Environmental Geology

, Volume 49, Issue 7, pp 946–959

Change of groundwater chemistry from 1896 to present in the Mid-Levels area, Hong Kong

Authors

    • Department of Earth SciencesThe University of Hong Kong
  • Jiu Jimmy Jiao
    • Department of Earth SciencesThe University of Hong Kong
Original Article

DOI: 10.1007/s00254-005-0133-9

Cite this article as:
Leung, C. & Jiao, J.J. Environ Geol (2006) 49: 946. doi:10.1007/s00254-005-0133-9

Abstract

In this study, groundwater quality information collected in 1896 (well waters), 1980/1981 (piezometric and seepage samples) and 2002/2003 (seepage samples) in the regions centered by the Mid-Levels area, Hong Kong Island, was compared to illustrate how groundwater quality has changed over a century and the processes controlling it. As shown by saline ammonia and nitrate levels in the late nineteenth century, groundwater was severely polluted by widespread and obvious leakage from poorly designed wastewater collection systems, although groundwater was still a drinking water source for local residents. The extremely high residual chlorines in groundwater demonstrated that large doses of disinfection agents were added to wells at that time. In view of the decline in saline ammonia and nitrate levels, groundwater became less organically polluted in the 1980s probably due to significant improvement of the design of underground sewers. However, more leakage from sources such as salty flushing water and fresh water pipes emerged in the past few decades which added complexity to groundwater chemical systems. Some chemicals were used to identify possible locations of leakages. The temporal variations of the distribution of these chemicals over the area may shed light on the rate of leakage. Leakage from service pipes seems to have improved from the early 1980s to 2002/2003. However, the area is still suffering from widespread and small-scale leakage from service pipes. More efforts should be paid to control small leakages in the future. The findings will be instructive to various government organizations such as the Water Supplies Department and Drainage Services Department to identify possible locations of unobvious leakages in the area.

Keywords

Groundwater qualityPollutionLeakage from service pipesHong Kong

Introduction

Hong Kong developed rapidly from a small fishing village with few thousands of people to a highly urbanized modern city with population over 6.7 millions from 1841 to the present. This development provides a unique opportunity to study the impacts of urbanization on the environment. Many routine monitoring programs on air, noise, surface water, solid waste, and radiation have been conducted in the territory. However, only very few surveys on groundwater quality have been done (Environmental Protection Department 1988–1994; Lam 1983; Leung et al. 2005). Moreover, these surveys were restricted to rural areas. Little is known about urban groundwater quality in Hong Kong.

Groundwater chemical data from 1896, 1980/1981 and 2002/2003 were gathered for two purposes: first, to investigate changes in groundwater chemistry in the urban area over a century and the processes controlling it; and second, to examine changes in rate of leakage from service pipes over the last 20 years.

The Mid-Levels and the surrounding areas were the first region to undergo urban development in Hong Kong. The area eventually combined with other regions on the northern Hong Kong Island to form the City of Victoria after British Colonization in 1841. In this study, the name “City of Victoria” is used to represent the area extending from Sai Ying Pun in the west to Sheung Wan in the east (Fig. 1). At the time of the 1896 survey, the City of Victoria was divided into ten Health Districts. Health Districts Number 3–10 in the City of Victoria were similar to the Mid-Levels, Sai Ying Pun and Sheung Wan areas at present day. The total population of these Health Districts was 136,653 in 1896 (Report on the Census of the Colony 1897). Numerous wells in the City of Victoria were used in 1896 to provide potable water for local communities. The well quality was of great concern and under regular monitoring by the Sanitary Board of the British Government.
https://static-content.springer.com/image/art%3A10.1007%2Fs00254-005-0133-9/MediaObjects/254_2005_133_Fig1_HTML.gif
Fig. 1

Location and geology of the study area. The area with ground surface contours is also the natural slope area with minimum development. The location of the geological cross section A–B shown in Fig. 2 is also presented. This map is based on the Geological Map of Hong Kong (GEO 1999) and the geological information from GCO (1982)

Fresh water has been imported from Mainland China since the 1960s. Now over 70% of Hong Kong’s potable water comes from Guangdong Province, China. The rest of the drinking water is from local reservoirs. Groundwater is not consumed anymore except in small villages in the remote areas of the New Territories, Hong Kong.

