The present study aims to evaluate the performance of the impervious surface as a control (O), sandy loam substrate (SL), gravel (G), gravel with geocell layer (GGE), rosemary (R), rosemary with geocell layer (RRE), turf (T), and turf with geocell layer (TGE) in the reduction of runoff volume, time-to-start runoff (TR), runoff coefficient (C), time-to end runoff (TER), peak flow rate (PF), time to peak (TP), and time base (TB) in the laboratory dimension under three different scenarios of rainfall intensity and two different slopes using a rainfall simulator. The results revealed a significant difference between the data at the level of 5% in all cases. Generally, three rainfall scenarios for all hydrological parameters TR, TER, TP, TB, C, and PF were classified into different groups. In all cases, GGE treatment performed better than that of the rest of the test groups in reducing runoff and cumulative volume. Further, treatments O and GGE experienced the highest and lowest flow rates, respectively. For a specific scenario of rainfall intensity and slope, the value of C is the lowest for GGE treatment. Finally, the implementation of geocell in the pavements was able to delay the time to start runoff.

  • Examining different rainfall scenarios can be effective in flood control on different lid methods.

  • Implementation of geocell to cover gravel, rosemary and turf can be effective in reducing floods.

  • Gravel with geocell layer (GGE) had the best performance in reducing floods and improving parameters affecting it.

Urbanization and the construction of buildings, pavements, roofs, and other surfaces have limited natural infiltration, as well as evaporation and transpiration. Therefore, these changes in land use lead to the rapid production of surface runoff, which is a significant threat to urban areas where millions of people live (Saraçoğlu & Kazezyılmaz-Alhan 2023). Low-impact development (LID) is an engineering method to reduce the negative effects of urbanization and protect water resources (Demirezen & Kazezyılmaz-Alhan 2022). LID is a design philosophy that provides flood management methods, plans, and technologies to minimize negative impacts commonly associated with urban flooding, including degradation of groundwater and surface water quality, erosion, and loss of aquatic diversity (Drake et al. 2013). The purpose of LIDs implementation is to simultaneously reduce runoff and flooding and improve water quality, in addition to decreasing costs in the construction, operation, and maintenance of infrastructure (Rodrigues et al. 2021).

Infiltration trenches (IT) and vegetated filters are two LID-best management practice (LID-BMP) methods that can provide nature-based solutions to control runoff volume (RV), time-to-start runoff (TR), peak flow rate (PF), and removal of stormwater pollutants (Dadrasajirlou et al. 2023).

As part of nature, turfs (T) are low-cost structures (Deletic 2005) and the use of turfs decreases the total volume of runoff through infiltration and storage. Further, the sand structure of the infiltration trench can reduce the volume of surface runoff and the amount of PF rate and increase the groundwater supply (Fach & Dierkes 2011). The field research reviewed in the following reveals the proper effectiveness of turf and IT in reducing RV. Bäckström (2002) reported a 33% reduction in average surface runoff after observing the performance of seven turfs in urban areas of Sweden. Fassman & Liao (2009) and Rujner et al. (2016) evaluated turf performance and showed that swale implementation leads to a reduction in average surface runoff by 63.7% and 40–55%, respectively. Hunt et al. (2010) monitored the hydrological performance of a surface plant spreading filter strip and found that this strip could eliminate the outflow of 20 out of 23 rainfall events and reduce the RV by 85%. Similarly, Line & Hunt (2009) investigated the turf filter surface in 14 rainfall events and reported a reduction of RV and PF rates by 49 and 23%, respectively. In some cases, peak currents are reduced by a high order of magnitude (Winston & Hunt 2009). For example, the efficiency and effectiveness of IT and grains of sand in flood control and reduction have been mentioned in research Heilweil et al. (2015) and Barber et al. (2003). The geocell (GE) family of pavement materials is defined by constructed plastic grids whose cells can be filled with aggregate or a planting substrate for turf (Ferguson 2005; Papakos et al. 2010).

Geocells are plastic panels that form a grid of box-like cells filled with soil material (Ferguson 2005). Dreelin et al. (2006) compared the performance of an asphalt parking lot and a porous pavement parking lot of turf with geocells in Athens (Georgia, USA) during rainfall events with low rainfall duration and intensity. The results indicated that porous pavements are a suitable option for reducing runoff and some pollutants caused by small floods in clay soils. By investigating the combination of different pavements with geocell, Rosen (2013), Sato & Kojma (2018) and Wesseloo et al. (2009) found that these compounds can be effective in slope protection and runoff and erosion control. In another similar research, Song et al. (2021a) used geocell structure to strengthen soil for growing vegetation and improving slope stability. The composition of geocell and soil was investigated in two small and large sizes and two slopes under three different rainfall intensities. The results were similar to the research Rosen (2013), Sato & Kojma (2018), Song et al. (2021b) and Wesseloo et al. (2009) and highlighted the appropriate performance of the geocell. In total, limited studies have been conducted in the field of combining geocells with plants in order to control and reduce the volume of urban runoff.

In past research, it seems that no study has been done on rosemary plants in the field of flood control methods. Also, the effect that this plant can have on hydrological parameters is not seen in past studies. The creation of variable rainfall intensity during the experiment (during rainfall) based on natural rainfall has been less seen in past research in the experimental dimension. In general, studies highlight the importance of developing different methods for urban runoff and flooding control in all parts of the world. Despite conducting numerous research on IT and turf plants in the field of runoff reduction, most research is related to field studies and few studies have been conducted on a laboratory scale to carefully evaluate the hydrological parameters of LIDs such as IT and turfs. The investigation of cloudbursts shows that the rainfall intensity varies during a rainfall event such that the maximum rainfall intensity may be several times the average rainfall of that cloudburst. Considering different scenarios for variable rainfall intensity, this research first introduces the rosemary plant as a new LID to check its efficiency in the field of reducing runoff in a laboratory using a rainfall simulator. To this aim, similar tests were performed for turf pavements and IT, and the results of their hydrological parameters were compared with those of the rosemary plant. Then, we evaluated the effect of combining the mentioned LIDs with the geocell layer in reducing runoff. Given the lack of testing and comparing effective hydrological parameters with control treatments in most studies, this research seeks to compare the hydrological parameters of time-to-start runoff (TR), time-to-end runoff (TER), time-to-peak (TP), time base (TB), runoff coefficient (C), and PF rate for treatments of infiltration trench, turf, and rosemary (with and without geocell) with an impervious surface (control) and bed soil using a rainfall simulator.

