated populations at urban cells, and the hospital that is most accessible to them through the road network.

This analysis was then completed for the urban profile at each time step, to predict how hospital usage may vary over time.

3 Results:

3.1 Results of the Urban Growth Model

The initial land use raster and model output shows that urban acreage constantly increases through each iteration of the model, and that the rate-of-increase increases as a result of more urban entities being appended to the existing cities. Over time, it is seen that the small, isolated municipalities in the rural reaches of the county gain large amounts of urban area, while the larger metropolitan area gathers proportionately less.

3.2 Population Growth Projections

As stated in the methods section, this analysis pairs urban expansion with population growth by assuming a constant population value to each new urban cell in the landscape resulting in paralleled population growth rates and urban expansion rates. Thus, the results of the model indicate that population growth and urban expansion results parallel one another in terms of quantity and rates of increase, as seen in figures 3 and 4.

3.3 Results of the Location Allocation Analysis

The resulting location allocation analysis indicates a disproportional amount of stress on the PeaceHealth – Sacred Heart hospital near downtown Eugene. This is true at the initiation of the land use model (in 2014), but as the population increases and urban areas expand, the stress becomes especially disproportional. Following from the results of the land-use model,  rural areas gain a disproportionately large amount of urban expansion due to the available forest and farmland surrounding a single urban cell, and thus demand points increase dramatically in the remote areas of the county, where many small, isolated residential and urban forms exist. As a result, the PeaceHealth, Sacred Heart location and Peace Harbor Hospital – the western-most hospitals – have the highest increases of demand over time. However, all hospitals have increased demands due to the rapid expansion of rural municipalities and neighborhoods (see figs 5 – 8). 

In an attempt to lessen the stress on the Sacred Heart medical facility, another large medical facility is proposed to accomodate the incoming population. A medical facility is placed North of Veneta, roughly in the center of Sacred Heart’s demand points. Another location analysis is performed on the new dataset to demonstrate the allocation of urban areas to the proposed facility.

The resulting analysis implies that the proposed facility dramatically reduces the number of urban points allocated to the Sacred Heart location, while slightly increasing the demand on the Mackenzie-Willamette and Riverbend Facilities. The induction of the proposed medical facility implies a more even distribution of urban points per medical facility.

3.4 Uncertainty

The results of this analysis are characterized by error in class generalization, or “grouping together entities with dissimilar properties” (Fisher et. al, 2005, p. 193). This error is generated through the large generalizations made of the data in this analysis.

Error is presented in the classification of urban and rural cells through the representation of land use in a binary dataset. Because a zone must be classified as either ‘rural’ or ‘urban’ for use in the model, population density variation is ignored over a large space. Thus, the rural and urban datasets may incorrectly categorize dense and sparse spaces in a dataset due to a generalization of the larger area.

The most profound error in this analysis is the assumption that all urban cells are associated with a uniform population density. This is an unrealistic assumption that results in population predictions that are inconsistent through both time and quantity. The simplification implies an unrealistically low population density in ‘urban’ spaces, and disproportionately high densities in ‘rural’ areas. The result of this is that the initial population is vastly underestimated to compensate for the incoming land use changes. By underestimating population in the initial timestep, the analysis still provides a spatially accurate representation of urban expansion throughout the duration of the temporal extent, and an accurate population estimation by the final timestep.

4. Conclusion

This analysis utilized spatio-temporal modeling and location analysis to predict the upcoming stresses to the medical facility system in Lane County, Oregon. The literature provided a framework for implementing a GIS to investigate medical facility accessibility, land use change, and population growth. The contributions of this analysis largely remain local, in that they provide a guide to assessing potential shortcomings of the current medical facility system.

More broadly, this analysis delineates an approach to medical facility allocation and population growth through a simple, lightweight model that has few data requirements – a process that is highly desired among organization that do not have access to large, maintained datasets. Despite the error and ambiguity presented in this model, this analysis is successful in that: (1) the model keeps a contiguous urban form throughout population growth with respect to both large and small municipalities – a predicted spatial distribution (Weitz et. al, 1998), (2) it incorporates smaller municipalities in the analysis and allocation of accessible medical facilities, (3) it still reflects a realistic pattern of land use change and urban growth around larger municipalities, while adhering to a realistic zoning resolution.

5. References

Aguda, A. (2013). Spatio-temporal assessment of urban growth of medium-size and nodal towns for sustainable management: Using GIS. Management of Environmental Quality, 24(1), 94-106.

Fisher, P F (2005). Models of Uncertainty in spatial data. In New Developments in Geographical Information Systems: Principles, Techniques, Management and Applications (Second ed., pp. 191 – 205). West Sussex, England: John Wiley & Sons.

Frumkin, H. (2002). Urban Sprawl and Public Health. Public Health Reports, 117, 201-217. Retrieved December 6, 2015, from http://www.cdc.gov/healthyplaces/articles/Urban_Sprawl_and_Public_Health_PHR.pdf

Hubbard, S. (2015, November 18). Oregon’s population hits 4 million; Lane County’s growth slightly below state average.The Register Guard.

Kline, J., Azuma, D., & Alig, R. (2004). Population Growth, Urban Expansion, and Private Forestry in Western Oregon. Forest Science, 50(1), 33-43. Retrieved November 28, 2015, from http://www.fsl.orst.edu/lulcd/Publicationsalpha_files/Kline_etal_2004_FS.pdf

PeuQuet, D. (2005). Time in GIS and geographical databases. In New Developments in Geographical Information Systems: Principles, Techniques, Management and Applications (Second ed., pp. 91 – 103). West Sussex, England: John Wiley & Sons.

Polo, G., Acosta, C., Ferreira, F., & Dias, R. (2015). Location-Allocation and Accessibility Models for Improving the Spatial Planning of Public Health Services. PLoS ONE PLOS ONE,10(3), 1-14. doi:10.1371/ journal.pone.0119190

Population Forecast for Lane County, its Cities and Unincorporated Area 2008 – 2035. (2009). Population Research Center, 1-96.

Trowbridge, M., Gurka, M., & O’connor, R. (2009). Urban Sprawl and Delayed Ambulance Arrival in the U.S. American Journal of Preventive Medicine, 37(5), 428-432. doi:10.1016/j.amepre.2009.06.016.

Weitz, J., & Moore, T. (1998). Development inside Urban Growth Boundaries: Oregon’s Empirical Evidence of Contiguous Urban Form. Journal of the American Planning Association, 64(4), 424-440. Retrieved November 30, 2015, from UO Libraries.