Various land uses accumulate pollutants during the dry period and these pollutants are eventually washed off with the stormwater during the rainfall event. Such NPS pollutants generated in urban areas cause water quality deterioration and affect the health of aqua-ecosystem, since they have high contents of particulates, heavy metals, chemicals and toxic materials (Boxall and Maltby, 1995; Kim et al., 2006; Maltby et al., 1995; Perdikaki and Mason, 1999; Son et al., 2008).

In order to reduce the water quality deterioration caused by NPS pollutants, the Korea Ministry of Environment (MOE) implemented various policies to manage the NPS discharges. NPS management policies and treatment facility installation in companies and application of total maximum daily load were the typically performed NPS management techniques. The pollutant mass loading of Korea's four major rivers as of 2010 was accounted for about 67% of the total water pollution, and by 2020 it was expected be greater than 70% (MOE, 2012). With continuous point source management, the biodegradable organic pollutants (e.g. BOD) in the rivers tend to decline, however the non-biodegradable organic matters continually increase. For the mentioned reason, implementation of the NPS management techniques became significantly essential (MOE, 2006).

In particular, application of low impact development (LID) techniques was ere necessary in order to control the NPS negative effects (MOE, 2012). The LID techniques have been adopted and applied by the U.S. and Europe since the 1990s, aiming at ecosystem preservation, water circulation restoration and environmental impact reduction through introducing decentralized stormwater techniques for various uses. The principle of a LID technique was to efficiently manage NPS pollutants generated after a land area development, while maintaining pre-hydrological function such as infiltration, retention and evapo-transpiration (DER, 1999). Most of the LIDs implemented were infiltration trenches, infiltration basins, vegetated swales, bioretention cells and constructed wetlands. The infiltration trenches and infiltration basins were regarded as the most basic LID structure compared to the other facilities. By adding vegetation and limiting the infiltration capabilities, a bioretention cell or a constructed wetland could be developed. This research aims to evaluate the applicability of an infiltration and filtration (IF) facility in various rainfall events based on the obtained volume and pollutant reduction capabilities from the monitored 29 events. Rainfall events in Korea were concentrated during summer and more than 80% of the annual rainfalls were 10mm and less. Based on this, the research also evaluates the volume and pollutant reduction of the IF facility during a 10 mm or less storm event. Finally, the results could suggest the design and maintenance/repair guidelines of an IF facility that could be implemented in Korea.

The IF facility considered in the research could primarily reduce runoff volume, adsorb and filter stormwater pollutants. To evaluate the effects of the facility's volume and NPS pollutants reduction, and the facility's performance in various rainfall events, this research built a test-bed. The test-bed was installed in the landscaping areas within a college campus for feasibility of monitoring so that it can treat rainfall runoff generated from the road surface. The IF facility structure consists of an initial stage sedimentation tank, a trench media tank to remove fine particles, soluble materials and a final stage sedimentation tank (Fig. 1). Table 1 shows the characteristics of the facility, and the facility area is 1.8% of catchment area. The initial sedimentation tank has infiltration function; woodchips and gravels were inserted to decrease the inflow (runoff) velocity and separate large particle matters. The trench media tank has multi-media layers of woodchip, sand and gravel to absorb and remove fine particles and soluble materials and to expand the active place of microorganisms.

Manual grab sampling technique was utilized for all storm events. Runoff samples were collected using a 4-L container in the inflow and outflow part of the IF facility. Four samples were taken every five minutes for the first 15 min with the first sample collected as soon as runoff was evident, and two samples after 30 min and one hour, and more samples hourly thereafter until a maximum of 12 samples. For most of the shorter events, the scheme was modified by adjusting the number of samples until the runoff flow ended (Kim and Kang, 2004a). Continuous measurements were also performed to monitor the inflow and outflow flow rates every five or ten minute interval using a 5-L capacity of graduated measuring container and a timer. The rainfall data were taken from the Korea Meteorological Administration (KMA) with reference from weather stations nearby the monitoring sites. Other in situ data gathered during the monitoring include antecedent dry day (ADD), rainfall duration, average rainfall intensity, and time before effluent starts of the hydraulic retention time (HRT). For volume reduction evaluation, a water balance determination method was used as shown in Equation (1). Cumulative inflow rainfall and runoff were measured, and the difference between rainfall volume and runoff volume was regarded as the volume of infiltration, retention and evaporation.

Where, Vol_in = inflow volume; Vol_out = outflow runoff volume; Vol_intil = infiltration volume; Vol_evop = evaporation volume; Vol_ret = retention volume; Vol_loss = other losses.

