Hydrological Performance Assessment of Low-Impact Development Practices: A Vegetated Swale Case Study

Abstract

The increase in rainfall intensity caused by climate change, combined with high levels of urban soil sealing and the limited capacity of drainage systems, is significantly increasing flooding risk. Integrated stormwater management is a key factor in dealing with the consequences of climate change by mitigating flood risk based on nature-based solutions. An experimental site was designed to assess the hydrological response of vegetated swales depending on different catchment surfaces, and to evaluate the impact of spatial variability of the hydraulic conductivity (K). A hydrological model was developed on EPA SWMM and calibrated based on measured data in two experimental swales with different active surfaces (N6—50 m² and N11—100 m²). The model validation was assessed with NSE scores higher than 0.7. The simulations considered various factors, such as the water level in the swales, the actual rainfall, the evapotranspiration, the swale geometry, the catchment area (Sa), and the hydraulic conductivities of the natural silty soil, to assess, for the heaviest rainfall event, the best swale morphological characteristics. The study examined the combined impact of K (8) and Sa (6) on swale storage capacity, designed with a 250 mm depth. The simulations showed that the 250 mm overflow limit was exceeded for N10 (90 m²) when K ≤ 2 × 10⁻⁶ m/s, and for N11 (100 m²) when K ≤ 4 × 10⁻⁶ m/s. These results provide valuable information on the optimal storage capacity based on the swales’ geometrical and physical properties.

1. Introduction

Climate change is a major challenge for the environment [1]. It is disrupting hydrological cycles and impacting rainfall frequency and intensity [2], thereby increasing natural hazards such as flooding, mudslides, and landslides [3].

Stormwater management is a critical issue when it comes to dealing with the destructive effects of heavy rainfall [4]. Today, stormwater management relies mainly on collection and treatment at wastewater treatment plants (WWTPs) [5]. This conventional approach, based on combined or separate sewer systems, leads to high energy consumption at WWTPs [6]. Moreover, sewer networks require regular maintenance and monitoring to prevent network leak contaminations [7,8]. In addition, the significant amount of urban soil sealing amplifies peak runoff, particularly considering the combination of topography and urban growth, combined with the more frequent and intense rainfall, posing a growing risk of urban floods [9]. Combined sewer overflows, storage reservoirs, and retention basins are built to mitigate flooding risk during heavy rainfall events [10,11]. However, these structures need extensive land areas, significant earth-moving operations, and the transportation of substantial amounts of building materials. Hence, a conventional grey infrastructure approach is ineffective to address the climate and environmental challenges [12].

To address these climate change-related challenges, integrated stormwater management relies on a holistic and sustainable approach [12,13]. Several concepts including “Low-Impact Development (LID)”, “Best Management Practices”, “Sponge City”, “Sustainable urban Drainage Systems”, and “Water Sensitive Urban Design” are used to describe integrated stormwater management [14]. LID practises enable multi-objective stormwater management, playing a crucial role in flood risk mitigation and runoff control, decreasing peak volumes at wastewater treatment facilities, incorporating biodiversity and green spaces into urban areas, and enhancing the life quality by mitigating urban heat islands and improving air quality [15,16]. Nature-based solutions, such as infiltration swales, green roofs, and bio-retention/biofiltration basins, serve several functions: stormwater runoff control, groundwater infiltration, sediment and pollution removal, and water transport [17]. LIDs’ hydrological performances, including, for instance, their runoff reduction and leakage rates, have been extensively examined in the literature [18,19,20,21]. Additionally, their effectiveness in runoff pollution and contaminant removal has been investigated by [22,23,24,25]. Furthermore, LIDs are cost-effective and have a limited carbon footprint compared to conventional sewage system approaches [26,27,28], thereby meeting sustainable development goals while considering the geographical limitations in an urban setting [29].

Nevertheless, several challenges limit the broad adoption of LIDs, especially in low hydraulic conductivity areas, such as silty or clayey soils. French guidelines suggest the use of drainage systems in infiltration swales for low hydraulic conductivity soils [30,31]. Consequently, it is crucial to empirically assess the hydrological functioning of swales in these regions to evaluate the impact of the factors affecting the performance of these green infrastructures [32] and enhance design guidelines concerning swales.

