How does proximity to the Ap Liu Flea Market influence the pattern of urban social stresses in Sham Shui Po, Hong Kong?

How does proximity to the Ap Liu Flea Market influence the pattern of urban social stresses in Sham Shui Po, Hong Kong?

Located in the Kowloon peninsula (Figure 2), Sham Shui Po (SSP) is an urban district within Hong Kong highly populated with low-income earners, low-skilled workers, and the temporarily unemployed. As such, SSP has earned the title of “the centre of poverty in Hong Kong” (Cheng 1). As one of Hong Kong’s earliest industrial and commercial areas, informal businesses and street markets, such as Ap Liu Flea Market (ALFM), have been and continue to be the lifeblood of this neighbourhood. Situated between Nam Cheong and Kweilin Street (Figure 3), Ap Liu Flea is one of the oldest street markets in Hong Kong and a popular destination for cheap, second-hand electronics.

SSP was designated as the fieldwork data collection site due to its physical and socio-economic features demonstrating apparent characteristics of an urban area, thus aligning with this exploration of urban social stresses. Furthermore, the grid-like pattern of this neighbourhood’s roads allows for primary data to be collected and recorded in an orderly manner.

The investigation pertains to the syllabus of Option G: Urban Environments and, more specifically, noise pollution and pedestrian congestion.

Urban social stresses will increase with proximity to the ALFM.

• Noise pollution will increase with greater proximity to the ALFM.
• Pedestrian congestion will increase with greater proximity to the ALFM.

The Peak Land Value Intersection (PLVI) is a region within a settlement with the highest land value and accessibility. As a result of its easy accessibility in terms of transport networks and overall walkability, the ALFM may be labelled as one of SSP’s Secondary Land Value Peaks (Figure 3). Hence, one may infer that it is an area with high urban social stress. Despite having a relatively low land value, the ALFM’s accessibility may result in it being heavily and densely populated, thus resulting in higher rates of pedestrian congestion and noise pollution.

Data was acquired on Friday, November 11, 2022, from 10:00 am to 14:30 pm. Prior to arrival, groups of students were assigned field paths for which they were responsible for collecting data. Systematic sampling was used to determine the data collection points for noise pollution (Figure 4) to allow for the consistent observation of changes in urban stress. Convenient sampling was used for pedestrian congestion data.

Investigating noise pollution

Materials used 

• Timer app
• Decibel X app

At all designated points, the Decibel X app was used to calculate the ambient noise at a distance - the average reading over one minute was recorded.

This methodology revealed the areas with heavy pedestrian and or traffic congestion, thus demonstrating its accessibility, one feature that a PLVI and or secondary land value peak must posses. Constant exposure to noise pollution commonly causes health problems including noise induced hearing loss, high blood pressure, heart disease, and stress.

Materials used -

• Timer app

A one minute pedestrian count was conducted at each intersection. Two students stood on either side of the street and counted the number of individuals arriving on their side; the two numbers were added together and recorded

This methodology’s use of convenience sampling revealed the distribution of pedestrian congestion in SSP, thus revealing the locations with higher accessibility which may therefore be labelled as one of SSP’s secondary land value peak (Figure 3).

To determine the strength of the correlations between specific indicators against proximity to the flea market, Spearman’s Rank Correlation Coefficient will be used. To calculate the Rs value from the rankings within the data sets, the following mathematical notation will be used -

\(R_s=1-(\frac{6Σd^2}{n^3-n})\)

Noise pollution will increase with greater proximity to the ALFM.

The bubble map shows that noise levels are highest in the centre sections (S5-8), with S7 being the only section with an average of 79-84 decibels. Contrarily, S9-12 had the least noise pollution, with S10-12 having an average of 61-66 decibels and S9 having an average of 55-60 decibels.

