Spatial Poverty Assessments Alex de Sherbinin Senior Staff Associate Center for International Earth Science Information Network (CIESIN) The Earth Institute at Columbia University Deputy Manager NASA Socioeconomic Data and Applications Center (SEDAC) GEOSS Science & Technology Stakeholder Workshop 30 August 2012 NASA Socioeconomic Data & Applications Center (SEDAC) • Focus on human dimensions of environmental change • Integration of social and Earth science data, especially with remote sensing • Direct support to scientists, applied and operational users, decision makers, and policy communities • Strong links to geospatial data community 2 Outline • • • • Spatial poverty data Remote sensing for poverty research Creating a human “observing system” Concluding Thoughts Measures of Well Being • Household income/consumption expenditures • Non-monetary indicators of well being – Malnutrition – Unsatisfied Basic Needs – Infant Morality Rates • Foster, Greer, Thorbecke measures: – Percent of population below the poverty line (Head Count Index; FGT_0) – Average shortfall between welfare levels and the poverty line measured as a percent of the poverty line (Poverty Gap Index; FGT_1) Spatial Poverty Data Why Map Poverty? • Understand spatial patterns and how poverty varies subnationally and across countries • Identify hotspots in need of intervention • Understand the spatial correlates of poverty – Biophysical correlates – Socioeconomic correlates – Spatial isolation or “poverty traps” In South Africa, the urban-rural poverty differential that applies to the country as whole is not necessarily reflected uniformly across all urban-rural gradients within the country In Bangladesh, the pattern of poverty rates is primarily shaped by proximity to the capital Dhaka. Poverty rates rise with distance from Dhaka. Coastal areas are less disadvantaged than the inland remote areas. In Malawi, Llongwe has less poverty within its limits, but is surrounded by regions of very high poverty. Blantyre, by contrast, has very high poverty within its limits, but is surrounded by regions of only moderate poverty. Spatializing Demographic and Health Survey Data Analysis of infant and child mortality For both infant and children, the chances of survival decrease monotonically the further one resides from a city (of 50,000 persons or more), in a 10-country West Africa study Balk, D., T. Pullum, A. Storeygard, F. Greenwell, and M. Neuman. 2004. A Spatial Analysis of Childhood Mortality in West Africa. Population, Space and Place, Vol. 10, No. 3. Global Hunger Map Identification of Hunger Hotspots • Defined by the Millennium Development Project Hunger TF as those sub-national units with rates of childhood malnutrition >20% and >100,000 children who are underweight • 75 sub-national units met this criteria Hunger by Farming Systems 2 1 3 Farming Systems Data Source: Dixon, J., A. Gulliver with D. Gibbon. 2001. Farming Systems and Poverty: Improving Farmers’ Livelihoods in a Changing World. United Nations Food and Agriculture Organization. (Available at http://www.fao.org/farmingsystems/). What are the biophysical correlates of malnutrition? de Sherbinin. 2009. “The Biophysical and Geographical Correlates of Child Malnutrition in Africa” Population, Space and Place Vol.15 Spatial Error Model Results (pseudo-r square) IMR Map IMR (2000) High : 208 Low : 2.0 • Sources – – – – – • • Demographic and Health Surveys (41 countries) Multiple Indicator Cluster Surveys (5 countries) National Human Development Reports (14 countries) National Statistical Offices (16 countries) UNICEF Childinfo – (115 countries) Subnational representation – – – – Source: de Sherbinin et al. AGU 2004. • Converting rates to counts – For each subnational unit, estimates of live births, infant deaths calculated based on gridded population, national fertility data, and subnational IMR. Calibration 8,029 units (6,886 in Brazil and Mexico alone) – Subnational IMR values 77 countries have subnational data; 115 national only adjusted to be consistent with 80% of world population has subnational data national UNICEF 2000 IMR Average 14 units per country (outside Brazil and Mexico) values Growing Season Growing Season (days) High : 365 Low : 0 Analysis for non-wealthy countries only IMR IMR by Growing Season (days) Cumulative Population by Growing Season (days) 36 5 60 -8 9 90 -1 19 12 014 9 15 017 9 18 020 9 33 036 4 24 029 9 30 036 5 0 159 kn o un 36 5 60 -8 9 90 -1 19 12 014 9 15 017 9 18 020 9 33 036 4 24 029 9 30 036 5 159 0 w n kn o un w n 100 90 80 70 60 50 40 30 20 10 0 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 Elevation Elevation High Low Analysis for non-wealthy countries only IMR Cumulative Population by Elevation (meters) IMR by Elevation (meters) 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 100 90 80 70 60 50 40 30 20 10 0 unknown 0-50 50-100 100-500 500-1000 1000- unknown 0-50 50-100 100-500 500-1000 1000- Malaria Malaria Transmission Index High : 37 Low : 0 Analysis for non-wealthy countries only IMR IMR by Malaria Risk Score Cumulative Population (billions) by Malaria Risk Score 100 90 80 70 60 50 40 30 20 10 0 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0 0-.08 .08-3.4 3.4-8.4 8.4-37 0 0-.08 .08-3.4 3.4-8.4 8.4-37 Soils Soil Constraints On Agricultural Production Undefined 1. No constraints 2. 