Model's explanations

In this research, 48 physical models were created at the laboratory level and on a small scale. Physical models were then classified into eight groups as follows: impervious pavement as the control (O) treatment, substrate soil (SL), gravel (G), gravel with geocell (GE), rosemary (R), rosemary with geocell (RGE), turf (T), and turf with geocell (TGE). Further, two types of slopes (0 and 5%) were considered for the treatments based on research (Li et al. 2020; Hou et al. 2022). To the best of our knowledge, limited studies have seemed to be conducted on different rainfall scenarios. Variable rainfall intensity may allow surface basins to drain to some extent before rain resumes. Therefore, experiments need to address variable rainfall intensity with a single peak. In this study, three different scenarios of rainfall intensity during rainfall were considered based on the rainfall conditions and the intensity–duration–frequency (IDF) diagram of the cities of Semnan, Tabriz, and Rasht (Karami et al. 2023; Samii et al. 2023). This method was considered based on research (Stone & Paige 2003; Parsons & Stone 2006). Considering the return periods of 5 and 10 years and time of 10 min, rainfall intensities of 45 and 55 mm/h were addressed as Semnan's IDF, rainfall intensities of 70 and 90 mm/h as Tabriz's IDF, and rainfall intensities of 170 and 200 mm/h as Rasht's IDF. This method was used in the research (Tasalloti et al. 2020). Each experiment was repeated three times simultaneously. Table 1 presents the details of the experiments. Figure 1 illustrates three different scenarios considered for the rainfall intensity. The duration of rain for each rainfall intensity was considered 15 min in all scenarios. Based on the table, six different modes of rainfall intensity and mentioned slopes were considered for testing in each test group.
Table 1

Test details

Sample testModels abbreviationSlope (%)Number of rainfall intensity scenarios
Control (without geocell) O1 
O2 
O3 
O4 
O5 
O6 
Base soil (sandy loam) SL1 
SL2 
SL3 
SL4 
SL5 
SL6 
Gravel G1 
G2 
G3 
G4 
G5 
G6 
Gravel with geocell GGE1 
GGE2 
GGE3 
GGE4 
GGE5 
GGE6 
Rosemary R1 
R2 
R3 
R4 
R5 
R6 
Rosemary with geocell RGE1 
RGE2 
RGE3 
RGE4 
RGE5 
RGE6 
Turf T1 
T2 
T3 
T4 
T5 
T6 
Turf with geocell TGE1 
TGE2 
TGE3 
TGE4 
TGE5 
TGE6 
Sample testModels abbreviationSlope (%)Number of rainfall intensity scenarios
Control (without geocell) O1 
O2 
O3 
O4 
O5 
O6 
Base soil (sandy loam) SL1 
SL2 
SL3 
SL4 
SL5 
SL6 
Gravel G1 
G2 
G3 
G4 
G5 
G6 
Gravel with geocell GGE1 
GGE2 
GGE3 
GGE4 
GGE5 
GGE6 
Rosemary R1 
R2 
R3 
R4 
R5 
R6 
Rosemary with geocell RGE1 
RGE2 
RGE3 
RGE4 
RGE5 
RGE6 
Turf T1 
T2 
T3 
T4 
T5 
T6 
Turf with geocell TGE1 
TGE2 
TGE3 
TGE4 
TGE5 
TGE6 
Figure 1

Histogram of different rainfall intensities during the duration of rainfall, (a): Scenario 1, (b): Scenario 2, and (c): Scenario 3.

Figure 1

Histogram of different rainfall intensities during the duration of rainfall, (a): Scenario 1, (b): Scenario 2, and (c): Scenario 3.

Close modal

Rainfall simulator

The rainfall simulator has two rain jets, along with a pan (basin) to place laboratory treatments (Ghazvinian & Karami 2023). These basins can change the slope from 0 to 15% and are located at a height of 70 cm above the ground and on metal bases. Basins are made of galvanized iron sheets and their dimensions are with a height of 50 cm. At a height of 30 cm from the basin, a pipe was placed to flow rate the runoff formed on the treatment, while graduated buckets were placed below these pipes to read the height and volume of the output water. The rain jets are located at a height of 2.5 m from the surface of the treatment basin. These jets can change their height, angle, and distance from each other. Moreover, this device can change the rainfall intensity from 20 to 220 mm/h. The device has a pump and a tank, and the water in the tank is pumped into the jets. The rainfall intensity is changed using the rotameter of the device and the taps available for adjusting the flow rate. Figure 2 shows the schematic of the rainfall simulator and its components. In general, the precipitation simulation device is similar to the research of Liu et al. (2020a, 2020b) and Aksoy et al. (2012) designed and implemented.
Figure 2

Details of the rainfall simulator.

Figure 2

Details of the rainfall simulator.

Close modal

Model's structure

Concerning the basins installed in the rainfall simulator, geotextile sheets were considered at the bottom and then, 15 cm of sandy loam soil (substrate soil) was placed at the bottom of the basin. The details of sandy loam soil are 4% clay, 19% silt, and 77% gravel, respectively. It is worth noting that sandy loam soil conditions were considered based on research conducted by Dehghani et al. (2019). A geotextile layer was placed on the sandy loam soil and then, 15 cm of experimental treatments were placed along with their soil. For tests containing geocell, geocell layers were placed in layers of sandy soil or plant soil. After each test, all the soils used in each test were removed from the basin and completely dry soil was used for the next test. Figure 3 shows an example of the structure of the model.
Figure 3

General layering of the model.

Figure 3

General layering of the model.

Close modal

Plants

Rosemary

Rosemary is a plant with fragrant, evergreen, needle-like leaves and white, pink, purple, or blue flowers (de Macedo et al. 2020). This plant belongs to the mint family, which also includes many plants (de Macedo et al. 2020).

Turf

Most grasses are perennial plants, although annual grasses are available in nature. Turf is mostly propagated through seed cultivation; some types of turfs are also propagated by non-seed methods. Among other characteristics of turf is the continuous production of fresh leaves from its growing center, which is performed throughout the growing season. However, the leaves are worn out and destroyed in cases like old age. Regardless of the natural and agricultural conditions, turf grows fast in any case and it is always more than the amount of seed sprayed on the land. Thus, turf has been widely used in green spaces, because of its characteristics such as being uniform, continuous, and having fast growth.

