Abstract

The aim of this study is to compare methane (CH4) and nitrous oxide (N2O) fluxes from horizontal subsurface flow (HSSF) and vertical subsurface flow (VF) systems treating municipal wastewater in tropical climates. The treatment performance from both systems was monitored simultaneously with CH4 and N2O fluxes to observe the relationships between them. Average CH4 fluxes of 5.4 mg·m−2·h−1 and 9.5 mg·m−2·h−1, and N2O fluxes of 0.32 and 0.21 mg·m−2·h−1 were measured from VF and HSSF systems, respectively. The average CH4 fluxes measured during the wet period were 63.5% and 44% less than those recorded during the dry period for HSSF and VF systems, respectively; those for N2O fluxes were 47% and 38% less than the dry period for these systems, respectively. The CH4 and N2O fluxes from both HSSF and VF dropped as a result of rainfall events and slowly increased after days with no rainfall. Influent total organic carbon (TOC) and total nitrogen (TN) concentrations are found to be the dominant factors regulating the fluxes where significant correlations between CH4 and N2O with the influent TOC and TN concentrations were found in both systems.

INTRODUCTION

Constructed wetlands (CWs) are artificial wetlands built for various purposes, of which wastewater treatment is one of the most common. Treatment can be achieved by creating a suitable environment for microorganisms responsible for degrading pollutants in wastewater. These microorganisms, however, emit substantial amounts of greenhouse gases during the process. Among the greenhouse gases emitted by CWs, methane (CH4) and nitrous oxide (N2O) are of particular concern owing to their persistence in the atmosphere and their much higher impacts on global warming relative to carbon dioxide (CO2), as measured by their global warming potential values.

The three common types of CW include free water surface (FWS) and two different subsurface flows (SF): vertical subsurface flow (VF) and horizontal subsurface flow (HSSF). Each type differs significantly in term of function and operation. For secondary wastewater treatment, HSSF and VF CWs are more commonly used, whereas FWS CW is more typically used for advance treatment. VF CW is commonly designed for unsaturated water conditions and to maximise O2 availability in the system. HSSF CW, on the other hand, is commonly designed to promote anaerobic and anoxic environments with saturated soil conditions (Kadlec & Wallace 2008).

Each type of CW hosts different groups of microorganisms that emit different amounts of CH4 and N2O. CH4 is produced predominantly by methanogenesis, which occurs in water-logged environments such as HSSF CWs. This process occurs to a lesser extent in VF CW environments, where aerobic conditions dominate. Such conditions are also favourable for the production of methanotrophic bacteria, which can turn CH4 to CO2 and thus lower CH4 emission. A comprehensive review by Mander et al. (2014) on greenhouse gas fluxes from CW revealed that CH4 fluxes from HSSF CWs range from 0.76 to 17.5 mg m−2 h−1 (Gui et al. 2007; Picek et al. 2007), whereas those from VF CWs range from 0.3 to 3 mg m−2 h−1 (Teiter & Mander 2005; Søvik et al. 2006; Gui et al. 2007; Mander et al. 2008).

N2O occurs in both aerobic and anoxic conditions in CWs. In the former, N2O is created by nitrification, a process in which ammonium (NH4+) is oxidised to nitrate (NO3) by nitrifying bacteria. In anoxic conditions, denitrification is responsible for N2O production. However, aerobic or anaerobic microsites can occur in various places within CWs. Therefore, nitrification can also occur in anaerobically dominated CWs, particularly near the root zone, where O2 can be transported from above ground to below ground via plant aerenchyma (Bodelier et al. 1996).

Recent studies have addressed the issue of greenhouse gas fluxes from CWs with various conditions, including those comparing the fluxes from VF and HSSF systems (Mander et al. 2005; Søvik et al. 2006; Gui et al. 2007; Liu et al. 2009). Among them, few studies have compared fluxes from CWs in different seasons, in which significant differences were found in the fluxes (Tai et al. 2002; Søvik et al. 2006). Several previous studies have addressed the issue of greenhouse gas emission from both natural and CWs in tropical climates (Bartlett & Harriss 1993; Chuersuwan et al. 2014). Although they suggest that greenhouse gas emissions from wetlands are higher in tropical climates, no concrete information is given on the effect of rainfall on the greenhouse gas emission from CWs operated in the tropics. Schouten et al. (2014) reported that fugitive emissions of CH4 and N2O from a wastewater treatment system varied widely among positions and occurred rapidly over time depending on the environmental conditions. Among them, humidity and temperature were given as important drivers for greenhouse gas flux from soil, which is directly influenced by the prevailing climate conditions (Hansen et al. 2013). Moreover, these fluxes can cause significantly larger variation in the emission data from CWs in tropical climate than from those in temperate and boreal climates, where most of the studies originated. A significant amount of N2O can also be formed in CWs under suboptimal condition, i.e., low O2 during nitrification and higher O2 or low C/N ratio during denitrification. Such emission could be reduced by maintaining a temporary high biochemical O2 demand (BOD)/N ratio and low O2/NO3 ratio during denitrification under intermittent wastewater feeding conditions (Huang et al. 2013).