1-20 Slight 3. 20-40 Moderate 4. 40-60 Constraints 5. 60-80 Severe 6. 80-99 Very Severe Analysis for non-wealthy countries only 7. 100 % severe constraints IMR by Soil Constraint Class Cum ulative Population by Soil Constraint Class (1=few constraints; 7=severe constraints) 100 90 80 70 60 50 40 30 20 10 0 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 Unknow n 1-3 4 5 6 7 Unknown 1-3 4 5 6 7 Drought Drought Index High : Low : Analysis for non-wealthy countries only IMR IMR by Drought Index Cumulative Population by Drought Index 100 90 80 70 60 50 40 30 20 10 0 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 Rails % of Grid within 2km of Railroad High : 99 Low : 0 Analysis for non-wealthy countries only IMR Cumulative Population (billions) by Rail Density IMR by Rail Density 100 90 80 70 60 50 40 30 20 10 0 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0 0-5 5-15 15-23 23-100 0 0-5 5-15 15-23 23-100 Ports Distance to nearest port High Low Analysis for non-wealthy countries only IMR Cumulative Population by Nearest Port (decimal degrees) IMR by Nearest Port (decimal degrees) 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 100 90 80 70 60 50 40 30 20 10 0 0-.5 .5-1 1-2 2-3 3-4 4-5 5-7 7-10 10-15 15-24 0-.5 .5-1 1-2 2-3 3-4 4-5 5-7 7-10 10-15 15-24 Compared with the non-poor, poor people are more likely to be found in drought-prone areas with shorter growing seasons Non-poor Poor For the Millennium Ecosystem Assessment CIESIN calculated average IMR within each MA ecosystem. We also calculated another measure of well-being, the ratio of the share of world population to share of world GDP. The drylands are the most disadvantaged. We further calculated rates of population growth within each ecosystem unit, and noted that the drylands had the highest rate of growth. To have fragile ecosystems with low levels of well-being experience the highest population growth is bound to make challenges more difficult in these regions. Millennium Ecosystem Assessment, 2005 Not Poor Somewhat Poor Moderately Poor Poor Extremely Poor The poor are at much greater risk of experiencing a drought Poverty Delhi, India: Multiple Deprivation Index and ASTER Nighttime Thermal Infrared Nighttime Temp MDI / Poverty Nighttime Temp Temperature Houston, Texas: Income Level and MODIS Nighttime Thermal Infrared Income PC Income PC Nighttime Temp Remote Sensing Applications for Poverty Research Night-time Lights Estimates of GDP / Population Comparison of HH Assets Index and Wealth Based on Mean Brightness of NTL Source: Noor et al., Population Health Metrics, 2008 http://www.ciesin.columbia.edu/confluence/display/slummap/Global+Slum+Mapping Dar Es Salaam, Tanzania, 1982 and 2002 Source: Data courtesy of Richard Sliuzas, ITC Neighborhood Mapping Damascus, Syria •Unstructured Settlements •Lowest to lower middle income •Rural migrants •Very loosely structured •Historical ethnic quarters/neighborhoods •Poor residents currently being displaced in some areas with urban development/tourism •Formal Urban Planning •Typical Urban Services •Middle to Upper Income Source: Slide courtesy Eddie Bright, ORNL Settlement characterization tool Source: Slide courtesy Eddie Bright, ORNL Source: Lela Prashad, www.nijel.org Creating a Human “Observing System” Source: www.benwilhelmi.com Frequency of Demographic and Health Surveys Subnational Poverty and Extreme Poverty Prevalence Source: Harvest Choice, http://harvestchoice.org/maps/sub-national-povertyand-extreme-poverty-prevalence Mean Number of Censuses 1970-2010 Migration Data Migration is one of the main demographic drivers of environmental change, yet there are very few data on human movements Source: Adamo, CODATA side event, Rio+20, June 2012 Source: Adamo, CODATA side event, Rio+20, June 2012 Concluding Thoughts • There is a growing availability of spatial poverty data, but – gaps remain • Integration of “bottom up” with “top down” data is possible, but – Development of globally integrated and harmonized subnational SE data is costly – It needs to be driven by specific research or decision making needs – One size fits all approaches for web services are unlikely to work • Growth in novel data sources – anonymized mobile phone records for human movement, crowd sourced data, etc. – are exciting developments, but – as yet have not provided a globally consistent view – Data quality issues may exist