Geospatial Analysis and Modeling

Analysis and modeling of urban dynamics

3706.2 - Estimating changes in urban land and urban population using refined areal interpolation techniques

Monday, July 3
4:30 PM - 4:50 PM
Location: Maryland A

Introduction
The analysis of changes in urban land and urban population is important because the majority of human population resides in urbanized or semi-urban areas, and this proportion is continuously increasing. Improved understanding of such changes has important implications in interdisciplinary contexts including the assessment of potential effects of such trends on climate change and energy consumption, crisis management and land-use planning, to name a few. This research employs areal interpolation methods coupled with spatial refinement to analyze how urban land and urban population have been altered from 1990 to 2010 over consistent small census units such as census tracts for the whole U.S. To delineate residential areas for spatial refinement over the whole U.S., nationally available ancillary datasets such as the National Land-cover Database (NLCD) and Global Human Settlement Layer (GHSL) are used. Spatial refinement has been shown to improve the accuracy of regular areal interpolation methods for the estimation of total population.
Methodology
At first, two regular areal interpolation methods, namely areal weighting (AW) and target density weighting (TDW) are used to transfer population values from source zones to target zones. Census tracts in 2010 are used as the target zones, and population estimates of census tracts in 1990 and 2000 (used as source zones in 1990-2010 and 2000-2010 time periods, respectively) are reapportioned to them. Therefore, two different sets of population changes are computed (using AW and TDW, respectively) for both 1990-2010 and 2000-2010 over consistent census tract boundaries in 2010.
Then, AW and TDW are dasymetrically refined and run on residential parts of source and target zones, identified by NLCD developed classes (21-23) and GHSL built-up areas. Expectation Maximization (EM) is also implemented using NLCD classes as control zones. By the end of this part, five sets of multi-temporal population estimates are computed for both 1990-2010 and 2000-2010. They are consistent population values derived by refined AW using NLCD and GHSL, refined TDW using NLCD and GHSL and EM using NLCD.
To validate, census block level population statistics for the source years (1990 and 2000, respectively) are used as the reference measures, and error metrics such as mean absolute error (MAE), median absolute error, root mean square error (RMSE) and 90th percentile of absolute error are calculated to compare the total seven areal interpolation methods and determine the best-performing approach.
Finally, different criteria for defining urban land and urban population are employed based on the population estimates of the best-performing method. These criteria include but are not limited to a threshold definition for population density per target zone, a threshold definition for urbanized areas per target zone based on its portion of developed/built-up land as well as contiguity and compactness of developed/built-up portions per target zone.
Discussion
The multi-temporal consistent population estimates computed through the described approach can be used to answer questions such as:
• To what extent has urban population changed nationally during 1990-2010 based on consistent small area census units such as census tracts used as analytical units?
• To what proportion have urban and rural lands been converted to each other nationally during 1990-2010 when using fine analytical scale?
• How can these changes be visualized in cartographic formats and analyzed through cluster analysis and spatial autocorrelation analysis?
• Which areas of the United States have observed the most and the least rural-urban conversion?
Answers to these questions can be utilized as inputs to large-scale interdisciplinary research, investigating the future interactions between social processes such as demographic trends and other processes such as climate change effects, land-cover/land-use transitions and energy consumption patterns.

Hamidreza Zoraghein

PhD Student
University of Colorado Boulder

Hamidreza Zoraghein is a Ph.D. student in the Geography Department at the University of Colorado Boulder. His main field of research is Geographic Information Systems (GIS) with a focus on population estimation. Specifically, He is interested in using more advanced spatial analytical techniques to estimate population in spatio-temporal applications.

Presentation(s):

Send Email for Hamidreza Zoraghein

Stefan Leyk

Associate Professor
University of Colorado Boulder

PhD University of Zurich, Department of Geography (Dr. sc. nat.), 2005
M.S. Dresden University of Technology, Department of Forest Sciences, 2002
B.S. Dresden University of Technology, Department of Forest Sciences, 2001
03/2016-present FACULTY ASSOCIATE - Institute of Behavioral Science. University of Colorado Population Center (CUPC)
07/2014-05/2015 VISITING PROFESSOR - Department of Geography, University of Zurich, Switzerland
07/2014-present ASSOCIATE PROFESSOR - Department of Geography, University of Colorado Boulder, U.S.A.
08/2007-06/2014 ASSISTANT PROFESSOR - Department of Geography, University of Colorado Boulder, U.S.A.

Presentation(s):

Send Email for Stefan Leyk

Send Email for Xiaobai Angela Yao


Assets

3706.2 - Estimating changes in urban land and urban population using refined areal interpolation techniques

Handout

Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for Estimating changes in urban land and urban population using refined areal interpolation techniques