Abstract

In this study, air temperatures were collected between 1985 and 2016 and compared with water temperatures in four locations in the distribution system of Pasadena Water and Power (PWP), which received surface water imported into Pasadena between 2001 and 2016 from the Metropolitan Water District. The concentrations of chloramine residual and nitrite concentrations were collected between 2001 and 2016 from these five locations. The results indicate that the median nighttime temperature of the period 2009–2016 was 1.6 °C warmer than the period 1985–2000 and 0.5 °C warmer than the period 2001–2008. The median water temperature in the four distribution system samples increased by 0.8–1.4 °C depending on the location over the study period (p < 0.001). The median chloramine concentration fell significantly (p < 0.001) at three distribution system locations, and the nitrite concentrations increased significantly at all four distribution system locations (p < 0.001). As air temperature in the study area increased, water temperatures also increased resulting in the loss of disinfectant residual and the increase in the activity of ammonia-oxidizing bacteria. As this represented an increased risk to public health, PWP took additional steps to increase disinfectant residuals by adding chlorine and flushing stale water. In localities where climate change is most measurable, local water purveyors must adapt to warmer water to ensure stable concentrations of disinfectants.

This article has been made Open Access thanks to the kind support of CAWQ/ACQE (https://www.cawq.ca).

INTRODUCTION

Research on the impact of anthropogenic climate change (ACC) on drinking water has focused entirely upon changes in water temperature, microbiology, and chemistry in source waters (Delpla et al. 2009). There has been no research to date on the impact of ACC on treated drinking water in the distribution system of a water purveyor. In the previous research, it has been shown that ACC has been occurring in the City of Pasadena. Records showed that air temperatures had increased on an average of 2.8 °C during daytime between the periods of 1911–1920 and 2011–2016 and 6.1 °C during nighttime. The daytime temperatures increased the most in January and the least in June, while the nighttime temperatures increased uniformly all year around. This change in air temperatures has been shown to affect stream flow (Kimbrough 2017) and thus water supply (Kimbrough 2018). The general rise in air temperatures can have various impacts on drinking water utilities, including the availability of water resources, the temperature of treated water in mains and stored in tanks, and the quality of water. Changing rainfall patterns, evapotranspiration rates, and customers' demand could influence the change in water temperatures (Melillo et al. 2014). However, it is not unreasonable to imagine that water temperature in all parts of the distribution system would increase as atmospheric temperature increases. This could have a number of important water quality implications. At higher water temperatures, disinfectant residuals decay more rapidly and bacterial growth is enhanced (Ndiongue et al. 2005; Michalak 2016). The purpose of this study is to determine if the increases in atmospheric temperatures are in fact affecting water temperatures and microbiological stability in the distribution system of Pasadena's Water and Power (PWP) Department.

PASADENA WATER AND POWER

The City of Pasadena, incorporated in 1886, has owned and operated a public water system (PWS) since 1914 after purchasing a number of privately held water companies. PWP operates a number of wells and has not used local surface water directly for the last 30 years, although it does divert local stream flow into percolation basins for groundwater recharge (Kimbrough 2017). PWP also receives imported surface water from both the Colorado River Aqueduct (CRA) and the California State Water Project (SWP) after the water is treated by the Metropolitan Water District of Southern California (MWDSC). The SWP is a system of dams, conveyances, and pumping stations spanning 1,000 km (600 miles) stretching almost the entire length of California from Lake Shasta in the north to Lake Silverwood in the south. The CRA takes water from Lake Havasu and moves it 389 km (242 miles) to Lake Matthews and then an additional 44 km (30 miles) to the F. E. Weymouth Treatment Plant (WTP). There the plant, operated by MWDSC, may blend the CRA and SWP water or treat 100% of either and then deliver the effluent to PWP and other agencies.

Imported surface water purchased from WTP must first enter the PWP system through one of the three reservoirs, Sunset, Jones, and Eagle Rock. The Sunset Reservoir has a capacity of 57 million liters (ML – 15 million gallons (MG)). The Sunset Reservoir can either blend WTP water with local well water or provide 100% WTP water. The maximum blending rate is 20% well water and 80% WTP water, but 90–100% WTP water is more typical, especially in the non-summer months. The local wells did not have any chlorine or monochloramines added, so the concentration of monochloramine fluctuated depending on the amount of well water blending.

The Jones Reservoir has a capacity of 189 ML (50 MG) and only held WTP water until late 2015 when some local well water was introduced and blended. Eagle Rock is considerably smaller with a volume of 3.6 ML (0.95 MG) and only uses WTP water. The WTP is 40 km (25 miles) from the Sunset Reservoir with a detention time of 1–2 days depending on the time of the year. The Jones Reservoir is 5 km (3 miles) closer to the WTP than the Sunset Reservoir and the Eagle Rock Reservoir is about 5 km further away. These reservoirs are primarily made of concrete and steel with the water in contact with concrete walls. The distribution system is summarized in Figure 1.

Figure 1

Schematic of the studied distribution system.

Figure 1

Schematic of the studied distribution system.

Water purchased from MWDSC by PWP contains monochloramine, so the Division of Drinking Water (DDW) of the State Water Resources Control Board (SWRCB) has required PWP to routinely sample different parts of its distribution system as part of a nitrification control plan for water temperature, total chloramine residual, and nitrite since 2001. The ‘Nitrification Monitoring and Action Plan’ (NMAP) dictates where PWP needs to test for indicators of nitrification and what PWP must do when those indicators are present. As a result, the PWP database consists of thousands of water temperature data from several locations between 2001 and 2016.

HYPOTHESIS

The hypothesis of this study was that the temperature of the water in PWP's distribution system has been increasing due to increasing atmospheric temperatures during the study period. This has resulted in the gradual decrease of monochloramine residual and an increase in bacterial nitrification where residence times are the longest. Such a relationship should not be unexpected, it has been reported that chloramine decay mechanisms and kinetics are temperature-dependent (Vikesland et al. 2001). The growth and activity of ammonia-oxidizing bacteria (AOB), which can consume monochloramine and release nitrite, are known to be temperature-dependent (Pintar & Slawson 2003).

