13. Application – Site Analysis Furniture Store

Once the layer structure is set up the next step is to perform a georeferencing on the raster map that will contain the information about where in NRW the existing sites of the company are located. A proper georeferencing connects points of the image coordinate system with their geographical coordinates of the given reference system. Either the geographical coordinates are picked out of already referenced material (shapefile, raster image, etc.) which is then called point-to-point-referencing or your unreferenced image data is overlaid by a cartographical grid that defines the coordinates of certain points in the map (point-to-coordinate-referencing). As the unreferenced raster map here covers the area of whole NRW as well as the shapefile of the municipalities the georeferencing can be processed via the point-to-point referencing method, using the georeferencer in QGIS. It is necessary to distribute the connection points homogenously over the whole map to make sure that the it is georeferenced equally accurate, as an algorithm is interpolating the coordinates for the rest of the map according to the chosen points. Clustering these points in a certain area will affect results in a negative way. As image coordinates are transferred to geographical coordinates, the order of setting the points is always first point on the unreferenced map, second point on the referenced material (Fig. 13.2).

After at least four well-distributed control points are set and the mean error is in an acceptable range, the software interpolates the coordinates for the rest of the referenced raster image and saves it as a new and overall referenced dataset (for more information on coordinate systems and projections see Part I, Sect. 1.​1.​2. Spatial Data Models).

After the raster map has geographic coordinates, the locations need to be extracted out of the map by digitizing the coordinates in a point shapefile as until now the locations are only pixels in a continuous dataset (see Part I, Chap. 1. Data Sources). To do this, a new point layer has to be created that has both the corresponding coordinate system and the necessary attribute table field to store the name information of the city where the existing sites are located (Fig. 13.3).

Once there is a new and empty shapefile that has the coordinate system and the table field, the edit session needs to be initiated for this particular layer. Only then features and their attributes can be added to it (here: city names). To digitize the features as well as the linked attributes it is necessary to add the point features via the ‘add features’-tool and then extend them by the information of the city name in the respective table field (Fig. 13.4). When this is done the edit session needs to be saved and toggled off.

One of the given requirements for the analysis is a population density of at least 800 inhabitants per square kilometer in each municipality. As the attribute table of the shapefile that contains the geometries of the municipalities stores neither the information of the relative nor the absolute number of inhabitants per municipality, these values have to be added.

They can be found in the csv-table that contains the (absolute) population data of NRW and needs to be imported into your GIS beforehand so that all the columns and lines are displayed properly.

For the import of the csv-file it is necessary to set several parameters. First there is the delimiter of the different fields (comma, tab, space, etc.), which ensures a proper conversion and puts each value into an individual cell. Second the encoding of the table content has to be set to make certain that all values are displayed correctly. Finally, one has to check if there are coordinates in the table that could be converted into geometries (Fig. 13.5) which isn’t the case here. After importing the table, it needs to be saved as either dbf- or spatial lite-format to provide that it’s editable.

To then add this imported data to the attribute table of the shapefile, it is necessary to perform a table join. This requires the identification of a key field (unique ID) that connects both tables – attribute and imported table – via identical values of each feature (Fig. 13.6).

Since a table join can only be done successfully if both key fields are of the same data type (text, number [integer, double, etc.], date, etc.), it can be necessary to copy the values of the key field of the imported table into a new field, setting a data type that matches the key field in the attribute table.

In this case, the key field is the field that stores the code of the particular municipality consisting of seven digits specifying the data type as a number (Fig. 13.7). As the data type of the field in the imported table is a text and there are missing several digits in some cells that prohibit an overall join, it is necessary to create a new field and copy the code values of the text field as numbers to this new field. Therefore the table was modified to dbf or spatial lite in the first place to make it editable.