Gravel

In this research, gravel (sand particles) was used to investigate an infiltration trench in the laboratory dimension. Gravel soil details, ≥15, 15–12, 12–9, 6–9, 4–6, and 4 mm, respectively, 100, 27.4, 35.2, 21.1, 15.2, and 1.1%.

Geocell

The geocells used in this experiment were made of high-density polyethylene (Monrose & Tota-Maharaj 2018) and their length and width were equal to those of the basin, i.e. . The length, width, height, and diameter of geocell openings are 100, 100, 5, and 1 cm, respectively (Ferguson 2005). Figure 4 shows the tested treatments.
Figure 4

Tested treatments: (a) O, (b) SL, (c) G, (d) GGE, (e) R, (f) RGE, (g) T, and (h) TGE.

Figure 4

Tested treatments: (a) O, (b) SL, (c) G, (d) GGE, (e) R, (f) RGE, (g) T, and (h) TGE.

Close modal

Statistical analysis

Concerning the purpose of statistical analysis, the values of PF, time-to-start runoff (TR), runoff coefficient (C), TER, time to peak (TP), and TB were analyzed using SPSS21 software after recording the results of measuring the data related to the pavements. In this research, the 3 × 3 factorial method was used for data analysis (Dadrasajirlou et al. 2022). Further, Duncan's method was utilized to categorize the independent data for each of the dependent data (Karami & Ghazvinian 2022). Equation (1) calculates the percentage of RV reduction, in which RRV represents the percentage of RV reduction, indicates the RV in the control treatment, and shows the RV in the geocell treatments (Liu et al. 2020a, 2020b).
(1)

By using a rainfall simulator, this study aims to evaluate the performance of impervious pavement, substrate soil, porous gravel, rosemary, and turf in two conditions with and without geocell in terms of TR, TER, C, PF, TP, and TB and under different rainfall intensities and slopes. To this aim, we measured the effective runoff parameters mentioned in the research (Liu et al. 2020a, 2020b).

Cumulative RV

Figure 5 shows the accumulated RV created from the start to the end of runoff from the basin for treatments O, SL, G, GE, R, RGE, T, and TGE. As shown, GGE performed better than the other test groups in reducing runoff and cumulative volume in all situations. However, the final value of the cumulative RV in the GGE and G treatments was the same only in scenario 1 with a slope of 5% (Figure 5(b)). Treatment O experienced the highest cumulative RV. In a certain rainfall intensity scenario, changing the slope from 0 to 5% increases the cumulative RV in all treatments O, SL, G, GE, R, RGE, T, and TGE at the end of the experiment. The diagram of cumulative RV in all treatments in scenario 1 differs from scenarios 2 and 3. In other words, the slope of the graph becomes steeper over time in scenario 1, as the rainfall intensity increases from 45 to 200 mm/h, resulting in an increase in the RV during the rainfall period. In scenario 2, variations in the cumulative volume graph follow a steep slope at the beginning of the runoff and gradually convert to a slow slope over time for all treatments, as the rainfall intensity changes from 200 mm/h (the highest rainfall intensity studied) to 45 mm/h (the lowest rainfall intensity studied) during the rainfall duration. Concerning the changes in the cumulative RV diagram at the beginning of the runoff in scenario 3, the slope of the diagram decreases over time. The difference between scenario 3 and scenario 2 is the rainfall intensity at the start of runoff in both scenarios. The reason is that scenario 2 starts with a rainfall intensity of 200 mm/h and over time (every 15 min), the rainfall intensity decreases until it reaches the lowest rainfall intensity (45 mm/h), while the rainfall intensity is 45 mm/s in scenario 3 at the beginning of rainfall, which reaches the rainfall intensity of 55 mm/s after 15 min. Given that scenario 3 starts with the lowest rainfall intensity in this research, the time to start runoff is later for all treatments, except treatment O, compared to scenario 2. Moreover, all the graphs are tangent on the time axis (horizontal axis), which indicates 0 RV. After the end of the rainfall, the slope of the cumulative volume graph decreases and finally becomes 0 (parallel to the time axis) and remains the same until the end of the experiment (t = 120 min).
Figure 5

Diagram of cumulative runoff volume in treatments O, SL, G, GE, R, RGE, T, and TGE: (a) Scenario 1, 0% slope, (b) Scenario 1, 5% slope, (c) Scenario 2, 0% slope, (d)) scenario 2, 5% slope, (e) scenario 3, 0% slope and (g) scenario 3, 5% slope.

Figure 5

Diagram of cumulative runoff volume in treatments O, SL, G, GE, R, RGE, T, and TGE: (a) Scenario 1, 0% slope, (b) Scenario 1, 5% slope, (c) Scenario 2, 0% slope, (d)) scenario 2, 5% slope, (e) scenario 3, 0% slope and (g) scenario 3, 5% slope.

Close modal

Changes in flow rate relative to time

Figure 6 compares flow rate changes over time under different test modes in terms of rainfall intensity and slope scenarios for treatments O, SL, G, GE, R, RGE, T, and TGE. As observed, treatments O and GGE experienced the highest and lowest flow rates, respectively. For both 0 and 5% slopes in scenario 1, the changes in flow rate over time have an increasing trend for all treatments, and an increase in rainfall intensity increases the flow rate with an upward trend. However, scenario 2 follows an opposite trend with respect to scenario 1. That is, the highest flow rate is reported at the initial times of the beginning of the runoff, and the flow rate decreases in a stepwise manner over time, which is due to the decrease in the rainfall intensity at certain times during the rainfall. In scenario 3, since the rainfall intensity in the first, second, third, fourth, fifth, and sixth 15 min is 45, 55, 200, 170, 90, and 70 mm/h, respectively, the flow rate has first an upward trend from the start of runoff to the end of the rainfall in the rainfall intensity of 200 mm/h. Then, the decrease in the rainfall intensity reduces the flow rate in a stepwise trend. Furthermore, the time to start runoff in the control treatment (O) occurred earlier compared to the other treatments. The results of this section, as well as the obtained graphs, are consistent with the research of Vafa (2018). Figure 6 exhibits the trend of flow rate changes over time, which is also observed in research Ahn et al. (2021). Rainfall intensity and duration affect PF reduction, arrival time, ponding depth, and shape of the hydrograph at the exit of the bioretention. Low peak outflow arrival time, high ponding, and high peak outflow are observed under high rainfall intensities. This issue was observed in the research of Gülbaz & Kazezyılmaz-Alhan (2017).
Figure 6

Hydrograph diagram in treatments O, SL, G, GE, R, RGE, T, and TGE: (a) scenario 1, 0% slope, (b) scenario 1, 5% slope, (c) Scenario 2, 0% slope, (d) scenario 2, 5% slope, (e) scenario 3, 0% slope, and (g) scenario 3, 5% slope.

