Merging statistics and geospatial information, 2018 projects - Iceland

Higher resolution Icelandic statistical geography standard for register-based census statistics; 2018 project; final report 18 December 2020

This article forms part of Eurostat’s statistical report on the Integration of statistical and geospatial information.

Full article

Problem

For the 2011 census data, Statistics Iceland created 42 statistical output areas (SOAs) with an average population size of 7 500 persons. Iceland aims to fulfil the small area European requirements for the population and housing census, in addition to the creation of 1 km x 1 km grid data. To do this, the statistical output areas can be subdivided into minor statistical output areas (MSOAs), with an average population size of approximately 1 500 to 2 000, increasing the number of areas from 42 to approximately 200.

Objectives

To create a higher resolution spatial statistics production system for the 2021 census, by:

building an improved Icelandic Statistical Geography Standard (ISGS), in other words, a hierarchy of small area geographies designed for register-based census data (statistical output areas, SOAs);
formulating a geospatial strategy and methodology;
using the Icelandic INSPIRE geospatial data together with the 2011 census data (used for testing) in order to build web atlases.

Method

The main work consisted of building the small areas of the urban and rural areas of Iceland. The methods are different for these two types of areas, due to the distribution of the population which is concentrated in very few urban areas, while most highlands and rural areas are extremely sparsely populated.

The resulting MSOAs need to fulfil a number of constraints which are expressed through finding an optimal value for an objective function. The following constraints were considered:

total population for each MSOA falls between 600 and 3 500;
average population among all MSOAs is close to 1 750;
high homogeneity inside each MSOA, for example in terms of education, income and occupation;
high heterogeneity across all MSOA;
contiguity is required for all MSOAs;
municipality boundaries are respected;
the MSOAs are a disaggregation of the existing SOAs.

For sparser populated municipalities, the MSOAs were constructed by expert opinion where it is often necessary to combine many municipalities into a single target zone in order to meet the minimum population criteria. In some cases, larger localities were split from their surrounding rural areas.

For more densely populated areas, source zones (from which the MSOAs were to be built) were defined in terms of available boundaries of planning zones obtained from the relevant authorities. Attributes based on the 2011 census data were averaged over households within each source zone giving extensive data available for combining them to form MSOAs: over 20 000 distinct solutions were generated. The solution selected for each MSOA was based on several weighted criteria and constraints:

population close to 1 750 but no less than 900 and no greater than 3 500,
compactness,
similarity in terms of type and age of buildings.

An open-source algorithm was developed within Statistics Iceland and improved throughout the project.

A web-based census atlas for the whole country was produced, with four higher resolution atlases where the preliminary MSOAs were employed. The atlases were static.

Record swapping and cell key perturbation methods were tested for statistical disclosure control (SDC) of the census data enriched with geospatial attributes of the newly built MSOAs.

Results

The main final results of the project are:

the MSOAs are shared as free, open data via the INSPIRE platform;
the interactive web atlases of register-based 2011 census data were developed for the new MSOAs;
various papers and presentations published/prepared;
collaboration of Statistics Iceland with the National Land Survey of Iceland (NLSI) and the Icelandic Regional Development Institute (IRDI) was strengthened;
information technology solutions at Statistics Iceland were improved by adopting the open source OSKARI – web mapping application and geospatial data analysis platform developed at Statistics Finland;
methodology at Statistics Iceland was improved with the testing and evaluation of R-based tools for SDC methods for the case study of small area disaggregated tabular census data.

Figure 1: The partitioning of the locality of Selfoss within the SOA “27 Árborg”