3) Methodology

Methodology: phase 1 (NUAs selection)

Phase 1 of the methodology consists in determining which non-urban areas (NUAs) are best suited to new forms of urban agriculture (NFUAs), according to several criteria. Land uses classified as lands without current use (LWCU) and forests could potentially be transformed into new forms of urban agriculture. The land uses in question have thus been selected and extracted out of the complete land use layer. The map below represents those two types of land uses in the district of Innsbruck. Note that the layer of the forests has been split in 4 parts to reduce the size of the file in order to be supported by Google Maps.

In total, there are 147 patches of lands without current use (LWCU) and 1357 patches of forests. Forests are abundant in Innsbruck, they represent in total an area of 716,85 km², while the lands without current use patches are less numerous and much smaller, representing a total area of only 1,04 km². Lands without current use are mostly located around urbanised areas.

In addition, forests patches must have an area of less than 0,15 km², as clearing too big surface areas would remove the advantages of evapotranspiration and biodiversity provided by the forest. These advantages are greater than the expected benefits brought by the future agricultural lands. Lands without current use must be within a distance of less than 1 km to urban areas, to be easily reachable by the population who will make use of the new agricultural land. 

Relating to the urban areas, one additional criterion has been defined. If not enough people live near the new agricultural infrastructures, they won't be useful and successful, since it is the role of the inhabitants to maintain the NFUAs. Moreover, as only a part of inhabitants would be interested in purchasing products, enough inhabitants are needed to make the NFUA economically viable or even profitable. For these reasons, not all urban plots have been kept, some isolated residential areas have been removed from the category "urban". 

To select the urban areas that are too isolated and in which the population is too sparse, a buffer of 500 meters around all urban areas (continuous and discontinuous urban fabrics) has been created to obtain groups of properties. This threshold of 500 meters was chosen in relation to another criterion that will be applied in the second phase of the methodology (future urban farms will have to be within 500 meters of urban areas) and it seemed to be a reasonable walking distance between the houses and the NFUAs.

The Urban Atlas classifies the urban land use in continuous urban fabric (S.L.: > 80%), discontinuous dense urban fabric (S.L. : 50% - 80%), discontinuous medium density urban fabric (S.L. : 30% - 50%), discontinuous low density urban fabric (S.L. : 10% - 30%) and discontinuous very low density urban fabric (S.L. < 10%). Continuous urban fabrics are constituted predominantly by residential use and are areas with a high degree of soil sealing. The different discontinuous types are characterised by a larger fraction of non-sealed and/or vegetated surfaces: gardens, parks, planted areas and non-surfaced public areas (European Environment Agency, 2011). We can thus assume that the discontinuous urban areas will accommodate less inhabitants than continuous urban areas.

An index has then been created to better approximate the population inside the urban groups. It's the product of the area (in square meters) of the urban plots inside each urban group times the soil sealing degree in these urban plots. For this weighting parameter, the average of the soil sealing degree has been considered (0,9 for the continuous urban fabric, 0,65 for the second category, 0,4 for the third, 0,2 for the fourth and 0,05 for the last category). The sum of the areas of urban plots multiplied by the weighting parameter in each urban group has been calculated. Finally, an arbitrary selection of only the urban groups containing at least 0,5 km² of total urban area (the newly created index) has been done. As a result, 3474 urban plots are remaining out of 3934 originally. We could observe that the number of plots inside the groups was equal to 99 urban entities minimum and 2528 entities maximum . Figure 3 represents these selected urban areas (in red), compared to the total urban fabric present in Innsbruck (in blue). In Model Builder, this layer corresponds to "urban_final.shp".

The buffer of 1 km inside which the lands without current use (LWCU) must fall was performed around this selection of urban plots.

Figure 3: Urban areas considered in the analysis

The two layers fulfilling the criteria have finally been merged together, to obtain one single suitability layer, in which the different land uses are distinguishable. This result will be displayed in the section 4 of the project. Finally, this polygon layer has been converted into a raster layer for the second part of the methodology.

Figure 4: First phase of the methodology - Model Builder (to be read from the top to the bottom)

Methodology: phase 2 (NFUAs selection)

Now we have to determine which new form of urban agriculture (NFUA) is best suited to each NUA. As mentioned in the introduction, we consider two possible types of NFUA: allotment gardens (AG) and urban farms (UF). 

The region group tool was used to form patches of NUAs in a raster format, based on the contiguity between cells of the same land use (lands without current use or forests). The contiguity based on four cells has been used, knowing that the pixel size is equal to 5m x 5m, an eight cells contiguity would have assembled several very close patches together. Even with this method, the rasterisation can create small distortions. This 5x5 pixel size has been chosen to have a sufficient degree of precision, without making the raster file too large, which would slow down the geoprocessing.

After that, we need to select only the patches that fulfil the criteria of allotment gardens or urban farms. AG can only have a maximal area of 5 000 m² and need to be within 250 m reach of urban areas. Allotment gardens need a very good accessibility because they are mainly provided for leisure and integration of older people and socially deprived groups, who can't travel long distances (La Rosa et al. 2014). Gardening is the principal activity in allotment gardens. The areas judged too small for being an allotment garden (areas under 50 m² according to Rubino (2007)), have been removed, resulting in the deletion of 23 plots of NUA). UF are bigger than allotment gardens, they need to have an area between 5 001 and 20 000 m², and be within a maximal distance of 500 m to urban areas. They are devoted to multi-functional agriculture. The production of fresh products near urban areas contributes to urban employment and the reduction of inequalities, to landscape conservation and socio-educational functions, to the city's waste recycling, and so on. The definition of "urban areas" is the same as in the first step of the methodology. We considered the plots of NUA completely within the defined distances to urban areas, not those partially inside, because forest patches can often be narrow and long, so a distance of 700-800 supplementary meters can be necessary to reach the opposite end of the land parcel. As a result, we obtain the plots of each NUA eligible to be converted to each type of NFUA. The result is presented in section 6 of this blog.

Methodology: phase 3 (sensitivity analysis)

A sensitivity analysis has been done in order to see the impact that input factors have on the results. 2 additional scenarios have been tested and compared together, and with the initial scenario, which consists in promoting a diversity in the urban agriculture.

The first scenario accepts only the creation of NFUAs at a shorter pedestrian distance from urban areas, because the population density is not high in the district of Innsbruck (equal to 1 211 hab./km² in the city centre and 86 hab./km² in Innsbruck-Land, i.e. the region surrounding the city). The Euclidean distance buffer has thus been reclassified to a maximum of 125 meters for allotment gardens and 250 meters of urban farms, the original maximum distances have been divided by two. All other conditions remain the same as the initial scenario (scenario 0).

In the original scenario, many forest patches selected in the first phase weren't considered for transformation into NFUAs because they were too big compared to the maximum size of 20 000 m² for urban farms. That's why in the second scenario, bigger urban farms are allowed. We rise the threshold to 50 000 m², to implement larger and more productive farmland areas where for example organic farming could be implemented. The urban distance criteria are kept unchanged with respect to scenario 0.