Proposed mechanism

The aforementioned three reservoirs, where WTP water enters PWP's distribution system, are made primarily of concrete and steel, which can absorb heat by direct absorption of sunlight and through the conduction of heat from the air to the outside surfaces as well as the air above the water in the reservoir which could affect the upper layer of water in the reservoir. This, in turn, can heat the water in the reservoirs before it leaves (Grayman et al. 2004). Pipes in the distribution system can likewise warm the water as the surrounding ground is warmed in the same fashion as that of the reservoirs. As the temperature of the air increases over the course of a year and over many years, the temperature of the water in reservoirs, pipes, and water would be warm as well. Moreover, the further the water has to move from the point of entry to the reservoirs, the more the increase in temperature.

Nitrification

The decay of monochloramines releases ammonia, which can be consumed by AOB, which releases nitrite (NO2) and utilization-associated products (UAPs). AOBs can also co-metabolize monochloramine, which also produce NO2 and UAP. UAP can react with monochloramine outside the cell wall, which reduces the concentration of monochloramine. This process of nitrification can result in the complete loss of chloramine residual, which can allow pathogenic and non-pathogenic bacteria to grow. Maintaining a disinfectant residual in a distribution system is an important barrier to the exposure of the public to waterborne pathogens and the loss of monochloramine can pose a significant threat to public health. It is very common for PWSs to flush water from their system when the residual is too low. The addition of chlorine to reservoirs where nitrification occurs is also widely practiced. Thus, water purveyors must work hard to prevent nitrification.

Expected results

If ACC is in fact warming both the local air temperature and the water temperature in the distribution system of PWP, two parallel trends should be observed:

  • (1)

    As temperatures in the water distribution system increase over time, the concentration of monochloramine should decrease and nitrite concentrations should increase.

  • (2)

    The above pattern should be more visible in the parts of the distribution furthest from the entry points into the distribution and less visible in the nearer points.

There is an important caveat to this hypothesis and expected results. During the study period, PWP staff were actively and vigorously trying to keep monochloramine concentrations high and nitrite concentrations low. According to the NMAP, PWP must flush water from locations when nitrite concentrations exceed 25 μg/L and/or add chlorine. Additionally, for compliance with the Total Coliform Rule and Surface Water Treatment Rule, a positive chloramine residual is required. As a result, many parts of the distribution system may be flushed when chloramine residuals are low and/or chlorine is added. These operational requirements and regulatory mandates influence the nature of the results seen in this study.

STUDY LOCATIONS

To test the above hypothesis, five sample locations were selected: four sample locations in PWP's distribution system and one from MWDSC's transmission system. All four of PWP's locations were routinely tested for water temperature and were fed from one of the three reservoirs mentioned above during the study period. Two of the locations are close to the reservoir influent and will be referred to as the proximal locations, and two locations were further away from the reservoir influents. These will be referred to as the distal locations. The sample locations are described as follows:

  • (1)

    Arroyo Terrace (272 m (897 ft) above mean sea level (AMSL)), which is fed from the Sunset Reservoir (383 m (1,264 ft) AMSL), which has a 4:1 blend of WTP water and local well water (which has no chlorine added), is located in the western edge of the Sheldon Zone. The sample location is 2.3 km (1.4 miles) from the outlet of the Sunset Reservoir through several different mains ranging from 20 to 30 cm (8–12 inches). Most of these mains are cast iron installed in or around 1930, although some of the older segments of the main are ductile steel. The final segment of the main where the sample point is located is 340 m (1,200 ft) of 15 cm (6 inch) ductile iron with only seven service connections, all 5 cm (2 inch) in diameter or smaller. The Arroyo Terrace main is a low demand area on a dead-end loop. This is a proximal sample location which means shorter detention times.

  • (2)

    Avenue 64 is in the Eagle Rock pressure zone, which is fed exclusively from the Eagle Rock Reservoir and uses only WTP water. The Eagle Rock Reservoir is 346 m (1,141 ft) AMSL and gravity feeds the entire Eagle Rock pressure zone. The distance from the outlet of the reservoir to the sample point at Avenue 64 (253 m (835 ft) AMSL) is 1.85 km (1.2 miles) through a 30 cm (12 inch) cast iron main that was installed in 1965. The sample location on Avenue 64 is at a point with significant flow, and it is not a dead-end. The final segment of the main where the sample point is located is 143 m (470 ft) of 15 cm (6 inch) cast iron with 10 service connections, all 2.5 cm (1 inch) in diameter or larger. This is a proximal sample location.

  • (3)

    Hill Avenue has an elevation of 227 m (749 ft) AMSL, which is also fed from the Sunset Reservoir. The sample location is 5.8 km (3.6 miles) from the Sunset Reservoir and the last few kilometers are made of 15 cm (6 inch) cast iron that was installed in 1917. This is a high demand area and is not a dead-end. The final segment of the main where the sample point is located is 162 m (532 ft) of 15 cm (6 inch) cast iron with 11 service connections, all 2.5 cm (1 inch) in diameter, but it is also only 178 m from a large 20 cm (8 inch) commercial service connection. This is a distal sample location.

  • (4)

    Tropical Avenue, like Avenue 64, received 100% WTP water, but it comes out of the Jones Reservoir (280 m (924 ft) AMSL)), which was then pumped up to a second reservoir, the Thomas Reservoir (5.3 ML (1.4 MG)), at a higher elevation (367 m (1,211 ft) AMSL) and then gravity fed into the Don Benito Reduced Pressure Zone. Tropical Avenue (326 m (1,076 ft) AMSL) is a moderate demand area and has a 10 cm (4 inch) dead-end cast iron main that was installed in 1951. The final segment of the main where the sample point is located is 176 m (576 ft) of 10 cm (4 inch) cast iron with 20 service connections, all 1.9 cm (0.75 inch) in diameter. The linear distance from the outlet of the Jones Reservoir to the Tropical Avenue sample location is only 2.1 km (1.3 miles). However, the amount of time that the water must travel from the influent to the Jones Reservoir to the Tropical Avenue sample location is significantly greater than the amount of time that the water must travel from the influent to either the Eagle Rock Reservoir or Sunset Reservoir to any of the other sample locations (see Figure 1). This is true because the Jones and Thomas Reservoirs have a combined volume of 190 mL (55 MG). This is a distal sample location. Figure 1 presents a schematic diagram of the WTP, the four reservoirs, and the four sample locations.

  • (5)

    Additionally, water temperatures, pH, and chloramine residual of the plant effluent of MWDSC's WTP (330 m (1,089 ft) AMSL) were also obtained for the study period. For nitrite, results were obtained from a service location nearest Pasadena designated FM-1 that delivers water from the WTP.

In summary, there are two pairs of sample locations, one pair that received 100% WTP water and one pair that received water from the Sunset Reservoir, which ranged from 100% WTP to an 80% blend of WTP and local groundwater. Each pair has one proximal sample location and one distal sample location.