The experimental site “ETAGEP” has been designed to address these challenges. The main purpose of the experimental site is to study the hydrological response of vegetated swales by evaluating the influence of hydraulic conductivity and impervious areas. Additionally, it aims to improve LIDs’ hydrological performance by analyzing the influence of fauna and flora on hydraulic conductivity.

This research focuses on the following:

• Evaluating the physical and hydrological properties on the experimental site;
• Assessing the hydrological response of the vegetated swales by calibrating and validating a numerical model;
• Exploring the combined effect of the catchment area and soil hydraulic conductivity on the swales’ hydrological functioning, and identifying the optimal swale geometrical parameters to avoid overflow.

2. Materials and Methods

2.1. Site Description

The ETAGEP experimental site is located on the “Cité Scientifique” campus of the University of Lille, commune of Villeneuve d’Ascq, Métropole Européenne de Lille (MEL), Hauts-de-France region (Figure 1). The research site’s GPS location is 50°36′52.138″ N, 3°8′35.395″ E, and its elevation is 35 m. The climate zone is classified as oceanic “Cfb” [32]. The mean annual precipitation in the MEL region is 740 mm, with an average of 126 rainy days per year, according to the average data from 1991 to 2020 [33].

The experimental site covers an area of 2500 m² (Figure 1). It consists of 12 vegetated swales, each 5 m in length and 2 m wide (a total area of 10 m² for each swale). The storage depth of each swale is 0.25 m, corresponding to a maximum storage capacity of 1.68 m³. To model various urban catchment situations, the swales gather rainwater from two distinct impervious surfaces. The N6 swales (Figure 2) collect rainwater from an impervious area of 50 m², and the N11 swales (Figure 3) collect rainfall from an impervious area of 100 m² (as shown in

Table 1. Total catchment area of the N6 and N11 swales.

Area	N6	N11
Impervious area	50 m2	100 m2
Swale area	10 m2	10 m2
Total catchment	60 m2	110 m2

).

2.2. Site Monitoring

Each vegetated swale is equipped with a water level pressure sensor (Seba DS22) that provides continuous measurements of the water level in the swale, with a time interval of 10 min and a high level of accuracy (0.05%). The sensors have been precisely calibrated to determine the exact position of the bottom of the ditch (Figure 4).

The experimental site is equipped with a meteorological station that includes a bucket rain gauge with a precision of 0.1 mm (Figure 5). The data are recorded at regular intervals of 10 min. Additional meteorological factors that are measured include air temperature, air humidity, wind speed, atmospheric pressure, and net radiation. The meteorological parameters are used to compute the potential evapotranspiration in the swales. The weather data were validated using the meteorological data from the Météo France “Lille-Lesquin” station [34], situated 7 km southwest of the experimental site.

The measured data on the site are stored in four dataloggers and sent remotely to the server, ensuring continuous data verification.

2.3. Geology and Soil Characterization

The geological structure of the northern region of France is characterized by the presence of Quaternary clayey and sandy silty layers [35] that overlay Senonian chalky layers from the Cretaceous period [36,37,38]. The chalk aquifer serves as a primary water source for the region’s drinking water supply [39,40]. The water table is located at a depth of 6 m; therefore, the swales are not affected by the water table (Figure 6).

Three samples collected from the experimental site were analyzed using a “Mastersizer 3000” laser particle size device to obtain a detailed characterization of soil texture. The analysis revealed a silty soil texture [41] as indicated in

Table 2. Sample particle size distributions.

	Sample 1	Sample 2	Sample 3
Sand %	17.9	13.2	13.9
Silt %	78.9	83.5	82.5
Clay %	3.2	3.3	3.6
Soil texture	Silt	Silt	Silt

.

2.4. Hydraulic Parameter Characterization

The soil hydraulic conductivity characterization was assessed using two approaches: a pedotransfert functions-based method (Rosetta), and an on-site hydraulic conductivity survey (Matsuo).

The initial approach relies on a soil texture-based estimation of the hydraulic conductivity. This approach relies on the USDA’s Rosetta tool [42,43], using pedotransfer functions derived from neural network predictions to estimate the hydraulic conductivity [44]. The hydraulic conductivity estimation varies between 7.05 × 10⁻⁶ m/s and 8.06 × 10⁻⁶ m/s, depending on sample’s soil texture (

Table 3. Estimated hydraulic conductivity.