Geology and hydrogeology of the study area

The region centered by the Mid-Levels area was chosen to be the study area because several groundwater chemical surveys had been undertaken in this area. The topographic surface varies from about 550 m above the Principal Datum (mPD) at the Victoria Peak to about sea level at the coastline at the Victoria Harbor. The geology and hydrogeology of the area has been described elsewhere in Geotechnical Control Office (GCO) (1982) and will only be briefly introduced here. The geology is dominated by two rock types: acidic volcanic rocks and a granitic intrusion. The volcanic rocks have been subject to low-grade regional metamorphism and deformation and affected by contact metamorphism where close to the granite. Both lithologies have been subsequently intruded by basaltic dykes. The irregular contact between the granite and volcanic rocks crosses the area and is disrupted by normal faults in several locations (Fig. 1).

Colluvium overlies several meters of decomposed rock above the bedrock. The granite underlies most of the developed area, composed of quartz (23–42%), potassium feldspar (31–42%), plagioclases (16–35%) and biotite (~5%) according to Allen and Stephens (1971). Volcanic rock underlies the upper undeveloped slopes.

A long history of engineering activities has resulted in hundreds of boreholes in this area. Using these borehole data, the authors have constructed a three-dimensional geological model (not shown here) in this area (Jiao et al. 2003). Figure 2 shows a geological cross section passing through the site. The study area is covered by colluvium, which was derived mostly from volcanic and granite on the higher slopes. The thickness of colluvium varies considerably from 0 to 5–20 m in mid-slope. The depth of weathering in the granite below the colluvium cover varies from 2 to 3 m in the upper slopes to depths of over 50 m in the lower slopes.
https://static-content.springer.com/image/art%3A10.1007%2Fs00254-005-0133-9/MediaObjects/254_2005_133_Fig2_HTML.gif
Fig. 2

Simplified geological section through A–B (see Fig. 1 for location) (Jiao et al. 2003)

Although it is likely that the lithologies in the subsurface are very heterogeneous and anisotropic, GCO (1982) have grouped them into three aquifer units corresponding to: (a) colluvium, (b) decomposed volcanic and granite rocks, and (c) volcanic and granite bedrocks. The colluvium contains transient and permanent perched water tables, whereas, as recently demonstrated by Jiao et al. (2003, 2004), the highly decomposed rock or saprolite below the colluvium is relatively impermeable due to its clay-rich content. The bedrock zone along the rockhead may be fairly permeable with confined groundwater contained within a well-developed fracture network.

The water level varies from tens of meter below the ground surface below the Victoria Peak to sea level near the coastline. This study concerns mainly the groundwater samples collected from the shallow or perched aquifers in the colluvium or highly to completely decomposed igneous rocks.

Methodology

This study compares groundwater chemical data collected in different periods of time by different institutions. The sources of data are described as follows.

Data from 1896 was obtained from the “Supplement to the Hong Kong Government Gazette, No. 37 of 14th August, 1897, Report of the Secretary, Sanitary Board, for 1896”, the annual report of work done by the Sanitary Board during the year 1896. It is one of the most comprehensive sources of well water quality in the late nineteenth century in the City of Victoria and villages on Kowloon. One task of the Sanitary Board was to analyze well waters in the colony to access their quality for potable use.

Data in 1980/1981 was collected from GCO (1982). In 1980 and 1981, GCO conducted a chemical survey of groundwater and seepages in the Mid-levels area to locate possible leakage points for slope stability study. Only those parameters which served their purposes were measured. Water samples from piezometers and seepage samples were collected and analyzed. Piezometric samples were collected from February to May 1981. Seepage samples were collected from March to August 1980 and also from January to June 1981.

Data in 2002/2003 was measured by the authors. Groundwater samples were collected in September 2002 (wet season) and January 2003 (dry season), mainly from natural and cut slopes. These were measured for physiochemical parameters and analyzed for chemical composition. Sampling and analytical techniques used are described in detail in Leung (2004). In brief, samples for chemical analyses were first filtered at the site by 0.45 μm cellulose papers and collected in 500 ml clean high-density polyethylene (HDPE) bottles. Anion analysis (including F, Cl and NO3) was done by high performance liquid chromatography (HPLC). A four-point calibration curve was employed for all anions and the reference standard was used to verify the results. Saline ammonia (NH4+) levels were measured in the field by HANNA HI93733 ammonia meter. Electrical conductivity was measured in situ by HANNA HI3292 ATC Conductivity Probe.