ANALYTICAL METHODS FOR WATER

  • (1)

    Water temperature – The water temperature was measured using an electronic thermometer using Standard Methods 2550 B (APHA 2005).

  • (2)

    Monochloramine – The concentration of total chlorine was determined by using a Hach field colorimeter using Method 4500-CI G N,N Diethyl-1,4 Phenylenediamine Sulfate (DPD) Colorimetric Method (APHA 2005).

  • (3)

    Nitrite (NO2) – The concentration of nitrite was determined by using a Hach field colorimeter using Standard Method 4500-NO2 B Diazotization Method Colorimetric Method (APHA 2005). This test was not performed in the field but in PWP's laboratory. A Hach 850 was used in the beginning of the study and a Hach 890 was used in the latter part.

  • (4)

    Water pH – The pH of the WTP water was determined using Standard Methods 4500-H+ (APHA 2005).

AIR TEMPERATURES

Air temperatures for the 1985–2016 study period were obtained from the National Oceanographic and Atmospheric Administration's National Climatic Data Center (NCDC). A database of the daily maximum air temperatures (all maximum temperatures occurred during the daylight hours temperature are referred to as ‘daytime temperatures’ here, so as to avoid confusion) and minimum air temperatures (referred to as ‘nighttime temperature’) were created and checked for accuracy against written records. For this study, only the nighttime air temperatures were used. Nighttime air temperatures were used because they are a more sensitive measure of climatic change than daytime temperatures. The air temperature was collected at Pasadena's City Hall located at the longitude and latitude +34.15, −118.14.

STATISTICAL PROCEDURES

  • (1)

    The distribution of each data set was assessed using the Shapiro–Wilk Test, and skewness and kurtosis were assessed. Data were considered non-normally distributed if the probability was less than 5% (p ≤ 0.05). All data in this study were non-normally distributed (De Muth 2014) for either skewness of kurtosis.

  • (2)

    There were 16 data sets, nighttime air temperature, the water temperature, total chlorine concentration, and nitrite concentration at the five locations covering the period of 2001–2016. Each database was divided into half, with approximately equal numbers of results covering the periods of 2001–2008 and 2009–2016. Each pair of sub-populations was compared using the Mann–Whitney Rank Sum Test (MWRST), the non-parametric equivalent to the Student's t-test for non-normally distributed data. Differences with a 5% or less level of significance (α ≤ 0.05) were considered significant. The two sub-groups were 2001–2008 and 2009–2016. For the air temperatures, a wider study period was used, 1985–2016, and there were three study periods, 1985–2000, 2001–2008, and 2009–2016. The test period of 2009–2016 was compared with both the 1985–2000 period and the 2001–2008 period by MWRST (de Muth 2014).

  • (3)

    The three air temperature populations were also compared with each other using the Kruskal–Wallis (KW) one-way analysis of variance on ranks. The KW test produces the Kruskal–Wallis Statistic (H). The threshold for significance was 5% (α = 0.05) (de Muth 2014).

  • (4)

    When different data sets collected over time were compared to determine whether they tended to follow correlated patterns, the Spearman rank-order correlation (SROC) test was used, which is the non-parametric equivalent of the Pearson Product-Moment Correlation. For the water data, the temperature, chloramine residual, and nitrite concentrations were compared (de Muth 2014).

  • (5)

    Nitrite results were not censored for this study but used as generated by the instrument. When the instrument generated a value of zero, a value of zero was used for statistical analysis.

RESULTS

Distribution of data

The distribution of all 16 data sets was tested for normality using the Shapiro–Wilk Test and all had a non-normal distribution (p < 0.001).

Air temperatures

The mean, standard deviation, 25th, 50th, and 75th percentile results for the entire study population (1985–2016) and each of the three sub-populations are shown in Table 1. The data on a yearly mean basis including the 99% confidence intervals are shown in Figure 2. The median air temperatures increased through the study period, both between the three sub-groups as seen in Table 1 and even on a year-to-year basis as seen in Figure 2. The median nighttime air temperature of the 2009–2016 period was 1.6 °C higher than the 1985–2000 period and 0.5 °C higher than the 2001–2008 period. The mean air temperature before 2003 was never higher than 12.5 °C but after 2003, it was never lower than 12.5 °C (the same was true for the median results except for 1992). In fact, 2016 was the coolest year since 2003, but it was still warmer than all years preceding 2003. These differences in the median air temperature were of a statistically significant nature with the MWRST having a probability of <0.05.

Table 1

Nighttime air temperatures in Pasadena between 1985 and 2016 (all results in °C)

PeriodnMeanSD25th50th75thSkewnessKurtosisp
1985–2016 11,531 12.3 4.5 8.9 12.2 15.6 0.008 −0.34 <0.001 
1985–2000 5,832 11.5 4.4 8.3 11.7 14.4 −0.04 −0.42 <0.001 
2001–2008 2,810 13.0 4.5 10.0 12.8 16.1 0.007 −0.48 <0.001 
2009–2016 2,889 13.4 4.5 10.6 13.3 16.7 0.04 −0.28 <0.001 
PeriodnMeanSD25th50th75thSkewnessKurtosisp
1985–2016 11,531 12.3 4.5 8.9 12.2 15.6 0.008 −0.34 <0.001 
1985–2000 5,832 11.5 4.4 8.3 11.7 14.4 −0.04 −0.42 <0.001 
2001–2008 2,810 13.0 4.5 10.0 12.8 16.1 0.007 −0.48 <0.001 
2009–2016 2,889 13.4 4.5 10.6 13.3 16.7 0.04 −0.28 <0.001 

Kruskal–Wallis one-way analysis of variance on ranks H = 389, p < 0.001.

Figure 2

Mean and 99% confidence intervals of nighttime air temperature in Pasadena California arranged by year 1985–2016.

Figure 2

Mean and 99% confidence intervals of nighttime air temperature in Pasadena California arranged by year 1985–2016.

Land use

Urbanized areas, like Pasadena, show changes in temperature; the landscape is changed. Constructed landscapes hold less water and capture more heat than open land and vegetation. Urbanized areas thus appear as an ‘island’ of higher temperatures compared with surrounding land or over time compared with previous land use. This can skew the analysis changes of air temperature over time. However, the areas of Pasadena included here were fully urbanized throughout the entire study period. None of the observed temperature changes were the result of changes in land use.

