Population Maps of Latin America
Glenn Hyman, Andy Nelson, German Lema
International Center for Tropical AgricultureCali, Colombia
Outline
• Why we need population maps
• The Database
• Gridding methods
• Verification
• Relationships between population and other variables
• Future Research
Why we need population maps
Raster Population Surface
Potential Agricultural Productivity
Population summedby zones
Note: Areas en black are cities.Zones in Purple are higher rural population densities. Zones in green are low rural population densities.
Table 7. Potential Agricultural Productivity and Rural Population in Latin America
103 kcal. Ha-1 yr-1 Area (km2) Rural Population
High level of inputs
cold climates 605,489 6,266,441< 500 185,006 14,324,680
501-1000 78,097 241,3711001-2500 640,171 10,612,3652501-5001 725,593 13,710,7785001-7500 482,869 6,522,0247501-10000 581,880 10,154,42610001-12500 423,531 28,478,42412501-15000 306,160 29,708,45215001-20000 390,005 29,968,818
> 20000 803,669 22,745,092
Low level of inputs
cold climates 613,040 8,254,366< 500 883,568 17,896,746
501-1000 120,468 134,432,5601001-2500 176,031 46,270,5042501-5001 400,196 46,998,612
Note: Low input level map is for Latin America and the Caribbean High input level map only includes South America
The Database
10,400 unitsmost are municipios
Municipiolevel
Parroquia level
CIAT Administrative Division Database
Population Density in Latin America and the Caribbean
< 1
1-10
10-25
25-50
50-100
>100
people/km2
Source: CIAT’s Population Database
r = ( ln ( pf / pi ) ) / t
where:r = rateln= logaritmo naturalpf = population at final timepi = population at initial timet = period of time
• Growth rates were calculated from ECLAC’s population database at the department level• An exponential growth rate function was used• Population was estimated to the year 2000• We used figures from the last 2 population censuses at department level to calculate population growth rates
Gridding Methods
• Equal distribution (total population / number of cells in unit)
• Pycnophylactic
• Accessibility
• Weight model
Roads and Rivers from DCW
Caminos y Rios
Pendientes
Luces Nocturnas
Nighttime Lights
Slope from 1 km DEM
Continental ScaleAccessibility Mapping
Model based on traveltime to populated places
Model uses roads, rivers,populated places, slope, and land cover to estimatetravel time
Population from Census
Accessibility Model
Verification
Municipio Distrito
Ecuador 203 976
Peru 191 1836
Verification data from different scales and independently collected
Model where municipal leveldata is verified from data aggregated to finer district level
PERU
Variable N Mean Minimum Maximum CV Sum ------------------------------------------------------------------------ DAREA 1781 196.40 0.13 16741.92 371.72 349797.39 DPYCNO 1781 190.89 0.17 20361.68 378.29 339991.89 DACCESS 1781 168.63 0.04 17139.46 381.31 300337.16 DACCESS_2 1781 160.16 0.02 15857.71 352.97 285248.06 DWEIGHT 1781 148.56 0.02 11166.89 338.47 264588.71
ECUADOR Variable N Mean Minimum Maximum CV Sum ---------------------------------------------------------------------- DAREA 918 235.06 0.01 17803.46 354.12 215788.54 DPYCNO 918 235.38 0.20 16337.44 342.67 216080.66 DACCESS 918 176.95 0.12 5001.63 248.19 162441.51 DACCESS_2 918 178.11 0.22 4116.35 233.95 163508.02 DWEIGHT 918 154.62 0.59 5022.37 235.05 141944.20
Modeled population counts compared to census datafor Peru and Ecuador verification data sets: percentagedifferences
Ecuador
BEST Frequency Percent ----------------------------- area 150 16.5 access 154 16.9 access_2 168 18.5 pycno 183 20.1
weight 255 28.0
Peru
BEST Frequency Percent ----------------------------- area 265 15.0 access 320 18.1 access_2 404 22.8 pycno 291 16.4 weight 492 27.8
Relationships between population and other variables
Department Population vs Night Light Area
R2 = 0.8103
0
10000
20000
30000
40000
50000
0.0E+00 5.0E+06 1.0E+07 1.5E+07 2.0E+07 2.5E+07 3.0E+07 3.5E+07
SAO PAULO
RIO DEJ ANEIRO
Department Population vs Summed Frequency
R2 = 0.8727
0
500000
1000000
1500000
2000000
0.0E+00 5.0E+06 1.0E+07 1.5E+07 2.0E+07 2.5E+07 3.0E+07 3.5E+07
Distribution of Rural Population in Latin America and the Caribbean
Central America, the Central America, the Andes, and the Andes, and the Caribbean have the Caribbean have the highest rural population highest rural population densitiesdensities..
0-11-1010-2020-3030-4040-5050-6060-7070-8080-9090-100>100No Data
People per km2
Source: CEPAL
Rural Model
Variable R2 Cumulative R2
URBANO 0.52 0.52 AGUA 0.23 0.75
PEN25_50 0.06 0.81
Variable Parameter F Prob > F Estimate
AGUA 441.14117780 346.35 0.0001 URBANO 4373.57019841 774.39 0.0001
PEN25_50 110.28701024 111.94 0.0001
Urban Model
Variable R2 Cumulative
R2 URBANO 0.7548 0.7548 LUC100 0.0376 0.7924 LUC25 0.0146 0.8069
AGUA 0.0133 0.8202
Variable Parameter F Prob>F Estimate
URBANO 36362.6 694.94 0.0001 LUC100 2564.6 80.49 0.0001 LUC25 -2558.5 34.26 0.0001 AGUA 873.8 23.48 0.0001
Future Work
• Gridding using statistical results • Further verification using data from Peru,
Honduras and Ecuador
• Compare to other work (eg. ORNL 1 km database)
• A Rural population map?
• New Map with 2000 Census round focusing onAmazon basin, Central American hillsides, Latin American savanas.