Subscribe to DeepL Pro to edit this document. Visit www.DeepL.com/pro for more information. Q 1 Analysis definitions space occupation The EU objective of no net land take by 2050 (NNLT) has its origins in environmental policy and is intended to protect agricultural and natural land from urbanization, as it is mostly an irreversible process. Although there is no official definition yet, it is primarily about converting (natural or agricultural) land use to (semi-)urban functions and, in particular, paving. If this becomes an obligation under European law, an unambiguous definition must be established. And if this obligation involves monitoring and reporting - it is expected that it will - then this definition must also be able to be operationalized with quantifiable indicators that are preferably collected and measured in the same way by all member states in order to make international comparisons. 1.1 Overview definitions Over the years, various definitions of land take and related concepts have been given within countries and in international forums (Marquard et al. 2020). It is clear from the definitions on land take that there are broadly two different interpretations regarding this topic. In part, this is due to the coincidence of the two related terms land and soil. The most common definitions concern land and look purely at the quantitative change of land use to urban or "artificial" functions. Land take or land take can thus be unambiguously expressed in hectares. The same applies to the opposite process (e.g., letting an industrial site go wild). Another term, land consumption, is usually understood as this quantitative/functional form of land take. The differences between these definitions then lie mainly in what functions are meant by "urban" or "artificial. Consensus exists on built-up areas such as residential and commercial areas, and many definitions also explicitly mention (transportation) infrastructure. However, there are several gray areas of human or urban use (think construction pits, landfills, greenhouse farming, reclamation areas, wind turbines and solar arrays). A special category on which there is no consensus is "urban green space," or parks, parklands, sports fields or recreational areas: these places are usually considered urban, but are largely unpaved and may host natural functions. There are also differences between definitions in terms of original function: nature, forest and agriculture are often mentioned, but also 'semi-natural' land such as an unused pasture that is occasionally mowed; EEA also mentions the loss of rural areas in general when describing land take.1 Following the broad EEA definition of land consumption, in Flanders land take is described as follows: 'It is a broader concept than building or paving, in fact it includes all parts of the settlement structure including agricultural structures, industry and non-built parts such as gardens, parks and recreational uses' (Building Shift 2021 Task Force, p. 5). The second type of definition concerns 'soil' and here consideration is given to the future use and ecological quality or fertility of the area. The most explicit proposition is in French environmental law where land encroachment (or "artifice mitigation") involves the degradation of ecological functions of soil. Because this implies a more qualitative approach (degradation suggests a sliding scale), it is not clear whether gradations of land take are possible, e.g., that a hectare of land that provides crucial ecosystem services should weigh more heavily than a hectare of land that does not or less so. Binder et al. (2021) recommend that even nearby natural areas be included in the calculation of additional land take to account for issues such as landscape fragmentation and noise pollution (Schatz et al. 2021, p. 7). Linked to this is the discussion about the disappearance of fertile agricultural land. Among other things, this requires knowledge about the type of agriculture (e.g., horticulture, arable, pasture) that will be converted to urban functions. A related term is soil sealing. It may differ from land sealing for at least two reasons. First, there are unpaved urban functions such as parks, public gardens and (some) sports grounds where there is little or no sealing. Second, there are paved non-urban functions such as livestock housing and greenhouse farming in agricultural areas and visitor centers in natural areas. Finally, the primary concern is not soil quality but the quantity of cover. Important in measuring soil cover is the accuracy of the analysis; at a higher resolution, paved roads and driveways come into focus, but also (non-tiled) backyards. Thus, different values for soil cover can be calculated for the same area depending on the resolution of the source data. 1.2 Analysis of quantitative definitions 1.2.1 Delineation and methods The two types of definitions quantitative and qualitative require separate analyses. The first concerns only the area of function change and can be calculated relatively easily. The second requires more interpretation (of, for example, soil quality or ecosystem services). All analyses primarily use Corine data (Corine Land Cover (CLC) for the existing situation and Corine Land Cover Change (CHA) for changes). This is the only source that has both European coverage and sufficient information on land use. A major drawback is that this dataset is coarse (25ha for CLC 1 One striking definition is the EEA's for land consumption: it is so broad that it even includes agriculture and forestry, that is, virtually any use for economic production (Marquard et al. 2020, p. 7). One consequence of this is that building a housing development on a former orchard would not represent a land take increase. and 5ha for CHA) to properly recognize small-scale spatial developments. As a result, there is both overestimation and underestimation of the extent of land take (see Section 3.5). Because of the relatively large-scale project development in the Netherlands, these data are sufficient for a global estimate (and hence this quick scan). The advantage is that it is possible to immediately calculate how the Netherlands scores relative to other member states (bearing in mind that results elsewhere may be less reliable). Land take increase is calculated in this study for the period 2000-2018: the longest time series of the Corine dataset; if necessary, the three sub-periods (2000-2006, 2007-2012, 2013-2018) can be zoomed in. To simplify calculations and presentation, the 11-category grouping of Corine use classes from the ESPON SUPER project is used. Again, the finer 44 Corine classes can be zoomed in as needed. 1.2.2 Dutch land take in a European perspective If the EEA's most extreme definition of land consumption (which includes intensive forestry and agriculture) is excluded, the broadest definition of land take in the Netherlands adds up to about 70,000 hectares (gross) for the period 2000-2018, or about 10 hectares per day since the turn of the century. By comparison, Germany has long had a goal of reducing urbanization to 30 hectares per day and Flanders 3 hectares per day by 2025 (and zero by 2040) and Luxembourg 0.25 hectares per day by 2035. Table 1 Gross land take increase Netherlands by use class in 2000-2018 in ha Use class ha Urban area (residential, retail, mixed) 10.685 Job sites (industry, business parks) 7.603 Infrastructure (including airports) 2.721 Reclamation (mining and quarrying). 