Data processing
To process the data, the open source softwares QGIS 3.10, GRASS GIS, as well as RStudio (RStudio team,
2020) were used. First, maps of the Berlin buildings
‘BB’, of its protected monuments
‘PM’, of the Berlin blocks
‘BL’ and of the urban greenspaces
‘UG’ were imported into QGIS. All layers were reprojected into the chosen coordinate reference system EPSG:25833 and the
fix geometries command was applied to allow for further processing.
Next, BB was filtered for several attributes that prevent buildings from the possibility of being greened: Building parts marked as either underneath the surface, destroyed or degenerated, or without walls (open warehouses) were removed. All remaining buildings were dissolved to lower the file size and processing time, resulting in a new layer ‘BB-diss‘.
The PM map, was also dissolved and unsuited objects for VGS, named as archeological and garden monuments, were deleted. Thereafter, the layer was dissolved again, to ‘PM-diss‘.
To not include urban greenspaces, which might be irritating for the visualized outcome, those were removed from the BB_diss, PM_diss and the BL layer. To do so, the difference command in QGIS was used, to delete overlapping parts of BB_diss and UG, PM_diss and UG and BL and UG, respectively.
Following, PM-diss, BB-diss and BL were imported to the GRASS GIS software. Larger data sets can be processed here with more computationally intensive commands (i.e., intersection), than in QGIS. PM-diss and BB-diss were then intersected (v.overlay command) which means, that overlapping parts of both layers are kept. The result was dissolved for easier further processing and exported as new layer ‘BB-PM-inter-diss’. This layer shows all parts of buildings in Berlin, that are under monument protection.
To determine the share of protected buildings on block-level, BB-PM-inter-diss and BB-diss were each intersected with the BL layer in GRASS GIS and exported as BB-PM-BL-inter and BB-BL-inter.
Next BB-PM-inter-diss, BB-diss, BB-PM-BL-inter, BB-BL-inter and IC were imported back into QGIS and the fix geometries command was applied again for all of those layers.
To calculate the ratio of protected monuments on city level, first, a column was added to the attribute tables of BB-PM-inter-diss and BB-diss, where information about the area size was generated. Joining the area field of BB-diss to the BB-PM-inter-diss layer allowed for creating a new column, where the area of protected monuments is divided by the area of relevant buildings, which equals the ratio of protected buildings on city-scale.
Those last two steps were almost similarly repeated for calculating the ratio for the inner city, after clipping both, BB-PM-inter-diss and BB-diss by IC, newly generating BB-PM-IC and BB-IC. Columns containing information on the area size were added for both layers. Joining the area field of BB-IC to the attribute table of BB-PM-IC allowed for adding the ratio-column, in which the area of protected monuments in the inner city was divided by the area of relevant buildings in the inner city.
Lastly, for calculating the ratio for each of the over 15,000 blocks, BB-PM-BL-inter and BB-BL-inter were dissolved by the block numbers, resulting in one attribute per block in both attribute tables of the respective layers. Those were enlarged by information on the area size and the area field of BB-PM-BL-inter was joined to the attribute table of BB-BL-inter by the column ‘block number’. Lastly, adding the ratio-column, as done before at both city scales, allows to visualize different levels of protected monument ratios on block level.
For an easier determination of particularly threatened areas, where both, shares of monument protection and the potential heat stress risk are high, the final block-level map was overlaid by the potential heat-stress risk map of Dugord et al. (
2014).
To develop this map, Dugord et al. (
2014) investigated the influence of land-use patterns on the land surface temperature and the location and concentration of vulnerable inhabitants (< 6 years old or > 65). Both factors were rated from 0 to 3 and then combined by multiplying them, leading to categorization of seven potential risk classes (Table
1).
Table 1
Potential risk classes according to Dugord et al. (
2014)
Extremely high potential risk | 9 |
High potential risk | 6 |
Medium potential risk | 4 |
Low potential risk | 3 |
Extremely low potential risk | 2 |
Negligible potential risk | 1 |
Residential uses without potential risk | 0 |
To enable the extraction of different combinations of potential heat stress risk classes and monument protection ratios, heat stress risk classes were aggregated to three levels (0–2 = no – extremely low potential risk; 3–4 = low – medium potential risk; 6–9 = high – extremely high potential risk) and monument protection ratios were aggregated to five classes (< 1 %; 1–25 %; 25–50 %; 50–75 %; 75–100 %). Afterwards, intersecting blocks were extracted for each possible combination of heat stress risk and protected monument ratio and the results were dissolved by the block numbers.
In the last step, the new layers were, again, intersected with a layer containing the population density. Therefore, a field with the area size was added to the attribute tables and multiplied with the number of inhabitants per area , that is part of the information in the population density layer. This had to be summed up to give the number of blocks and inhabitants for each possible combination of heat stress risk and monument protection ratios.