The scatterplot shows that noise levels decrease as the distance from the flea market increases. For example, 60 m away from the ALFM, the noise level was at 72.3 decibels (green point), whereas 528 m away from the market, the noise level was at 53.3 decibels (purple point). The average level of noise is approximately 75 decibels. Judging from the r-squared value of 0.127, the model cannot account for a significant amount of variability, and the correlation between noise levels and distance to the ALFM is relatively weak.

The high noise levels in S5-8 in Figure 8 and the patterns in Figure 9 can be attributed to how the ALFM is an economic hub, as well as one of SSP’s secondary land value peaks. Dragon Centre, SSP’s PLVI, is also situated nearby. With reference to the PLVI model, theoretically, this implies that SSP’s highest land value and commerce are in the sections that surround ALFM. Considering this area’s high accessibility (ie. high density of transport links) and how it consists of the neighbourhood’s primary economic activities, it makes sense that these sections would attract more people, thus creating a higher and denser population within the area. Correspondingly, urban stress, including noise pollution from pedestrians and cars alike, would also increase.

The negative correlation of the trend in Figure 9 showing how noise levels decrease with distance from the ALFM can also be explained through the acknowledgement of the Bid Rent Theory. The Bid Rent Theory is a concept suggesting that price and demand for land declines with increasing distance to the PLVI due to limited communication links. Therefore, the primary factor that attracts retailers to establish their businesses in districts occupied with economic activities is high accessibility. As such, it can be assumed that there are significant levels of both pedestrian and traffic congestion in this location, especially during the late morning to early afternoon, the typical peak operating hours for a business. As a result, the noise emissions from vehicles combined with the loud voices from business transactions being made and people walking to and from the market, noise pollution increases with closeness to the ALFM. Moreover, as S9-12 (Figure 8) are further from the PLVI, the retail sector would not compete for this land as it would not maximize profits; fewer people would come here to either work or purchase goods. Instead, these sections have been designated to be residential areas which one could assume to lack pedestrian and or traffic congestion during the time of data collection, thus explaining the lower noise levels.

Furthermore, at around 380m from the market, an anomaly of around 35 decibels is present (Figure 9). The reason behind this is that the data point had been collected on the street adjacent to the SSP park. In addition to the absorption of noise pollution by the park’s surrounding greenery (Kong 27), the low noise level may also be due to the data being collected midday. As the data collection period was synchronous to school/work hours, and the elderly tend to utilise such parks in the early mornings or evenings where temperatures are milder, it can be assumed that the park was relatively empty at the time. As a result of an absence of people occupying the park, the anomalous noise level was significantly lower than the rest despite its proximity to the market. Because the SSP park occupies the majority of S9 (Figure 8), the section with the least noise pollution, this reasoning may also be used to explain why noise levels averaged to only 55-60 decibels in this region.

Overall, the analysis of the bubble map and scatterplot justifies the sub-hypothesis that noise pollution will increase with greater proximity to the ALFM. Additionally, it can be said that levels of social urban stress are influenced by the level of economic activity and the location of greenspace and residential areas.

Pedestrian congestion will increase with greater proximity to the ALFM.

As shown on the choropleth map, pedestrian congestion is more apparent in areas surrounding the ALFM and the Dragon Centre (PLVI). Around that area, the streets are mostly populated with over 32 pedestrians, as revealed by the concentration of dark green. Perversely, the least congested streets are found in the south of SSP, as revealed by the average count of 0-8 pedestrians.

The histogram also supports the sub-hypothesis that pedestrian congestion will increase with proximity to the flea market. From 0 to 14.38 metres from the market, the average pedestrian count is approximately 38 (red column), after which the following distances, up to 43.13 metres, reveal a reduced but still relatively high count of 15 and 9 pedestrians, respectively. The pedestrian count continues to decrease at a constant rate until 100.63 to 115 metres, the furthest range of distance from the ALFM within the fieldwork area, which demonstrates an average pedestrian count of 1 (yellow column).

