1. SciDataCon 2016 – 13 – September
Improving Interoperability of Statistical Data: The Case for a CensusML
SciDataCon 2016 – 13 – September
Geospatial Data and Key Characteristics of Geospatial Data Analysis and Science: The way forward
Kytt MacManus (CIESIN) & Vilni Verner Holst Bloch (EFGS/Statistics Norway)

2. Lessons Learned from the production of Gridded Population of the World Version 4 (GPW4)

3. Gridded Population of the World
Gridded (raster) data product developed to provide a spatially disaggregated population layer that is compatible with data sets from social, economic, and Earth science fields.
Population data are transformed from their native spatial units to a global grid of quadrilateral latitude-longitude cells
(2.5 arc minutes in GPW3)
(Balk et al. 2010)

4. Number of Input Units (subnational geographic units)
History of GPW
GPWv1 was an outgrowth of a Global Demography Workshop held at CIESIN in 1994
Consensus that a consistent global database of population totals in raster format would be invaluable for interdisciplinary study (Deichmann et al., 2001)
GPWv1
GPWv2
GPWv3
GPWv4
Publication Year
1995
2000
2005
2014/2015
Years of Estimation
1994
1990, 1995
1990, 1995, 2000
2000, 2005, 2010, 2015, 2020
Number of Input Units (subnational geographic units)
19,000
127,000
c. 400,000
~ 12,500,000
Grid Resolution
2.5 arc-minute
30 arc-second (1 km)
Census variables
Total Population
Total Population, Sex, Age, Urban/Rural status
Details for each of our versions… not much change between v1-3, but in 4, we are making substantial improvements in # of input units, grid res. And census variables,
I’ll speak about these later in the presentation

5. Challenges with census data and boundaries
Room for ongoing improvements with online dissemination
Different levels of availability of tabular census data and census geography
Significant research required to reconcile census areas with those present in the GIS data and to assign common identifiers
Primary method of census data dissemination
Percent of respondents (121 countries)
Paper publications
52%
Static web pages (html, excel, PDF)
28%
Interactive online databases
14%
CD/DVD 4%
Other 2%
Although improvements in techonology have resulted in increased use of internet for census data dissemination, there is still a lot of room for improvement. Survey
High res. Tabular data is often not as frequently available when bureaus only disseminate via PDF or published. Requires direct contact with census bureaus – time-consuming and challenging to locate, and in some cases is not possible, limiting the accuracy of those countries.
Asymmetry – partially due to separate NSOs and Nat mapping agencies.Particularily small and developing countries still develop skills and capacity with GIS. 64% use GIS
Static maps require time-consuming digitization
Asymmetry – compromise (use non-census – problems to talk about in next slide OR use coarser res. Boundaries, meaning we are unable to grid to the highest level possible, to the detriment of the grid’s accuracy
Source: 2011/2012 survey for the review of the 2010 World Programme on Population and Housing Censuses; UN Statistics Division, 2013.

6. Challenges with integrating census data
Ideally, the NSO releases census population data and digital census geography, with a common identifying code
GIS data from non-Census Source
Refer to the same unit by different names
GIS data might not capture changes in boundaries over time and therefore requires editing
Significant effort is needed to reconcile census areas with those present in the GIS data and to assign common identifiers

7. India Census Atlas Documents boundary change over time
Rich historical data ranging back to the 1800’s
Makes the idea of a Census time-series imaginable

8. Census Drama
Mauritania attempts Census in 2011 but is blocked by protest from minority groups who claim they are not being enumerated

9. Census Drama
Chile releases Census in 2013 which is later withdrawn due to serious errors

10. Lessons Learned from the European Forum on Geography and Statistics

16. Organizational Barriers to Statistical Interoperability
NSIs and NMCAs usually work within different frameworks.
The number of geographers and statisticians are unequally distributed among national mapping and cadaster agencies (NMCAs) and national statistical institutions (NSIs). NMCAs have few statisticians and vice versa.
Two different worlds
one with focus on geographic descriptions with little emphasis on temporal aspects,
another with focus on statistical descriptions with little emphasis on geographical aspects.
Each “world” has developed a huge set of different standards and specifications; at local, national, regional and global level.
Some of the standards are not open, not free or hardly understandable for other than experts in the respective fields.

17. Organizational Issues
NSIs report statistics at all regional levels, however, usually only as tabular data with attributes that have some geographic linkage.
Although mostly consisting of civil servants working for public interests, statistical acts are sometimes a barrier for the exchange and publishing of data.
Statistical acts often gives NSIs wide and free access to microdata on one side
On the other side statistical acts also set restrictions with respect to e.g. protection of privacy, confidentiality and copyrights of data etc…

18. Toward Global Demographic Interoperability

19. Existing Efforts
UNGGIM Working Group on the Integration of Statistical and Geospatial Information
Convened in 2014, ongoing…
Information in the European Community (INSPIRE)
“to advance the development and adoption of geospatial data standards in Europe”
GeoSUR
Regional Initiative to integrate and disseminate spatial data in Latin America and the Caribbean

21. The way forward…
Adoption of standards should be community driven and will likely consist of a combination of Markup Language, Linked Data and Semantic Web, and data development standards.
5 years of person effort to integrate the 2010 round of census is not acceptable…how do we improve international interoperability?