2.143 Landfills 217 Construction sites 38.446 Urban green space and recreation 7.875 Total gross land take increase Source: CHA 2000-2018, own edit 69.690 Figure 1 Source: CHA 2000-2018, own edit Figure 1 shows the position of the Netherlands within Europe. The major EU countries have the largest increase in land take in absolute terms in the period 2000-2018: The Netherlands occupies a sixth position after Spain, France, Germany, Poland and Italy. This comparison should be taken with a grain of salt because, as mentioned, the European data are not based on accurate measurements so small-scale urban developments are not accurately reflected. As a result, figures for Belgium, among others, must be considered unreliable. This problem is explained in more detail in section 3.5. However, when the land take increase is plotted against the total area2 (an indicator of intensity), the Netherlands is the European leader. In terms of relative growth (land take increase relative to the size of urban area in 2000), the Netherlands ranks sixth, far behind Spain where extreme urban area expansion has taken place.3 2 In this calculation, the territory of the Netherlands is taken without inland water (IJsselmeer). If inland water is included (4,154,300 hectares instead of 3,497,464 hectares) little changes in intensity except that the Netherlands consistently ranks second, behind Cyprus. 3 These figures differ slightly from earlier PBL publications on the subject. The reason is that they follow the most common definitions of land take that consider building sites as land take. In the ESPON SUPER project, among others, building sites are seen as an intermediate category: only upon completion was a development registered as urban. Moreover, Iceland, Norway, Switzerland and the United Kingdom were also included in the analysis there. Figure 2 Source: CHA 2000-2018, CLC2000, CLC2018, own editing All figures so far refer to gross land take increase. To calculate the net increase in land take, we must also look at the decrease in land take, functions converted from urban to non-urban in the same period. For the most part, this involves abandoned building sites. To a lesser extent it involves reclaimed land or landfills (such as Diemerpark in Amsterdam). Only 5 hectares concern the disappearance of urban areas and 25 hectares for work sites. Altogether, the decreasing land take was 5,833 hectares in the period 2000-2018. Together, for the Netherlands this means a net land take of 63,857 hectares or around 9.7 hectares per day. The increase in land take in the Netherlands is almost 14 times greater than the decrease in land take, which is not exceptional in Europe. There is wide variation among member states. The Czech Republic, Germany and Luxembourg have only four times as much positive land take, so the "net" aspect is very important in meeting their target. In contrast, Slovenia, Romania and Malta have relatively little declining land take. Table 2 Declining land take in the Netherlands in 2000-2018 in ha urban agriculture building urban area purview infrastructure mining landfill site greenery 5 22 35 57 33 885 13 nature (vegetated) 191 2697 nature (bare) wetlands 3 Total 5 25 Source: CHA 2000-2018, own edit 44 273 70 1505 79 521 103 5088 13 In the Netherlands, non-implementation of construction projects on building sites is the main cause of decreasing land take. But within the EU27, the disappearance of reclaimed land is more important. Indeed, a significant part of the declining land take in Austria, Estonia and Romania can be explained by this. In absolute terms, Germany, Spain and Poland have 'returned' many hectares of such 'artificial' land to agriculture and nature. Incidentally, correcting gross land take increases does not change the relative position of the Netherlands in Europe in terms of intensity and growth. 1.2.3 Sensitivity analysis definitions space occupation The picture of the Netherlands as a country with relatively large land take changes with the application of different categories in the definitions. Always subtracting categories from the broadest definition reduces the gross increase in land take anyway, but also the possible decrease in land take. To calculate the gross land take increase when deleting a category, two steps are required. We take the deletion of urban green space as an example. First, the number of acres from this category is subtracted from the total gross land take increase, because all acres of rural functions developed to urban green space no longer count as land take (69,690 - 7,875 = 61,815). But because urban green space has now become a "rural" category, all change from it to the other urban functions such as housing, work and infrastructure (this involves 1,123 acres) is now land take and must therefore be added (61,815 + 1,123 = 62,938). Next, the net increase in land take must be calculated. As urban green space is no longer considered urban/land use, any change from this to rural functions is no longer a decreasing land take (5,833 - 13 = 5,820). The net land take increase for the first alternative definition (without urban green space) is then: 57,118 hectares (62,938 - 5,820). The results of this calculation are the starting point for calculating the next definition in the same way. Table 3 shows the results for all definitions. Table 3 Gross and net space take increase in 2000-2018 in the Netherlands by definition Definition of land take Gross Net Wide 69690 63857 - Urban greenery 62938 57118 - Urban greenery, artificial 60625 55429 - Urban green, artificial, construction site 47597 47488 - Urban green space, artificial, construction site, infrastructure Source: CHA 2000-2018, own edit 45008 44978 The first alternative category involves removing urban green space and recreation from the broad definition. One argument for this is the low percentage of urbanization of these functions; in fact, the ground is hardly covered by buildings, if at all. Looking separately at these two Corine classes, the Netherlands laid out the most urban green space (parks, public gardens, urban forests) - in absolute terms - between 2000 and 2018, just slightly more than Spain. The Netherlands is then the European leader. Recreational land (sports fields, golf course, allotments, race track) is a much larger category and there the Netherlands scores somewhat lower in absolute terms, fifth in Europe. Removing recreational areas would make the biggest difference in Austria where this category represents one-third of the total land take increase. Removing both classes (urban green space and recreation) from the definition would be relatively favorable for the Netherlands. Because it is a relatively small category, however, not much changes. In the ranking, the Netherlands remains first in terms of intensity and sixth in terms of growth. Figure 3 The second definition involves excluding "artificial land" such as landfills and reclamation sites (these categories are not always mentioned and are not urban). Because these categories occur relatively infrequently in the Netherlands, they have little impact on the domestic land take increase. They hardly change the ranking of the Netherlands within Europe. The situation is different for other countries: clearing is an important component of additional land take in Bulgaria, Estonia and Latvia. There, removing this category will significantly reduce their land take increase. This places the Netherlands fifth in terms of growth (Greece is now sixth) and still ranks first for intensity. The third definition, which also removes the 'building sites' class, has a major impact on Dutch land take increase. Building sites account for more than half of the Dutch land take increase; in fact, this is also the case in Lithuania and Slovakia. In Spain (known for abandoned building sites) it accounts for 48 percent of the additional land take. One argument for not including building sites is that it is uncertain whether they are actually built on; they are an intermediate stage, which can complicate monitoring. This also appears to be the case in the calculation: in some member states the increase in land take becomes larger in this definition because there was more change from building land to urban functions (realization) than building land added from rural functions. In terms of the ranking, the Netherlands returns to sixth place, behind Portugal. For intensity, the Netherlands is at the top as always. Less likely is the exclusion of infrastructure. One argument for this may be that it cannot be accurately measured at a low(er) resolution (like Corine). Incidentally, this definition would mean little for the Netherlands (less than 5 percent of total land take increase), but a lot for Croatia (more than a third). Next come Greece, Malta and Portugal with just above 10 percent. This leaves the Netherlands in sixth place in terms of growth. Interestingly, the Netherlands is just overtaken by Cyprus (which has always ranked second) for intensity. 