There is no well depth in the well water survey in 1896. It is believed that the depth is small (a few meters) since at that time the wells were all dug by hand. The water quality from this survey is then believed to represent the groundwater chemistry in the shallow aquifer. The water samples collected in the survey of 2002/2003 are from horizontal drains installed in slopes and natural seeps, so they also represent the water chemistry of the shallow groundwater. Therefore, this paper concerns mainly the groundwater chemistry of the shallow groundwater in the study area.

Results

Change of groundwater chemistry from 1896 to recent years

Groundwater chemical data in the City of Victoria in 1896 was compared with that of from Mid-Levels area collected in recent years to illustrate how groundwater chemistry changed over long periods of time. The concentration of saline ammonia and residual chlorine/chloride of groundwater in 1896 and 1980/1981 were compared. However, since no nitrate was measured in 1980/1981, the nitrate level of seepage samples in 2002/2003 was used for comparison. Figure 3 shows the locations of well samples collected in 1896 and piezometric samples collected in 1981.
https://static-content.springer.com/image/art%3A10.1007%2Fs00254-005-0133-9/MediaObjects/254_2005_133_Fig3_HTML.gif
Fig. 3

Locations of water samples collected in 1896 and 1981

The results from 1896 are expressed in grains per imperial gallon. They were converted to parts per million (ppm) by multiplying 14.286. In 1896, there was no direct measurement on the nitrate levels in well waters. Nitrate level (in grains per imperial gallon) in 1896 was calculated by multiplying the result of “nitrogen in nitrate and nitrite” by a factor of 4.427 for those samples with absence of nitrite.

Saline ammonia (NH4+)

Saline ammonia is a useful indicator for sewage pollution. Table 1 shows the comparison of saline ammonia levels of well water samples from the City of Victoria in 1896 and the samples collected from piezometers in Mid-Levels area in 1981. Saline ammonia level in 1896 ranged from 0 to 20 mg/l with an average of 2.18 mg/l. In 1981, levels ranged from 0 to 7.32 mg/l with an average of 0.78 mg/l. The average saline ammonia level in 1896 was about three times higher than in 1981. This may reflect that the sewage pollution of groundwater in 1896 was much more serious than that of in 1981.
Table 1

Statistics of the saline ammonia levels of well water samples from the City of Victoria in 1896 and the Mid-Levels area in 1981

 

N

Min

Mean

Max

Standard deviation

Coefficient of variation

1896 (whole year)

64

0

2.18

20

4.75

2.18

1981 (February–May)

33

0

0.78

7.32

1.30

1.67

Remarks: Except standard deviation, the results are expressed in mg/l. N denotes “number of samples”

As shown in Table 2, about 55% of wells in 1896 showed indication of sewage contamination with saline ammonia concentration higher than 0.13 mg/l (European Community 1988), while in 1981, about 73% of waters in piezometers were polluted. The high average value in the 1896 may be caused by abnormally high values at certain sites. This assumption is further indicated by the larger coefficient of variation in 1896 (2.18) compared with that in 1981 (1.67). These figures seem to reflect that saline ammonia pollution in 1981 was wide spread over the entire area, but the amount of leakage was relatively smaller than in 1896.
Table 2

Comparison of the percentage of sampling points with concentration of saline ammonia in the range indicated between 1896 and 1981

Saline ammonia concentration (mg/l)

1896

1981

Frequency

%

Frequency

%

Clean

<0.05

20

31.25

5

15.15

0.05 to <0.13

9

14.06

4

12.12

Sewage pollution

0.13 to <0.3

12

18.75

5

15.15

0.3 to <1

4

6.25

11

33.33

1 to <10

14

21.88

8

24.24

10 or above

5

7.81

0

0.00

Residual chlorine (Cl2)/chloride (Cl)

Because well waters were heavily polluted in the City of Victoria in the late nineteenth century, people added chlorine to waters for disinfection before daily consumption (Ho 2001). Chlorine is the most commonly used chemical for disinfection of water. Chlorine can be added to waters in two forms (ATSDR 2002). Liquid chlorine is sodium hypochlorite (NaOCl), which can react with acids or ammonia to release chlorine or chloramines. Chlorine in solid form is calcium hypochlorite (CaCl2O2), which is more commonly used. Calcium hypochlorite decomposes in water to release chlorine and oxygen.