Water temperatures

The mean, standard deviation, 25th, 50th, and 75th percentile results for the entire study population (2001–2016) and each of the two sub-populations (2001–2008 and 2009–2016) for all five locations are shown in Table 2. The WTP shows no change in median water temperatures when the two sub-populations are compared by MWRST with the median water temperature actually 0.9 °C lower. However, all four locations in the distribution show statistically significant increases in water temperature (p < 0.001). The median temperatures increased at Avenue 64 by 1.0 °C, Tropical Avenue by 1.4 °C, Arroyo Terrace by 0.9 °C, and Hill Avenue by 0.7 °C. These increases parallel the increase in nighttime air temperature, which was 0.5 °C. To better assess the relationships between air and water temperature, the monthly median water temperatures of each of the five locations were plotted on a monthly basis, which is summarized in Figure 3.

Table 2

Water temperatures in Pasadena at five locations between 2001 and 2016 (all results in °C)

LocationPeriodnMeanSD25th50th75thSKp
Arroyo Terrace 2001–2016 830 19.9 4.2 16.2 20.0 23.7    
Arroyo Terrace 2001–2008 417 19.4 4.1 15.8 19.5 23.3 −0.03 −1.2 <0.001 
Arroyo Terrace 2009–2016 413 20.4 4.2 16.7 20.4 24.2 −0.03 −1.1 <0.001 
Mann–Whitney U Statistic = 74,811, p = 0.001 
 Avenue 64 2001–2016 832 19.9 4.3 16.0 20.0 24.0    
 Avenue 64 2001–2008 414 19.3 4.3 15.2 19.3 23.4 0.04 −1.3 <0.001 
 Avenue 64 2009–2016 418 20.6 4.3 16.9 20.3 24.4 0.04 −1.2 <0.001 
Mann–Whitney U Statistic = 71,496, p < 0.001 
 Hill Avenue 2001–2016 832 21.8 3.3 19.6 22.0 24.3    
 Hill Avenue 2001–2008 418 21.1 3.2 19.0 21.7 23.8 −0.54 −0.4 <0.001 
 Hill Avenue 2009–2016 414 22.4 3.2 20.0 22.5 25.2 0.08 0.2 <0.001 
Mann–Whitney U Statistic = 69,260, p < 0.001 
 Tropical Avenue 2001–2016 832 21.8 3.9 18.8 21.4 24.8    
 Tropical Avenue 2001–2008 419 21.0 3.6 18.1 20.8 24.0 0.01 −0.7 <0.001 
 Tropical Avenue 2009–2016 413 22.6 4.1 19.6 22.2 25.0 0.40 −0.4 <0.001 
Mann–Whitney U Statistic = 67,706, p < 0.001 
 WTP 2001–2016 2,548 20.1 4.8 15.8 20.1 24.6    
 WTP 2001–2008 1,206 19.9 4.9 15.4 20.6 24.4 −0.19 −1.25 <0.001 
 WTP 2009–2016 1,342 19.4 5.9 15.3 19.5 25.0 −0.76 0.67 <0.001 
Mann–Whitney U Statistic = 782,378, p = 0.15. 
LocationPeriodnMeanSD25th50th75thSKp
Arroyo Terrace 2001–2016 830 19.9 4.2 16.2 20.0 23.7    
Arroyo Terrace 2001–2008 417 19.4 4.1 15.8 19.5 23.3 −0.03 −1.2 <0.001 
Arroyo Terrace 2009–2016 413 20.4 4.2 16.7 20.4 24.2 −0.03 −1.1 <0.001 
Mann–Whitney U Statistic = 74,811, p = 0.001 
 Avenue 64 2001–2016 832 19.9 4.3 16.0 20.0 24.0    
 Avenue 64 2001–2008 414 19.3 4.3 15.2 19.3 23.4 0.04 −1.3 <0.001 
 Avenue 64 2009–2016 418 20.6 4.3 16.9 20.3 24.4 0.04 −1.2 <0.001 
Mann–Whitney U Statistic = 71,496, p < 0.001 
 Hill Avenue 2001–2016 832 21.8 3.3 19.6 22.0 24.3    
 Hill Avenue 2001–2008 418 21.1 3.2 19.0 21.7 23.8 −0.54 −0.4 <0.001 
 Hill Avenue 2009–2016 414 22.4 3.2 20.0 22.5 25.2 0.08 0.2 <0.001 
Mann–Whitney U Statistic = 69,260, p < 0.001 
 Tropical Avenue 2001–2016 832 21.8 3.9 18.8 21.4 24.8    
 Tropical Avenue 2001–2008 419 21.0 3.6 18.1 20.8 24.0 0.01 −0.7 <0.001 
 Tropical Avenue 2009–2016 413 22.6 4.1 19.6 22.2 25.0 0.40 −0.4 <0.001 
Mann–Whitney U Statistic = 67,706, p < 0.001 
 WTP 2001–2016 2,548 20.1 4.8 15.8 20.1 24.6    
 WTP 2001–2008 1,206 19.9 4.9 15.4 20.6 24.4 −0.19 −1.25 <0.001 
 WTP 2009–2016 1,342 19.4 5.9 15.3 19.5 25.0 −0.76 0.67 <0.001 
Mann–Whitney U Statistic = 782,378, p = 0.15. 

K, kurtosis; S, skewness; p, probability.

Figure 3

Mean water and air temperatures at six locations with 99% confidence intervals 2001–2016.

Figure 3

Mean water and air temperatures at six locations with 99% confidence intervals 2001–2016.

Chloramine residual

The chloramine residual is shown in Table 3 including the mean, standard deviation, 25th, 50th, and 75th percentile results for the entire study population (2001–2016) and each of the two sub-populations (2001–2008 and 2009–2016) for all five locations. The WTP showed a slight increase in residual when the two sub-populations are compared by MWRST by 0.05 mg/L (p < 0.001). This was due to an operational target concentration of chlorine dosing at the WTP. However, some locations in PWP's distribution show statistically significant decreases in the residual concentration (p < 0.001). The median chloramine residuals decreased at Avenue 64 by 0.03 mg/L, Tropical Avenue by 0.27 mg/L, Arroyo Terrace by 0.51 mg/L, but Hill Avenue showed a slight increase in residual concentration of 0.07 mg/L; however, this was not statistically significant.

