Skip Navigation

UNC Carolina Population Center

 

About the Spatial Analysis Unit

The overall objective of the Spatial Analysis Unit is to promote the use of spatial data and tools in state-of-the-art population research.

As the first population center to propose a Spatial Analysis Unit in 1994, CPC has been a leader in this regard.  The unit’s specific objectives are to promote the incorporation of spatial perspectives and assist in project development, provide cost-effective support to CPC projects using spatial data and methods, encourage innovation in the blending of social and spatial tools and perspectives, and provide training as needed.

The Spatial Analysis Unit provides the following services to the CPC community in support of population-based research: 

  1. Consultation for Proposal Development and Ongoing Research
  2. Spatial Database Development
  3. Spatial Analyses and Measurements
  4. Remotely Sensed Image Analysis
  5. Spatially Explicit Simulation Modeling
  6. Spatial Statistics
  7. Cartographic Design and Visualization
  8. Training 

Selected Project Highlights

Innovative data development. Diverse data collections carried out under the aegis of CPC have included the use of the GPS and other field mapping techniques.

  • The Add Health project collected residential locations that have been used, among other purposes, to study the effects of neighborhood contexts on achievement and delinquency, the relationship between residential segregation and social segregation in schools, and the impact of the physical environment on obesity.
  • The Pregnancy, Infection, and Nutrition project worked with spatial staff to develop sampling and GPS strategies for characterizing neighborhood characteristics related to birth outcomes.
  • Unit staff worked with researchers in the Understanding Change in Physical Activity Postpartum project to develop a methodology for sampling and characterizing neighborhoods around respondents for analysis in relation to exercise opportunities and postpartum weight loss.  In addition, staff merged, cleaned and updated road data from four counties to create a comprehensive road dataset for the study area.  The result was a dataset more accurate and complete than any other available, free or for purchase.
  • In Nang Rong, Thailand, coordinates for villages, household residences, and field plots are making possible new analyses of population and land use dynamics. Techniques developed for the Thai 2000 data collection were the first to successfully link households to distant field plots for a large sample.
  • In the Ecuador Amazon, Spatial Analysis staff developed field survey instrument maps and protocols for field collection efforts in 2000, 2001, 2002, and 2003 as part of a larger study of migration and land use and land cover change in this region.
  • In Nicaragua, GPS training and data collection protocols were developed for the MEASURE health and family planing facility surveys conducted there and used for evaluation of routine health intervention strategies and Hurricane Mitch recovery efforts.
  • In Bangladesh, MEASURE Evaluation collaborated with the Centre for Urban Studies and the National Institute of Population Research and Training to physically identify boundaries of slums and squatter settlements in order to facilitate the implementation of the urban health survey throughout the country.
Satellite and aerial remote sensing.  The Ecuador and Nang Rong studies have made extensive use of remotely sensed data.  Spatial Analysis staff have assembled a time series of satellite images to characterize deforestation and the expansion of agriculture into formerly forested areas.  In the Nang Rong study, this picture is substantially enhanced with aerial photographs dating to 1954.  Generally, investigators have used classified remote imagery to develop plot- or village-based (Nang Rong) or farm-based (Ecuador) measures of land use and land cover, linking these variables to households and villages or communities in SAS-based data sets, respectively, and incorporating them into statistical analyses of population determinants and consequences.

Spatial analyses and measurements.  The Spatial Analysis Unit uses state-of-the-art GIS software tools and custom programming to carry out analyses on project spatial databases.   Some examples:
  • In their study of obesity and the environment using Add Health and CARDIA data, Faculty Fellows Gordon-Larsen and Popkin are developing first-of-the-kind spatial analytic databases of measures of neighborhood factors affecting adolescent physical activity, including data on number and type of recreational opportunities, distances (Euclidean and travel) to these opportunities, walkability of neighborhoods, and so on.
  • The PIN Understanding Change in Physical Activity Postpartum project has measured a variety of neighborhood characteristics - including crime incidences, density of roads and intersections, land use metrics, accessibility to recreational opportunities, and food resources - to analyze effects on physical activity, nutrition, and postpartum weight status.
  • For the MEASURE Nicaraguan Health Facility Survey (NHFS), Unit staff assisted in the collection of health facility locations and characteristics to analyze the staffing and provision of services in relation to population in potential catchment areas.  A new data analysis technique, Kernel Density Estimation (KDE), was used to calculate ratios of health facilities and staff members to nearby populations.

