EUROSTAT
Directorate E: Sectoral and regional statistics
Unit E4: Regional statistics and geographical information
E4/REG/2015/70
(Only available in EN)
Working Group on Regional, Urban and Rural Development statistics
to be held in Luxembourg on 19.10.2015 and 20.10.2015
Rural development indicators and Degree of Urbanisation
Item 7.2 of the agenda
The members of the Regional, Urban and Rural Development Statistics Working Group are asked:
· To take note of the recent developments and the planned future actions with respect to the rural development statistics including the Degree of Urbanisation;
· To express their views on the dissemination activities of Eurostat;
· To give a feedback on the Degree of Urbanisation update.
Abstract
The document is intended to inform the delegates on the state of play of the rural development statistics in particular the major update of the European Degree of Urbanisation as well as the forthcoming update of the regional typologies in particular the Urban-Rural Typology.
1. Introduction
Rural development is a broad concept mainly covering the improvement of the quality of life and encouraging diversification of the economy in rural areas. Rural development represents the second pillar of the Common Agricultural Policy (CAP). Moreover, the urban-rural aspects of EU regions play an important role in the EU regional policy.
Rural development statistics focus especially on socio-economic fields such as demography, the labour market, employment, education, economic accounts or economics activities (e.g. agriculture, industry, tourism, etc.). "Rurality" for statistical purpose has been defined at NUTS level 3 using the Urban-Rural typology. This typology is based on the Degree of Urbanisation concept that classifies the Local Administrative Units level 2 (LAU 2) into cities, towns and suburbs and rural areas[1].
In order to meet the policy needs to support the monitoring and evaluation of the CAP 2014-2020 from the beginning of the approval process Eurostat provided its observations to 160 Interservice consultations concerning national and regional rural development programs that will be supported by the European Agricultural Fund for Rural Development. As a result from the consultations, most of the programs’ final versions make a wide use of Eurostat’s data and employ the Eurostat’s definition of Rural region.
2. The update of the Degree of Urbanisation
While the existing version of the Degree of Urbanisation (DEGURBA) was based on the population grid 2006, a new population grid based on the 2011 census results became available in summer 2015. The new population grid not only provides much more recent population data, it is also much more accurate for those countries which switched from a disaggregated to a bottom-up grid for the reference year 2011.
In order to benefit from the new 2011 population grid data, a first version of the updated DEGURBA classification has recently been produced by the GISCO team of Eurostat unit E4. In addition to the 2011 population grid, this update uses the 2014 LAU 2 boundaries (European Boundary Map Version 9). The underlying methodology to calculate the DEGURBA has not changed compared to the previous version. However, for some exceptional cases a refinement procedure had to be applied. For detailed information on the algorithm, please refer to Annex I and II.
While Eurostat will continue its own quality checks on the updated DEGURBA, it will provide Member States with the DEGURBA for their countries for evaluation purposes. Member States are invited to give feedback on the new Degree of Urbanisation by 13 November to: .
Eurostat will then release the updated DEGURBA on its website as soon as there is an agreement on the classification.
Eurostat will also thoroughly analyse the impact of the new DEGURBA on the data published by this classification, in cooperation with the domains that include the DEGURBA variable in their surveys. The grid effect and the impact of the LAU 2 boundaries’ changes will be a subject of the analysis as well. Data for the first reference period using the updated DEGURBA will need to be flagged as “break in time series” when disseminated.
3. New datasets of rural development indicators and by Degree of Urbanisation
With respect to the implementation of the ESA 2010, 3 new regional accounts datasets by urban-rural typology have been recently published. The previous table based on ESA 95 will remain available in separate folders.
A new metadata file of regional typologies has been published recently. It contains detailed information on data by Urban-Rural typology (methodology, definitions, statistical domains coverage, data quality aspects).
The availability of online datasets by Degree of Urbanisation has been expanded significantly for the domains LFS, Education, EU-SILC and ICT, with links to relevant metadata files published by the production units. More information on the datasets available can be found in the overview on regional statistics on the Eurostat website[2] (see also the Annex for item 6 on the agenda).
4. Further Work
The production units in Eurostat publish many regional variables and indicators by NUTS level 3. NUTS level 3 are the building blocks of the Urban-Rural typology maintained by unit E.4. Not all data available at NUTS level 3 have been aggregated and published by urban, rural and intermediate regions. Policy analysts and other data users are preparing the aggregates of interest manually.