When chlorine is added to water, the chlorine molecules combine with water in a reaction called hydrolysis (Eq. 1). The hypochlorous acid, HOCl (aqueous), is a powerful oxidizing substance. It rapidly oxidizes organic and inorganic matter, including bacteria in the water. The chlorine is converted to chloride and is no longer available as a disinfectant. The organic matter as well as such material as iron and manganese consumes the chlorine.
$$ {\text{H}}_{2} {\text{O}}({\text{l}}) + {\text{Cl}}_{2} ({\text{g}}) \to {\text{HOCl}}({\text{aq}}) + {\text{HCl}} ({\text{aq}}). $$
(1)

Residual chlorine is the amount of unreacted chlorine remaining after chlorine compound is added. The amount of residual chlorine should be much lower than the chloride in heavily polluted water after chlorine disinfection.

Residual chlorine (Cl2) levels of well waters were determined in 1896. However, the residual chlorine was not measured but only the chloride (Cl) level was determined in the Mid-Levels study (GCO 1982). Although a direct comparison between these two parameters cannot be made, some clues may be observed since the concentration of chloride in water is usually higher than the residual chlorine.

Residual chlorine in 1896 ranged from 12.86 to 620 mg/l, with an average of 86.72 mg/l (Table 3). In 1981, the chloride level ranged from 11 to 467 mg/l with the mean of 80.33 mg/l. The residual chlorine level in 1896 was generally higher than the chloride concentration in 1981. As suggested, in view of the high residual chlorine level, it is reasonable to assume that chloride levels could be higher in 1896 than in 1981. People in 1896 added chlorine to well water for disinfection. According to measured residual chlorine level in well water, huge amounts of chlorine were used. However, in 1981, no chlorine disinfection was taking place. The chloride observed in groundwater seems mainly contributed from leaking salty flushing water pipes and sewage pipes (GCO 1982).
Table 3

Statistics of the chlorine levels of well water samples from City of Victoria in 1896 and chloride level from the Mid-Levels area in 1981

 

N

Min

Mean

Max

Standard deviation

Coefficient of variation

Residual chlorine (Cl2) in 1896 (whole year)

64

12.86

86.72

620

103.72

1.20

Chloride (Cl) in 1981 (February–May)

33

11

80.33

467

123.73

1.54

Remarks: Except standard deviation, the results are expressed in mg/l. N denotes “number of samples”

Nitrates (NO3)

As stated, no direct information about the nitrate levels of groundwater in 1981 is available. Information from the seepage sample in 2002/2003 was considered for comparison. Table 4 shows the comparison of nitrates from well water in 1896 and seepage samples in 2002/2003. For the 1896 dataset, 17 wells were determined to have nitrate. Measurements were not done in some heavily polluted sites (as indicated by high albuminoid ammonia and saline ammonia levels). The “actual” average nitrate in 1896 should probably be much higher than the one presented.
Table 4

Comparison of statistics of nitrates in 1896 and 2002/2003

 

N

Min

Mean

Max

Standard deviation

Coefficient of variation

1896 whole year (well water)

17

4.68

41.82

109.28

25.48

0.61

2002 wet season (seepage)

38

0.58

10.42

36.60

11.05

1.06

2003 dry season (seepage)

26

0.90

9.74

31.88

9.84

1.01

Remarks: Except standard deviation, the results are expressed in mg/l. N denotes “number of samples”

The concentration of nitrate in 1896 ranged from 4.68 to 109.28 mg/l, with a mean of 41.82 mg/l. In the 2002 wet season, nitrate ranged from 0.58 to 36.6 mg/l, with a mean of 10.42 mg/l. In the 2003 dry season, nitrate ranged from 0.90 to 31.88 mg/l, with a mean of 9.74 mg/l. Nitrate contamination in 1896 was very severe compared with the conditions in 2002/2003 as shown in Table 5. Nitrate concentration above 25 mg/l is regarded as indication for direct contamination (European Community 1988), and above 50 mg/l is regarded as unsuitable for drinking (World Health Organization 1993). Although well waters in the City of Victoria were consumed by people in 1896, about 35% of them are now considered unsuitable for drinking (Table 5). Also, about 76% showed contamination by sewage (Table 5). All data imply that the groundwater in 1896 was seriously contaminated by human and animal manure and their decomposition products.
Table 5

Comparison of the percentage of sampling points with the concentration of nitrate in the range indicated among 1896, 2002 wet season and 2003 dry season

Nitrate concentration (mg/l)

