Estimation of poverty rates based on satellite observed nighttime lights Chris Elvidge, NOAA National Geophysical Data Center (NGDC) Boulder, Colorado.

Estimation of poverty rates based Estimation of poverty rates based on satellite observed nighttime on satellite observed nighttime

lightslights

Chris Elvidge, NOAA National Geophysical Data Center (NGDC)Boulder, Colorado Tel. 1-303-497-6121 [email protected]

Kimberly Baugh, Ben Tuttle, Daniel Ziskin, Tilo GhoshCIRES University of Colorado

August 6, 2008

The U.S. Air ForceDefense Meteorological Satellite Program (DMSP)Operational Linescan System (OLS) has aUnique capability to collect low-light imagery.

Polar orbiting3000 km swath2.7 km ground sample

distance (GSD)Two spectral bands:

visible and thermalNightly global coverageFlown since 1972Will continue till ~2012

Visible

Thermal

Originally designed Originally designed for the detection for the detection of moonlit clouds, of moonlit clouds, the OLS detects the OLS detects lights from cities, lights from cities, towns, villages, towns, villages, gas flares, fires, gas flares, fires, and heavily lit and heavily lit fishing boats.fishing boats.

Fires in Africa

Gas flares – Persian Gulf

Fishing boats & city lights - Japan

Can lighting per person be Can lighting per person be used as an estimator for used as an estimator for

poverty levels?poverty levels?

Landscan Population Count as red, lights as green and blue.

The World Bank’s World The World Bank’s World Poverty MapPoverty Map

• Index based on $2 per person per day not a valid poverty indicator in developed countries. • Wide disparities in survey years and methods.• Governments can influence the outcome.• Infrequent updates.• Spatial variation of poverty levels within countries not revealed.

CalibrationCalibration

First Satellite Derived First Satellite Derived Global Poverty MapGlobal Poverty Map

Population Count in Poverty Population Count in Poverty Green 1-10, Yellow 11-50, Red > 50Green 1-10, Yellow 11-50, Red > 50