Table 3

Total chloramine residuals in Pasadena at five locations between 2001 and 2016 (all results in mg/L)

LocationPeriodnMeanSD25th50th75thSKp
Arroyo Terrace 2001–2016 830 0.82 0.57 0.35 0.73 1.20    
Arroyo Terrace 2001–2008 417 1.06 0.57 0.60 1.00 1.40 0.5 −0.3 <0.001 
Arroyo Terrace 2009–2016 413 0.58 0.46 0.20 0.49 0.85 1.0 0.6 <0.001 
Mann–Whitney U Statistic = 74,811, p = 0.001 
 Avenue 64 2001–2016 832 2.03 0.38 1.90 2.09 2.20    
 Avenue 64 2001–2008 414 2.09 0.40 1.82 2.10 2.30 −0.5 1.2 <0.001 
 Avenue 64 2009–2016 418 1.97 0.35 1.92 2.07 2.19 −2.9 9.9 <0.001 
Mann–Whitney U Statistic = 72,211, p < 0.001 
 Hill Avenue 2001–2016 832 0.86 0.5 0.50 0.74 1.02    
 Hill Avenue 2001–2008 418 0.83 0.48 0.50 0.70 1.00 2.0 5.2 <0.001 
 Hill Avenue 2009–2016 414 0.90 0.53 0.50 0.77 1.20 0.7 −0.4 <0.001 
Mann–Whitney U Statistic = 81,953, p = 0.19 
 Tropical Avenue 2001–2016 832 1.07 0.62 0.52 1.04 1.50    
 Tropical Avenue 2001–2008 419 1.22 0.63 0.70 1.20 1.70 0.2 −0.8 <0.001 
 Tropical Avenue 2009–2016 413 0.92 0.58 0.40 0.93 1.38 0.2 −0.7 <0.001 
Mann–Whitney U Statistic = 63,310, p < 0.001 
 WTP 2001–2016 2,548 2.60 0.09 2.55 2.60 2.65    
 WTP 2001–2008 1,206 2.58 0.09 2.51 2.57 2.62 1.7 8.9 <0.001 
 WTP 2009–2016 1,342 2.63 0.07 2.60 2.62 2.67 4.7 21 <0.001 
Mann–Whitney U Statistic = 474,479, p < 0.001 
LocationPeriodnMeanSD25th50th75thSKp
Arroyo Terrace 2001–2016 830 0.82 0.57 0.35 0.73 1.20    
Arroyo Terrace 2001–2008 417 1.06 0.57 0.60 1.00 1.40 0.5 −0.3 <0.001 
Arroyo Terrace 2009–2016 413 0.58 0.46 0.20 0.49 0.85 1.0 0.6 <0.001 
Mann–Whitney U Statistic = 74,811, p = 0.001 
 Avenue 64 2001–2016 832 2.03 0.38 1.90 2.09 2.20    
 Avenue 64 2001–2008 414 2.09 0.40 1.82 2.10 2.30 −0.5 1.2 <0.001 
 Avenue 64 2009–2016 418 1.97 0.35 1.92 2.07 2.19 −2.9 9.9 <0.001 
Mann–Whitney U Statistic = 72,211, p < 0.001 
 Hill Avenue 2001–2016 832 0.86 0.5 0.50 0.74 1.02    
 Hill Avenue 2001–2008 418 0.83 0.48 0.50 0.70 1.00 2.0 5.2 <0.001 
 Hill Avenue 2009–2016 414 0.90 0.53 0.50 0.77 1.20 0.7 −0.4 <0.001 
Mann–Whitney U Statistic = 81,953, p = 0.19 
 Tropical Avenue 2001–2016 832 1.07 0.62 0.52 1.04 1.50    
 Tropical Avenue 2001–2008 419 1.22 0.63 0.70 1.20 1.70 0.2 −0.8 <0.001 
 Tropical Avenue 2009–2016 413 0.92 0.58 0.40 0.93 1.38 0.2 −0.7 <0.001 
Mann–Whitney U Statistic = 63,310, p < 0.001 
 WTP 2001–2016 2,548 2.60 0.09 2.55 2.60 2.65    
 WTP 2001–2008 1,206 2.58 0.09 2.51 2.57 2.62 1.7 8.9 <0.001 
 WTP 2009–2016 1,342 2.63 0.07 2.60 2.62 2.67 4.7 21 <0.001 
Mann–Whitney U Statistic = 474,479, p < 0.001 

K, kurtosis; S, skewness; p, probability.

Nitrite concentration

The nitrite measured in μg/L is shown in Table 4 including the mean, standard deviation, 25th, 50th, and 75th percentile results for the entire study population (2001–2016) and each of the two sub-populations (2001–2008 and 2009–2016) for all the five locations. Three of the distribution system samples showed significant increases in the median concentration of nitrite (p < 0.001). At Avenue 64, the median concentration of nitrite doubled from 5.0 to 10.0 μg/L and the mean concentrations showed a parallel increase. At the Tropical Avenue location, the median concentration increased from 0 to 3.0 μg/L, while the mean concentration more than doubled from 2.3 to 5.9 μg/L. At the Hill Avenue location, the median concentration increased from 0 to 3.0 μg/L, while the mean concentration almost tripled from 1.6 to 4.3 μg/L. The Arroyo Terrace data are more complicated. The median concentration of nitrite decreased slightly from 14.5 to 14.0 μg/L, while the 25th percentile remained constant. However, the mean increased from 22.1 to 28.4 μg/L and the 75th percentile increased from 29.0 to 42.0 μg/L. So, the concentration increased overall but the p-value was only 0.028, which is statistically significant but much less so as compared with the other locations. Arroyo Terrace also had a considerably higher median and mean concentration of nitrite as compared with the other sites. No nitrite was ever detected at the WTP effluent, but at the FM-1 distribution location there were 798 samples collected although only 21 had nitrite at a concentration of greater than 5 μg/L and only seven of those were 6 μg/L or greater and the highest value was 9 μg/L. With so few quantifiable results, which are all quite low, there is no evidence that the rate of nitrification increased between the two study periods (although 13 of the measurable nitrite results occurred in 2016). A number of the nitrite results were reported at the lower end of the linear dynamic range for this method, which might increase the uncertainty in the data. However, over 700 data points were collected over a 15-year period, which significantly increased the robustness of the statistical analysis.

Table 4

Nitrite concentrations in Pasadena at four locations between 2001 and 2016 (all results in μg/L)

LocationPeriodnMeanSD25th50th75thSKp
Arroyo Terrace 2001–2016 825 25.3 29.3 5.0 14.0 35.0     
Arroyo Terrace 2001–2008 416 22.1 26.8 5.0 14.5 29.0 2.6 10.4 <0.001  
Arroyo Terrace 2009–2016 409 28.4 31.9 5.0 14.0 42.0 1.6 2.0 <0.001  
Mann–Whitney U Statistic = 77,576, p = 0.028 
 Avenue 64 2001–2016 750 9.7 11.7 3.0 6.0 12.0    
 Avenue 64 2001–2008 414 6.3 8.4 0.0 5.0 8.0 4.4 31.4 <0.001 
 Avenue 64 2009–2016 336 13.9 13.7 5.0 10.0 19.0 3.4 21.7 <0.001 
Mann–Whitney U Statistic = 38,075, p < 0.001 
 Hill Avenue 2001–2016 705 2.7 3.3 0.0 2.0 5.0    
 Hill Avenue 2001–2008 418 1.6 2.4 0.0 0.0 3.0 1.3 0.33 <0.001 
 Hill Avenue 2009–2016 287 4.3 3.7 2.0 3.0 5.0 2.1 6.6 <0.001 
Mann–Whitney U Statistic = 28,484, p < 0.001 
 Tropical Avenue 2001–2016 705 3.7 10.7 0.0 1.0 4.0    
 Tropical Avenue 2001–2008 419 2.3 5.2 0.0 0.0 3.0 4.3 23.4 <0.001 
 Tropical Avenue 2009–2016 286 5.9 15.3 1.0 3.0 5.3 8.8 90.1 <0.001 
Mann–Whitney U Statistic = 31,410, p < 0.001 
LocationPeriodnMeanSD25th50th75thSKp
Arroyo Terrace 2001–2016 825 25.3 29.3 5.0 14.0 35.0     
Arroyo Terrace 2001–2008 416 22.1 26.8 5.0 14.5 29.0 2.6 10.4 <0.001  
Arroyo Terrace 2009–2016 409 28.4 31.9 5.0 14.0 42.0 1.6 2.0 <0.001  
Mann–Whitney U Statistic = 77,576, p = 0.028 
 Avenue 64 2001–2016 750 9.7 11.7 3.0 6.0 12.0    
 Avenue 64 2001–2008 414 6.3 8.4 0.0 5.0 8.0 4.4 31.4 <0.001 
 Avenue 64 2009–2016 336 13.9 13.7 5.0 10.0 19.0 3.4 21.7 <0.001 
Mann–Whitney U Statistic = 38,075, p < 0.001 
 Hill Avenue 2001–2016 705 2.7 3.3 0.0 2.0 5.0    
 Hill Avenue 2001–2008 418 1.6 2.4 0.0 0.0 3.0 1.3 0.33 <0.001 
 Hill Avenue 2009–2016 287 4.3 3.7 2.0 3.0 5.0 2.1 6.6 <0.001 
Mann–Whitney U Statistic = 28,484, p < 0.001 
 Tropical Avenue 2001–2016 705 3.7 10.7 0.0 1.0 4.0    
 Tropical Avenue 2001–2008 419 2.3 5.2 0.0 0.0 3.0 4.3 23.4 <0.001 
 Tropical Avenue 2009–2016 286 5.9 15.3 1.0 3.0 5.3 8.8 90.1 <0.001 
Mann–Whitney U Statistic = 31,410, p < 0.001 

K, kurtosis; S, skewness; p, probability.

pH

The pH of the WTP water during this study ranged from 8.0 to 8.1. The pH was controlled through the addition of sodium hydroxide.

Correlation

If the hypothesis is correct, as air temperatures increase, water temperatures should also increase which could cause the concentration of chloramine residuals to decrease and nitrite concentrations to increase. In other words, there ought to be a negative or inverse correlation between water temperature and chlorine concentration and a positive correlation with nitrite. Additionally, there ought to be an inverse relationship between chlorine concentration and nitrite concentration. To test this, the data from the five water locations were analyzed using the SROC. The correlation coefficient and probability for each paired set of data are shown in Table 5. No correlation was observed in the data from the WTP, which is not surprising since the concentrations of chloramine residual are high and there is little opportunity for the chlorine to decay. Given the fact that there was very little evidence of nitrification, no attempts were made to correlate nitrite concentrations and either water temperature or chloramine residual. At Avenue 64, Tropical Avenue, and Arroyo Terrace, there was similarly no correlation between water temperature and chloramine residual. Hill Avenue, however, showed a weak (R = −0.20) but statistically significant (p < 0.001) negative correlation between water temperature and chloramine residual concentration. In contrast, at Avenue 64, Tropical Avenue, and Arroyo Terrace, there was a correlation between water temperature and nitrite concentration, which ranged from weak to moderate, but all of it was significant. The Hill Avenue sample location showed no pattern with no measurable correlation between water temperature and nitrite concentration. The relationship between chloramine residual concentration and nitrite concentration is more complex. The sample location on Avenue 64 did indeed produce the expected negative correlation, which, while not strong (R = −0.20), was statistically significant (p < 0.001). Tropical Avenue had similar results but with a slightly weaker correlation coefficient. On the other hand, the Arroyo Terrace sample location did not show a statistically significant correlation between chloramine residual and nitrite concentration, while the Hill Avenue location showed a weak (R = 0.12) but positive and statistically significant correlation (p < 0.0016).

Table 5

Spearman's rank-order correlation between water temperature, chlorine concentration, and nitrite concentrations in Pasadena at five locations between 2001 and 2016

LocationParametersSROCp
Arroyo Terrace Cl2 and temperature −0.048 0.17 
Arroyo Terrace NO2 and temperature 0.37 <0.0001 
Arroyo Terrace Cl2 and NO2 −0.015 0.66 
Avenue 64 Cl2 and temperature 0.016 0.65 
Avenue 64 NO2 and temperature 0.18 <0.0001 
Avenue 64 Cl2 and NO2 −0.20 <0.0001 
Hill Avenue Cl2 and temperature −0.22 <0.0001 
Hill Avenue NO2 and temperature 0.052 0.16 
Hill Avenue Cl2 and NO2 0.12 0.0016 
Tropical Avenue Cl2 and temperature 0.052 0.13 
Tropical Avenue NO2 and temperature 0.21 <0.0001 
Tropical Avenue Cl2 and NO2 −0.14 <0.0001 
WTP Cl2 and temperature 0.006 0.83 
LocationParametersSROCp
Arroyo Terrace Cl2 and temperature −0.048 0.17 
Arroyo Terrace NO2 and temperature 0.37 <0.0001 
Arroyo Terrace Cl2 and NO2 −0.015 0.66 
Avenue 64 Cl2 and temperature 0.016 0.65 
Avenue 64 NO2 and temperature 0.18 <0.0001 
Avenue 64 Cl2 and NO2 −0.20 <0.0001 
Hill Avenue Cl2 and temperature −0.22 <0.0001 
Hill Avenue NO2 and temperature 0.052 0.16 
Hill Avenue Cl2 and NO2 0.12 0.0016 
Tropical Avenue Cl2 and temperature 0.052 0.13 
Tropical Avenue NO2 and temperature 0.21 <0.0001 
Tropical Avenue Cl2 and NO2 −0.14 <0.0001 
WTP Cl2 and temperature 0.006 0.83 