Figure 6

Hydrograph diagram in treatments O, SL, G, GE, R, RGE, T, and TGE: (a) scenario 1, 0% slope, (b) scenario 1, 5% slope, (c) Scenario 2, 0% slope, (d) scenario 2, 5% slope, (e) scenario 3, 0% slope, and (g) scenario 3, 5% slope.

Close modal

Statistical analysis

Table 2 presents the results of the factorial analysis of parameters TR, TER, TP, TB, C, and PF based on independent variables i and s and treatments O, SL, G, GGE, R, RGE, T, and TGE. The results were then categorized using Duncan's method. Table 2 indicates the factorial results of the effect of parameters i and s and sample tests on TR, TER, TP, TB, C, and PF. In all cases, there is a significant difference at the level of 5% between the data. It is worth noting that the t-student method was used for the six dependent parameters TR, TER, C, TP, TB, and PF, as the independent parameter s has been tested at two levels of 0 and 5% in this research. The results revealed no significant difference between the two slopes in the outputs in all six parameters .

Table 2

Results of analysis of independent variables

SourceType III sum of squaresdfMean squareFSig.
TR 239.218 239.218 48,517.408 0.000 
ST 14,024.412 2,003.487 406,341.111 0.000 
RIS 23,520.105 11,760.053 2,385,137.408 0.000 
s * ST 73.410 10.487 2,126.970 0.000 
s * RIS 3.484 1.742 353.296 0.000 
ST * RIS 3,472.517 14 248.037 50,306.085 0.000 
s * ST * RIS 60.338 14 4.310 874.117 0.000 
Error 0.473 96 0.005   
Total 146,921.480 144    
Corrected Total 41,393.958 143    
TER 58.650 58.650 104.797 0.000 
ST 6,359.253 908.465 1,623.265 0.000 
RIS 71.708 35.854 64.065 0.000 
s * ST 128.776 18.397 32.871 0.000 
s * RIS 21.526 10.763 19.232 0.000 
ST * RIS 286.917 14 20.494 36.619 0.000 
s * ST * RIS 136.306 14 9.736 17.397 0.000 
Error 53.727 96 0.560   
Total 1,523,975.950 144    
Corrected Total 7,116.863 143    
588.063 588.063 90,086.170 0.000 
ST 59,721.036 8,531.577 1,306,964.930 0.000 
RIS 29.042 14.521 2,224.511 0.000 
s * ST 112.148 16.021 2,454.292 0.000 
s * RIS 42.012 21.006 3,217.915 0.000 
ST * RIS 428.478 14 30.606 4,688.511 0.000 
s * ST * RIS 67.868 14 4.848 742.632 0.000 
Error 0.627 96 0.007   
Total 356,254.720 144    
Corrected Total 60,989.273 143    
PF 798,908.234 798,908.234 9,337,888.445 0.000 
ST 45,026,996.860 6,432,428.123 75,184,224.820 0.000 
RIS 269,540.004 134,770.002 1,575,233.791 0.000 
s * ST 286,231.903 40,890.272 477,938.243 0.000 
s * RIS 54,817.821 27,408.910 320,363.889 0.000 
ST * RIS 716,681.944 14 51,191.567 598,342.996 0.000 
s * ST * RIS 221,768.369 14 15,840.598 185,149.844 0.000 
Error 8.213 96 0.086   
Total 448,969,873.400 144    
Corrected Total 47,374,953.350 143    
TP 286.456 286.456 38,194.083 0.000 
ST 1,075.318 153.617 20,482.253 0.000 
RIS 97,657.401 48,828.700 6,510,493.370 0.000 
s * ST 719.427 102.775 13,703.374 0.000 
s * RIS 86.805 43.403 5,787.000 0.000 
ST * RIS 1,249.623 14 89.259 11,901.169 0.000 
s * ST * RIS 840.354 14 60.025 8,003.370 0.000 
Error 0.720 96 0.008   
Total 387,348.070 144    
Corrected Total 101,916.103 143    
TB 64.937 64.937 15,082.081 0.000 
ST 17,846.026 2,549.432 592,126.210 0.000 
RIS 22,136.045 11,068.023 2,570,637.500 0.000 
s * ST 184.278 26.325 6,114.302 0.000 
s * RIS 32.998 16.499 3,831.984 0.000 
ST * RIS 4,039.174 14 288.512 67,009.334 0.000 
s * ST * RIS 114.619 14 8.187 1,901.514 0.000 
Error 0.413 96 0.004   
Total 867,687.390 144    
Corrected Total 44,418.490 143    
SourceType III sum of squaresdfMean squareFSig.
TR 239.218 239.218 48,517.408 0.000 
ST 14,024.412 2,003.487 406,341.111 0.000 
RIS 23,520.105 11,760.053 2,385,137.408 0.000 
s * ST 73.410 10.487 2,126.970 0.000 
s * RIS 3.484 1.742 353.296 0.000 
ST * RIS 3,472.517 14 248.037 50,306.085 0.000 
s * ST * RIS 60.338 14 4.310 874.117 0.000 
Error 0.473 96 0.005   
Total 146,921.480 144    
Corrected Total 41,393.958 143    
TER 58.650 58.650 104.797 0.000 
ST 6,359.253 908.465 1,623.265 0.000 
RIS 71.708 35.854 64.065 0.000 
s * ST 128.776 18.397 32.871 0.000 
s * RIS 21.526 10.763 19.232 0.000 
ST * RIS 286.917 14 20.494 36.619 0.000 
s * ST * RIS 136.306 14 9.736 17.397 0.000 
Error 53.727 96 0.560   
Total 1,523,975.950 144    
Corrected Total 7,116.863 143    
588.063 588.063 90,086.170 0.000 
ST 59,721.036 8,531.577 1,306,964.930 0.000 
RIS 29.042 14.521 2,224.511 0.000 
s * ST 112.148 16.021 2,454.292 0.000 
s * RIS 42.012 21.006 3,217.915 0.000 
ST * RIS 428.478 14 30.606 4,688.511 0.000 
s * ST * RIS 67.868 14 4.848 742.632 0.000 
Error 0.627 96 0.007   
Total 356,254.720 144    
Corrected Total 60,989.273 143    
PF 798,908.234 798,908.234 9,337,888.445 0.000 
ST 45,026,996.860 6,432,428.123 75,184,224.820 0.000 
RIS 269,540.004 134,770.002 1,575,233.791 0.000 
s * ST 286,231.903 40,890.272 477,938.243 0.000 
s * RIS 54,817.821 27,408.910 320,363.889 0.000 
ST * RIS 716,681.944 14 51,191.567 598,342.996 0.000 
s * ST * RIS 221,768.369 14 15,840.598 185,149.844 0.000 
Error 8.213 96 0.086   
Total 448,969,873.400 144    
Corrected Total 47,374,953.350 143    
TP 286.456 286.456 38,194.083 0.000 
ST 1,075.318 153.617 20,482.253 0.000 
RIS 97,657.401 48,828.700 6,510,493.370 0.000 
s * ST 719.427 102.775 13,703.374 0.000 
s * RIS 86.805 43.403 5,787.000 0.000 
ST * RIS 1,249.623 14 89.259 11,901.169 0.000 
s * ST * RIS 840.354 14 60.025 8,003.370 0.000 
Error 0.720 96 0.008   
Total 387,348.070 144    
Corrected Total 101,916.103 143    
TB 64.937 64.937 15,082.081 0.000 
ST 17,846.026 2,549.432 592,126.210 0.000 
RIS 22,136.045 11,068.023 2,570,637.500 0.000 
s * ST 184.278 26.325 6,114.302 0.000 
s * RIS 32.998 16.499 3,831.984 0.000 
ST * RIS 4,039.174 14 288.512 67,009.334 0.000 
s * ST * RIS 114.619 14 8.187 1,901.514 0.000 
Error 0.413 96 0.004   
Total 867,687.390 144    
Corrected Total 44,418.490 143    