In the frame of an IT project all NUTS 3 available data will be aggregated be the three main regional typologies including the Urban-Rural typology. 24 new datasets will be published in EuroBase by the middle of 2016. A new production database will be created for this purpose. With the new database it will be possible to create, in a very flexible manner, any regional aggregate of regions under request, check its reliability and, if the aggregate is reliable enough, to make it available to the public (for instance aggregates of island, mountain or border regions).
With respect to the maintenance of the regional typologies, the next step is to update the Urban-rural typology on the basis of the new Degree of Urbanisation and the NUTS 2013 classification. The impact of the update on the relative share of people living in different types of regions will be analysed.
Annex I
Methodological Note
on Determining the degree of urbanisation at LAU2 level
Table of contents
Introduction 2
GEOSTAT Grid 2
Degree of Urbanisation 3
Coherence with Urban Audit city definition 4
Introduction
This document contains technical notes explaining the workflow (using ArcGIS Desktop 10.0) to determine the Degree of Urbanisation of Local Administrative Units (LAU2) in Europe.
The initial phases of the workflow focus on managing the 1 km² grid cells and the related population data to produce 1° Urban Clusters and 2° High density Clusters (= Urban Centres).
1. GEOSTAT Grid
The GEOSTAT grid is composed of regular 1 sqkm cells. These are the basic units of analysis when determining urban and rural areas at grid level.
The reference layers of the GEOSTAT grid:
2 spatial layers are available:
· GEOSTAT_GRD_RG: polygon layer containing the full extent of all 1 square kilometre cells in ETRS LAEA projection.
· GEOSTAT_GRD_PT: point layer showing the centres of the 1 square kilometre cells in ETRS LAEA projection.
To these spatial layers can be joined (on GRD_ID field) an attribute table:
· GEOSTAT_GRD_AT: this table contains several attributes describing each cell and can be used to collect various additional characteristics of the grid cells.
The join between this attribute table and the to spatial layers (preferably the point layer) can be used to generate several raster layers:
· POPL_DENS_GR_1KM_2006_GEOSTAT (population density, float)
· POPL_GR_1KM_2006_GEOSTAT (population counts, integer)
Pixels of both rasters have the same value when they are fully located on land surface. But they differ when located partially above water, because POPL_DENS_GR_1KM_2006_ GEOSTAT has been calculated using a land-proportion factor.
Based on these 2 rasters, the Urban Clusters and High density Clusters can be computed.
Tools (in Model Builder) were created to automatize cluster generation.
Urban Cluster Generation:
Urban clusters are contiguous cells with a density of at least 300 inhabitants/km² and a total population ≥ 5000 inhabitants. The model URBAN_cluster_generation in CLST_CALCULATION.tbx produces: 1° a raster of Urban Clusters with, for each one, its total population; 2° a mask raster giving the location of Urban Clusters (with values 0 or 1).
High density Clusters ( = Urban Centres) Generation:
High density clusters are produce thanks to the 3 following tools in CLST_CALCULATION .tbx:
· The model HDENS_cluster_generation selects all grid cells with a density of more than 1500 inhabitants/km². The contiguous high-density cells are then clustered and gaps inside them are filled. Only the clusters with a minimum population of 50 000 inhabitants are kept as an "Urban Centre".
· The model MakeRST4Smoothing creates a raster of high-density clusters used as input in the next tool.
· With the model SmoothIteration, small bays in the high-density clusters are smoothed using the majority rule iteratively. The majority rule means that if at least five out of the eight cells surrounding a cell belong to the same high-density cluster it will be added. This tool must be used many times, until no more cells are added (until HD_CLST (t) = HD_CLST (t-1))
New raster layers are (in ArcSDE):
· URB_CLST_2006_GEOSTAT
· HDENS_CLST_2006_GEOSTAT
Once generated, clusters can be converted to polygons. The point feature class GEOSTAT_GRD_PT is then intersected with these polygons to identify urban cluster cells or high-density cluster cells. This information is part of the attributes of GEOSTAT_GRD_AT.
2. Degree of Urbanisation
With the population grid, the high-density clusters and the urban clusters, the degree of urbanisation can be calculated for LAU2s.