DISCUSSION

The air in Pasadena has been warming significantly since 1985 as seen in Figure 2 and Table 1, more so than most parts of the world (Kimbrough 2018). In parallel with the increasing air temperature, water temperatures in PWP's distribution system have increased as well. While the median air temperature in Pasadena increased by 0.5 °C in the period of 2009–2016 as compared with 2001–2008, when comparing the two periods, the median water temperature at the WTP changed by −1.1 °C, although this was not statistically significant. This is also a bit misleading as the 75th percentile of the water temperature was actually higher in the second period than that in the first period by 0.6 °C and the 25th percentile only decreased by 0.1 °C while the mean is 0.5 °C lower. Obviously, the water temperatures are distributed in a complex fashion that is not easily captured in a single measure of the central tendency. Suffice it to say, there is no evidence that the water temperate of the effluent of the WTP has increased between the two study periods.

In contrast, the median water temperature at Arroyo Terrace increased by 0.9 °C, Avenue 64 by 1.0 °C, Hill Avenue by 0.8 °C, and at Tropical Avenue by 1.4 °C. These were all statistically significant increases (p < 0.001). That the water temperature should increase more than the air temperature is not necessarily surprising, as the heat capacity of water is five times higher from that of air and can thus retain more heat much longer than air. It is also not a surprising fact that the water at Tropical Avenue showed a larger median increase in water temperature as compared to the other three sites. This is because the water first enters PWP's distribution system through the Jones Reservoir (by far the largest of PWP's reservoirs) from which it is pumped up to the Thomas Reservoir before reaching the sample tap. As a result, the detention time is considerably longer and much of that detention time is in above grade reservoirs as opposed to the other three sample locations where the water, once it leaves the smaller reservoirs, travels through subsurface mains so detention times are considerably shorter and there is much less contact with the atmosphere.

The vast majority of the warming of the water occurs in the distribution system itself. The influent water from the WTP showed no increase in water temperature during the study period and no loss of chloramine residual (although 2014 and 2016 were the two warmest years in the study period). This is because of the difference in the surface area to volume ratio. The influent water moves through large conveyances, open channels, and large diameter below grade pipes, where the surface area in contact with the atmosphere is small compared with the volume of water. In PWP's distribution system, the surface area to volume ratio is far more favorable for heat exchange, as are the above grade reservoirs. Moreover, what is clear is that as the water moves further from the reservoirs where the water is taken from WTP, the water temperature changes. In winter, the WTP water that comes is cold and is warmed as it passes through the distribution system. In summer, the exact opposite is observed at most locations; the WTP enters the system and is slightly cooled, except for Tropical Avenue where it warms very slightly. The two distal locations showed this pattern more than the two proximal locations as can be seen in Figure 3.

In examining the monthly data a pattern emerges. The month of December had the lowest mean air temperature, 8.3 °C, while August had the highest, 18.3 °C, a range of 10 °C. The mean temperature of the effluent of the WTP ranged from a low in January from 13.1 °C to 26.2 °C, a 13.0 °C difference. Arroyo Terrace showed a range of monthly averages of 10.9 °C, 14.4 °C in January, and 25.3 °C in August. Avenue 64 presented a range of monthly averages of 11.3 °C, 14.4 °C in January, and 25.7 °C in August. Hill Avenue had a range of monthly averages of only 7.7 °C, 17.8 °C in February, and 25.5 °C in August. Tropical Avenue had a range of monthly averages of only 8.8 °C, 18.6 °C in December, and 27.4 °C in August. The two distal sample locations showed a range of mean temperatures that were very similar to the changes in the local air temperature. The WTP and the two proximal sample locations showed a wider range of mean monthly temperatures. The two proximal locations were more influenced by changes in water temperature from the WTP, while the two distal locations were more influenced by changes in local air temperature. There are only small differences between the maximum mean water temperatures and the differences in the range of monthly averages are almost entirely due to differences in the minimum monthly temperatures.

Changes in the concentration of chloramine residuals were largely parallel to that seen in water temperatures. There was no decrease in the median concentration of chlorine in the effluent of WTP during the study period. Avenue 64 showed a small decrease, while Arroyo Terrace and Hill Avenue both showed larger decreases. Arroyo Terrace and Hill Avenue have longer detention times than Avenue 64, so this is not unexpected. In addition, the Arroyo Terrace sample tap is on a dead-end loop and has tuberculation, so this site could be expected to show the greatest median chlorine loss. The Tropical Avenue sample location, however, showed the least chlorine loss, which would seem counterintuitive, given the fact that that location showed the greatest water temperature increase. However, at both Jones and Thomas Reservoirs, chlorine gas is routinely fed during the warmer months of the year. At no point was enough chlorine gas added to break over the water to free chlorine and the pH is too high to form dichloramine or trichloramine. As seen in Figure 4, the median chlorine concentrations at each of the five locations are arranged by month. The chlorine concentration increases significantly at Tropical Avenue in the summer months because of the addition of chlorine gas in the reservoirs, while the pattern does not hold at any other location.

Figure 4

Median, 25th and 75th percentiles chloramine residual from WTP and in the distribution system of PWP at four locations at arranged by month 2001–2016.

Figure 4

Median, 25th and 75th percentiles chloramine residual from WTP and in the distribution system of PWP at four locations at arranged by month 2001–2016.