1.3 Analysis qualitative definitions 1.3.1 Methods The best way to estimate how much valuable soil disappeared during the period 2000-2018 would be to compare the land take increase cartographically with the local soil quality in 2000. Unfortunately, such mapping material is not available, but 16 maps are available for the situation in the present. This makes it technically possible to monitor qualitative land take increase from now on and perform scenario analysis on future land take by testing the exact change to urban use in a GIS system against the various ecosystem maps. In carrying out this analysis, it turned out that this method required more computing power than was practically possible for this quick scan. As a result, an alternative analysis was performed. Maps on ecosystem services 1. The following 16 maps were read into a GIS. perc_service_1D_energy production_wood perc_service_1D_wood production_wood perc_service_1D_carbon storage_forest perc_service_2A_Fertility_hydrology perc_service_2D_refrigeration perc_service_2 _water treatment_retNE perc_service_2 _water treatment_retPE perc_service_2F_plague suppression perc_service_2G_pollination perc_service_2H4_CO2emissions peat perc_service_2J_AQ_conc perc_service_2J_AQ_pop perc_service_2L_infiltration_mm perc_service_2L_infiltration_pop perc_service_3A_green_recreation perc_service_3B_MNP Note: These maps are normalized stocks (from 0-100 percent). They are always about the effect of vegetation in providing an ecosystem service. Water purification is nitrate and phosphate. AQ is Air Quality, which is determined as the percentage of particulate matter PM2.5 filtered to bring exposure below the WHO standard. Both Air Quality and infiltration maps are included, idem for air purification. MNP is a biodiversity indicator and is about the number of species for which there are good conditions. Figure 4 Impression composite ecosystem map Source: PBL/WUR The alternative analysis involves first assessing the 44 land use classes from Corine 2018 for their current ecological values using the current map material. First, the legends of the 16 maps are normalized and stacked on top of each other. The result is an aggregate map that represents an index value of all 16 components (see Figure 4). Then, for each CLC land use class, the average value is calculated. In principle, this average value can be multiplied by the increase in land take over the period 2000-2018 to get an impression of the change in soil quality. 1.3.2 Dutch land take in European perspective Agriculture provides most of the land for functions covered by land take in Europe. In contrast, the share of wetlands is usually negligible: in only 10 countries does more than 1 percent of urbanization take place on wetlands. The northern countries Estonia (8.5 percent), Latvia (4.7 percent) and Finland (3.7 percent) do the most (the Netherlands sits at 2.5 percent). There is wide variation in the share of nature as a source of additional land take, however. Finland (82 percent), Sweden (65 percent), Croatia (62 percent) and Portugal (61 percent) use natural areas the most for (semi-)urban functions while the Netherlands (3 percent), Ireland (4 percent) and Denmark (5 percent) do so the least. At the EU27 level, the figure is 21 percent. The share of "barren nature" hardly appears as a source of additional land take: only Portugal, Croatia and Spain have a share greater than 1 percent (but still below 2 percent). Figure 5 Source: CHA 2000-2018, own edit In the subcategories of agriculture and nature, the differences between member states become even greater. This is partly explained by territorial diversity: after all, there will be little or no land take increase in the category of olive groves in Finland or beach areas in Slovakia, because Finland has no olive groves and Slovakia has no beaches. From most agricultural categories, it is not immediately clear whether that category would have high or low soil quality. The exception is code 243 (agriculture with natural vegetation); urbanization on this soil class is found mostly in Malta, Bulgaria and Greece. For "bare" nature, a few classes probably have very low soil quality: 332 (bare rock), 334 (burned areas) and 335 (permanently snow-covered). For Europe as a whole, this is less than 1,000 hectares of land take increase in these categories. The alternative analysis as described in 3.3.1 provides more insight into the relative value of soils changed to (semi-)urban functions. This shows that indeed agricultural class 243 (agriculture with natural planting) scores relatively high. Planted nature scores by far the best, as the most valuable soil. After planted and unplanted nature, for example, urban green space remarkably scores highest by far. Agriculture, on the other hand, scores relatively low on the composite ecosystem services indicator: about the same as the urban area, landfills and construction sites. The categories of industry, reclaimed land and infrastructure score the worst. Table 4 shows the results using the SUPER classification. Table 4 Gross land take increase Netherlands by use class in 2000-2018 in ha Use class Average sum* Grade** Urban area 245,4 5,6 Job sites (industry, business parks) 220,2 5,1 Infrastructure (including airports) 236,1 5,4 Reclamation (mining and quarrying). 221,9 5,1 Landfills 241,3 5,6 Construction sites 240,5 5,5 Urban green space and recreation 310,4 7,1 Agriculture 244,8 5,6 Nature - planted 434,4 10,0 Nature - bare 296,8 6,8 Water - marsh and water bodies 248,7 * Average sum is the total score on the 16 ecosystem services (on a scale of 0-1600) ** Figure is calculated as share of best scoring category (x10) Source: CHA 2000-2018, own edit 5,7 This analysis should emphatically be considered a rough indication of soil quality, made in the context of this quick scan. It requires further refinement and elaboration. Not all 16 maps are (equally) relevant to soil quality and it is questionable how permissible the unweighted aggregation of the maps is. Moreover, these results cannot be used for other European countries, because both agriculture and nature are unique in the Netherlands and therefore incomparable with those in other countries. That said, the results are very striking and thus important for policy discussion. For example, if land take were to count only for soil quality of a grade 6 or higher (the values used from the method in Table 4), total Dutch additional land take in the period 2000-2018 would be negative. This is a substantial difference from the quantitative approach. In addition, this analysis provides substantiation for the removal of urban green space from the definition of land take and to strongly downplay the contribution of agriculture to land take. 1.3.3 Sensitivity analysis definitions space occupation Because there is so much uncertainty about the qualitative approach, a sensitivity