An Overview of Methods for Estimating Urban Populations Using Nighttime Satellite Imagery Paul Sutton Department of Geography University.

Slides:



Advertisements
Similar presentations
Multiple Indicator Cluster Surveys Survey Design Workshop
Advertisements

DEPARTMENT OF GEOMATIC ENGINEERING Mapping Anthropogenic Activities from Earth Observation Data Christopher Doll, Jan-Peter Muller Workshop on Gridding.
An Exploration of using Nighttime Satellite Imagery from the DMSP OLS for Mapping Population and Wealth in Guatemala Paul C.Sutton Department.
BEN ANDERSON PROJECT MANAGER UNIVERSITY OF LOUISVILLE CENTER FOR HAZARDS RESEARCH AND POLICY DEVELOPMENT Using Dasymetric Mapping.
Multi-Scale Analyses Using Spatial Measures of Segregation Flávia Feitosa New Frontiers in the Field of Segregation Measurement and Analysis Monte Verita,
Relationships between Nighttime Imagery and Population Density for Hong Kong Qing Liu Paul C. Sutton Christopher D. Elvidge Asia Pacific Advanced Network.
Improving intraurban land use characterization with nighttime imagery Sharolyn Anderson, Assistant Professor, University of Denver Paul C. Sutton, Associate.
Vegetation and Population Density in Urban and Suburban Areas in the U.S.A. Francesca Pozzi Center for International Earth Science Information Network.
19 th Advanced Summer School in Regional Science Overview of advanced techniques in ArcGIS data manipulation.
Correlation and Autocorrelation
Estimation of GDP at Sub-National Scales using Nighttime Satellite Imagery Paul Sutton Department of Geography University of Denver April, 2005 Presentation.
Using Impervious Surface as a spatially explicit Proxy Measure of CO 2 Emssions Dr. Paul C. Sutton Dr. Sharolyn Anderson Dr. Sharolyn Anderson Department.
Paving the Planet: Mapping and Monetizing Human Impact on the Earth Presentation Footprint Forum June 11, 2010 Colle di Val d’Elsa Paul C. Sutton Department.
Mapping the Dollar value of natural production and the Dollar cost of human consumption globally Paul C. Sutton & Benjamin Tuttle Department of Geography.
Using Impervious Surface as a spatially explicit Proxy Measure of CO 2 Emssions Dr. Paul C. Sutton Department of Geography University of Denver AAG presentation.
Progress in Empirical Measurement of the Urban Environment: An exploration of the theoretical and empirical advantages of using Nighttime Satellite Imagery.
Global and continental population databases “Supply side view” What has been done Related developments Possible next steps.
Remote Sensing of Urban Landscapes and contributions of remote sensing to the Social Sciences.
An Empirical Environmental Sustainability Index derived solely from Nighttime Satellite Imagery and Ecosystem Service Valuation Paul Sutton
Geog 458: Map Sources and Errors Uncertainty January 23, 2006.
GIS in Spatial Epidemiology: small area studies of exposure- outcome relationships Robert Haining Department of Geography University of Cambridge.
Stratified Simple Random Sampling (Chapter 5, Textbook, Barnett, V
Estimating Gross Domestic Product, Informal Economy and Remittances of Mexico using Nighttime Satellite Imagery Tilottama Ghosh Dr. Paul C. Sutton Dr.
GIS 2, Final Project: Creating a Dasymetric Map for Two Counties in Minnesota By: Hamidreza Zoraghein Melissa Cushing Caitlin Lee Fall 2013.
Gridded Population Workshop: New York: May 2000 High resolution and local scale: national population surface models from the UK Censuses David Martin Department.
A Century of Classification: The Census Bureau’s Urban and Rural Classification, Michael Ratcliffe Geography Division U.S. Census Bureau.
Lecture 15 Basics of Regression Analysis
Spatial Data Analysis Areas I: Rate Smoothing and the MAUP Gilberto Câmara INPE, Brazil Ifgi, Muenster, Fall School 2005.
Food Store Location Analysis Albuquerque New Mexico, 2010 Prepared for: Geography 586L - Spring Semester, 2014 Larry Spear M.A., GISP Sr. Research Scientist.
An Object Oriented Algorithm for Extracting Geographic Information from Remotely Sensed Data Zachary J. Bortolot Assistant Professor of Geography Department.
Collaborative Tool for Collecting Reference Data on the Density of Constructed Surfaces Worldwide Chris Elvidge NOAA-NESDIS National Geophysical Data Center.
Scale Effect of Vegetation Index Based Thermal Sharpening: A Simulation Study Based on ASTER Data X.H. Chen a, Y. Yamaguchi a, J. Chen b, Y.S. Shi a a.