Furthermore, the use of CWs is considered to be an economical alternative to wastewater treatment technology and is appropriate for developing countries (Kivaisi 2001), many of which are located in tropical climates. Therefore, the aim of this study is to provide additional information on the effect of short-term seasonal change and rainfall events on CH4 and N2O fluxes from CWs operated in tropical climates.

MATERIALS AND METHODS

Experimental setup and operation

VF and HSSF CW units were used in this study. The units were kept outdoors under ambient conditions with daily average temperatures ranging from 18.3 °C in January to 31.9 °C in April. The climate information of the experimental location is given in Table 1.

Table 1

Monthly climate data

 Temperature (°C) Rainfall (mm) Rainfall (days) Evaporation rate (mm) 
January 21.32 79.5 
February 24.27 108.4 
March 27.71 5.9 140.4 
April 29.58 34.9 157.7 
 Temperature (°C) Rainfall (mm) Rainfall (days) Evaporation rate (mm) 
January 21.32 79.5 
February 24.27 108.4 
March 27.71 5.9 140.4 
April 29.58 34.9 157.7 

Both units were operated simultaneously to treat municipal wastewater with an average chemical oxygen demand (COD) concentration of 126.8 mg L−1 and an average total nitrogen (TN) concentration of 28.0 mg L−1. Both units had the same cross-section area of 1.4 m2, hydraulic loading rate of 70 L day−1, and plant density of 43 plants m−2.

The VF system included a rectangular concrete tank and a wastewater storage tank. A peristaltic pump was used to deliver the influent from the storage tank to the CW tank. The VF CW was 1.4 m long, 1 m wide and 1.2 m high. The tank was filled with fine sand as the main media and two layers of small (10–20 mm in diameter) and large (40–80 mm in diameter) gravels of approximately 10 cm thickness on the top and bottom of the tank, respectively. The total media depth was approximately 1 m. The setup for the HSSF system also consisted of a wastewater storage tank, a peristaltic pump and a rectangular concrete tank. The tank was 0.6 m wide, 2.3 m long and 0.6 m high. Small gravel was used as the main media, and large gravel was placed at the inlet and outlet zones (Figure 1). Both tanks were planted with young stems of umbrella sedge (Cyperus alternifolius L.) approximately 10 cm tall, which were obtained from a natural area near the experimental site. The VF system was fed intermittently every 4 h, and the HSSF system received wastewater continuously.

Figure 1

Side view of the VF (top) and the HSSF (bottom) systems.

Figure 1

Side view of the VF (top) and the HSSF (bottom) systems.

Water sampling and analysis

Water samples from the HSSF system were collected at four different locations including the inlet, middle and outlet zones and the effluent outlet; those from the VF system were collected only at the effluent outlet. The quality of the influent and effluent was measured on a weekly basis. COD, total Kjeldahl nitrogen (TKN), ammonium nitrogen (NH4-N), and nitrate (NO3-N) and nitrite nitrogen (NO2-N) were analysed by using the APHA/AWWA/WEF (2005). The TN was estimated from the sum of TKN, NO3-N and NO2-N. The total organic carbon (TOC) was monitored on a bi-weekly basis and was analysed by using a Shimadzu TOC analyser. Dissolved O2 (DO) was measured simultaneously through gas sampling by using Clean Instrument® DO200 DO meters. The Redox potential (OPR) was measured by using a Clean Instrument® OPR 30 tester. TOC, DO and ORP in the influent and effluent from both systems were monitored simultaneously with CH4 and N2O on a bi-weekly basis. The experiment was conducted from the middle of winter to the beginning of summer, which was the dry period with zero rainfall, and from the beginning to the middle of the rainy season.

Per cent pollutant mass removal (Equation (1)) was used to represent the water treatment performance of the CWs; this method considers the differences of water loss in each tank (Kadlec & Wallace 2008). In this equation, %MR is the per cent pollutant mass removal, Qi is inflow rate (L day−1), Qo is the outflow rate (L day−1), Ci is the influent concentration (mg L−1) and Co is the effluent concentration (mg L−1). 
formula
(1)

Gas sampling and analysis

Gas fluxes were measured at 2-week intervals during steady operation of CWs. This flux measurement was performed for observing the effects of short-term seasonal change and rainfall events on gas emissions. A closed static chamber (Figure 2) was used for gas sampling following the study by Chiemchaisri et al. (2009). The chamber was constructed from clear acrylic, which is inert to N2O, and its dimensions were 0.3 × 0.3 × 1.0 m in length, width and height, respectively. A fan and a thermometer were mounted inside the chamber to ensure even mixing of the gas and to record the air temperature inside the chamber during sampling. Acrylic bases were fixed permanently at each sampling point and consisted of open grooves upon which the chamber could be placed and sealed with water to prevent gas leaks during sampling. Gas samples were drawn from the chamber via rubber septa at the top of the chamber.

