Towards sustainable development exemplified by monitoring land use efficiency in Europe using SDG 11.3.1

The empirical investigation of SDG 11.3.1 comprises 39 Eionet (European environment information and observation network) countries, both member (32 nations) and cooperating (seven nations) countries (Fig. 1).

Figure 1

Information on urbanized areas was derived from the CORINE land Cover (CLC) datasets of 2006, 2012 and 2018 via Copernicus Land Services. The population counts at the country level referred to the years 2006, 2012 and 2018, and were obtained from the Eurostat web page. The countries’ boundaries (NUTS 0) at the scale 1:60,000,000 came from CISCO data repository maintained by the Eurostat.

The methodological framework of cartographic modelling was the basis of LUE computing, analysis and mapping. Cartographic modelling provides a standardized procedure for data integration, spatial operations, and analyses, along with the results visualization in the form of thematic layers, maps, tables and reports by means of GIS tools. The study was conducted in four consecutive phases (Fig. 2).

Figure 2

Phase 1 comprised data acquisition and pre-processing, including selection by attributes of artificial area from CORINE Land Cover (land cover type indicated as the artificial surfaces at the level 1), overlay analysis of CLC artificial surfaces and countries’ polygon, and spatial join with descriptive population data. Phase 2 relied on LCR, PGR, and LUE computation according to the formula given in the next section, while statistical analysis of data trends (the Mann-Kendall tau), statistical dispersion (i.e. standard deviation, interquartile range, variance to mean ratio), and spatial pattern by Global Moran’s I was undertaken in phase 3. Finally, in phase 4, the results were presented in the form of choropleth maps.

The LUE indicator is calculated using the formulas 1–3: (1) LUE=(LCR)/(PGR) {\rm{LUE}} = (LCR)/(PGR)

Land consumption rate: (2) LCR=(lnBt+nBt)/t LCR = (ln{{{B_{t + n}}} \over {{B_t}}})/t

Population growth rate: (3) PGR=(lnpopt+npopt)/t {PGR} = (ln{{{pop}_{t + n}} \over {{pop}_t}})/t where: ln denotes the natural logarithm; B_t+n B_t indicate the built-up area in the final year and initial year of the measurement, and Pop_t+n and Pop_t specify the total population within the built-up area in the final and initial year; while t defines the number of years between the final and initial years, LCR indicates built-up area enlargement, and PGR illustrates the changes in demography.

LUE equal to or lower than −1 implies inefficient land use; a level in the −1<= LUE <0 and above 1 range, it is moving away from efficiency; while in the range of 0< LUE <=1, it points to efficient land use or moving towards efficiency. SDG 11.3.1 is achieved when land use and population growth rate are in balance, as indicated by a LUE value close to 1 (Melchiorri et al. 2019; Calka 2021).

The Mann-Kendall tau examines whether the regular time series show monotonic upward or downward trends. It is an inferential statistic, where the null hypothesis: H0: ô = 0 states that there is no observed trend between analysed data series. An alternative hypothesis is tested for HA: ô < 0 or HA: ô > 0. The Kendall tau is calculated based on numbers of concordant and discordant pairs based on formula elaborated by Brossart at al. (2018): (4) τ=nc−nd12n(n−1) \tau = {{{n_c} - {n_d}} \over {{1 \over 2}n(n - 1)}} where: n_c, n_d denote the number of concordant and discordant pairs, respectively, and n indicates the total number of observations in the series. The τ values are bound between −1 and +1; negative values of τ implies that the data series are inversely related (negative monotonous), while positive values show positive monotonous relations.

Statistical dispersion was analysed by standard deviation (σ), interquartile range (IQR=Q₁−Q₃), and variance to mean ratio (VMR=σ²/μ).

The geographical pattern of the countries’ LUE value was investigated by Global Moran’s I statistics, which measures spatial autocorrelation based on feature locations and feature quantitative values concurrently throughout a whole dataset. The null hypothesis of complete spatial randomness says that the attribute being investigated (LUE values) is randomly distributed among the features (countries) in the whole study area. A detailed description of Global Moran’s I is given in ESRI ArcGIS Help or in Moran (1950).