For the collected water samples, TSS, COD and metals (Cr, Fe, Ni, Cu, Zn, Cd, Pb) were analysed based on Standard Methods for the Examination of Water and Wastewater (APHA, AWWA, and WEF, 1992). Event mean concentration (EMC) for each monitored rainfall was calculated using Equation (2) (Irish et al., 1998; Sansalone and Buchberger, 1997).

Where, C(t) and Q_TRu (t) are the pollutant concentration and discharge rate at time t, respectively. The NPS pollutant removal efficiency was calculated on a basis of mass reduction as shown in Equation (3).

To evaluate the volume reduction and NPS pollutants removal efficiency of the IF facility, 29 times of monitoring in total were carried out from May 2009 to August 2013. Table 2 summarizes hydrological characteristics of the measured rainfall events. Mean rainfall depth of the 29 rainfall events was 9 mm. This monitoring result is judged to properly reflect climate characteristics of the study area, since 70~80% of Korea's annual rainfall events are small-scale with 10mm and less of rainfall (Maniquiz et al., 2012).

Fig. 2 showed the hydro-pollute graph of inflow and outflow part of the IF facility. The runoff entering the inflow part of the facility from the road showed a typical first flush phenomenon in which highly concentrated pollutants are discharged at the earlier part of a runoff (Kim and Kang, 2004b). The outflow concentration of the pollutants is very low and stable, compared to the inflow. Such characteristic means that the IF facility is properly constructed for pollutant removal. Reduction of water volume and peak flow can be estimated through flow curve change of inflow and outflow. The peak flow of inflow was very high, but the peak flow of outflow was very low. Thus, IF facility can reduce delay runoff time, peak flow time, runoff and peak flow volume.

Table 3 shows the EMC calculation results of the IF facility in the three rainfall ranges (R<5mm, 5mm10mm). The inflow pollutant EMC was 174.6~476.5 mg/L in TSS, 155.6~316.4 mg/L in COD and 66.7~12,438 μg/L in metals. In the rainfall event with 5mm and less, the concentrations of NPS pollutants were high due to the first flush phenomenon. On the other hand, in the rainfall event of more than 20 mm, a dilution effect arises from high rainfall (Wu et al., 1998). The outflow pollutant EMC was 71.8~160.3 mg/L in TSS, 65.4~153.7 mg/L in COD and 85.0~6,703 μg/L in heavy metals. The outflow EMC showed lower concentrations than the inflow EMCs. TSS, COD, Total Fe, Total Zn and Total Pb were highly reduced to more than half of its value in the inflow.

The regression plot displaying the relationship between the discharged and reduced volume with rainfall depth is presented in Fig. 3. The runoff volume reduced by the IF facility was assumed to have infiltrated the ground through the drain pipes, evaporated, and retained or stored in the system. The amount of volume reduced by the IF facility was higher compared with the volume discharged by the system up to approximately 5.5 mm rainfall wherein beyond this value, the percentage of volume discharged by the system was increased with a corresponding decrement in volume reduced by the system. Based on the storm events monitored, for rainfall of less than 5 mm, the system reduced 52% of the total runoff volume which entered the system. Meanwhile, for rainfall between 5 and 10 mm, the mean percentage of runoff volume that was reduced by the system was decreased to 36%. Beyond 10 mm, the average volume which was reduced by the system was further decreased to 26%. Since 70~80% of the total numbers of storm events per year in Korea were mostly below 10~20 mm, the IF facility is appropriate to be applied in Korea (Maniquiz et al., 2010).

Fig. 4 shows average removal efficiency of each pollutants in the IF facility for different ranges of the rainfall depth. The average removal efficiency of pollutants in the IF facility was 83% in TSS, 80% in COD and 67~79% in metals. The removal efficiency was very high, compared to a filtration facility or constructed wetland (MOE, 2014). In the case of 5 mm and less in the rainfall range, more than 80% of removal efficiency was shown in all pollutants. On the other hand, over 10 mm in the rainfall range, at least 60% of removal efficiency was exhibited in all pollutants. This finding suggested that volume reduction through infiltration and retention mechanisms in the facility plays an important role in reducing the pollutant loads from road runoff (Geronimo et al., 2013).

Urbanization arises from many environmental, hydrological and ecological problems such as distortion of the natural water circulation system, increase in nonpoint source pollutants in stormwater runoff, degradation of surface water quality, and damage to the ecosystem. Due to the increase in impervious surface by urbanization, developed countries apply low impact development (LID) techniques as an important alternative to reduce the impacts of urbanization. Therefore, this research aims to evaluate the applicability of an infiltration and filtration (IF) facility in various rainfall events based on the obtained volume and pollutant reduction capabilities. The following conclusions were drawn through this research: