|
nif:isString
|
-
Methods to Estimate Demand for Industrial and Commercial Land: The objective of this study is to develop and validate approaches to estimate demand for industrial and commercial land. The approaches should be relatively straightforward so that they can be easily replicable and applicable to large spatial regions (e.g. countries and continents) in the context of land use modelling. We propose to explore in particular those approaches based on land use intensity measures. The main reason for this choice relies on the fact that, as reviewed in the previous section, these measures are not especially intensive to calculate, requiring only a few aggregate variables, characteristics which become relevant when working at very large spatial extents. Still, intensity measures are informative and conceptually easy to interpret. Moreover, they link to sector-specific processes of economic development that are expected to be relevant for the land uptake of industrial and commercial use. Using intensities allows linking land use simulations to regional economic projections. Below in this section, several variants of the land use intensity approach are formally introduced. By definition, intensity measures integrate one driver of land use change at a time. In this study, sector gross value added (GVA) is used as a proxy for sector economic output. In addition to the land use intensity approaches, trend extrapolations are also tested. Trend extrapolations can be seen as the simplest way to make estimations because they do not specifically address drivers of land use change, but rather apply observed growth trends to describe possible future conditions. The main reason to consider trend extrapolations in this study is to create an adequate term of comparison for the estimations based on intensity measures. Once introduced, the models will be applied to estimate the demand for industrial and commercial land for a set of countries in Europe. Trend extrapolation (models 1 & 2): Two trend extrapolation methods are considered: a linear extrapolation (model 1) and an exponential extrapolation (model 2). The linear extrapolation is formulated in equation 3:(3)where A refers to the industrial and commercial area, t0 and t1 correspond to the starting and ending years of the calibration, respectively, and ε is the error term. In this method, the average yearly absolute growth of the calibration period (t0 to t1) is multiplied by the total forecasting years (t2–t1) to obtain the estimate for desired year t2. In the exponential extrapolation, the average yearly growth rate G observed between t0 and t1 is firstly obtained through equation 4, and it is then applied to estimate the industrial and commercial land in t2 (equation 5). The graph in Figure 1 shows the application of both models to a hypothetical region with 200 ha of industrial and commercial land in 1990, and 300 ha in 2000.
Figure data removed from full text. Figure identifier and caption: 10.1371/journal.pone.0091991.g001 Extrapolation models for hypothetical region with 200 and 300 hectares of industrial and commercial land in 1990 and 2000. (4)(5) Region specific land use intensity (models 3 & 4): The economic product and surface area of commercial and industrial uses are consistently highly correlated at the regional level. If we take the sum of the regional gross value added of the industrial, commercial and service sectors and relate it to the respective surface area as reported by the CORINE Land Cover (CLC) datasets, correlation coefficients ranging between 0.74 and 0.76 can be found for the years 1990, 2000 and 2006, in Europe. This suggests that, in general, the higher the economic product of a region, the more physical infrastructure is required to support the economic activity. Models 3 and 4 are characterized by using economic output or product P of regions as the driver of development of industrial and commercial areas. In both models, a land use intensity approach is used to relate the economic product with the respective area of industrial and commercial units. In model 3, the land use intensity LUI is computed for the year t1 and measured as economic output per hectare of industrial and commercial land (eq. 6). Then, assuming a stable land use intensity in time, and knowing the product P for t2, the total industrial and commercial land is predicted (eq. 7). (6) (7) This model takes the whole regional product and the whole existing area of industry and commerce per region in t1 to compute the land use intensity. However, the total amount of industrial land is strongly related to historic developments and only partly dependent on current economic performance. In fact, as existing industrial and commercial land is likely to remain (with or without actual economic activity), this inertia is not captured by a single and static snapshot of the land use intensity. So we should perhaps focus especially on changes in economic development and related changes in the amount of land needed. This implies that the land use intensity of new developments is important in order to capture shifts in the production structure. Model 4 builds upon this idea. It measures land use intensity only of the industrial and commercial land developed during the calibration period t0 and t1 (eq. 8). The ‘land use intensity of the recently developed land’ is then used to estimate the extra land related to the growth of the product in the subsequent period (t1:t2) (eq. 8). Contrary to the model 3, this approach ignores the land use intensity of the industrial and commercial land developed prior to t0. (8) (9) We call these approaches ‘region specific’ because the intensity measures described above can be computed separately for any set of regions composing the whole of any area of interest. Consequently, regional differences in land use intensity (which are underpinned by differences in productivity and production structure) are captured.
Region and sector specific land use intensity (models 5 & 6): Industrial and commercial land is a rather broad and heterogeneous land use class. For example, in the CORINE Land Cover nomenclature, the homonymous land use class includes factories of all different kinds of industries, facilities for energy production and telecommunication networks, facilities related to defence and security, shopping malls and exposition sites, and a wide range of facilities related to public or private services likes schools, university and research campuses and hospitals [41]. Trying to model such a heterogeneous class as a whole poses obvious limitations. Most obvious of all, land use intensities vary considerably among industries [28], [30], [31], let alone the differences between the various economic sectors. To address this limitation, one could think of making the land use intensity measures both region and sector specific. In this case, the economic product of a given sector s would be related to the land area A known to be used by sector s in year t1 (eq. 10). At this point, it would be possible to estimate the aggregated industrial and commercial land for a given t2 (eq. 11). Conceptually, this formulation is more robust than models 3 and 4 because it allows the integration of land use intensities specific to n number of sectors (model 5). In addition, a factor ω could be used to transform the land use intensities when calculating A in t2, as a function of the observed changes in LUI between t0 and t1, that is ωs = f(ΔLUIs,t0,t1). In this study, we let ω = 1 for all sectors. (10) (11) This model can be also combined with the concept of ‘land use intensity of the recently developed land’, as introduced in model 4. This is done by applying equations 12 and 13 (model 6):(12) (13)
|
![[This page as RDF]](http://data.gesis.org/somesci/static/rdf-icon.gif){kind=icon}