The high pedestrian count around the ALFM and Dragon Centre (Figure 10) is due to how land value and commerce are highest in these two locations due to their high accessibility in terms of transport networks and walkability; ALFM is one of SSP’s secondary land value peaks whereas Dragon Centre is the PLVI. As a result, more individuals would be present in this area due to the extensive retail activities available. Conversely, due to the south of SSP being a residential area, it is not only significantly less accessible but there are also little to no shops, therefore explaining the lack of pedestrians in this region (Figure 10).

Additionally, the patterns of pedestrian congestion demonstrated in Figures 10 and 11 are also influenced by the surrounding public transport. As the ALFM and Dragon Centre are closely surrounded by a plethora of SSP’s Mass Transit Railway (MTR) stations, as circled in Figure 10, there is not only increased accessibility to these areas as compared to other

locations in SSP but there is also a strong likelihood that the entrances and exits of the MTR stations will have high pedestrian traffic. As a result of citizens frequently coming in and out of the SSP stations, pedestrian counts will increase, even if those individuals are not necessarily crossing the road to access the ALFM or Dragon Centre. Likewise, the primary reason for the number of pedestrians significantly decreasing from 14.38 metres (Figure 11) may be that most SSP MTR stations are located from 0 to 14.38 metres from the ALFM.

The expanse of the darkest green around the ALFM (Figure 10), along with the moderate pedestrian congestion from 14.38 to 43.13 metres, can be explained by the layout of the ALFM. As seen in Figure 12, instead of the market being limited to a single point on the map, it stretches across multiple blocks. Hence, up to at least 43.13 metres from the centre point, individuals are still walking through the ALFM, thus contributing to the pedestrian count. Despite the ambiguous start and ending of the market (Figure 12), one might assume that from 43.13 metres away from the ALFM point and onwards (Figure 11), fewer shops are present as those streets start to transition from the market to areas in SSP with different land uses.

According to the Spearman’s Rank Correlation Coefficient of -0.65 (Figure 13), there is a correlation between pedestrian congestion and distance from the ALFM. As the correlation coefficient is less than 0, the relationship is negative; as one variable increases (distance from the flea market), the other variable will decrease accordingly (pedestrian congestion).

Given that the range of Spearman’s Rank Correlation Coefficients is from -1 to +1, with these values indicating that a perfect association of ranks is present, The Rs of -0.65 suggests a strong correlation between pedestrian congestion and distance from the AFLM. As seen in Figure 14, the Rs value being notably higher than the 0.1% significance level reveals how the probability of the correlation between this indicator of urban social stress and the distance from the ALFM being a chance event is less than 1 in 100. Although these results

cannot prove that the increased pedestrian congestion is a result of the distance to the market, it allows one to be over 99.9% confident that a relationship exists.

Ultimately, from the choropleth map, histogram, and Spearman’s Rank Correlation Coefficient, it can be deduced that the sub-hypothesis of pedestrian congestion increase with greater proximity to the ALFM is accurate. Moreover, the data collected reveals how accessibility to public transport and the overall layout of the ALFM influences urban social stresses.

Overall, there appears to be a correlation between urban social stress and distance from the ALFM, thus proving both sub-hypotheses. Firstly, the bubble map (Figure 7) shows that average noise levels were loudest surrounding the flea market due to it being a secondary land value peak. The scatter plot (Figure 8) showed a correlation between these two variables and demonstrated how proximity to green space and residential areas might reduce urban stress. Despite this data proving the first sub-hypothesis, the graph’s r-squared value of 0.127 revealed that the relationship between noise pollution and distance from the flea market was not as strong as anticipated. Secondly, the choropleth map (Figure 10) and histogram (Figure 11) revealed that pedestrian congestion decreases as the distance from the flea market increases. This pattern was due to the ALFM’s proximity to SSP MTR stations and the linear layout of the market. The Spearman’s Rank Correlation Coefficient of -0.63 further substantiated the existence of such a relationship. All in all, this investigation revealed that urban stress in SSP, notably noise pollution and pedestrian congestion, increases with proximity to the ALFM.