Figure 2

Gas sampling chamber.

Figure 2

Gas sampling chamber.

Gas samples were collected between 09:00 and 10:00 local time (LT) at 10-min intervals for 30 min, i.e., at 0, 10, 20 and 30 min at each sampling point. With the pre-inserted base, the media were disturbed only slightly when the chamber was placed over the plants. Approximately 1 min was allowed for the gaseous content to stabilise before taking the sample for the 0-min interval. The gas samples were contained in vacuum blood collection tubes (BD Vacutainer®) and were analysed to determine the CH4 and N2O concentrations by using a gas chromatograph (PerkinElmer Clarus 580 GC) equipped with an auto sampler in addition to a flame ionisation detector and an electron capture detector. The flux could then be determined using Equation (2). 
formula
(2)

In this equation, F is the flux (mg·m−2·h−1), C is the gas concentration (mg m−3), V is the chamber volume (m3), A is the surface area inside the chamber (m) and t is the time of sampling (h). To estimate the daily flux, sampling was conducted during the hours in which the flux was believed to represent the daily mean because fluxes are influenced by ambient temperature (Livesley et al. 2008; Alves et al. 2012); thus, 09:00 and 10:00 LT were determined to be the most suitable times (Alves et al. 2012). The daily average flux can be estimated from the change of gas concentration within the chamber over the 30-min interval discussed previously.

Statistical analysis

All statistical tests were performed by using the SPSS® software package (Version 19.0) for Windows and Microsoft Excel® 2013. The means of replicate pairs were used for flux calculations. The coefficient of determination for all fluxes plotted as linear regression graphs of concentration against time were greater than 0.67. The SPSS® software was used to determine Pearson's correlation and its P-value, and analysis of variance analysis was performed to determine the significance of the results.

RESULTS AND DISCUSSION

Treatment performance of CWs

Table 2 shows the influent and effluent characteristics of the samples obtained from the HSSF and VF units during the study. The VF unit showed better removal efficiency than the HSSF unit for TOC, COD and TN. The average removal efficiencies were 91.5% and 83.1% for COD, 86.6% and 83.1% for TOC, 84.3% and 86.04% for TN for the VF and HSSF units, respectively.

Table 2

Influent and effluent characteristics from the two CW systems

Parameter Influent
 
VF
 
HSSF
 
Range (dry) (mg·L−1Range (wet) (mg·L−1Average (mg·L−1 ± SD) Range (mg·L−1Average (mg·L−1 ± SD) Range (mg·L−1Average (mg·L−1 ± SD) 
COD 89.0–236.0 72.0–182.3 111.0 ± 41.8 4.0–32.0 11.0 ± 8.6 4.0–56.0 25.4 ± 16.7 
TOC 55.2–105.6 62.9–74.3 77.9 ± 16.4 9.2–15.8 12.9 ± 2.2 12.1–21.1 17.3 ± 2.9 
TKN 20.7–39.8 12.3–24.6 26.0 ± 6.8 0.1–1.2 0.5 ± 0.3 0.3–15.1 4.9 ± 4.4 
NH3 14.7–30.8 11.2–21.3 21.5 ± 5.2 0.0–0.6 0.5 ± 0.2 0.5–12.9 4.3 ± 3.9 
NO3 0.03–0.09 0.00–0.04 0.04 ± 0.02 1.90–5.75 4.22 ± 1.10 0.01–0.14 0.05 ± 0.04 
DO 0.1–0.6 0.2–0.7 0.3 ± 0.2 3.0–5.8 4.1 ± 0.9 1.5–4.2 2.3 ± 0.7 
Parameter Influent
 