To now add the missing digits to the respective values, all cells that need to be updated are selected by an SQL expression (see Part I, Sect. 3.​1. Simple Spatial Analysis). Looking at the values in the cells it can be stated that either they are complete and ready for a join or they are missing three zeros on their right to match the ones in the attribute table of the spatial layer. The task now is to find an expression that filters the table and only selects the specific values that need an update and then multiply them by one thousand which will add three zeros. As all affected values are in between a range of five thousand and six thousand the right expression for the selection here is “<field name>” > 5000 AND “<field name>” < 6000. Via the field calculator the selection is multiplied by one thousand as mentioned above (Fig. 13.8).

After that the table join can be done by navigating to the properties of the municipality shapefile and add a join, choosing the key field in the attribute table, then choosing the table whose values should be added and then choosing the key field in that particular table.

To prevent the result dataset from being changed it is then exported into a new single and persistent shapefile that contains the geometries of the municipalities as well as the absolute population data.

The first positive layer for the analysis contains all areas that are in range of 1000m within freeways. That is a valuable approach to provide an easy access to the new store avoiding small roads as most customers use to go shopping with their own car. To create this layer as a shapefile, all freeway features of the street network layer need to be selected, buffered by 1000 m and then exported to a new shapefile.

The selection is done most accurately and effectively via an expression in structured query language (SQL, see Part I, Sect. 3.​1. Simple Spatial Analysis). The relevant information for this selection is stored in a certain field that implies abbreviations of the street classes combined with the individual numeric value of the street. All freeways in NRW are tagged with an A (freeway in German: Autobahn) and an individual number. As the expression should manage to select all highways at once, the individual number of the freeways is replaced by ‘%’, changing the ‘=’ to a ‘LIKE’, as now the value does not equal a certain value but is implying all the different freeways at once, what then leads to the expression “<field name>” LIKE ‘A%’.

The result layers of all upcoming calculations are saved in two different directories – one called positive (for the requirement-layers) and one called negative (for the restrictive layers).

To then buffer all selected freeways by the value of 1000 m and save it in the positive-folder as an independent shapefile, one has to open the buffer tool, select the input layer, set the beforehand selected features as the only ones that should be used and define a buffer distance with a value in the unit of the used coordinate system.

If the result features should be accumulated to have one mutual border in case of overlaps, it is necessary to dissolve them (Fig. 13.10). In some versions of the tool (depending on the GIS-software and its version), it is as well necessary to define the number of segments that build the outline of the buffered area (the more segments, the rounder the shape).

As there are not only freeways but also federal highways that should be in direct range of the new store to assure a good accessibility, the whole process can be repeated in almost the same manner. The only thing that needs an adjustment is the expression for the selection of the particular features. In this case to “<field name>” LIKE ‘B%’ OR “<field name>” LIKE ‘N%’ since the federal highways are tagged with two different letters. To not select the features that contain both – B and N – the logical operator OR connects the two expressions, for only one of the two commands has to be true to select the corresponding feature.

Afterwards the selection is buffered by 500 m (Fig. 13.11) and then saved into the positive-folder.

The UTM coordinate system is in the metric unit meter, which can require another important step (depending on the used GIS) for the calculation of the population density per square kilometer, as the area has to be multiplied by 1.000.000 (1 km² = 1 m² × 1.000.000) in case the unit of the area can’t be defined before the calculation.

After the population density is calculated properly, all municipalities that are greater or equal than 800 inhabitants per square kilometer need to be selected by another expression.

As the only reasonable operator here is the greater than or equal to operator (>=) the expression for this selection is “<field name population density>” > = 800. The selection then is exported to a separate and persistent layer into the positive-folder (Fig. 13.12).

The combination of all three files from the positive-folder to a single layer is done via an intersection, which is equal to a logical AND (‘&’). After simply setting the three layers as input, the output of the intersect-tool only contains areas where all requirements are fulfilled concurrently. Areas where only one or two requirements are accomplished are excluded from the positive result layer (Fig. 13.15).

For the overall negative layer the single layers have to be combined to an overarching layer via the merge or union tool, which is equal to a logical OR (‘|’). No matter if one or all restrictions are regarded – there is no way the new site will be built in an area that is located within a single negative layer (Fig. 13.16).