This can been done using the model:
DGUR_CALCULATION.tbx\CALCULATE_DGUR_CODE_2011
DGUR_CODE Description
1 densely populated area (cities)
2 intermediate (town and suburbs)
3 thinly populated area (rural areas)
9 no data
If the data to be processed does not contain High density Cluster, the following model must be used (instead of CALCULATE_DGUR_CODE_2011): DGUR_CALCULATION.tbx\CALCULATE_DGUR_CODE_2011_NO_HD_CLST
Nevertheless, it is possible to calculate degree of urbanisation without geoprocessing, only using GEOSTAT_GRD_AT once a LAU2 code column has been added to this table. The proportion of rural population by LAU2 can then be computed by SQL queries using criteria on fields HDENS_CLST_2006, URBAN_CLST_2006 and LAU2_CODE_2011.
This process results in the degree of urbanisation for LAU2s, and can be added to the LAU2 attribute table COMM_AT_2011.
3. Coherence with Urban Audit city definition
In order to ensure coherence between the Urban Audit city definitions and the degree of urbanisation, it has been ensured that all LAU2 defined as densely populated according to the degree are part of one of the cities defined for the Urban audit, and vice-versa.
Annex 2
Refinement of the classification and highlighting exceptional cases
The methodology presents some problem and exceptions due to the spatial resolution and in some cases the quality of the data used for the classification.
The proposed solutions to identify the exceptions and to re-classify possible misclassifications are mainly based on the share of urban cluster.
There are 3 cases:
1. LAU2 without a raster equivalent: by creating the LAU2s grid, some very small LAU2s have no raster equivalent due to their small size. In consequence, these LAU2s could not have been classified with the "main method".
2. Border effects: due to the coarse spatial resolution (1km), compared to the relatively small size of some LAU2s, mismatching of the grid population to the right LAU2 may happen.
3. LAU2 with skewed distribution: it may happen that some LAU2s have no population according to the population grid but are highly populated according to the census data.
LAU2 without a raster equivalent
Inputs:
Urban Cluster (UC)
High Density Clusters (HDC)
LAU2 layer (LAU2)
Workflow:
The first thing to do is to spot these LAU2s: Join the LAU2 layer with the rasterized LAU2 layer. Query for the LAU2s that don't exist in the rasterized LAU2 layer and extract them as layer. We will call it Very small LAU2s (VS_LAU2).
The UC and HDC are transformed to polygons (respectively pUC and pHDC).
Intersect resulting polygons: pUC x VS_LAU2 and pHDC x VS_LAU2.
Join the resulting attribute tables with Very small LAU2s and using the area field calculate the percentage of their area inside UC and HDC.
The rules to classify these LAU2s are:
~ "Densely populated area" if LAU2 area inside HDC > 50 % of the LAU2's area
~ "Intermediate" if LAU2 area inside UC > 50 % of the LAU2's area
~ "Thinly populated area" if none of the precedent rule applies
Border effects
The mismatches happen mainly during the rasterization of the LAU2 layer. The value of a cell is assigned to the LAU2 with the largest area within it; it's easy to figure that small LAU2s with non-regular or elongated area might not be very well represented in large cells (or not represented at all as in the previous case).
Inputs:
Urban Cluster (UC)
High Density Clusters (HDC)
LAU2 layer (LAU2)
Workflow:
Intersect: pUC x LAU2 and pHDC x LAU2.
Join the resulting attribute tables with LAU2 layer and using the area field calculate the percentage of their area inside UC and HDC.
The rules to spot and reclassify these issues are:
· If a LAU2 is classified as "Thinly populated area", its total area < 5 km² and its area inside UC and/or HDC > 30%; it can be reclassified as following:
~ "Densely populated area" if LAU2 area inside HDC > 50 % of the LAU2's area
~ "Intermediate" if LAU2 area inside UC > 50 % of the LAU2's area
· If a LAU2 is classified as "Densely populated area" or "Intermediate", its area inside UC or HDC < 10% and its population < 5000 inhabitants; it should be reclassified as "Thinly populated area"
LAU2 with skewed distribution
Input:
Population Census table, joinable to the LAU2 layer (CENSUS).
LAU2 layer (LAU2)
POPL_GR_1KM_2006_GEOSTAT (POP_grid)
Workflow:
Rasterize the LAU2 layer using 1km cell, clip on POP_grid (called rLAU2).
Zonal stat : rLAU2 x POP_grid (result called LAU2_pop)
Join LAU2_pop and CENSUS.