Table 3 reports the classification of parameters i and for dependent data of TR, TER, TP, TB, C, and PF based on Duncan's method. Three rainfall scenarios for all hydrological parameters TR, TER, TP, TB, C, and PF were placed in different groups. For the dependent variables of TR, C, PF, and TB, eight treatments were each placed in a different group. Concerning the dependent variable TER, the performance of treatments RGE and O, treatments R and GGE, and treatments T and TGE were similar to each other. For the dependent variable TP, treatments R and SL were placed in one group of Duncan's classification.

Table 3

Classification of parameters TR, TER, TP, TB, C, and PF based on the independent variables of rainfall intensity scenario and treatments with Duncan's method

NSubset
12345678
TR RIS 48 9.8958        
48  30.7833       
48   40.5333      
Sig.  1.000 1.000 1.000      
ST 18 2.0500        
SL 18  25.3556       
18   28.0333      
TGE 18    29.6889     
18     31.0222    
18      32.3667   
RGE 18       33.0278  
GGE 18        35.0222 
Sig.  1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 
TER RIS 3.00 48 101.9021        
2.00 48  102.4125       
1.00 48   103.5875      
Sig.  1.000 1.000 1.000      
ST RGE 18 96.3722        
18 96.7000        
18  98.0111       
GGE 18  98.0278       
18   98.7000      
SL 18    109.0333     
18     111.8722    
TGE 18     112.3556    
Sig.  0.192 0.947 1.000 1.000 0.056    
RIS 3.00 48 44.6542        
1.00 48  45.5125       
2.00 48   45.6792      
Sig.  1.000 1.000 1.000      
ST GGE 18 23.3000        
18  26.4944       
RGE 18   37.5111      
18    41.0111     
TGE 18     41.7944    
18      46.3556   
SL 18       52.1000  
18        93.6889 
Sig.  1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 
PF RIS 2.00 48 1,620.6500        
3.00 48  1,663.3146       
1.00 48   1,725.9937      
Sig.  1.000 1.000 1.000      
ST GGE 18 1,056.7444        
18  1,116.8722       
RGE 18   1,490.6278      
18    1,525.777     
TGE 18     1,652.7333    
18      1,706.0222   
SL 18       1,821.2556  
18        2,989.8556 
Sig.  1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 
TP RIS 48 14.5229         
48  41.0188        
48   78.0229       
 1.000 1.000 1.000       
ST 18 41.3556        
18  42.3389       
SL 18  42.3556       
18   43.3722      
RGE 18    44.3500     
TGE 18     45.6778    
GGE 18      46.3500   
18       50.3722  
Sig.  1.000 0.565 1.000 1.000 1.000 1.000 1.000  
TB RIS 1.00 48 63.1500        
3.00 48  71.1604       
2.00 48   92.5250      
Sig.  1.000 1.000 1.000      
ST GGE 18 63.0333        
RGE 18  64.0333       
18   66.3722      
18    67.0111     
TGE 18     82.6889    
18      83.3611   
SL 18       83.6944  
18        94.7000 
Sig.  1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 
NSubset
12345678
TR RIS 48 9.8958        
48  30.7833       
48   40.5333      
Sig.  1.000 1.000 1.000      
ST 18 2.0500        
SL 18  25.3556       
18   28.0333      
TGE 18    29.6889     
18     31.0222    
18      32.3667   
RGE 18       33.0278  
GGE 18        35.0222 
Sig.  1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 
TER RIS 3.00 48 101.9021        
2.00 48  102.4125       
1.00 48   103.5875      
Sig.  1.000 1.000 1.000      
ST RGE 18 96.3722        
18 96.7000        
18  98.0111       
GGE 18  98.0278       
18   98.7000      
SL 18    109.0333     
18     111.8722    
TGE 18     112.3556    
Sig.  0.192 0.947 1.000 1.000 0.056    
RIS 3.00 48 44.6542        
1.00 48  45.5125       
2.00 48   45.6792      
Sig.  1.000 1.000 1.000      
ST GGE 18 23.3000        
18  26.4944       
RGE 18   37.5111      
18    41.0111     
TGE 18     41.7944    
18      46.3556   
SL 18       52.1000  
18        93.6889 
Sig.  1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 
PF RIS 2.00 48 1,620.6500        
3.00 48  1,663.3146       
1.00 48   1,725.9937      
Sig.  1.000 1.000 1.000      
ST GGE 18 1,056.7444        
18  1,116.8722       
RGE 18   1,490.6278      
18    1,525.777     
TGE 18     1,652.7333    
18      1,706.0222   
SL 18       1,821.2556  
18        2,989.8556 
Sig.  1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 
TP RIS 48 14.5229         
48  41.0188        
48   78.0229       
 1.000 1.000 1.000       
ST 18 41.3556        
18  42.3389       
SL 18  42.3556       
18   43.3722      
RGE 18    44.3500     
TGE 18     45.6778    
GGE 18      46.3500   
18       50.3722  
Sig.  1.000 0.565 1.000 1.000 1.000 1.000 1.000  
TB RIS 1.00 48 63.1500        
3.00 48  71.1604       
2.00 48   92.5250      
Sig.  1.000 1.000 1.000      
ST GGE 18 63.0333        
RGE 18  64.0333       
18   66.3722      
18    67.0111     
TGE 18     82.6889    
18      83.3611   
SL 18       83.6944  
18        94.7000 
Sig.  1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 