Hydraulic Conductivity	Rosetta Sample 1	Rosetta Sample 2	Rosetta Sample 3	Matsuo In Situ
K (mm/h)	29.02	25.66	25.38	23.59
K (m/s)	8.06 × 10−6	7.13 × 10−6	7.05 × 10−6	6.55 × 10−6

). These values are in accordance with the typical silt hydraulic conductivity values, ranging from 10⁻⁵ to 10⁻⁷ m/s [45].

The second approach is the “Matsuo” percolation test. This infiltration test is recommended in the French technical guides for vegetated swale design [46]. This test is recommended for closely matching the swale morphological characteristics. This model estimates the hydraulic conductivity as follows:

$$K={\displaystyle \frac{-C}{60\times t}}\times \ln \left({\displaystyle \frac{h+C}{h_0+C}}\right)$$

(1)

$$C={\displaystyle \frac{L\times l}{2\times (L+l)}}$$

(2)

where K (m/s) is the hydraulic conductivity, C (m) is a geometric parameter, L (m) is the ditch length, l (m) is the ditch width, h₀ (m) is the initial water level, and h (m) is the measured water level.

A 1.2 m long, 0.7 m wide, and 0.70 m deep ditch was excavated for this test. The trench was filled with water to saturate the soil. Then, the test began with an initial water level of 0.45 m. Based on the calculations, the estimated hydraulic conductivity is 6.55 × 10⁻⁶ m/s (Table 3).

2.5. EPA SWMM Model

EPA SWMM is widely used as a numerical simulation tool for evaluating the performance of LIDs [47,48]. This software offers a LID tool that allows for the configuration of the geometric and hydraulic parameters based on the chosen LID simulation. More specifically, the software includes a “Vegetative Swale” option to simulate swale hydrological performance. This tool has been extensively studied to assess the hydrological response of vegetated swales in the context of urban planning [49,50,51,52]. SWMM calibration and sensitivity analysis models were studied by researchers to enhance the model performances to simulate LID hydrological responses [53,54,55].

The initial model parameters corresponding to the characteristics of the experimental swales are shown in

Table 4. Initial SWMM parameters.

Layer	Parameter	Swale
	Berm height (mm)	250
1	Vegetation volume fraction	0.02
	Surface roughness (Manning)	0.01
	Surface slope (%)	0.01
	Swale side slope	3

.

The infiltration model chosen is the Green-Ampt model [56]. This model considers the flow in the soil following a homogeneous and continuous moisture front. It allows for the evaluation of soil infiltration based on the load and suction as follows:

$$f\left(t\right)=Ks\left(1+{\displaystyle \frac{h_0+h_s}{L}}\right)$$

(3)

where f(t) is the potential infiltration rate, Ks the saturated hydraulic conductivity, h₀ is the pressure head, and h_s is the suction head at the wetting front.

The swale vegetation is designed to improve the hydrological response through evapotranspiration. The potential evapotranspiration estimation for green infrastructure have been extensively studied [57,58]. The chosen evapotranspiration model is the Penman Monteith FAO-56 physical model [59]. This model evaluates the potential evapotranspiration by considering the presence of grass with an albedo of 0.23, which is representative of vegetated swales.

This model estimates the potential evapotranspiration as follows:

$${ET}_0={\displaystyle \frac{0.408\times ∆\times \left(Rn-G\right)+\gamma \times \frac{900}{T_{moy}+273}\times {ff}_{2m}\times \mathrm{m}\mathrm{a}\mathrm{x}(e_s-e_{a }; 0)}{∆+\gamma \times (1+0.34\times { ff}_{2m})}}$$

(4)

where ff (m/s) is wind speed, Tmoy (°C) the mean temperature, Rn (MJ/m²/day) is the net radiation, ∆ and γ are parameters based on temperature and atmospheric pressure.

The potential evapotranspiration was calculated using the meteorological data from the site’s weather station. The estimated potential evapotranspiration values have been implemented into the SWMM model.