Spatial modeling.  The use of spatially explicit simulation models in the study of relationships between population, land use, and the environment has become a regular practice within the Unit.  One such approach is the Cellular Automaton (CA) model which is composed of a regular grid of cells, each in a finite state, that are interactively updated in discrete time steps according to a set of transition rules.  In these models, change at any given location is at least partially dependent on conditions at nearby locations.  As part of the Ecuador project, CA models are being developed to simulate the migration-driven land use change in the region and the effects of urbanization on land use patterns on household farms.  A similar approach was taken in the Nang Rong project, and is an explicit example of the cross-fertilization of projects facilitated by the Spatial Analysis Unit.  Through the ongoing progression of these models, we are able to design and test a wide range of scenarios on land use change, from both endogenous and exogenous shocks.  Examples of these shocks include droughts, floods, and new road development.

Another spatially explicit approach is Agent-Based Mmodeling, which examines characteristics and activities of individual agents as they interact and change over time, as they adapt (or not) to their environment, and learn (or not) from experience.  With funding from the UNC Vice Chancellor for Research and Economic Development, Unit staff developed a set of agents and tools that can be used in the initial creation of a population/environment ABM.  This in turn has directly led to the development of ABMs for both the Ecuador and Nang Rong projects.  In the Ecuador project, an ABM is being developed that models over 2000 farms across a subset of the study area, with a focus primarily on land use change decision making and interactions between households in the region.  In the Nang Rong project, the ABM under development is focused on land use change decisions, sharing of land parcels, and societal interactions modeled within the context of a social network.

Cartographic design and visualization.  In addition to their use in data development and fieldwork as described above, maps are key components of data validation, description, analysis, and presentation of results.  Maps are also an important product that we can provide to help decision makers.  Some examples:

  • MEASURE Evaluation is working with key decision makers in nations across Africa, Asia and the Caribbean to create maps locating populations of orphans and vulnerable children (OVC) , as well as health facilities and aid services that can help these children.  In addition, guidance will be provided to help strengthen OVC data infrastructures and use those data to make better decisions.
  • Spatial Analysis staff assisted CPC Fellow Sharon Weir in the development of the PLACE method, which uses available demographic, epidemiologic, and contextual data to identify areas of high transmission for sexually transmitted disease, especially HIV/AIDS, and then within these areas, identifies, maps, and characterizes where people meet new sexual and injecting drug use partners.
  • Mapping can be very useful  in assessing the validity of survey data as well.  The 1994 village surveys in Nang Rong asked about connections to other villages by way of shared schools, temples, and the like.  Mapping these social network connections directly revealed errors in the data not otherwise visible.

Center initiatives.  In addition to specific research projects, the Spatial Analysis Unit is also part of two Center-level initiatives.  One is the Data Sharing for Demographic Research (DSDR) Center (CPC is a subcontractor to ICPSR at the University of Michigan), in which the role of the Unit is to investigate questions of study design and dissemination related to confidentiality, security, and disclosure risk.  Spatial Analysis staff will help develop protocols to incorporate spatial and spatially referenced data into social survey data sets and protocols for the spatial representation of social survey data as maps and other displays.  The Spatial Analysis Unit is also central to the Integrative Graduate Education and Research Training (IGERT) program in Population and Environment.  Graduate students in this program are required to obtain training in geographic information systems, remote sensing, and spatial analysis.  The unit reinforces this formal training and provides opportunities for informal learning.