Evaluation of the effective parameters in runoff

Figures 79 compare the graphs of PF rate, runoff coefficient, and the time-to-start and TER for treatments O, SL, G, GGE, R, RGE, T, and TGE for all cases of the experiments. As shown, the value of PF is the highest in treatment O in all cases. The lowest PF rate in the first scenario was observed in treatment G in two levels of 0 and 5% slope while the lowest PF rate in scenario 3 was in 0% slope. In the other three scenarios, the lowest PF rate was related to the GGE treatment, indicating that gravel could provide better performance compared to other treatments in reducing the PF rate. When the rosemary plant is implemented with the geocell layer, the PF rate decreases in most scenarios of rainfall intensity and levels of bed slope. This issue is also observed for the turf plant. The results highlight that the SL treatment alone cannot be effective in reducing the PF rate. In Liu et al.’s (2020a, 2020b) research, which investigated the use of turf in the control of runoff, it was concluded that in the impervious parts, the peak discharge and the volume of runoff are more than in the turf land parts. Du et al. (2015) showed that increasing the upstream impervious surface increases the peak discharge approximately 14 times more than the same increase in the downstream impervious surface. However, in the current research, the implementation of turf can reduce the runoff between 40 and 50% compared to the impervious surface in all test cases. The above findings are somewhat contradictory due to the effects of the impervious surface on the peak discharge of the watershed depending on the characteristics of the watershed and rainfall. Therefore, the lower part of the meadow was shown as a potential area for reducing storm water runoff. In practice, the present results reaffirmed the importance of the urban pasture plan in reducing urban flood risk.
Figure 7

Radar chart to compare peak flow rate for different test treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and 0% slope, and (f): scenario 3 and 5% slope.

Figure 7

Radar chart to compare peak flow rate for different test treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and 0% slope, and (f): scenario 3 and 5% slope.

Close modal
Figure 8

Comparison of the runoff coefficient for different test treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and slope 0%, and (f): scenario 3 and 5% slope.

Figure 8

Comparison of the runoff coefficient for different test treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and slope 0%, and (f): scenario 3 and 5% slope.

Close modal
Figure 9

Comparison chart of time to start runoff and time-to-end runoff for different test treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and 0% slope, and (f): scenario 3 and 5% slope.

Figure 9

Comparison chart of time to start runoff and time-to-end runoff for different test treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and 0% slope, and (f): scenario 3 and 5% slope.

Close modal

Figure 8 illustrates the runoff coefficient diagram for treatments O, SL, G, GGE, R, RGE, T, and TGE for the scenarios of rainfall intensity and different slopes in terms of percentage. For a specific scenario of rainfall intensity and slope, the value of C is the highest for the O treatment and the lowest for the GGE treatment. Regarding a specific rainfall intensity scenario, changing the slope from 0 to 5% causes the runoff coefficient to increase in all treatments O, SL, G, GGE, R, RGE, T, and TGE. In general, the results of the runoff coefficient in this research can be considered close to the results of research (Hunt et al. 2002; Mullaney & Lucke 2014). It can be concluded that the implementation of the geocell layer in treatments G, T, and R in all scenarios of rainfall intensity and studied slopes can reduce the runoff coefficient. In treatment O, changes in the runoff coefficient in different conditions of the tests are not large and are around 10%. The runoff coefficient for the control treatment in scenario 1 and 5% slope was higher among all the tests and its value was 0.97. Ultimately, the runoff coefficient for the GGE treatment in scenario 2 and the slope of 0% is the lowest value and equal to 0.19. In Liu et al.’s (2020a, 2020b) research, it was observed that grass vegetation can reduce the runoff coefficient between 40 and 50%, which is consistent with the present research. Dayaratne & Perera (2008) concluded that urban impervious areas are able to produce a faster hydrological response compared with the natural pervious areas, even for a low-intensity rainfall. Due to the limited initial loss, more rainfall was transferred into the concrete impervious surface runoff under more significant rainfall events and thus caused a higher runoff coefficient.

Figure 9 shows the time to start runoff (TR) and TER for treatments O, SL, G, GGE, R, RGE, T, and TGE under different test conditions. The time-to-start runoff for an experiment occurs earlier in the control treatment (O) compared to treatments O, SL, G, GGE, R, RGE, T, and TGE for all cases. The time-to-start runoff occurs later in treatments GGE, RGE, and TGE than in treatments G, R, and T under the same test conditions. In other words, at a certain rainfall intensity and slope, combining the treatments with the geocell layer allows them to perform better in delaying the time to start runoff. The time-to-start runoff for all test treatments in scenario 1 occurs later than that of scenarios 2 and 3, because the rainfall intensity changes from the lowest to the highest rainfall intensity in scenario 1, and the creation of runoff occurs later. In scenario 2, runoff occurs earlier in all investigated treatments, as the rainfall starts with the highest intensity and reaches the lowest amount of intensity with the passage of time, which makes the capacity of water infiltration in the soil to be completed faster than that of other scenarios and thus, runoff occurs earlier. Further, it is possible to create runoff faster by increasing the slope from 0 to 5% in a specific scenario for a treatment. However, the TER varies between 92 and 114 min for all treatments. In Saraçoğlu & Kazezyılmaz-Alhan (2023) and Demirezen & Kazezyılmaz-Alhan (2022), because of the larger test basin, the TER in the infiltration trench and turf treatments was reported to be longer than that of the current research. In Liu et al.’s (2020a, 2020b) research, it was shown that the grassland surface effectively delays the onset of runoff. Grassland surface represented that the time to runoff was about 25 times than that of the impervious surface. In the present research, it is between 15 and 20 times for turf and rosemary plants and between 20 and 25 times for gravel. So it was showcased as a potential storm water retention area.