1896

Wet 2002

Dry 2003

Frequency

%

Frequency

%

Frequency

%

Clean

<5

1

5.88

20

52.63

14

53.85

5 to <10

1

5.88

2

5.26

2

7.69

10 to <25

2

11.76

10

26.32

7

26.92

Contamination

25 to <40

6

35.29

6

15.79

3

11.54

40 to <50

1

5.88

0

0.00

0

0.00

50 or above

6

35.29

0

0.00

0

0.00

On the other hand, all seepage samples collected in 2002 and 2003 were found to be suitable for drinking. About half of the seepage samples show low nitrate concentration (<5 mg/l). Only about 15 and 11% of samples in the 2002 wet season and the 2003 dry season, respectively, show sewage contamination (Table 5). Although groundwaters in 2002/2003 were still contaminated by sewage, the extent was lowered significantly from 1896 to 2002/2003.

Interpretation of data in 1896 and recent years

Saline ammonia, residual chlorine/chloride and nitrate levels of well waters in the City of Victoria in 1896 were compared with data collected in the Mid-Levels area from recent years (1981 and 2002/2003 datasets). The results seem to reflect that groundwater was much more polluted in 1896 than in recent years.

In 1896, contamination might have been caused by significant leakage from wastewater collection system as indicated by saline ammonia and nitrate levels. The wastewater collection system in 1896 was made of stone (Ho 2001). Serious leakage from the gaps between rock units and fractures or cracks of rocks was likely. Chlorine disinfection was common throughout the wells in the City of Victoria as corroborated by the abnormally high residual chlorine levels. Chlorine can effectively oxidize organic matters and pathogens in water. However, potentially carcinogenic compounds, the trihalomethanes (THMs), are generated during this process (Environmental Protection Agency 1999). Moreover, chlorine residual, even at low concentrations, is corrosive and toxic to organisms (Environmental Protection Agency 1999). “Uncontrolled” chlorine disinfection might have generated another type of well water contamination in the City of Victoria during the late nineteenth century.

In recent years, the nature of pollution has changed and sources seem more complex. Needless to say, the study area has changed significantly during the century of time from the late 1890s to recent years. Many high-rise buildings and roads were constructed. Numerous underground pipes and wires were installed. One major source of contamination could be smaller but widespread leakage from salty flushing water pipes and sewage pipes in urban areas. Compared with the situation in 1896, water distribution systems have been improved greatly. Service pipes are usually made by either metals (such as stainless steel, cast iron with internal cement lining) or plastic materials [such as polyvinyl chloride (PVC)] (Water Supplies Department 2001). The extent of leakage has obviously improved; however, the systems have become more widely distributed, spreading the “contaminates” to larger areas.

Comparison of leakage from services in the study area in 1980/1981 and 2002/2003

As mentioned, GCO conducted a simple groundwater chemical survey in 1980/1981 aiming at identifying possible locations and extent of leakage from service pipe (including fresh water pipes, flushing water pipes and sewage pipes). There were substantial amounts of leakage in the Mid-Levels area which in turn may affect slope stability. The corresponding government departments, such as the Water Supplies Department (responsible for the drinking water and salt water pipes) and Drainage Department (responsible for the sewage system), were asked to improve leakage problems. Attempts were made to assess whether the leakage problems have been improved or not over the last 20 years. A combination of parameters, including saline ammonia, chloride, electrical conductivity, fluoride and locality of sampling sites were considered.

Leakage from sewage pipes

Leakage from sewage pipes can be identified by saline ammonia because, except from sewage pipes, no other major source for saline ammonia exists in the Mid-Levels area. The results in the 1980/1981 dataset are expressed in ammonia nitrogen (NH3–N). To convert to saline ammonia (NH4+), ammonia nitrogen results were divided by 0.778.