There were no significant problems regarding the methodology of data collection. Regarding measuring noise levels, taking the average reading over one minute helped minimise anomalies. Moreover, the risk of miscounting pedestrians was reduced by having two people stand on opposite ends of a crosswalk. However, a weakness of both these methodologies is that data was collected over 4.5 hours. Assuming that rush hours and lunch breaks play a role the urban social stress an area experiences, this time frame allowed for variances in the data. Alternatively, the methodologies could be carried out simultaneously at the data collection point, meaning that students would conduct the methodology individually instead of in small groups.

The data presentation was justified based on the assumption that the ALFM is a secondary land value peak. Despite its accessibility, it is a street market and therefore lacks the land value that a secondary land value peak should have. This raises the concern that the data does not sufficiently align with this theory, thus hindering the overall validity of the investigation. Instead, the research question could be revised to: “How might land use impact urban stress in Sham Shui Po, Hong Kong?”. In this case, the Hoyt’s Sector Model and the Multiple Nuclei Model could be explored. Additionally, unlike SSP’s other land value peaks, such as the MTR stations and Dragon Centre, the location of the ALFM is linear instead of a point on the map. Hence, it is challenging to see the impacts of distance on urban stress, which may impact the accuracy of data presentation methods.

Barcelona Field Studies Centre. "Spearman's Rank Correlation Coefficient." Barcelona Field Studies Centre, 6 May 2022, geographyfieldwork.com/SpearmansRank.htm. Accessed 6 Dec. 2022.

Cheng, Christopher. "Sham Shui Po: The Centre of Poverty in Hong Kong." JSTOR, 2013, www.jstor.org/stable/23891236. Accessed 6 Dec. 2022.

DeWolf, Christopher. "The Second Coming of Sham Shui Po: Revelation or Revolt at Gentrification?" South China Morning Post, May 2015, www.scmp.com/lifestyle/article/1805321/second-coming-sham-shui-po-revela tion-or-revolt-gentrification. Accessed 6 Dec. 2022.

Google Maps. Hong Kong SAR. Google, 1 Feb. 2023, www.google.com/maps. Accessed 1 Feb. 2023.

Habitattt. Apliu Street Flea Market 鴨寮街. Trip Advisor, 4 Mar. 2013, en.tripadvisor.com.hk/ShowUserReviews-g294217-d2226330-r153655375-Ap _liu_Street_Market-Hong_Kong.html. Accessed 1 Feb. 2023.

Hong Kong Tourism Board. Sham Shui Po Park. Hong Kong Tourism Board, www.discoverhongkong.com/uk/interactive-map/sham-shui-po-park.html. Accessed 7 Feb. 2023.

Koster, Brecht. Parts of an Urban Area Central Business District CBD. SlidePlayer, slideplayer.com/slide/14103705/. Accessed 1 Feb. 2023.

LT. Flea Market in Aerial View in Apliu Street in Hong Kong. Adobe Stock, stock.adobe.com/hk/images/flea-market-in-aerial-view-in-apliu-street-in-hong -kong/221699066. Accessed 9 Feb. 2023.

Margaritis, Efstathios, and Jian Kang. "Relationship between Green Space-related Morphology and Noise Pollution." ScienceDirect, Jan. 2017, www.sciencedirect.com/science/article/abs/pii/S1470160X16305635#:~:text= Green%20spaces%20have%20been%20proved,explored%20on%20the%20ur ban%20level. Accessed 6 Dec. 2022.

Sen, Koushik. Significance Graph for Spearman's Rank Correlation Coefficient. Research Gate, 2006, www.researchgate.net/figure/Significance-graph-for-Spearmans-rank-correlati on-coefficient-r-N-represents-the_fig5_223238852. Accessed 9 Feb. 2023.

Tsoi, Grace. "No Homes in Po Town." South China Morning Post, 10 Mar. 2011, www.scmp.com/magazines/hk-magazine/article/2033295/no-homes-po-town. Accessed 6 Dec. 2022.