Figure 10 shows the graph of time to peak (TP) and TB for different treatments under various test conditions. As shown, TP occurs later in the RGE treatment in the first scenario, followed by the GGE treatment in the second scenario, and the G treatment in the third scenario. In the second scenario, treatments T and TGE had the longest TB value. In the first and third scenarios, this issue was observed for the O treatment, which can be due to the rapid onset of runoff in the O treatment. Regarding the values of TB in most of the tests performed in the three studied rainfall scenarios, it can be said that changing the slope from 0 to 5% increases the value of TB, implying that the runoff spends more time on a steeper slope. In scenario 2, TB values are higher in all cases compared to the corresponding cases in the first scenario. This means that when the rainfall starts with the highest rainfall intensity and ends with the lowest rainfall intensity, it will result in a longer runoff time compared to its opposite state (starting with the lowest rainfall intensity and ending with the highest rainfall intensity). In the research of Gülbaz & Kazezyılmaz-Alhan (2017), it was shown that the implementation of different plants, including grass, can help in reducing the peak discharge time and the runoff base time. This issue was observed in the present study.
Figure 10

Comparison chart of time to peak (TP) and time base (TB) for different test treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and 0% slope, and (f): scenario 3 and 5% slope.

Figure 10

Comparison chart of time to peak (TP) and time base (TB) for different test treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and 0% slope, and (f): scenario 3 and 5% slope.

Close modal

Percentage of RV reduction

Figure 11 shows the percentage of RV reduction in treatments SL, G, GGE, R, RGE, T, and TGE for different test modes. As observed, treatment GGE has a higher percentage of RV reduction than that of other treatments in all scenarios of rainfall intensity and slopes, followed by treatment G. Further, treatment GGE can reduce the RV by 78%. In general, using the geocell layer leads to a higher percentage of RV reduction in all treatments compared to the treatments without geocell. The results of this research revealed the higher effectiveness of the treatments in reducing the RV for the first scenario compared to the other two scenarios. In this regard, treatment SL has the weakest performance in reducing RV. In constant rainfall intensity, changing the slope from 0 to 5% decreases the percentage of RV reduction. In general, the results of this part of the research are consistent with those of research (Wang et al. 2012). Jayasuriya & Kadurupokune (2010) compared the geocell and turf treatment with ordinary asphalt and found that the turf and geocell treatment could reduce the RV by 45–60%, which is consistent with the results of the present study.
Figure 11

The results of reducing runoff volume in different treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and zero slope percentage, and (f): scenario 3 and slope 5%.

Figure 11

The results of reducing runoff volume in different treatments, (a): scenario 1 and 0% slope, (b): scenario 1 and 5% slope, (c): scenario 2 and 0% slope, (d): scenario 2 and 5% slope, (e): scenario 3 and zero slope percentage, and (f): scenario 3 and slope 5%.

Close modal

This research sought to quantitatively evaluate the effect of rosemary, turf, and infiltration trench treatments on some parameters of the catchment basin, such as the amount of RV reduction, time to start runoff, and TER, runoff coefficient, runoff base coefficient, time to peak, and PF rate in the laboratory dimension. To this aim, 48 tests were conducted on eight treatments of control (impermeable), sandy loam, gravel, gravel with geocell layer, rosemary, rosemary with geocell layer, turf, and turf with the geocell layer in three scenarios of rainfall intensity and two slopes. By using the geocell layer in a certain treatment, the cumulative RV and the reduction percentage of the RV were the lowest and the highest in all test cases, respectively. The gravel with geocell treatment can reduce the RV by about 78%, which is the best treatment in this regard, although different types of vegetation and gravel without geocells may be preferred for normal rainfall events due to practical implementation. The presence of vegetation and gravel can also delay the time to PF rate. By comparing the turf and rosemary pavement and gravel with sandy loam pavement, it is found that the reduction percentage in RV is higher and the PF rate is less. Thus, it is recommended to use medium or low clay and silt in the implementation of sandy loam bed soil. By performing statistical analysis, it can be found that the change of slope and rainfall intensity creates a significant difference in the parameters of time to start and TER, runoff coefficient, TB, time to peak, and PF rate. Changes in the slope from 0 to 5% in all treatments caused the runoff to start a little earlier. The runoff coefficient should increase to about 5%. The present study was conducted on a laboratory scale and the dimensions and size of the plot and pond are limitations of this research. Water flow infiltration was allowed in all tests; however, its quantity was not measured. Future research is suggested to investigate the soil erosion rate and the quality of runoff with and without geocell layers under the soil conditions studied in this research. Also, it is suggested to conduct experiments on a field scale and with larger dimensions.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.