Table 6 shows the change in saline ammonia in seepage samples from 1980/1981 to 2003. The data show exceptional variability. The concentration of saline ammonia in 1980/1981 ranged from 0 to 45.33 mg/l, with a mean of 3.25 mg/l. In the 2003 dry season, levels ranged from 0 to 2.3 mg/l, with a mean of 0.33 mg/l. The average and maximum concentrations of saline ammonia in 1980/1981 are about 10 times and 20 times, respectively, higher than those in 2003. The overall saline ammonia levels in the Mid-Levels area have reduced significantly in the last 20 years.
Table 6

Comparison of saline ammonia in seepage samples from the Mid-Levels area in (1980/1981) and 2003

 

N

Min

Mean

Max

Standard deviation

Coefficient of variation

1980 (March–August) and 1981 (January–June)

35

0

3.25

45.33

9.20

2.83

2003 dry season (January)

13

0

0.33

2.3

0.62

1.88

Remarks: Except standard deviation, the results are expressed in mg/l. N denotes “number of samples”

Table 7 shows the percentage distribution of saline ammonia in 1980/1981 and 2003. Saline ammonia levels higher than 0.13 mg/l are an indication for sewage contamination (European Community 1988). In 1980/1981, about 83% of samples have saline ammonia concentrations lower than 0.13 mg/l, while in 2003 about 46% were lower than 0.13 mg/l. A higher percentage of samples in 2003 shows sewage contamination. However, the data suggested that the overall degree of contamination in 2003 was lower than in 1980/1981, as there were fewer sites with abnormally high saline ammonia level in 2003. The coefficient of variation in 2003 (1.88) is less than in 1980/1981 (2.83).
Table 7

Comparison of the percentage of sampling points in the urban area with the concentration of ammonium in the range indicated between 1980/1981 and 2003 dry season

Saline ammonia concentration (mg/l)

1980/1981

2003 dry season

Frequency

%

Frequency

%

Clean

<0.05

15

42.86

6

46.15

0.05 to <0.13

14

40.00

0

0.00

Sewage pollution

0.13 to <0.3

0

0.00

3

23.08

0.3 to <1

0

0.00

3

23.08

1 to <10

2

5.71

1

7.69

10 or above

4

11.43

0

0.00

These results may indicate that the extent of major leakage from sewage pipes in the Mid-Levels area has been lowered in the last 20 years. Nevertheless, the problem of “small leakage” is becoming more serious and widespread. According to information provided by Drainage Service Department of the Hong Kong SAR Government, old sewage pipes along the main roads in the Mid-Levels area were replaced from 1980 to 2002. Replacement work may improve major leakage from the sewage water system as demonstrated in Fig 4. Sites with abnormally high saline ammonia levels in 1980/1981 have either disappeared or have lower saline ammonia levels in 2003. However, small-scale leakage from joints or micro-cracks may increase due to the continuous aging of the older parts of the system.
https://static-content.springer.com/image/art%3A10.1007%2Fs00254-005-0133-9/MediaObjects/254_2005_133_Fig4_HTML.gif
Fig. 4

Comparison of saline ammonia of seepage samples in 1980/1981 and 2003 in the Mid-Levels area. The solid black line represents the sections of sewage pipes that had been replaced from 1980 to 2002

Leakage from flushing water pipes

Seawater has been used for flushing in the Western Hong Kong Island including the study area since the mid-1960s (Ho 2001). In 2001, about 80% of the population was supplied with seawater for flushing in all of Hong Kong (Water Supplies Department 2002). The distinctive chemical signature of seawater, i.e., high salinity, may assist in identification of leakage from flushing water pipes.

In GCO (1982), identification of flushing water pipe leakage was mainly performed by electrical conductivity and sample locations. Chlorine concentration was used sometimes because not all samples were determined to have Cl. According to GCO (1982), it was expected that flushing water would be highly saline, but samples from three domestic premises were only slightly brackish. The electrical conductivity and chloride level of flushing water were 2,200 μS/cm at 20°C and 600 mg/l, respectively. Flushing water was regarded as slightly brackish as compared with the water sample collected from the saltwater tank, with an electrical conductivity and chloride level of 61,000 μS/cm at 20°C and 16,600 mg/l, respectively. GCO (1982) concluded at least two seepage points with electrical conductivities of 1,845 and 1,710 μS/cm at 20°C, respectively, which are close to 2,200 μS/cm at 20°C were believed to be from leakage of flushing water. Many other seepage points in the urban area were also suspected to be contaminated by flushing water to varying degrees because of their high electrical conductivities, as shown in Fig. 5a.
https://static-content.springer.com/image/art%3A10.1007%2Fs00254-005-0133-9/MediaObjects/254_2005_133_Fig5_HTML.gif
Fig. 5

Estimated percentage of flushing water in seepage samples in 1980/1981 (a) and 2002/2003 (b) in the Mid-Levels area. The estimated percentage is calculated by (sampleconductivity/flushconductivity)×100%