2.6. Model Performance Index

The NSE indicator (Equation (5)) is widely used to assess hydrological model performance [60,61]. This performance index is used for LID simulation performance analysis [62]. The models’ performances were computed on Python.

$$NSE=1-{\displaystyle \frac{{\sum }_{t=1}^T(H_{obs}^t-H_{sim}^t)2}{{\sum }_{t=1}^T(H_{obs}^t-\overline{H_{obs}})2}}$$

(5)

The calculation of water level error and drainage time error are evaluated as follows:

$$H_{\% error}={\displaystyle \frac{H_{max}obs-H_{max}sim}{H_{max}obs}}$$

(6)

$$T_{\% error}={\displaystyle \frac{T_{vidange}obs-T_{vidange}sim}{T_{vidange}obs}}$$

(7)

3. Results

3.1. Observed Data

The study focused on the 09/20/2023 to 10/22/2023 period. This period was characterized by four heavy rainfall events (Figure 7), resulting in a total cumulative rainfall of 79 mm, with a major event of 31.5 mm. The mean and maximum intensity of the rainfall events are detailed in

Table 5. Rainfall events with their mean and maximum intensities.

Day	Rainfall	Duration	Mean Intensity	Max Intensity
21 September	31.5 mm	14.5 h	2.2 mm/h	27.0 mm/h
13 October	10.1 mm	8.2 h	1.2 mm/h	27.0 mm/h
18 October	9.8 mm	4.7 h	2.1 mm/h	7.8 mm/h
20 October	10.8 mm	10.2 h	1.1 mm/h	4.8 mm/h

.

During this study period, the potential evapotranspiration, as determined by the Penman Montheith FAO-56 model, did not exceed a daily value of 2 mm (Figure 8).

The hydrological response needs to meet specific geometric and regulatory requirements. These include ensuring that the maximum level does not exceed the overflow limit (250 mm) and that the draining time is lower than 48 or 72 h, depending on the guidelines [30].

The swales stored and managed variable volumes during this period, depending on the impervious areas. The September 21st rainfall event was the heaviest. During this rain event, rainfall reached a cumulative height of 31.5 mm in 14.5 h, resulting, for the N11 swale, in a peak water level of 229 mm (Figure 9), corresponding to a volume of 1.49 m³. For the N6 swale, the maximum water level reached 153.1 mm (Figure 10), corresponding to a volume of 0.81 m³ during this rainy episode.

Three other heavy rainfall events occurred during the study period, varying between 9.8 mm and 10.8 mm. Consequently, these events generated significant runoff in the two swales. A peak water level value of 125.8 mm (Figure 10) was reached on October 13 for N11, corresponding to a volume of 0.61 m³. In N6, the peak water level was 58.6 mm (Figure 9), corresponding to a volume of 0.19 m³.

The drainage times during the four rainfall events were short, not exceeding a duration of 9 h and 40 min for the N11 swale and 8 h and 20 min for N6 during the heaviest rain event of 21 September.

3.2. SWMM Simulation

SWMM v.5.2.4 was used to carry out the simulations. Adjustments were made during the calibration process to enhance the performance of the SWMM model. The hydraulic conductivity, swale side slope and vegetation rate (indicated in Table 4) were iteratively modified to achieve a satisfactory model calibration.

The swale water level simulations provided prediction results that closely matched the observed values. The NSE indicator results validate the model’s ability to accurately simulate the hydrological responses of the swales. The NSE scores achieved during the calibration phase were 0.89 for N6 and 0.95 for N11 (

Table 6. Calibration and validation NSE scores.

Swale	Calibration NSE	Validation NSE
N6	0.89	0.76
N11	0.95	0.73

). During this phase, adjustments were made to the hydraulic conductivity. The adjusted hydraulic conductivity values for N6 and N11 were estimated as 11 mm/h and 27 mm/h respectively. The validation phase confirmed the model’s performance, achieving an NSE score of 0.76 for N6 and 0.73 for N11 (Table 6).

Figure 11 and Figure 12 show the measured and simulated water levels during the calibration process for the N6 and N11 swales.

Figure 13 and Figure 14 show the measured and simulated water levels during the validation process for the N6 and N11 swales.

The simulation results were compared to the observed water level (

Table 7. Water level simulations and errors.

Swale	Rainfall	Observed Peak Level	Simulated Peak Level	Peak Level Error
N6	31.5 mm	153.1 mm	164.2 mm	−7.3%
	10.1 mm	58.6 mm	77.4 mm	−32.1%
	9.8 mm	57.2 mm	67.1 mm	−17.3%
	10.8 mm	42.4 mm	63.0 mm	−48.6%
N11	31.5 mm	229.5 mm	225.6 mm	1.7%
	10.1 mm	125.8 mm	117.3 mm	6.8%
	9.8 mm	96.7 mm	96.4 mm	0.3%
	10.8 mm	73.9 mm	77.0 mm	−4.2%

) and draining time (

Table 8. Draining time simulations and errors.

Swale	Rainfall	Observed Emptying Time	Simulated Emptying Time	Draining Time Error
N6	31.5 mm	8h20	9h10	−10.0%
	10.1 mm	4h20	6h20	−46.2%
	9.8 mm	3h50	4h10	−8.7%
	10.8 mm	5h20	6h40	−25.0%
N11	31.5 mm	9h40	9h30	1.8%
	10.1 mm	2h50	4h10	−47.3%
	9.8 mm	2h20	2h50	−21.5%
	10.8 mm	3h10	3h00	5.3%

) values of the experimental swales. For N11, the water level error is satisfactory, with a maximum peak level calculated error of less than 7%. N6’s simulated peak water levels are accurate for the heaviest rainfall event (7.3%). However, the simulated peak water level precision is lower for observed peak levels lower than 60 mm (from 17.3% to 48.6%). The errors can be attributed to slight differences in the morphological characteristics in the swales’ bottom surface.

The simulated swales’ draining time is reliable for high rainfall mean intensity events (09/21 and 10/18 with a mean intensity of 2.2 mm/h and 2.1 mm/h respectively). During these rainfall events, the maximum calculated draining time error was 10.0% for N6 and 21.5% for N11 (Table 8). The draining time error was higher for the 10/13 rainfall event, with a calculated error of 46.2% for N6 and 47.3% for N11.

3.3. Combined Effect of Hydraulic Conductivity and Impervious Areas on Swale Hydrological Responses

The hydrological response is influenced by the catchment area and the soil hydraulic conductivity. To assess the impact of these factors, we conducted simulations for the heaviest rainfall event.

Six impervious areas (

Table 9. Simulated impervious surfaces.

Swale	Total Impervious Surface
N6	50 m2
N7	60 m2
N8	70 m2
N9	80 m2
N10	90 m2
N11	100 m2

) and eight hydraulic conductivities were simulated. During the simulations, we systematically adjusted the hydraulic conductivity within the range of 1 × 10⁻⁶ m/s to 8 × 10⁻⁶ m/s. These values are commonly observed in silty soils. We incremented the conductivity by 1 × 10⁻⁶ m/s in each simulation. Neither the morphological properties of the swales nor the calibration parameters were modified.

Peak water levels were simulated in SWMM by systematically modifying the hydraulic conductivity and the impervious area. Subsequently, we constructed linear regression curves for every catchment surface scenario (Figure 15).

The simulations provide insights into the impacting factors of overflow. The swales with smaller impervious surfaces (N6 to N9) did not reach the critical overflow level of 250 mm, despite having the lowest simulated hydraulic conductivity of 1 × 10⁻⁶ m/s. However, for N10 (90 m²) or N11 (100 m²), the critical water level was attained with hydraulic conductivities below 2 × 10⁻⁶ m/s and 4 × 10⁻⁶ m/s, respectively.

To assess the most appropriate geometry for managing this rainfall, we evaluated the peak water level for the N10 and N11 scenarios (Figure 16). This allowed us to identify the optimal swale geometry based on the assessed parameters.

To prevent any overflow during the simulated rainfall event, for a hydraulic conductivity of 1 × 10⁻⁶ m/s, the N10 storage depth needs to be set to 258 mm, which is 4% higher than the designed depth. N11 requires a storage depth of 272 mm, which is approximately 9% higher than the swale’s actual storage depth.

4. Discussion and Future Work

4.1. Hydrological Response Simulation

Effective sustainable stormwater management involves understanding several factors such as the impervious area, hydraulic conductivity, geometric parameters, and weather conditions. The objective of managing these parameters is to optimize the hydrological functioning of infiltration systems, such as swales. To analyze the performance of the instrumented swales N6 and N11, the SWMM model was adjusted using real rainfall data from four heavy rainfall events. During the calibration, the model’s generated hydraulic conductivity was adjusted iteratively to enhance calibration performance. The calibration confirmed the observed spatial variance in hydraulic conductivity in the two swales. As an illustration, N6 provides an estimated hydraulic conductivity of 11 mm/h (equivalent to 3.06 × 10⁻⁶ m/s) for a measured value on site of 23.6 mm/h (equivalent to 6.55 × 10⁻⁶ m/s). In contrast, the estimated hydraulic conductivity of N11 was 27 mm/h, which closely matched the measured value. Conducting laboratory investigations, such as oedometer testing, is crucial for comprehending the elements that affect the spatial variability of hydraulic conductivity, specifically soil compaction [63,64].

