**4. Results**

The first stage of VPO identification and listing resulted in a long list of 42 VPOs containing more localized and non-conventional visual pollution objects including hanging wires, electricity transformers, broken poles, dilapidated buildings, etc. Such VPOs have not been considered in earlier studies with such emphasis. The listed VPOs have been classified by the panel experts on the basis of the similarity of objects which resulted in 10 wider groups as presented in Table 2. A few glimpses presenting some key VPOs from the streets of Lahore, Pakistan are presented in Figure 4. More graphical evidence of these VPOs can be accessed at https://urbanvisualpollution.wordpress.com.


**Table 2.** Classification of VPOs in Major VPO Groups.

**Figure 4.** Pictures from the streets of Lahore, Pakistan capturing some key VPOs including outdoor advertisements, poles, hanging and cluttered wires, architecturally poor structures, dilapidated building and encroachments.

The identified groups were ranked by the experts and their weights were calculated. The consolidated weights and ranks generated by experts through AHP reveal that open dumping of solid waste is marked as the largest contributor to visual pollution (23.8%), followed by outdoor advertisements and signage (20.1%). Dilapidated buildings have been ranked as the third major contributing VPO (13.8%) followed by hanging and cluttered wires (11.1%). The list continues with overflown sewers and drains at fifth place (10.4%), graffiti/wall chalking at sixth place (6.9%), various poles and transformers at seventh place (4.5%), encroachments at eighth place (3.8%), and broken roads/ditches at ninth place (3.5%). The VPO group of architecturally poor structures is ranked at the tenth place with a score of 2.1%. Figure 5 presents the consolidated matrix generated from the individual responses of panel experts while Figure 6 represents the final weights and ranks for VPO groups.


**Figure 5.** Consolidated AHP matrix generated from individual responses of panel experts.


**Figure 6.** Final weights and ranks for VPO groups.

In order to understand the pattern of ranking by each expert, an AHP consensus indicator was calculated using Shannon alpha and beta entropy [60]. The consensus indicator ranges from 0% (no consensus) to 100% (full consensus). The calculated consensus turned out to be 80.1%, which reflects a high overall level of consensus among the experts. Figure 7 represents the mapping of VPO weights given by each panel member. Each line represents one expert while the bold red line shows the average value. The dispersion in the opinion of experts on certain VPOs reflects the diversity which comes in opinion because of their experience, knowledge or professional background.

**Figure 7.** Mapping of weights assigned to VPO groups by each panel expert.

After ranking, rubrics were prepared to systematically measure the characteristics of VPOs. Table 3 shows some of the characteristics and rubric values listed for billboards. Similar tables were prepared for each VPO under study and vetted by the panel of experts.


**Table 3.** Listing of VPOs Characteristics and Preparation of Rubrics for "Billboards".

After the AHP based ranking, the weighting of VPO and preparation of rubrics, VPOs and their characteristics were arranged in the form of a scorecard. This VPA scorecard is a condensed resource (available at https://urbanvisualpollution.files.wordpress.com/2019/02/visual-pollution-assessmenttool-scorecard.png) that can be used to record the prevalence of various VPOs and their characteristics on a site under observation. In addition to VPO related information, the tool records the elements related to place character (number of road legs, dominant land-use, nature of activity, average height of buildings, average road width, average distance between facing building lines, area type (planned/ unplanned), and socio-economic status of the place along with the geospatial coordinates. Place character is particularly useful to generate correlations at the analysis stage. The data collected through the scorecard is processed through a visual pollution score calculator sheet (available at https://urbanvisualpollution.files.wordpress.com/2019/02/visual-pollution-assessment-tool-scorecalculator-sheet.png) that presents the sequential stages of assigning inter-group weights, rubric values for VPOs and then the contribution of the total of those assigned numbers in the VPO score calculator.

The final form of the VPA tool has been made available under GNU General Public License v3.0 at GitHub with open public access at https://github.com/khydijawakeel/UrbanVisualPollution. Furthermore, the tool has been placed at https://urbanvisualpollution.wordpress.com as well where other researchers can access and use for similar studies.

As discussed previously, the tool has been tested at 20 locations to assess its validity and reliability through IRR analysis. Figure 8 shows the level of calculated visual pollution on those 20 assessed sites. Table 4 presents the number of inter-observer agreements for each location/site. Furthermore, agreemen<sup>t</sup> ratio has been calculated for each observer pair and then the mean of each row has been calculated to see the overall agreemen<sup>t</sup> ratio at each site, as presented in Figure 9.

**Figure 8.** Map showing the spatial spread of visual pollution on the 20 assessed sites (heat map color scale from red to green represents highest to lowest visual pollution).


**Table 4.** Number of inter-observer agreements.