A flushing water sample was collected in 2003. Content was similar to seawater with an electrical conductivity of about 70,000 μS/cm at 20°C and a chloride level of about 19,000 mg/l. From the results in 2002/2003, there was an absence of seepage points chemically comparable to seawater. This may imply that no major leakage from flushing water was found in 2002/2003, as shown in Fig. 5b. The highest and second highest chloride concentrations of seepage samples measured in the study area were only about 2,800 and 1,770 mg/l, which are equivalent to about one-seventh and one-eleventh of seawater levels. A number of sites in the urban area had chloride concentrations ranging from about 300 to 750 mg/l. These sites are suspected to be contaminated by flushing water, suggesting that small leakage from flushing water pipes was commonly occurring in urban area in 2002/2003 (Fig. 5b).

Leakage from fresh water pipes

Leakage from fresh water pipes was identified mainly by fluoride concentration. The drinking water in Hong Kong has been fluoridated since 1961 at the level of 1.0 mg/l (Chu et al. 1999). In June 1978, the fluoride concentration in Hong Kong water supplies was reduced from 1.0 to 0.7 mg/l (Evans 1989). The concentration was reduced in several stages to 0.5 mg/l in 1988 (World Health Organization 1994). The concentration of fluoride in drinking water in Hong Kong in the early 1980s (~0.6 mg/l) was higher than in 2002/2003 (~0.5 mg/l). Analysis on a tap water sample in 2003 gave a value of 0.48 mg/l fluoride, which is closed to the normal dose, and the measured conductivity was 166 μS/cm at 20°C.

According to GCO (1982), high fluoride levels (up to ~2.5 mg/l) have been measured in samples in the natural area where no mains could be found nearby. Such high fluoride concentrations may be derived from the granite. F could have originated from F-bearing minerals such as fluorite in the rocks. Apambire et al. (1997) suggested that the main source of groundwater fluoride in granitic rocks is the dissolution of and the anion exchange with micaceous minerals and their clay products. Analyzes on samples taken from the natural slopes where no service pipes were installed in the upper Mid-Levels area gave fluoride levels close to the levels of drinking water. This finding adds some difficulties for the identification of leakage from fresh water pipes. Fluoride content alone may not be sufficient for justification of the presence of fresh water main leakage. In GCO (1982), electrical conductivity and their localities (whether in natural slope or urban area) were also taken into account. At least five seepage points were found to be substantially derived from fresh water mains in 1980/1981 (Figs. 6a, 7a) largely because (1) they were all in urban areas; (2) their fluoride levels were close to 0.6 mg/l and (3) their electrical conductivities were close to 160 μS/cm at 20°C. Major leakages were likely occurring around these places. Some other freshwater seepage sites in the urban areas were also suspected. GCO (1982) concluded that “the predominance of samples throughout the study area with fluoride concentrations close to the dosing level for mains water suggests that a significant amount of recharge is coming from services”.
https://static-content.springer.com/image/art%3A10.1007%2Fs00254-005-0133-9/MediaObjects/254_2005_133_Fig6_HTML.gif
Fig. 6

Distribution of fluoride of seepage samples collected in 1980/1981 (a) and 2002/2003 (b), respectively, in the Mid-Levels area. Samples with fluoride of 0.4–0.6 mg/l are suspected to be largely tap water

https://static-content.springer.com/image/art%3A10.1007%2Fs00254-005-0133-9/MediaObjects/254_2005_133_Fig7_HTML.gif
Fig. 7

Distribution of electrical conductivity (in μS/cm at 20°C) of seepage samples collected in 1980/1981 (a) and 2002/2003 (b), respectively, in the Mid-Levels area. Samples with electrical conductivity of 100–300 μS/cm at 20°C are suspected to be largely tap water