Exploring Metropolitan Dynamics with an Agent- Based Model Calibrated using Social Network Data Nick Malleson & Mark Birkin School of Geography, University.
Census Data for GIS and Planning Professionals GIS in Action April 15, 2014 Charles Rynerson Census State Data Center Coordinator Population Research Center.
Department of Cognitive Science Michael J. Kalsher PSYC 4310 COGS 6310 MGMT 6969 © 2015, Michael Kalsher Unit 1B: Everything you wanted to know about basic.
Regression Analysis. Scatter plots Regression analysis requires interval and ratio-level data. To see if your data fits the models of regression, it is.
Role of Statistics in Geography
Health Datasets in Spatial Analyses: The General Overview Lukáš MAREK Department of Geoinformatics, Faculty.
Today: Our process Assignment 3 Q&A Concept of Control Reading: Framework for Hybrid Experiments Sampling If time, get a start on True Experiments: Single-Factor.
The Spatial GINI Coefficient
Spatial Data Analysis Yaji Sripada. Dept. of Computing Science, University of Aberdeen2 In this lecture you learn What is spatial data and their special.
Basic Concepts of Correlation. Definition A correlation exists between two variables when the values of one are somehow associated with the values of.
MDG data at the sub-national level: relevance, challenges and IAEG recommendations Workshop on MDG Monitoring United Nations Statistics Division Kampala,
November 19, The City and Citizenship. General Definitions  a large and densely populated urban area; may include several independent administrative.
Summary of Tract-to-Tract Commuter Flows by Type of Geographic Area. A useful way of comparing the general pattern of tract-to-tract commuter flows across.
Chapter 6: 1 Sampling. Introduction Sampling - the process of selecting observations Often not possible to collect information from all persons or other.
United Nations Workshop on Revision 3 of Principles and Recommendations for Population and Housing Censuses and Evaluation of Census Data, Amman 19 – 23.
Inference: Probabilities and Distributions Feb , 2012.
Using the National Land Cover Database and LIDAR to reveal urban abandonment in Detroit Emily Thompson, Kirsten de Beurs Department of Geography and Environmental.
Spatial Smoothing and Multiple Comparisons Correction for Dummies Alexa Morcom, Matthew Brett Acknowledgements.
GIS September 27, Announcements Next lecture is on October 18th (read chapters 9 and 10) Next lecture is on October 18th (read chapters 9 and 10)
Psychology 202a Advanced Psychological Statistics October 22, 2015.
1 Part09: Applications of Multi- level Models to Spatial Epidemiology Francesca Dominici & Scott L Zeger.
Impervious Surface Area of the Conterminous United States Christopher D. Elvidge John B. Dietz Paul S. Sutton.
Statistical Concepts Basic Principles An Overview of Today’s Class What: Inductive inference on characterizing a population Why : How will doing this allow.
INTRODUCTION Despite recent advances in spatial analysis in transport, such as the accounting for spatial correlation in accident analysis, important research.
Lessons Learned from the production of Gridded Population of the World Version 4 (GPW4) Columbia University, CIESIN, USA EFGS October 2014.
Institute for Transport Studies FACULTY OF ENVIRONMENT CQC Efficiency Analysis Concepts and approach Dr Phill Wheat Senior Research Fellow 14 th October.
David Housman for Math 323 Probability and Statistics Class 05 Ion Sensitive Electrodes.
Michael Xie, Neal Jean, Stefano Ermon
Regression Analysis.
Workshop on Land Accounts and urban morphology, ETC-CE, 12 july 2006
CHAPTER 29: Multiple Regression*
By Lewis Dijkstra, PhD Deputy Head of the Economic Analysis Unit,
Outline. A 2010 Mapping of the Constructed Surface Area Density for China Preliminary Results.
Paul C. Sutton & Benjamin Tuttle Department of Geography
The Statistics Canada population centre and rural area definition and the proposed European and Global version of the degree of urbanization: a short comparative.
Expert Expert Group Meeting on Statistical Methodology for Delineating Cities and Rural Areas Iven M. Sikanyiti 28th-30th January 2019 United Nations:
Department of Geography
Paul C. Sutton & Benjamin Tuttle Department of Geography
Presentation transcript:

An Overview of Methods for Estimating Urban Populations Using Nighttime Satellite Imagery Paul Sutton Department of Geography University of Denver May, 2000

Outline ‘Known’ Population Data how good/bad is it?‘Known’ Population Data how good/bad is it? Data: brief description of DMSP OLS imageryData: brief description of DMSP OLS imagery Estimating the Population of cities/urban clustersEstimating the Population of cities/urban clusters Estimating intra-urban population densityEstimating intra-urban population density ‘Temporally-Averaged’ Population Density‘Temporally-Averaged’ Population Density Questions of spatial and temporal scaleQuestions of spatial and temporal scale Summary/ConclusionsSummary/Conclusions

How Good are the numbers and who cares? When did the world population reach 6 billion? Absolute population of Cities Mexico City Mexico City –U.S. Census Bureau 28 million –United Nations 16 million Sao Paulo Sao Paulo - U.S Census Bureau 25 million - U.S Census Bureau 25 million - United Nations 16 million - United Nations 16 million Shanghai Shanghai - U.S. Census Bureau 8 million - U.S. Census Bureau 8 million - United Nations 15 million - United Nations 15 million Istanbul Istanbul - Nat. Geog. Atlas* (1999) - Nat. Geog. Atlas* (1999) 2,938,000 [3,258,000] 2,938,000 [3,258,000] - Nat. Geog. Atlas* (1995) - Nat. Geog. Atlas* (1995) 6,620,200 [7,309,200] 6,620,200 [7,309,200] * Cited PRB and U.S. Census Bureau Percent of Population Urban Models described here produce national population estimates very sensitive to these numbers. Errors inflate with increasing rural fraction of population Models described here produce national population estimates very sensitive to these numbers. Errors inflate with increasing rural fraction of population Spatial Accuracy Spatial Accuracy The 1994 Guatemala census included hundreds of populated places never previously enumerated. Nevertheless, the spatial characteristics of these data were rudimentary. The “maps” supplied to enumerators in some frontier districts were generally hand drawn and based on anecdotal information. As a consequence, we have better information than ever before regarding the size and character of the Guatemalan population, we still lack a clear sense of where these people are. The 1994 Guatemala census included hundreds of populated places never previously enumerated. Nevertheless, the spatial characteristics of these data were rudimentary. The “maps” supplied to enumerators in some frontier districts were generally hand drawn and based on anecdotal information. As a consequence, we have better information than ever before regarding the size and character of the Guatemalan population, we still lack a clear sense of where these people are.

Nighttime Satellite Imagery (DMSP OLS) ‘Percent Observation’ This hyper-temporal imagery used to measure urban areal extent

Aggregate Estimation of Total City Populations Method I: Conterminous U.S. Imagery: DMSP OLS “Percent Observation” ‘Truth’: Wall to wall grid of Pop. Den. From 1990 Census Block Groups Method: Cluster adjacent pixels & Count them to measure Areal Extent of Cluster, overlay to obtain corresponding Population Method II: All Nations of the World Imagery: DMSP OLS “Percent Observation” ‘Truth’: Point Dataset of over 3000 cities with known population Method: Threshold, Cluster, & Count pixels for Area, Geo-reference & Overlay to obtain nationally specific slope & intercept parameters for the Ln(Area) vs. Ln(Popualtion) relationship from known cities, apply to all clusters

Method 1: Proof of Concept with U.S. Data (Note: This also worked well with Mexico Data)