VF
 
HSSF
 
Range (dry) (mg·L−1Range (wet) (mg·L−1Average (mg·L−1 ± SD) Range (mg·L−1Average (mg·L−1 ± SD) Range (mg·L−1Average (mg·L−1 ± SD) 
COD 89.0–236.0 72.0–182.3 111.0 ± 41.8 4.0–32.0 11.0 ± 8.6 4.0–56.0 25.4 ± 16.7 
TOC 55.2–105.6 62.9–74.3 77.9 ± 16.4 9.2–15.8 12.9 ± 2.2 12.1–21.1 17.3 ± 2.9 
TKN 20.7–39.8 12.3–24.6 26.0 ± 6.8 0.1–1.2 0.5 ± 0.3 0.3–15.1 4.9 ± 4.4 
NH3 14.7–30.8 11.2–21.3 21.5 ± 5.2 0.0–0.6 0.5 ± 0.2 0.5–12.9 4.3 ± 3.9 
NO3 0.03–0.09 0.00–0.04 0.04 ± 0.02 1.90–5.75 4.22 ± 1.10 0.01–0.14 0.05 ± 0.04 
DO 0.1–0.6 0.2–0.7 0.3 ± 0.2 3.0–5.8 4.1 ± 0.9 1.5–4.2 2.3 ± 0.7 

N = 15 for COD, TKN, NH3 and NO3 (dry: n = 10, wet: n = 5); N = 8 for TOC (dry: n = 5, wet: n = 3); N = 12 for DO (dry: n = 8, wet: n = 5).

Most studies that compare the treatment performance between HSSF and VF CWs reveal that VF systems are likely to treat domestic wastewater more efficiently (Zurita et al. 2009; Abou-Elela et al. 2013; Pandey et al. 2013). The VF system is also expected to remove organic C (measured as COD, BOD and TOC) better owing to higher O2 availability in the system and the filtration capability, particularly in new systems (Zurita et al. 2009). Contrasting results were reported by Yalcuk & Ugurlu (2009), in which an HSSF system removed organic C more efficiently than a VF system, with both treating landfill leachate. This occurred because the HSSF has a higher hydraulic retention time, which is more crucial for the removal efficiency under high organic loading conditions. In this study, the COD removal efficiencies in both systems fell in the same range as those reported in other studies under similar conditions. The VF showed similar removal efficiency as that in other studies, at 91.5% compared with 92.9% in Abou-Elela et al. (2013) and 88.9% in Zurita et al. (2009). The HSSF's removal efficiency of 83.1%, however, is slightly higher than the results obtained from other studies, i.e., 75.5–77.2% in Zurita et al. (2009) and 81.1% in Abou-Elela et al. (2013).

The TOC and DO concentrations were measured at different zones in the HSSF CW. The average TOC concentrations from the inlet, middle and outlet zones were 72.6, 32.8 and 22.5 mg L−1, respectively. The average effluent DO concentration in the HSSF was slightly lower than that of the VF; however, the DO concentrations with the HSSF unit were significantly lower than that of the VF. The average DO concentrations of 0.55 mg L−1 and 0.58 mg L−1 were measured near the inlet and in the middle of the vegetation zone, respectively. The DO concentration increased rapidly only near the outlet zone, at an average of 1.18 mg L−1 before reaching an average of 2.3 mg L−1 in the effluent.

The VF system was able to remove organic N and ammonia N better than the HSSF system. However, a considerable increase was noted in the VF effluent NO3 concentration from nitrification, whereas, lower nitrification occurred in the HSSF system with only a trace amount of NO3 found in the HSSF effluent. In this aspect, effluent NO3 concentration can be controlled better in a combined HSSF and VF CW system with internal recirculation to further denitrify the NO3 (Tunçsiper 2009). The TN removal efficiencies in the VF system and in the HSSF system in this study are higher than those reported in other research (Zurita et al. 2009). The higher efficiency may be attributed to the low N loading rate or the warm climate of the system location. Tunçsiper (2009) reported that lower HLR and warmer climates are two main factors that can significantly increase the TN removal efficiency in CWs.

The high nitrification rate was a factor in the VF CW's high TKN removal efficiency (Vymazal 2007). However, nitrification left N in the form of NO3 which prevented the system from achieving such high TN removal efficiency. Under typical VF aerobic conditions, the denitrification reaction that reduces NO3 to N2O and N2, is prohibited. However, it has been suggested that some nitrifiers can perform the denitrification process in a pathway known as nitrifier denitrification, in which NO2 is converted to N2O and N2 without producing NO3 (Wrage et al. 2001; Kool et al. 2011). Nitrifier denitrification occurs in aerobic but low O2 conditions, which could be a pathway for N transformation in the VF CW because low O2 conditions may occur in some places within the system.

Greenhouse gas emission

Table 3 shows the average observed CH4 and N2O fluxes from the two CW systems. The average CH4 flux from three sampling points in the HSSF system was 9.5 mg·m−2·h−1 and an average CH4 of 5.4 mg·m−2·h−1 was found from the VF system. The CH4 fluxes in the HSSF CW declined from the inlet to the outlet zones, which correspond to the TOC concentrations.