In 2002/2003, using similar criteria (1) fluoride concentration was close to 0.5 mg/l; (2) electrical conductivity was close to 160 μS/cm at 20°C and (3) the proximity to pipes for justification, it seems that there is only one site in the urban area with fluoride concentration and electrical conductivity comparable to that of drinking water (Figs. 6b, 7b). However, at that site, strong water seepage was discovered out coming from a weephole in a retaining wall after a rainstorm in the early January 2003. The amount of flow decreased gradually with time and was completely dried after several weeks. This characteristic seems to suggest that the seepage may not be from leakage of drinking water, although fluoride and electrical conductivity of the sample were close to drinking water. Identification of leakage from drinking water is not as straightforward as other leakages. Leakages from flushing water and sewage pipes have chemical signatures, seawater salinity and high saline ammonia, respectively, which are significantly distinctive from the natural groundwater in the area. However, this is not the case for leakage from drinking water pipes. Leakage of drinking water may have already mixed with natural groundwater or other leakages before discharging from the weepholes of retaining wall. The fluoride level and electrical conductivity of the “resultant” seepage may be changed which in turn makes the justification becomes difficult. Even so, leakage from drinking water pipes is believed to be occurring in the developed area as zero leakage is hardly possible. However, based on the fluoride levels and electrical conductivities of the seepage samples collected, no substantial leakage from drinking water pipes was noticed in 2002/2003. The extent of major or substantial leakage from drinking water pipes has been improving in comparison with the situation in 1980/1981.

Interpretation of 1980/1981 and 2002/2003 data

Saline ammonia, chloride, electrical conductivity, fluoride of seepage samples in 1980/1981 and 2002/2003 were investigated to examine whether there was any improvement in the extent of leakage from service pipes in the Mid-Levels area. The results indicate that there is an overall reduction in major or substantial leakage from services pipes, possibly because many of the underground service pipes have been replaced and repaired over the 20 years after the recommendation by GCO (1982). This work could reduce the amount and risk of major leakage. However, small-scale leakage may be increasing because of the continuous aging of the old system.

Conclusion

Some groundwater chemical results measured from the late nineteenth century to recent years are presented. These data may provide information about changes in the subsurface environment over a long period of time. Different stages of development in the study area may affect the groundwater chemistry. Since this study involves two phases of study, we need to briefly summarize the findings of the first phase here as well.

In 1896, groundwater in the City of Victoria was mainly affected by the disposal of human and animal manure, as well as the addition of a disinfection agent (chlorine). The pollution problem was very serious at that time mainly because of high population density and improper installation of the sewage system in the city. Large amount of sewage was leaked from poorly installed sewerage liner. To “solve” the organic pollution problems, large doses of disinfection agent were added. However, the observed organic pollutants such as saline ammonia and albuminoid ammonia were still at high levels. Moreover, uncontrolled chlorine disinfection resulted in exceptionally high residual chlorine level in well waters. Such huge amount of chlorine may lower the pH of groundwater, although no measured pH value was recorded. Potentially carcinogenic substances may also be produced by extensive chlorine disinfection. These may significantly deteriorate the subsurface environment. In recent years, due to the great improvement of the sewerage system compared with the situation in the late nineteenth century, lesser amount of leakage from sewage pipes was observed as indicated by the saline ammonia and nitrate levels. However, since the sewerage system became more widespread as the urban area expanded, the area contaminated by sewage seems to be increased. Comparing the present with the situation in the late nineteenth century, more recharge sources existed in the study area in previous times. Examples of these sources are leakage from salty flushing water and fresh water pipes, which undoubtedly alter the groundwater regime and increase the overall complexity of the urban hydrological system.

The extent of substantial leakage from service pipes has been improved in the last 20 years in the Mid-Levels area. Major leakage from services virtually disappeared over the entire area. These results were achieved by pressure reduction, repairing aging mains, establishing suitably sized metering areas, repairing reported leaks and actively finding and fixing hidden leaks over the last 20 years. However, small-scale leakage seems to be a common problem in the water distribution systems mainly because of the gradual deterioration of the pipes. Significant initial reductions in leakage can usually be achieved relatively easily, but subsequent reductions will become increasingly difficult and expensive because the remaining, unfixed leaks will tend to be those that are small or hard to detect or access. Therefore, measures adopted on controlling major leakage are effective, but more efforts are needed to control smaller leakages in the entire area. This study may provide some information to aid in finding potential locations of small leakage from different types of pipes, and can therefore be used as one reference for further pipe improvement works.

The groundwater chemical data presented in this paper were collected by different authorizations serving for different purposes and some were even collected at the end of nineteenth century, therefore the data contains several limitations. Firstly, the location, number and period of sampling are not exactly the same among the three sets of data. Secondly, analytical methods employed may be different, especially for the data measured in 1896. Every effort has been made to present the different types of data with clarity.

Acknowledgements

The work described was partially supported by Seed Funding in the Faculty of Science, the University of Hong Kong and by a grant from Research Grants Council of Hong Kong (Project No.: HKU7013/03).

Copyright information

© Springer-Verlag 2005