analysis is premature. It is questionable whether this approach will be included at all in the final definition of land take; therefore, so far most attention has focused on the quantitative approach. But if a decision is nevertheless made to include soil quality in monitoring and reporting the increase in land take, there are two points of interest: (1) what indicators are used to determine soil quality and (2) whether and how soil quality is weighed against acres of additional land take. These two points are briefly elaborated below without empirical analysis. Soil quality is both a technical issue and a political one. The normative aspect is about the question: soil quality for whom? If the soil is (made) very fertile, this can be detrimental to certain ecosystems: for different types of nature, a different chemical composition of the soil will be beneficial. This fact alone makes our composite map from this quick scan insufficient for policy practice: political choices must first be made and environmental expertise consulted to arrive at a set of indicators. Probably the choice will also have to be area-specific, because which soil quality needs the most protection will vary from place to place. Perhaps the indicators can also change as new issues arise or need to be resolved over time. Finally, both the method of measurement and the weighting of indicators will have to be part of the discussion. The second issue concerns how soil quality information is used in the NNLT target. Above, the extreme example of removing agriculture as a category from the definition of land take has been mentioned, so that the Netherlands automatically meets the target. It is more prudent to apply weighting. Occupying natural land can be "penalized" by making it count twice and agriculture by half. And transformation from agriculture to urban green space can then be considered neutral or positive. Such a key works out favorably for the Netherlands, which uses relatively little natural land for urban functions while also developing a relatively large amount of urban green space. 1.4 Soil sealing analysis Although not strictly the same as land take, the degree of land cover is also important when discussing NNLT. In contrast to the quantitative approach, it is not about the function of the area, but only about paving or construction. This basically means that one hectare of private garden in the urban area, if unpaved, is equivalent to one hectare of arable land or natural area. Important in calculating soil cover is scale. Only with high-resolution data can backyards or roads, for example, be accurately measured. As with the other calculations, there is a preference to work with global or European data. Several files are available in this context. The recently made available World Settlement Footprint (WSF) from Evolution of the European Space Agency (ESA) has the advantage that a time series from 1986 to 2015 is available. Also, the even more accurate WSF2019 file is available for that year. Unfortunately, the periods do not quite match Corine so a comparison of equal periods is impossible. Figure 6 Soil sealing in Randstad (1986-1999 and 2000-2015) According to the WSF Evolution data, 102,365 hectares of soil cover were added in the Netherlands in the period 2000-2015 (a 28 percent increase in area compared to 1999), see Figure 6. If the soil cover is taken over the entire stock period (1986-2015), the amount of hardening in the Netherlands increased by 211,666 hectares, so the degree of cover was more or less the same for both 15-year periods. This is also the case for many EU27 countries, with some outliers such as Ireland, Poland and Slovakia where there was a lot of urban development after the turn of the century. In addition to this analysis, a calculation was also made based on CBS data (BBG: Soil Use File). This shows a soil cover increase between 2000 and 2017 of 50,081 hectares (this is 10 percent of the area compared to 2000). There is a big difference between the results of the two files (WSF Evolution and BBG): WSF is almost twice as much as BBG. The exact reason for this is beyond the scope of this quick scan, but does indicate that the choice of a particular data file can have a great deal of influence on the results. 1.5 Side notes and reflection There has been criticism of the use of CLC data to measure space occupancy. A major problem is the relatively low resolution. As a result, two types of errors occur: (1) areas designated as urban but with little or no built-up area and (2) small-scale built-up areas not recorded as urban. The ESPON SUPER project investigated these two types of errors by comparing the Corine-2012 map with a much more accurate built-up map (Global Urban Footprint) (Van Schie et al. 2020); see Figures 7 and 8. Figure 7 Urban area registered in CLC with hardly any construction near Liège Source: CLC2012, GUF2012 Figure 8 Buildings not recorded in CLC near Warsaw Source: CLC2012, GUF2012 If Corine is used for follow-up analyses, wrong conclusions may be drawn. In Flanders, for example, many provincial roads with occasional houses are designated as urban areas in Corine 2000. As a result, further ribbon development in the 2000-2018 period is considered densification and not land take or land take contrary to own observations and policy. The delineation of areas in Corine is not only problematic for development. Renewable energy production is also a tricky category. A wind turbine takes up relatively little space on the ground, but needs surrounding infrastructure that may require paving of the soil. There is also no separate category for solar arrays, which can affect soil quality. Sometimes nearby trees need to be cut down or other vegetation removed, which can affect soil (quality). If in CLC an area is designated as 'energy production' then the (relatively broadly defined) area changes function; this helps explain the significant transformation from nature to industry in the UK (Cole et al. 2022), see Figure 9. Figure 9 CLC Delineation of a wind farm in England Source: Cole et al., 2022 Another difficult category in CLC is greenhouse farming. Corine registers this as ordinary agriculture (non-irrigated arable land - see Figure 10 where this is indicated by a light yellow color, but is sometimes also registered as industrial land (purple)). But greenhouse farming certainly has urban aspects in terms of soil; it is covered and often (partially) paved. According to Corine and the most common definitions, replacing greenhouse horticulture with a recreational area or a villa district would be labeled as additional land take. But according to the key figures, the minimum unpaved area of greenhouse horticulture is only 10 percent while for a villa district it is as much as 75 percent and 95 percent for recreational landscape (Ruimte met Toekomst - Area Typification, 2022). While converting greenhouse horticulture to a villa district would be considered land take. Figure 10 Greenhouse agriculture invisible in CLC at Municipality of Westland Source: Corine2018 (left) and Google Maps (right) Apart from the data source, general comments can be made about the accounting offsetting between gross and net land take increase, because the original land take (greenhouse cultivation) may have caused a destruction of the natural substrate. A change in function does not mean that the soil has been returned to its original condition. In that respect, it is not "fair" to count land take decrease as much. The qualitative approach can also be reflected upon. The makeshift analysis in this quick scan study shows that including soil quality can have major implications for determining land take increase . Despite the fact that the qualitative approach comes closest to the spirit of the NNLT