Table 3

Average CH4 and N2O fluxes (mg·m−2·h−1) from the two CW systems

 HSSF inlet HSSF middle HSSF outlet HSSF VF 
CH4 Dry period Range 9.8–17.8 8.9–16.6 7.1–15.3 8.6–15.5 4.3–8.9 
Average 14.1 ± 2.9 13.1 ± 3.8 10.2 ± 3.3 12.5 ± 2.8 6.5 ± 1.7 
Wet period Range 5.4–5.9 3.7–5.8 2.9–3.7 4.1–5.1 3.4–3.7 
Average 5.6 ± 0.3 4.9 ± 1.1 3.2 ± 0.4 4.5 ± 0.5 3.6 ± 0.2 
N2Dry period Range 0.18–0.25 0.10–0.36 0.23–0.40 0.18–0.32 0.15–0.49 
Average 0.21 ± 0.04 0.23 ± 0.11 0.32 ± 0.07 0.26 ± 0.05 0.37 ± 0.16 
Wet period Range 0.10–0.16 0.10–0.11 0.11–0.21 0.11–0.16 0.21–0.27 
Average 0.12 ± 0.03 0.1 ± 0.0 0.18 ± 0.06 0.13 ± 0.03 0.23 ± 0.03 
 HSSF inlet HSSF middle HSSF outlet HSSF VF 
CH4 Dry period Range 9.8–17.8 8.9–16.6 7.1–15.3 8.6–15.5 4.3–8.9 
Average 14.1 ± 2.9 13.1 ± 3.8 10.2 ± 3.3 12.5 ± 2.8 6.5 ± 1.7 
Wet period Range 5.4–5.9 3.7–5.8 2.9–3.7 4.1–5.1 3.4–3.7 
Average 5.6 ± 0.3 4.9 ± 1.1 3.2 ± 0.4 4.5 ± 0.5 3.6 ± 0.2 
N2Dry period Range 0.18–0.25 0.10–0.36 0.23–0.40 0.18–0.32 0.15–0.49 
Average 0.21 ± 0.04 0.23 ± 0.11 0.32 ± 0.07 0.26 ± 0.05 0.37 ± 0.16 
Wet period Range 0.10–0.16 0.10–0.11 0.11–0.21 0.11–0.16 0.21–0.27 
Average 0.12 ± 0.03 0.1 ± 0.0 0.18 ± 0.06 0.13 ± 0.03 0.23 ± 0.03 

N2O fluxes of 0.32 and 0.21 mg·m−2·h−1 were measured from the VF and HSSF CWs, respectively. No significant difference was found between N2O fluxes measured in the inlet and middle zones. N2O fluxes measured from HSSF outlet zone were significantly higher than those of the other two measuring points in the system.

CH4 can be generated during methanogenesis whereby organic C is consumed by microbes that release gaseous CH4 as a metabolic by-product. Methanogenesis occurs in anaerobic conditions and is prohibited by O2 under aerobic conditions. Therefore, higher CH4 flux from the HSSF system can be attributed to the system's low O2 environment, which is also supported by the lower DO concentration in the HSSF effluent. N2O can be produced under both aerobic (nitrification process) and anoxic (denitrification process) conditions, which may explain the lack of significant differences between fluxes from different CW types reported in most studies (Mander et al. 2014). Specifically, significant amounts of N2O are formed under suboptimal conditions, i.e., low O2 during nitrification and higher O2 or a low C/N ratio during denitrification (Fuchs et al. 2011). In this study, N2O emission from the VF CW was not detected at significantly higher levels than that in the HSSF system despite its highly aerobic condition. Its formation could have been avoided likely by maintaining a temporary high BOD/N ratio and a low O2/NO3 ratio during denitrification under intermittent wastewater feeding conditions.

During the dry period, when absolutely no rainfall occurred during the entire period, both CH4 and N2O fluxes were found to be higher than those recorded during the wet period, which covers the later period of the experiment. The differences in fluxes between the dry and wet periods were significantly larger in the HSSF system than those in the VF systems for both CH4 and N2O emissions. The average CH4 fluxes measured during the wet period were 63.5% and 44% less than those during the dry period for the HSSF and VF systems, respectively; the N2O fluxes in the wet period were 47% and 38% less than those in the dry period for the systems, respectively. CH4 fluxes measured on a day with rain were found to be lower than those measured a few days before and after a rainy event (Figure 3). However, the same effect was not found in N2O fluxes from both systems (Figure 4).

Figure 3

CH4 fluxes measured days before (negative numbers) and after (positive numbers) rain events in the HSSF CW (▴) and in the VF CW (○).

Figure 3

CH4 fluxes measured days before (negative numbers) and after (positive numbers) rain events in the HSSF CW (▴) and in the VF CW (○).

Figure 4

N2O fluxes measured days before (negative numbers) and after (positive numbers) rainfall in the HSSF CW (▴) and in the VF CW (○).

Figure 4

N2O fluxes measured days before (negative numbers) and after (positive numbers) rainfall in the HSSF CW (▴) and in the VF CW (○).