The nitrite results are also informative. The FM-1 water had no significant increase in nitrite concentration, while all four PWP sample locations showed significant increases in the median concentration of nitrite (p < 0.001). Both Hill Avenue and Tropical Avenue had almost no nitrification prior to 2009 but afterwards showed a very noticeable and statistically significant increase. Both showed an increase of ∼3 μg/L in both the mean and median concentrations. This is not a large increase in absolute value, but it does represent a dramatic shift in the water quality of these two locations. What this means is that nitrification had not been occurring before but has occurred later. Avenue 64 likewise showed a definite increase in median nitrite concentration. The mean and median concentrations both double in value, from 5.0 to 10.0 μg/L and 6.3 to 13.9 μg/L, respectively. Arroyo Terrace actually showed a slight decrease in the median concentration of nitrite from 14.5 mg/L to 14.0 μg/L, but the mean increased by 6.3 μg/L and there was a 13 μg/L increase in the 75th percentile concentration. The absolute difference in the means and medians was generally 3–6 μg/L.

The hypothesis is that as water temperatures rise, the concentration of chlorine should fall and if this is the case, there ought to be a negative correlation between these two variables.

However, no such correlation is observed for four of the locations. At Hill Avenue there was a correlation, which while significant (p < 0.001) was not strong (R = −0.22). For WTP, this is not surprising since there has been no increase in water temperature or loss of chloramine residual. Similarly, Tropical Avenue is fed by two reservoirs in tandem where chlorine is added so the lack of correlation is not surprising. Avenue 64 showed only a minor loss of chlorine like the WTP, and a lack of correlation might be expected. Nonetheless, Arroyo Terrace showed considerable chlorine so a lack of correlations is unexpected.

On the other hand, Arroyo Terrace, Avenue 64, and Tropical Avenue all showed significant positive correlations between water temperature and nitrite concentration (p < 0.0001). Generally, as water temperatures rose, nitrite concentrations rose at these three locations. Why Hill Avenue should show no correlation between nitrite concentrations and water temperature but show the expected negative correlation between chloramine residual and water temperature is not clear.

Finally, both Avenue 64 and Tropical Avenue showed the expected negative correlation between chloramine residual concentration and nitrite concentration, while Arroyo Terrace showed no correlation and Hill Avenue actually showed a weak (R = 0.12) but significant (p = 0.0016) positive correlation, the opposite of what would be expected. Of 13 possible correlations among the variables in the water samples, seven were statistically significant and six were of the expected direction. Therefore, the data generally support the hypothesis, increasing air and water temperatures are affecting the quality of water in the form of decreasing monochloramine concentrations and increasing nitrite concentrations. The pattern is not entirely consistent with the hypothesis but as was noted above, operational practices may have influenced the association between water temperature and distribution water quality. The data were gathered from two working distribution systems (WTP and PWP) where staff were working diligently and vigorously to prevent the loss of monochloramine residual and the increase in nitrite concentrations. Without the addition of chlorine gas at Jones and Thomas Reservoirs and a very active program of flushing numerous parts of the distribution system, a much greater impact and clearer correlations would no doubt have been observed. Flushing during the study period occurred at all times of the year but especially in the warmer months. The effect of flushing and chlorination is to destroy any correlation between water temperature and monochloramine and nitrite concentrations. As noted above, chlorine addition was largely practiced in summer.

It is important to note that during part of the study period, California suffered a period of intense drought (Kimbrough 2017, 2018). This resulted in unprecedented reductions in water demand and increases in water age in the distribution system. This may well have exaggerated the impact of increasing air temperatures on water temperatures.

SUMMARY AND CONCLUSIONS

Local climatic change has resulted in significant and measurable increases in the temperature of the nighttime air in Pasadena, which in turn has increased the water temperature in the distribution system of PWP. This has caused increased rates of chlorine decay and increased rates of nitrification. As noted above, such a finding is entirely consistent with previously published research (Vikesland et al. 2001; Pintar & Slawson 2003). As the temperature of the air in the study area increased, water temperatures increased as well resulting in the loss of disinfectant residual and the activity of AOB increased. As this represented an increased risk to public health, PWP took additional steps to increase disinfectant residuals by adding chlorine and flushing stale water. Additionally, PWP purchased a portable water treatment plant that allowed water to be pumped from the distribution system, filtered, chlorinated, and then returned to the distribution without loss of pressure or exposure to the atmosphere. In localities where climate change is most measurable, local water purveyors will need to adapt to warmer water to ensure stable concentrations of disinfectants. This might mean more labor dedicated to flushing low residual or nitrified water, more chlorination of the distribution system, and overall increased labor and chemical costs. Local climate change will challenge local water purveyors to maintain water quality.

ACKNOWLEDGEMENTS

The author thanks Tae Yun and JR Rhoads of the Metropolitan Water District of Southern California's Water Quality Section.

REFERENCES

REFERENCES
APHA
,
Eaton
A. D.
,
Clesceri
L. S.
&
Greenberg
A. E.
, (eds)
2005
Standard Methods for the Examination of Water and Wastewater
, 21st edn.
American Public Health Association
,
Washington, DC
.
De Muth
J. E.
2014
Basic Statistics and Pharmaceutical Statistical Applications
, 3rd edn.
Chapman and Hall/CRC
,
Boca Raton, FL
, pp.
242
244
.
Delpla
I.
,
Jung
A.-V.
,
Baures
E.
,
Clement
M.
&
Thomas
O.
2009
Impacts of climate change on surface water quality in relation to drinking water production
.
Environment International
35
,
1225
1233
.
Grayman
W. M.
,
Rossman
L. A.
,
Deininger
R. A.
,
Smith
C. D.
,
Arnold
C. N.
&
Smith
J. F.
2004
Mixing and aging of water in distribution system storage facilities
.
Journal of the American Water Works Association
96
(
9
),
70
80
.
Kimbrough
D. E.
2017
Local climate change in Pasadena, Calif., and the impact on streamflow, water and climate change
.
Journal of the American Water Works Association
109
(
10
),
2
13
.
https://doi.org/10.5942/jawwa.2017.109.0114
.
Kimbrough
D. E.
2018
Changes in temperature and rainfall as a result of local climate change in Pasadena, California
.
Hydrology
5
(
2
),
25
.
https://doi.org/10.3390/hydrology5020025
.
Melillo
J. M.
,
Richmond
T. C.
&
Yohe
G. W.
, (eds.)
2014
Climate Change Impacts in the United States: The Third National Climate Assessment
.
U.S. Global Change Research Program
,
841
pp. .
Michalak
A. M.
2016
Study role of climate change in extreme threats to water quality
.
Nature
535
,
349
350
.
doi:10.1038/535349a
.
Vikesland
P. J.
,
Ozekin
K.
&
Valentine
R. L.
2001
Monochloramine decay in model and distribution system waters
.
Water Research
35
,
1766
1776
.