objective (not affecting the soil), this has hardly been elaborated on as yet. 2 Sustainable urbanization The goal of reducing land take to zero (NNLT) is a good example of a generic measure that does not take into account local and regional circumstances. Areas where the pressure to build is high will be most affected by this policy: halting additional land take is easier in a shrinking region than in a growing one. To shed some light on this, below we relate land take to demographic pressure. Land take increase per capita is often used as an indicator when describing sustainable urbanization under SDG 11.3 (UN Habitat 2021). Only urban developments (land consumption) are mentioned in the description of this target, so functions such as mining, landfills, recreation and so on may not count. In addition to population development, we can also calculate household development. This has more influence on the real demand for housing than people: in fact, there may be a stable population but increasing number of households (due to divorces, fewer children, for example) the so-called "household thinning" (Hamers 2020). A major objection to linking urbanization sec to demographics is that much building is being erected for other reasons (see Colsaet et al. (2018) for a review). After all, the increase in logistics centers in the Netherlands has little to do with population development. Ideally, a split should be made between residential areas (with demographic drivers) and employment areas (with economic drivers). An analysis conducted as part of the ESPON SUPER project finds that it is very difficult to explain the land take of work locations by economic indicators (Van Schie et al. 2020). For example, an office tower has a huge number of jobs per hectare compared to a data center or a distribution center. Moreover, much economic activity takes place within the urban area (urban fabric) category that is usually understood as residential. For these reasons, we refrain from refining the analysis with economic drivers - the relationship is too insignificant to warrant scientific analysis. 2.1 Dynamic land take analysis increase When calculating sustainable urbanization, a dynamic "relative land take" indicator is sometimes proposed: the percentage growth rate of land take divided by the percentage growth rate of population. The idea is that if the number is less than one, then the land take is sustainable (Schiavina et al. 2019, p. 5). In the broad definition of land take, the Netherlands scores 1.71 in the most common definition of sustainable urbanization (14 percent net land take increase divided by 8 percent population growth), so not exactly sustainable. But the Netherlands is far from the only problematic country in Europe in this regard. At the top of the list is Portugal whose population barely grew from 20002018 but where 15 percent additional urban area was added. Slovakia, Germany and Spain also rank above the Netherlands. On the other side is Greece which had just slightly more urban growth than the Netherlands but also had negative population growth at the same time: not exactly sustainable. Among the most "sustainable" countries are Belgium, Luxembourg, Sweden and Austria. As mentioned above in section 3.5, these figures are not entirely reliable because of the under- and overestimation of urban area in small-scale developments (especially in Belgium). But if the number of households is used, the picture can be quite different: the area of (semi-)urban functions is growing (much) slower than the number of households. Table 5 Dynamic land take growth in the Netherlands per person and household by definition Definition of land take %R/%∆bev %R/%∆hh Wide 1.71 0.91 - Urban greenery 1.50 0.80 - Urban greenery, artificial 1.46 0.77 - Urban green, artificial, construction site 0.56 0.30 0.49 0.26 - Urban green space, artificial, construction site, infrastructure Source: CHA 2000-2018, EUROSTAT, own editing The table above shows that per capita, only two definitions of land take (i.e., those that do not include building sites) are sustainable. Indeed, according to the broad definition of land take, the "excess" land take increase is almost twice what is considered sustainable. However, all householdlevel calculations do fall within the sustainability boundary: even under the broadest definition, the two variables are more or less equal. This analysis is consistent with the findings of a similar, global study of densities based on Global Human Settlement Layer data (Li et al. 2022)data (Li et al. 2022). There it can be seen for the Netherlands that the population in some large cities grew faster than built-up areas (densification) while immediately outside the Randstad the opposite happened (thinning). Incidentally, this fits the global trend: except in the major cities, the increase in land take is not sustainable (Li et al. 2022, p. 9). This indicator does have the problem of path dependence. A metropolitan region that is densely populated and grows 1 percent in area and population gets the same score (1) as a suburban region with lots of building per capita that also grows by 1 percent While both areas score very differently on a number of sustainability indicators. 2.2 Analysis relative land take increase As an alternative to dynamic land take increase, Schiavina et al. (2019) recommend the indicator the marginal land take increase (MLCNI). The table below presents a calculation of MLCNI for both per capita and per household (the latter being a better predictor of housing demand). Thus, this indicator (MLCNI) indicates the number of acres of land take added per new resident or household. Table 6 Relative land take increase in the Netherlands per person and household by definition Definition of land take Per person Per household Wide 0.0485 0.0601 - Urban greenery 0.0425 0.0527 - Urban greenery, artificial 0.0413 0.0513 - Urban green, artificial, construction site 0.0159 0.0197 - Urban greenery, artificial, construction site, infrastructure 0.0139 0.0172 Source: CHA 2000-2018, EUROSTAT, own editing Converted to square meters, this analysis shows that almost 500 square meters per new resident and over 600 square meters per new household were added to urban area in the period 20002018. Again, the building land category makes the main difference: about 300 square meters of the land take increase per new resident and 400 square meters per household is explained by this land use category. In the narrowest definition (the one that excludes urban green space, artificial, building land and infrastructure, or just residential and commercial areas), it is 138 square meters of additional land take per new resident and 172 square meters per new household. It is important to note that this is land take (footprint) that does not include stacked buildings. It is again possible to compare the Netherlands with other EU countries, with the same caveat about the CLC file. First of all, there are nine member states experiencing negative population growth, making any additional land take in those countries inherently unsustainable. This imbalance is greatest in Greece, Poland and Estonia. The Netherlands also performs reasonably well in positive terms, with Portugal topping the list with a marginal land take increase of about 1 hectare (10,000 square meters) per new resident. Slovakia and Cyprus score twice as high as the Netherlands. The most "efficient" countries are Malta, Belgium, Ireland and Austria. In the ESPON SUPER project, a similar analysis was made at the regional level for Europe. There, the urban area (as this concerns mostly residential areas) was compared to the European average (Van Schie et al. 2020, p. 60). Most regions in the Netherlands are in the relatively efficient category of above-average population growth and below-average urbanization. Figure 11 Average urban area growth per person in European regions Source: Van Schie et al., 2020, p. 60 2.3 Side notes and reflection NNLT appears to be the same as sustainable urbanization but there are important differences between the two perspectives. One is that sustainable urbanization is not an absolute value but an aspiration within current conditions. For example, a relative indicator is used in SDG 11.3: urbanization per capita. This has the advantage that shrinking and growing areas are not lumped together. In this respect, this approach is more similar to the ladder of sustainable development in the Netherlands than to the NNLT target (Evers & Blom 2016). As with different conceptions and operationalizations of land take, this can be problematic. For example, it is implicitly assumed that urbanization occurs to accommodate population growth, but there are many more factors (Colsaet et al. 2018). Household development appears to be a better predictor of housing demand. Many urban functions (such as job sites) have very different drivers. Also, different calculation methods offer different outcomes and insights. As with land take, the inaccuracies in CLC data can lead to wrong conclusions: for example, Belgium seems very efficient because the population grows faster than the (incorrectly) observed urbanization. In short: the dynamic or relative approach is also an obvious panacea. A second difference between additional land take and sustainable urbanization is that the latter leaves more solution space open. NNLT implies that urbanization an sich is not sustainable. Sustainable urbanization involves optimal combinations of functions, knowing that there are all kinds of trade-offs. Sometimes it can actually be sustainable to develop urban functions in order to address certain challenges, think of the nitrogen debate, renewable energy and making the housing stock more sustainable. Moreover, there are also sustainability disadvantages to compaction; soil is just one of many factors involved. 3 Future land take 3.1 Analysis plan stock to 2030 Many (hard) plans for urban functions are currently in place. Some are already far advanced (e.g., an adopted zoning plan or an issued permit) while others are relatively "soft. Thus the question arises to what extent those plans in the pipeline will lead to new land take. Assuming that hard plans are likely to be realized around 2030, an analysis of these plans dovetails with the expectation of a European obligation to report progress by 2030. The website ruimtelijkeplannen.nl contains all adopted spatial policies and plans, but does not provide information on preliminary phases of plan making. In the past, the initiative 'De Nieuwe Kaart van Nederland' collected plans and intentions and continuously put them on the map (and removed cancelled plans from the map). This initiative was terminated in 2010 for financial reasons. In 2018 an attempt was made to compile a (New) New Map again (CRa 2018) but that file is also no longer maintained. Currently, there is no national registry of soft plans; some provinces such as North Holland have published this information, but others have not because of perceived secrecy or that it is seen as too sensitive information, among other reasons. In addition, there is little regularity in how proposed plans are reported.4 This makes it difficult to calculate future land take in the medium term. At this time, provinces have received so-called starter packages that contain various living environment tasks (De Jonge 2022). Provinces are expected to have their "spatial 'puzzles' ready during 2023. If that result is reported in a consistent way with so-called geo-information, it should be possible to make a calculation of the spatial impact for the medium term. Such an analysis can probably be made only during 2024 . Another technical possibility is to make an additional land 4 Sometimes plan data were provided with precise geographic information, sometimes only tabular (number of dwellings within a municipality) without an indication of density and whether the plans involve infill or expansion (additional land take). take analysis of the known hard plans and supplement it with a Space Scanner simulation based on the targets per province. However, that is beyond the scope of this quick scan. 3.2 European scenarios for 2050 As for the longer term, the Joint Research Centre (JRC) has developed the LUISA model that, among other things, can be used to estimate future land use, including outputs on (net) urbanization and urbanization per capita. For the period 2010-2050, the baseline scenario calculated the average urbanization intensity at 1.6 square meters per person per year at the EU28 level, with Ireland, Finland, Belgium, Cyprus, Luxembourg and Sweden as high outliers and Bulgaria, Germany, Latvia, Croatia and Greece as low outliers (Kompil et al. 2015, p. 30). For the average resident, the perceived population density (weighted population density) in 2010 was 70 persons per hectare and 95 for metropolitan areas. According to modeling, these densities will decrease slightly by 2050, but with many spatial differences. For example, people in Spanish metropolitan regions (278 residents per hectare) will experience a decrease of about a third, but will still live in densities among the highest in Europe. The Dutch live in densities around the EU average and, according to model calculations, will live just slightly more spaciously in 2050. In the ESPON SUPER project, the open source version of the LUISA model was used to calculate three urbanization scenarios. These were intended to simulate the spatial distribution of the urban area under various assumptions. Here it is possible to understand what kind of areas would be urbanized under each scenario (Evers et al. 2021). Because the emphasis was placed on mapping the possible urban form, the total land take increase was made very large (for the sake of visibility on the map) and the variation in additional land take between scenarios was made small. For this reason, they are not suitable for making ex-ante analyses of quantitative land take increase. However, an analysis of qualitative land take increase (in the same way as in Section 3.2) should in principle be possible, as the exact areas of additional land take are known. It should also be technically possible to draw up new scenarios with different assumptions. These options are also beyond the scope of this quick scan. Figure 12 Impression of the urbanization scenarios in the ESPON SUPER project Source: PBL 3.3 Dutch scenarios for 2050 In the PBL Spatial Explorations 2019 project, six 2050 scenarios for urbanization were developed and published in the PBL report Grote opgaven in een beperkte ruimte (Hamers et al. 2021). The variables of the urbanization scenarios run along two axes. The first concerns the demand for urban functions. The second axis deals with location choices: densification in and building near larger cities (Near), building near ov stops (Connected) and car-related building (Spacious). The scenarios are thus very similar in content to those of the SUPER project but the implementation is more realistic because they take into account the current plan stock, have a higher resolution and assume a less extreme demand for urbanization. A drawback is that they are only available for the Netherlands making a European comparison impossible. 3.3.1 Methods To estimate the demand for urban functions in the future, the Space Scanner model used the updated Welvaart en Leefomgeving (WLO) scenario study (CPB/PBL 2015, pp. 29-30). There, two scenarios are described in which the development of the number of people, households and jobs is calculated through to 2050 in a 'high' and a 'low' variant. In general, the greatest pressure for urban functions is expected to occur in and around the Randstad and least in the border areas. To calculate the quantitative land take increase, the output of the Spatial Scanner outputs by summing the size of all new urban functions. A GIS is then used to look at the original function of the urbanized area in the Corine2018 database to see which CLC class the land take falls into (e.g. agriculture or nature). 3.3.2 Dutch land take in 2050 In the PBL urbanization scenarios, only new residential and commercial sites are modeled. Other categories usually included in land take (infrastructure, reclamation, landfills, and urban green space) are excluded. The output of the Space Scanner (see Table 7) is then similar to the narrowest definition of land take. Table 7 Additional land take in the Netherlands in 2050 by scenario Scenario Close to Connected Large High 21.616 40.217 67.650 5.168 42.230 26.293 Low Source: PBL This table shows that the range is very wide both in terms of external pressure (high versus low) and urbanization variance. There is more than a factor of 10 difference between extremes. Since demand (low/high) cannot be effectively controlled through spatial planning, it is about the choice of urban form (shown in the columns). There, too, the scenarios differ quite a bit: in the lowpressure scenario, Spacious has three times more land take increase than Near, while this difference is five times greater in the high-pressure scenario. The major differences become apparent when the original function is considered. Figure 13 shows the number of hectares in which the function has changed in all scenarios (only agriculture and nature count as land take). Thus it is clear that in Nearby there are still many housing and employment areas being added, but mainly within the existing urban area. It is also striking how little nature disappears in all scenarios. New land take occurs almost exclusively on agricultural land: this also occurred in the 2000-2018 period, but the Netherlands does so exceptionally much by European standards. It should be emphasized that these results were largely a conscious choice within the scenario design: for example, in Near, the Space Scanner actively tries to find space in the existing urban area, while this is not the case in Spacious. In any case, the main message of Big tasks in a limited space also applies to future land use: there is much to choose from. Figure 13 Source: PBL 3.4 Side notes and reflection According to the National Environmental Vision, the Netherlands is on the eve of a major transformation. Many functions will have to change between now and 2050 to provide a solution for the major spatial tasks of our time. It seems inevitable that this will involve new land take. How much land take there will be depends on the choices that are made now, for example about where which housing locations are envisioned. To support political decision-making, scenarios are an appropriate tool. They can provide direction and insight into possible consequences of policy choices, but they are not predictions. Indeed, the outcomes are largely embedded in the model inputs and serve as a basis for further discussion. For example, both the SUPER scenarios and PBL's assume a situation where top-down control is possible. Ultimately, planning decisions are political decisions where local factors play a major role (Mazzoleni 2022). 4 Monitoring space occupation Regardless of the exact details of the European Commission's proposal to introduce the no net land take target, it is expected that there will be an obligation to monitor and report (change in) land take (European Commission 2021b). In response, some member states have already established monitoring systems, including France and Italy. Member States that already had their own national targets (Flanders/Walloon, Germany, Luxembourg) have been monitoring for some time. The question is what kind of monitoring system will (or should) be established at the European level and to what extent the Netherlands is able to set up its own system to measure domestic land take. 4.1 Quantitative land take An important issue when measuring quantitative land take is the data. For the sake of comparability and verifiability, there should preferably be one data source for the whole of Europe. At present, land take in Europe is mostly calculated using Corine data, including by the European Environment Agency (EEA). On the other hand, there are a number of already known problems with the Corine database . The main objection, elaborated in Section 3.4, is that the resolution is too coarse to accurately observe small-scale urbanization. Because urbanization in the Netherlands usually occurs through relatively large projects (large areas), the problem is less acute for the Netherlands with Corine. GIS analyses performed as part of this study show that national land take figures according to Corine and national (CBS) data do not differ much.5 But a European comparison based on these data could be detrimental to the Netherlands. Work is underway to improve the Corine2018 file by increasing the resolution (to 0.5 hectares) and refining the classification (Cole et al. 2022, p. 22). A relevant change for land use monitoring concerns the recreational area category: it is refined into built-up and unbuilt-up areas. In addition, the division of urban area into four different density categories can help with land take monitoring, especially in a qualitative sense (Rosina & Vizcaino-Martinez 2018). It seems that the European Commission sees bread in such activities. The 2030 Soil Strategy states, "Enhance the use of digital tools and Copernicus and rely on the JRC to further develop the European Soil Observatory (EUSO) and the EEA to develop the Land Information System for Europe (LISE), supported by geospatial analytical products" (European Commission 2021b, p. 20). There are several EU-coverage alternatives to Corine, but all have their own drawbacks. The Global Urban Footprint (GUF) and World Settlement Footprint (WSF) provide accurate data on built-up areas, for example. The Global Human Settlement Layer (GHSL) does distinguish between residential and commercial areas (by linking to population density), but offers nothing else. Data derived from the Sentinel-2 satellite with ten (mostly natural) land use classes are available in highresolution tiff files (10x10 meters). While the Urban Atlas and LUCAS sources both have higher resolution than Corine and also contain many use classes, both do not cover the entire European territory (Decoville & Schneider 2015). In addition, there are initiatives such as the Atlas of Urban Expansion that randomly analyzes the urbanization of various urban areas for a variety of indicators (Angel 2016). National sources may be used for monitoring and reporting. The Soil Use File (BBG) is a digital map by land use function of the Netherlands published by CBS. It shows the dominant land use function at ground level. It uses a classification with 38 land-use categories (e.g. residential area, business park and forest) which are in turn divided into nine main groups. For most categories, this uses a lower limit of one hectare (sometimes 0.5 hectares) in terms of the degree of "contiguity. When producing the BBG, various sources are used: the previous version of the BBG, aerial photos, BRT, 5 This does not necessarily mean that the data are the same, it may be that overestimates in one area are offset by underestimates elsewhere; further research should confirm this. Top25raster, BAG, ABR and Locatus (retail data). A so-called mutation series (years 1996, 2000, 2003, 2006, 2008, 2010, 2012, 2015, 2017) is also published by CBS for the purpose of