The TOC and TN concentrations in the influent were found to affect the CH4 and N2O fluxes substantially because the drop in these fluxes correlates significantly with the influent TOC and TN concentrations. Figure 5 shows the relationship between influent TOC and CH4 flux. Pearson's R between the influent TOC and CH4 fluxes were 0.75 and 0.87 for the HSSF and VF, respectively. Similar relationship was observed between influent TN and N2O flux (Figure 6). The R values for the TN and N2O correlations were 0.76 and 0.81, respectively, with P-value less than 0.05. The same effect was reported in studies measuring the CH4 and N2O fluxes from subsurface flow CWs in temperate and warm climates (Gui et al. 2007; Liu et al. 2009). Luo et al. (2013) suggested that the available C and N are related to the significant and more direct control of CH4 and N2O fluxes, whereas the effects from other factors such as temperature and moisture can be more complicated. Similar TOC to CH4 and TN to N2O ratios were found between this study and other research in different climates (Mander et al. 2014). These results suggest that the effect of C and N availabilities on CH4 and N2O prevails regardless of the climate condition.

Figure 5

TOC and CH4 relationship in the HSSF CW (▴) and in the VF CW (○).

Figure 5

TOC and CH4 relationship in the HSSF CW (▴) and in the VF CW (○).

Figure 6

TN and N2O relationship in the HSSF CW (▴) and in the VF CW (○).

Figure 6

TN and N2O relationship in the HSSF CW (▴) and in the VF CW (○).

A comparison of CH4 and N2O fluxes from the VF and HSSF CWs in different climate zones (Tables 4 and 5) revealed that the fluxes measured in this study were slightly higher than those measured in colder climate conditions. However, because the C and N availability is a dominant factor affecting the fluxes, as previously suggested, the CH4 and TOC and N2O and TN ratios were compared. The CH4 to TOC ratios determined in this study for both HSSF and VF CWs are similar to those of studies in temperate/warm climates and are higher than those from temperate/boreal climates during the wet season. Moreover, the present ratios are higher than other CWs from both temperate/warm and temperate/boreal climates, which suggests that a higher CH4 flux for the same TOC loading rate in the influent can be expected in warmer climates.

Table 4

Comparison of CH4 fluxes from the CW systems

Climate CH4 (mg·m−2·h−1CH4/TOC (%) Plant CW type Reference 
Temperate/boreal 0.88 Phragmites australis VF Teiter & Mander (2005), Søvik et al. (2006) and Mander et al. (2008)  
Temperate/warm 0.3 1.68 P. australis VF Gui et al. (2007)  
Temperate/warm 0.76 4.3 P. australis HSSF Gui et al. (2007)  
Temperate/warm 1.73 P. australis VF Liu et al. (2009)  
Temperate/warm P. australis HSSF Liu et al. (2009)  
Tropical 2.9 – Cyperus spp. SF Chuersuwan et al. (2014)  
Tropical 3.4–8.9 Dry season 2.8 Cyperus alternifolius L. VF This study 
Wet season 1.8 
Tropical 4.6–15.5 Dry season 5.3 Cyperus alternifolius L. HSSF This study 
Wet season 2.3 
Climate CH4 (mg·m−2·h−1CH4/TOC (%) Plant CW type Reference 
Temperate/boreal 0.88 Phragmites australis VF Teiter & Mander (2005), Søvik et al. (2006) and Mander et al. (2008)  
Temperate/warm 0.3 1.68 P. australis VF Gui et al. (2007)  
Temperate/warm 0.76 4.3 P. australis HSSF Gui et al. (2007)  
Temperate/warm 1.73 P. australis VF Liu et al. (2009)  
Temperate/warm P. australis HSSF Liu et al. (2009)  
Tropical 2.9 – Cyperus spp. SF Chuersuwan et al. (2014)  
Tropical 3.4–8.9 Dry season 2.8 Cyperus alternifolius L. VF This study 
Wet season 1.8 
Tropical 4.6–15.5 Dry season 5.3 Cyperus alternifolius L. HSSF This study 
Wet season 2.3 
Table 5

Comparison of N2O fluxes from the CW systems

Climate N2O (mg·m−2·h−1N2O/TN (%) Plant CW type Reference 
Temperate/boreal 0.225 0.021 P. australis VF Teiter & Mander (2005), Søvik et al. (2006) and Mander et al. (2008)  
Boreal 0.222 0.11 No vegetation VF Søvik et al. (2006)  
Boreal 0.89 0.894 No vegetation HSSF Søvik et al. (2006)  
Temperate/warm 0.123 0.96 P. australis VF Gui et al. (2007)  
Temperate/warm 0.073 0.42 P. australis VF Liu et al. (2009)  
Temperate/warm 0.4 0.4 P. australis HSSF Liu et al. (2009)  
Tropical 1.0 – Cyperus spp. SF Chuersuwan et al. (2014)  
Tropical 0.15–0.49 Dry season 0.20 Cyperus alternifolius L. VF This study 
Wet season 0.16 
Tropical 0.11–0.32 Dry season 0.14 Cyperus alternifolius L. HSSF This study 
Wet season 0.09 
Climate N2O (mg·m−2·h−1N2O/TN (%) Plant CW type Reference 
Temperate/boreal 0.225 0.021 P. australis VF Teiter & Mander (2005), Søvik et al. (2006) and Mander et al. (2008)  
Boreal 0.222 0.11 No vegetation VF Søvik et al. (2006)  
Boreal 0.89 0.894 No vegetation HSSF Søvik et al. (2006)  
Temperate/warm 0.123 0.96 P. australis VF Gui et al. (2007)  
Temperate/warm 0.073 0.42 P. australis VF Liu et al. (2009)  
Temperate/warm 0.4 0.4 P. australis HSSF Liu et al. (2009)  
Tropical 1.0 – Cyperus spp. SF Chuersuwan et al. (2014)  
Tropical 0.15–0.49 Dry season 0.20 Cyperus alternifolius L. VF This study 
Wet season 0.16 
Tropical 0.11–0.32 Dry season 0.14 Cyperus alternifolius L. HSSF This study 
Wet season 0.09 

CONCLUSIONS

The VF CW showed better overall treatment performance and significantly lower CH4 emission than the HSSF CW. However, the high effluent NO3 concentration and the slightly higher N2O emission in the VF CWs must be addressed because they may be limitations in actual application. Although the effluent NO3 concentration can be better controlled in a combined HSSF and VF CW system with internal recirculation, N2O emission from a VF CW could be reduced by maintaining a high BOD/N ratio and a low O2/NO3 ratio during denitrification under intermittent feeding conditions. Emissions during a dry period were higher than those recorded during a rainfall event. Reductions in the CH4 and N2O fluxes correlated to the dilutions of influent TOC and TN concentrations during the rainfall event. The results suggest that C and N availability, which fluctuates seasonally, is the prevailing factor influencing CH4 and N2O fluxes, respectively. In actual application, the difference between seasonal fluctuation of C and N availability owing to dilution by rainfall would likely be higher, and thus may result in higher fluctuation of CH4 and N2O emission from HSSF and VF CWs operating in tropical climates. The CH4 and N2O fluxes from both systems were higher than those presented in other research conducted in temperate and boreal climates.

ACKNOWLEDGEMENTS

This study was supported by the Center of Excellence on Hazardous Substance Management (HSM), Chulalongkorn University; the Department of Environmental Engineering, Chiang Mai University; and Faculty of Engineering, Kasetsart University.

REFERENCES

REFERENCES
Abou-Elela
S. I.
,
Hellas
M. S.
,
Golinelli
G.
&
Abou-Taleb
E. M.
2013
Municipal wastewater treatment in horizontal and vertical flows constructed wetlands
.
Ecological Engineering
61
,
460
468
.
Alves
B. J. R.
,
Smith
K. A.
,
Flores
R. A.
,
Cardoso
A. S.
,
Oliveira
W. R. D.
,
Jantalia
C. P.
,
Urquiaga
S.
&
Boddey
R. M.
2012
Selection of the most suitable sampling time for static chambers for the estimation of daily mean N2O flux from soils
.
Soil Biology and Biochemistry
46
,
129
135
.
APHA/AWWA/WEF
2005
Standard Methods for the Examination of Water and Wastewater
,
21st edn
.
American Public Health Association/American Water Works Association/Water Environment Federation
,
Washington, DC
,
USA
.
Bartlett
K. B.
&
Harriss
R. C.
1993
Review and assessment of methane emissions from wetlands
.
Chemosphere
26
(
1–4
),
261
320
.
Bodelier
P.
,
Libochant
J. A.
,
Blom
C.
&
Laanbroek
H. J.
1996
Dynamics of nitrification and denitrification in root-oxygenated sediments and adaptation of ammonia-oxidizing bacteria to low-oxygen or anoxic habitats
.
Applied and Environmental Microbiology
62
(
11
),
4100
4107
.
Chiemchaisri
C.
,
Chiemchaisri
W.
,
Junsod
J.
,
Threedeach
S.
&
Wicranarachchi
P. N.
2009
Leachate treatment and greenhouse gas emission in subsurface horizontal flow constructed wetland
.
Bioresource Technology
100
(
16
),
3808
3814
.
Chuersuwan
S.
,
Suwanwaree
P.
&
Chuersuwan
N.
2014
Estimating greenhouse gas fluxes from constructed wetlands used for water quality improvement
.
Songklanakarin Journal of Science and Technology
36
(
3
),
367
373
.
Gui
P.
,
Inamori
R.
,
Matsumura
M.
&
Inamori
Y.
2007
Evaluation of constructed wetlands by waste water purification ability and greenhouse gas emissions
.
Water Science and Technology
56
(
3
),
49
55
.
Hansen
R.
,
Mander
Ü.
,
Soosaar
K.
,
Maddison
M.
,
Lohmus
K.
,
Kupper
P.
,
Kanal
A.
&
Sober
J.
2013
Greenhouse gas fluxes in an open air humidity manipulation experiment
.
Landscape Ecology
28
(
4
),
637
649
.
Huang
L.
,
Gao
X.
,
Guo
J.
,
Ma
X.
&
Liu
M.
2013
A review on the mechanism and affecting factors of nitrous oxide emission in constructed wetlands
.
Environmental Earth Sciences
68
(
8
),
2171
2180
.
Kadlec
R. H.
&
Wallace
S.
2008
Treatment Wetlands
.
CRC Press
,
Boca Raton, FL
,
USA
.
Kool
D. M.
,
Dolfing
J.
,
Wrage
N.
&
Van Groenigen
J. W.
2011
Nitrifier denitrification as a distinct and significant source of nitrous oxide from soil
.
Soil Biology and Biochemistry
43
,
174
178
.
Liu
C.
,
Xu
K.
,
Inamori
R.
,
Ebie
Y.
,
Liao
J.
&
Inamori
Y.
2009
Pilot-scale studies of domestic wastewater treatment by typical constructed wetlands and their greenhouse gas emissions
.
Frontiers of Environmental Science and Engineering in China
3
(
4
),
477
482
.
Livesley
S. J.
,
Kiese
R.
,
Graham
J.
,
Weston
C. J.
,
Butterbach-Bahl
K.
&
Arndt
S. K.
2008
Trace gas flux and the influence of short-term soil water and temperature dynamics in Australian sheep grazed pastures of differing productivity
.
Plant and Soil
309
(
1–2
),
89
103
.
Mander
Ü.
,
Teiter
S.
&
Augustin
J.
2005
Emission of greenhouse gases from constructed wetlands for wastewater treatment and from riparian buffer zones
.
Water Science and Technology
52
(
10–11
),
167
176
.
Mander
Ü.
,
Lõhmus
K.
,
Teiter
S.
,
Mauring
T.
,
Nurk
K.
&
Augustin
J.
2008
Gaseous fluxes in the nitrogen and carbon budgets of subsurface flow constructed wetlands
.
Science of the Total Environment
404
(
2
),
343
353
.
Mander
Ü.
,
Dotro
G.
,
Ebie
Y.
,
Towprayoon
S.
,
Chiemchaisri
C.
,
Nogueira
S. F.
,
Jamsranjav
B.
,
Kasak
K.
,
Truu
J.
,
Tournebize
J.
&
Mitsch
W. J.
2014
Greenhouse gas emission in constructed wetlands for wastewater treatment: a review
.
Ecological Engineering
66
,
19
35
.
Pandey
M. K.
,
Jenssen
P. D.
,
Krogstad
T.
&
Jonasson
S.
2013
Comparison of vertical and horizontal flow planted and unplanted subsurface flow wetlands treating municipal wastewater
.
Water Science and Technology
68
(
1
),
117
123
.
Søvik
A. K.
,
Augustin
J.
,
Heikkinen
K.
,
Huttunen
J. T.
,
Necki
J. M.
,
Karjalainen
S. M.
,
Kløve
B.
,
Liikanen
A.
,
Mander
Ü.
,
Puustinen
M.
,
Teiter
S.
&
Wachniew
P.
2006
Emission of the greenhouse gases nitrous oxide and methane from constructed wetlands in Europe
.
Journal of Environmental Quality
35
(
6
),
2360
2373
.
Tai
P. D.
,
Li
P. J.
,
Sun
T. H.
,
He
Y. W.
,
Zhou
Q. X.
,
Gong
Z. Q.
,
Mizuochi
M.
&
Inamori
Y.
2002
Greenhouse gas emissions from a constructed wetland for municipal sewage treatment
.
Journal of Environmental Sciences
14
(
1
),
27
33
.
Vymazal
J.
2007
Removal of nutrients in various types of constructed wetlands
.
Science of the Total Environment
380
(
1
),
48
65
.
Wrage
N.
,
Velthof
G. L.
,
Van Beusichem
M. L.
&
Oenema
O.
2001
Role of nitrifier denitrification in the production of nitrous oxide
.
Soil Biology and Biochemistry
33
(
12
),
1723
1732
.