The validation phase reflected the model’s performance, with an NSE value of 0.76 for N6 and 0.73 for N11. To improve the model’s performance, measured data, such as that of the water level, from our experimental swales need to be implemented to provide a more comprehensive understanding of the variability and performance of vegetated swales under different conditions, such as the catchment area and hydraulic conductivity. The model should employ innovative tools for optimal calibration and sensitivity analysis, such as harmony-based and copula-based calibration methods [53,54]. To enhance the model’s performance, the study duration needs to be extended, by integrating longer time series, a wider range of weather conditions, and different rainfall intensities, especially heavy rainfall events higher than 10 mm; heavy rainfall events are relevant for assessing swale overflow, and therefore assess the optimal swale morphological characteristics. Another factor to take into consideration for improving the model is to evaluate the temperature effect on the hydraulic conductivity: a low temperature decreases the hydraulic conductivity, hence impacting the hydrological performance [64].

The simulation of the swales’ response to the heaviest rainfall event (31.5 mm in 14 h) was conducted by considering the combined effects of hydraulic conductivity and impervious surfaces. The simulations explore a range of permeability values, varying from 1 × 10⁻⁶ m/s (equivalent to 3.6 mm/h) to 8 × 10⁻⁶ m/s (equivalent to 28.8 mm/h), and impervious areas, ranging from N6 (50 m²) to N11 (100 m²). These results enable the identification of the finest optimal geometry for sizing the channels, while minimizing the impact on the land. Spatial allocation optimization indexes, suggested by [65], need to be studied based on these results to enhance our understanding of spatial catchment area optimization.

4.2. Hydrological Performance Enhancement

This study does not consider the impact of fauna and flora on the experimental swales. Researchers studied the influence of macrofauna and vegetation on soil water infiltration [64,66]. For this purpose, earthworms and specific vegetations will be implemented in the swales to investigate fauna and flora impacts on enhancing the swales’ hydrological performance.

The hydrological response enhancement can be evaluated by non-invasive techniques specifically designed to assess the influence of fauna and flora [67] developed a Ground Penetrating Radar (GPR)-based model to investigate the influence of vegetation roots and earthworms on soil texture. The established model provides a correlation between the GPR scan signal and earthworm-induced soil hydraulic conductivity enhancement. This GPR model will be tested on the experimental site to investigate the impact of earthworm activity. The findings (the combined effect of macrofauna and flora) will help in overcoming the guideline limitations related to low hydraulic conductivity soils and boarding vegetated swale deployment.

5. Conclusions

This research examines the hydrological performance of vegetated swales. The implementation of these LID solutions is crucial for achieving sustainable stormwater management as they play an essential role in flood risk mitigation, groundwater recharging, and soil unsealing.

A SWMM model was developed to evaluate the hydrological response of two experimental swales by adjusting the model using experimental measurements at the ETAGEP site. The model was initially calibrated based on rainfall, evapotranspiration, hydraulic and geometric parameters, and impervious surfaces. The hydrological model’s performance was assessed during the validation phase, yielding an NSE score of 0.76 for N6 and 0.73 for N11.

The combined effect of hydraulic conductivity and impervious areas was evaluated. The simulations were employed to emphasize the crucial thresholds that result in overflow, as well as to determine the most effective shape for vegetate swale designs to control stormwater through storage, infiltration, and evapotranspiration.

Future work will further explore several aspects to enhance the hydrological model’s performance. Longer time series and heavy rainfall events will be studied. More measured water level data from other experimental swales will be added to improve our model. The effect of physical parameters such as temperature on hydraulic conductivity needs to be examined. The implementation of innovative calibration tools will be incorporated for model optimization. Lastly, there will be large-scale implementation to emphasize integrated stormwater planning strategies for urban flooding risk mitigation, specifically for low hydraulic conductivity areas.