comparisons over longer periods. The BBG is currently being updated. This will involve as much automation as possible using available (open source) sources. The automation will also make it possible to increase the frequency of publication. Because of the long lead time and set-up of a renewed BBG, a gap in the BBG series of at least 5 years has currently arisen (the last BBG dates from 2017). This may pose a problem for past land use analyses. Other land use records in the Netherlands are: Landelijk Grondgebruiksbestand Nederland (LGN): this is a (semi-)commercial file produced by the WUR with an update frequency of 3 to 6 years. The emphasis is on rural areas, nature and agriculture and then in particular on crop type. It consists of 48 land use classes with a resolution of 5x5 meters. The file is created on the basis of the Basic Registration of Plots (BRP) and the Basic Registration of Topography (BRT-Top10NL). The urban area is virtually undifferentiated. Basic Registration Topography (BRT): also called Top10NL from the Land Registry with an update frequency of around two years (the ambition is to appear annually). This is primarily a topographic file and not a land use file. The expansion of buildings can be monitored well with it, but the classification into different classes is limited (a lot falls in the class 'other'). In the Netherlands there is no data deficit but - as at the European level - there are choices to be made in terms of accuracy, consistency through time and land use classes. Currently, there is no formal land use monitor. This will have to be set up. 4.2 Qualitative land take Monitoring qualitative land take is less straightforward, as there is so much ambiguity about definition and operationalization, in addition to the choice yet to be made about whether and how to combine this with quantitative land take. Here again, for the sake of comparability, it is good to look at European sources first. Work on a monitor for ecosystems in Europe has been underway since 2013. Two recent initiatives are of interest: the Integrated system for Natural Capital and ecosystem services Accounting (INCA) and the Mapping and Assessment of Ecosystems and their Services (MAES). A joint report on possible methodologies and existing technical and political bottlenecks was published in 2018 (Maes et al. 2020). Currently, all member states are mapping their ecosystem services. A European Soil Observatory has now been established and is in the process of setting up a monitoring system. As with Corine's improvements, the ambition seems to be to make monitoring EU-wide. The mission of the ESO is to develop into: 'the principal provider of reference data and knowledge at EUlevel for all matters relating to soil' (Maréchal et al. 2022). This also in anticipation of the proposal for a European soil health law in 2023, the same one that gives rise to this quick scan (European Commission 2022). National data sources may also be chosen. Which geographic information is then needed will depend on further elaboration. It is not yet clear to what extent the necessary data are already available. The degradation of soil quality is often calculated based on drought, reduction of planting, erosion, salinization and carbon removal (Prăvălie et al. 2021) so these could be requested indicators. In any case, the 16 ecosystem services maps are available that were also used in this study. As discussed in Section 3.3, there is currently only information on the current situation. This makes it possible to assess the impact of future land take, but not the evolution of additional land take in the past. 4.3 Other points of interest Setting up a space occupation monitoring system is technically possible. There are still many unanswered questions and potential bottlenecks. Some are of a technical nature while others involve more of a policy choice. In summary: Data and expertise It is recommended to ensure uniform and equal classification over time. Interim changes in classification definitions make monitoring more complex and more difficult to compare with previous monitoring years. If national data sources are chosen, the update frequency must be taken into account. Monitoring should not take place more frequently than data updates: if an indicator is updated once every two years, there is little point in reporting annually. With composite indicators with multiple update frequencies, this can become complex. Working with composite indicators from multiple sources means a loss to the lowest common denominator. If you average raster files from a higher resolution (to a lower resolution), information will be lost. This happens because when averaging the file, the pixels become larger and the number of pixels per square meter decreases. This makes the image coarser and grainier, and small details that were visible on the original file are lost. Some member states including the Netherlands have very high resolution data on, for example, ecosystem services. As a result, editing this GIS data can be technically challenging and require a lot of computing power. A monitor requires the necessary substantive knowledge about both land use changes (understanding increasing and decreasing land take) and GIS technical knowledge. Preferably, this knowledge will have to be present within the same institute or group of people. Definitions There needs to be clarity on some land uses that are not well defined in Corine. These are sometimes difficult because they are temporary and/or multifunctional. These include renewable energy production (wind farms and solar arrays), housing (private green space) and agriculture (stables and greenhouse farming). A choice must be made whether paving should also be calculated in land take. For example, a distinction must be made between a fully built-up residential area and a spacious residential area with lots of greenery between the houses. And whether nonurban buildings (e.g. glasshouses or livestock housing) should be included as land take. Choices must be made about land take reduction (recultivation). It may take a long time to recover soil. Is a change of function sufficient to consider the area non-urban? A choice must be made about the extent to which soil quality should be part of the monitor. This requires different data sources and clear choices about which indicators to use and how to apply them. What role does land reclamation (natural or otherwise) or coastal erosion play in calculating additional land take? Does encroachment on (inland) water count as additional land take, or does this only concern soil? At what level of scale should NNLT take place? Is the goal to be achieved by each state, or can land use increases in one country be offset by greening elsewhere? The same is true within countries: should states, provinces or municipalities all pursue NNLT (as is often done now with CO2 ) or not? Organization and Policy Land and soil have been a mandatory part of environmental impact assessments since 2014 but national practices differ (Vargas 2019). Is land use monitoring intended to bring more unity to this? Urban form is not taken into account in land take increases; only acres (quantitative approach) or soil characteristics (qualitative approach) count. However, urban morphology has many implications for related issues such as landscape fragmentation. Moreover, there is a link here to sustainable urbanization. Land use planning consists of several phases. For project decisions, the location has usually already been determined. It is then recommended that land take increase be taken into account at an earlier stage (for example, in the preparation of visions). Thus, it makes more sense to conduct the environmental impact assessment at plan-mer level (SEA) rather than mer level (EIA).