Method II: Going Global Use 1,383 Known Urban Populations to Estimate Populations of the 22,920 urban clusters found in DMSP OLS imagery Thresholding: Trade-off between too much conurbation and ability to see small settlementsThresholding: Trade-off between too much conurbation and ability to see small settlements Geo-Location : Provide each identified urban cluster with Country ID and related national StatsGeo-Location : Provide each identified urban cluster with Country ID and related national Stats Regression: Using Ln(Area) vs. Ln(Population) relationship to identify nationally specific slope and intercept parameters for every nationRegression: Using Ln(Area) vs. Ln(Population) relationship to identify nationally specific slope and intercept parameters for every nation Estimation: Estimate population of all 22,920 cluster with parameters and use % urban statistic to get total national population estimateEstimation: Estimate population of all 22,920 cluster with parameters and use % urban statistic to get total national population estimate

Thresholding: As thresholding increases intercomparisons of parameters become increasingly difficult Medium & High GDP/Capita – 80 Low GDP/Capita - 40

Regression Scatterplot of All Cities/Urban Clusters of the World w/ Known Populations Scatterplot of All Cities/Urban Clusters of the World w/ Known Populations All Cities (N= 1,404): Ln(pop) =.850* Ln(Area) R 2 = 0.68 High Income Cities (N=471): Ln(pop) = 1.065*Ln(Area) R 2 = 0.77 Medium Income Cities (N=575): Ln(pop) = 1.011*Ln(Area) R 2 = 0.78 Low Income Cities (N=358): Ln(pop) = 0.989*Ln(Area) R 2 = 0.80 Venezuelan Cities (N=15):Venezuelan Cities (N=15): Ln(pop) = 1.164*Ln(pop) R 2 = 0.84

Example of estimating nationally specific regression parameters for Venezuela

Some Results The Big Ugly Table that you can’t read…. Estimated and actual populations, regression parameters etc.

Some more results…. A smaller table you might be able to read

How did it go with the Biggest Cities?

Disaggregate or ‘Intra-Urban’ estimates of Population Density Allocate aggregate estimate of total city population to pixels within urban clusterAllocate aggregate estimate of total city population to pixels within urban cluster Use linearly proportional relationship between light intensity and population densityUse linearly proportional relationship between light intensity and population density Compare to residence and employment based measures of population densityCompare to residence and employment based measures of population density

Radiance Calibrated DMSP OLS images of Denver aka ‘Low-Gain’ or Light Intensity This imagery used to model intra-urban population density

Formal & Graphical Representation of the Model

Actual, Modeled, and Smoothed Representations of Minneapolis

Some Results….

What do the Errors look like?

Temporally Averaged Population Density Census data is typically a residence based measure of population density People, work, shop, go to school, and entertain & transport themselves outside of the home Is a temporally averaged measure of population density useful? (e.g. for a given 1 km 2 area with 600 people in it for 8 hrs, 300 in in the next 8 hours, and 0 people in it the last 8 hours it has a temporally averaged population density of 300 persons/km 2 )

Are DMSP OLS based estimates of population density a temporally averaged measure of population density?

Questions of Spatial & Temporal Scale Is a population density dataset at a 1 km 2 spatial resolution useful forIs a population density dataset at a 1 km 2 spatial resolution useful for –Vulnerability studies? –Land-use Land-cover change studies? –Environmental Modeling? What kind of temporal resolution of population density representations are useful and needed?What kind of temporal resolution of population density representations are useful and needed? What measures other than simple density are needed and what means are there to acquire them?What measures other than simple density are needed and what means are there to acquire them? When are errors of population numbers and/or spatial location unacceptably large?When are errors of population numbers and/or spatial location unacceptably large?

What’s Going on in 1 km 2 ?

Summary/Conclusions Nighttime Satellite imagery from DMSP OLS can be used to: 1) Estimate the population of urban agglomerations around the world 1) Estimate the population of urban agglomerations around the world 2) Estimate intra-urban temporally averaged measures of population density 2) Estimate intra-urban temporally averaged measures of population density Continuing research will shed light on improved means of delineating areal extent of cities using the radiance calibrated datasets, better explanations of the national variations in the slope and intercept parameters, and a greater understanding of the spatio-temporal nature of the population density estimates produced by these methods Future research should be informed by the potential users of these datasets as to the spatial and temporal scale required, and the numerical and spatial accuracy required There is potential for inclusion of these methods into the suite of tools used for conducting national censuses throughout the world