The geovisualization of the results was the final stage of the research. Choropleth maps with six class intervals were adopted to portrayal the land use efficiency in Europe in the analysed periods. Class 1 takes LUE values greater than or equal to 2, emphasizing that the LCR is twice the PGR, while class 2 represents an LUE in the range (1, 2] showing the LCR slightly greater than the PGR. Class 3, with index values in the range 0 < LUE ≤ 1, indicate a population growing at a higher rate than the built-up area. Classes 4–6 present decreasing population, with −1 <LUE ≤ 0 (class 4) showing the negative pace of population is faster than built-up area rate; class 5 and 6, with −2 < LUE ≤−1 and LUE ≤ −2, where the negative pace of population or slower than the built-up rate. The colour scheme was chosen so that warm (orange–red) colours indicate faster population growth than built-up areas, while cold (blue) colours indicate that LCR is preferred over PGR.

Both the PGR and LCR show monotonic upward trends, with the Mann-Kendall tau equal to 0.800 and 0.247 (with significance p ≤ α = 0.05) for two consecutive analysed data series (2006–2012 and 2012–2018), which results in a similar trend in the LUE data series (tau = 0.570, p ≤ α = 0.05). A decline in the population growth rate in both periods was observed in 14 countries, 13 of which belong to Central and Eastern Europe. Changes in land consumption pace were also small; the country average annual increase in built-up land in 2006–2012 was 67.8 km², while in 2012–2018, it was 16.6 km². Selected measures of descriptive statistics show the greater statistical dispersion of the LUE values in 2006–2012 than 2012–2018 (Table 1). In the years 2006–2012, the standard deviation is more than twice as high as in the years 2012–2018, and the quartile range is even four times higher. Also, VMR indicates a greater degree of LUE disparity in 2006–2012 than 2012–2018, taking the value of 145.96 and 129.62, respectively.

In both analysed periods, LUE values reached extremes in five countries: in 2006–2012, they were Portugal, Poland, Greece, Bosnia and Herzegovina, and Croatia; in 2012–2018, they were Poland, Estonia, Spain, Macedonia and Montenegro. Very slight decreases in the rate of land consumption were recorded in Finland, Ireland and San Marino, which, with modest population growth, suggested a trajectory of compacted land change and synergies between the country’s development policy and the EU’s ‘no net land take by 2050’ strategy (COM 2011). Former post-communist countries, both Central European and Balkan, were characterized by a decline in the number of inhabitants with a large increase in the built-up area, which ultimately indicated that the country’s development policy, supported by EU funds, was not fully in line with the Efficient Europe strategy. However, changes in the LUE between 2012 and 2018 finally suggested some trade-off between national and EU development policies. The remaining countries are characterized by both population and built-up area growth, indicating a sprawl-type development trajectory, also quite distant from the EU strategy. The dependence of LUE on LCR and PGR is shown in Figure 3.

Figure 3

Considering all 39 analysed countries, the Global Moran’s I analysis shows that in both periods the spatial pattern of land use efficiency in the areas of nations did not appear to be significantly different from the random one, yielding the z-score of 0.0367 (p-value: 0.97069) in 2006–2012 and 1.266 (p-value: 0.205549) in 2012–2018. Also, the land consumption rate showed a random distribution, while the population growth ratio in both periods revealed a cluster pattern, with a z-score of 5.578 and 4.422 (p-value: 0.00001), successively.

In 2006–2012, the land efficiency index reached a value close to 1 in Liechtenstein (0.96) and Sweden (1.07), and in 2012–2018 in Turkey (0.92) and Slovenia (1.03), indicating significant land use efficiency, with a comparable increase in land consumption and population growth rate. As many as eight countries made little progress towards efficiency (LUE <−2 in 2006–2012); in 2012–2018 there was still an insufficient increase in urbanized area in relation to population growth, although it was less than in the period 2006–2012 (Table 2). Relatively large progress in achieving SDG 11.3.1 was observed in Portugal and Germany (see Fig. 4). Almost all Balkan countries are moving towards more efficient land use. Poland was characterized by an increase in urban areas and a decline in population, which manifested itself in ineffective land use. Estonia showed both a PGR growth and an LCR reduction, which indicates an insufficient increase in built-up area in relation to population growth, with the result that the LUE values for both periods were below −2. Therefore, both countries are shown in the choropleth map in dark blue (Fig. 4).

Figure 4