Bäckström
M.
2002
Sediment transport in grassed swales during simulated runoff events
.
Water Science and Technology
45
(
7
),
41
49
.
Barber
M. E.
,
King
S. G.
,
Yonge
D. R.
&
Hathhorn
W. E.
2003
Ecology ditch: A best management practice for storm water runoff mitigation
.
Journal of Hydrologic Engineering
8
(
3
),
111
122
.
Dadrasajirlou
Y.
,
Ghazvinian
H.
,
Heddam
S.
&
Ganji
M.
2022
Reference evapotranspiration estimation using ANN, LSSVM, and M5 tree models (Case study: Of Babolsar and Ramsar Regions, Iran)
.
Journal of Soft Computing in Civil Engineering
6
(
3
),
103
121
.
Dayaratne
S. T.
&
Perera
B. J. C.
2008
Regionalisation of impervious area parameters of urban drainage models
.
Urban Water Journal
5
(
3
),
231
246
.
Dehghani
H.
,
Haji Agha Bozorgi
H.
&
Ghaemi
A. A.
2019
The effect of different irrigation regimes under subsurface drip irrigation system on soil moisture distribution in pistachio orchard
.
Water and Soil
33
(
3
),
391
404
.
Deletic
A.
2005
Sediment transport in urban runoff over grassed areas
.
Journal of Hydrology
301
(
1–4
),
108
122
.
de Macedo
L. M.
,
dos Santos
É. M.
,
Militão
L.
,
Tundisi
L. L.
,
Ataide
J. A.
,
Souto
E. B.
&
Mazzola
P. G.
2020
Rosemary (Rosmarinus officinalis L., syn Salvia rosmarinus spenn.) and its topical applications: A review
.
Plants
9
(
5
),
651
.
Drake
J. A. P.
,
Bradford
A.
&
Marsalek
J.
2013
Review of environmental performance of permeable pavement systems: State of the knowledge
.
Water Quality Research Journal of Canada
48
(
3
),
203
222
.
Fassman
E. A.
&
Liao
M.
2009
Monitoring of a series of swales within a stormwater treatment train
. In:
H2009: 32nd Hydrology and Water Resources Symposium, Newcastle: Adapting to Change: Adapting to Change
.
Engineers Australia
,
Barton, ACT
, Australia, pp.
368
378
.
Ferguson
B. K.
2005
Porous Pavements
.
CRC Press
,
Boca Raton
, FL, USA.
Ghazvinian
H.
&
Karami
H.
2023
Effect of rainfall intensity and slope at the beginning of sandy loam soil runoff using rain simulator (Case study: Semnan City)
.
JSTNAR
26
(
4
),
319
334
.
Heilweil
V. M.
,
Benoit
J.
&
Healy
R. W.
2015
Variably saturated groundwater modelling for optimizing managed aquifer recharge using trench infiltration
.
Hydrological Processes
29
(
13
),
3010
3019
.
Hunt
B.
,
Stevens
S.
&
Mayes
D.
2002
Permeable pavement use and research at two sites in eastern North Carolina
. In:
Global Solutions for Urban Drainage
.
American Society of Civil Engineers
,
Reston, VA
, USA, pp.
1
10
.
Hunt
W. F.
,
Hathaway
J. M.
,
Winston
R. J.
&
Jadlocki
S. J.
2010
Runoff volume reduction by a level spreader–vegetated filter strip system in suburban Charlotte, NC
.
Journal of Hydrologic Engineering
15
(
6
),
499
503
.
Jayasuriya
N.
&
Kadurupokune
N.
2010
Comparative Performance of Permeable and Porous Pavements
.
Karami
H.
&
Ghazvinian
H.
2022
A practical and economic assessment regarding the effect of various physical covers on reducing evaporation from water reservoirs in arid and semi-arid regions (Experimental study)
.
Iranian Journal of Soil and Water Research
53
(
6
),
1297
1313
.
Line
D. E.
&
Hunt
W. F.
2009
Performance of a bioretention area and a level spreader-grass filter strip at two highway sites in North Carolina
.
Journal of Irrigation and Drainage Engineering
135
(
2
),
217
224
.
Mullaney
J.
&
Lucke
T.
2014
Practical review of pervious pavement designs
.
CLEAN–Soil, Air, Water
42
(
2
),
111
124
.
Papakos
T. H.
,
Gould
M.
&
Brunner
J.
2010
Redeveloping brownfields with LID design
. In:
Low Impact Development 2010
.
American Society of Civil Engineers
,
Reston, VA
, USA, pp.
1684
1697
.
Rodrigues
A. L. M.
,
da Silva
D. D.
&
de Menezes Filho
F. C. M.
2021
Methodology for allocation of best management practices integrated with the urban landscape
.
Water Resources Management
35
(
4
),
1353
1371
.
Rosen
A.
2013
The use of geocells for slope protection under special conditions
. In:
Proceedings of the International Symposium on Design and Practice of Geosynthetic-Reinforced Soil Structures
,
October
,
Bologna, Italy
.
Rujner
H.
,
Leonhardt
G.
,
Perttu
A.-M.
,
Marsalek
J.
&
Viklander
M.
2016
Advancing green infrastructure design: Field evaluation of grassed urban drainage swales
. In:
Novatech 2016-9ème Conférence internationale sur les techniques et stratégies pour la gestion durable de lEau dans la Ville/9th International Conference on Planning and Technologies for Sustainable Management of Water in the City
.
GRAIE
,
Lyon
,
France
.
Samii
A.
,
Karami
H.
,
Ghazvinian
H.
,
Safari
A.
&
Dadrasajirlou
Y.
2023
Comparison of DEEP-LSTM and MLP models in estimation of evaporation pan for arid regions
.
Journal of Soft Computing in Civil Engineering
7
(
2
),
155
175
.
Saraçoğlu
K. E.
&
Kazezyılmaz-Alhan
C. M.
2023
Determination of grass swale hydrological performance with Rainfall-Watershed-Swale experimental setup
.
Journal of Hydrologic Engineering
28
(
3
).
Sato
T.
&
Kojma
K.
2018
Development of slope protection work using geocell and soil nailing
.
Japanese Railway Engineering
(
4
),
1
3
.
Song
G.
,
He
S.
,
Song
X.
,
Duan
Z.
,
Zhang
Y.
,
Kong
D.
&
Huang
M.
2021a
The use of geocell as soil stabilization and soil erosion countermeasures
.
Geomatics, Natural Hazards and Risk
12
(
1
),
2155
2169
.
Song
X.
,
Huang
M.
,
He
S.
,
Song
G.
,
Shen
R.
,
Huang
P.
&
Zhang
G.
,
2021b
Erosion control treatment using geocell and wheat straw for slope protection
. In:
Advances in Civil Engineering
(
Ren
Q.-X.
, ed.). pp.
1
12
. .
Stone
J. J.
&
Paige
G. B.
2003
Variable rainfall intensity rainfall simulator experiments on semi-arid rangelands
. In:
Proc. 1st Interagency Conference on Research in the Watersheds
.
Citeseer
, pp.
83
88
.
Tasalloti
A.
,
Marshall
A. M.
,
Heron
C. M.
&
Hashemi
M. A.
2020
Geocellular railway drainage systems: Physical and numerical modelling
.
Transportation Geotechnics
22
,
100299
.
Vafa
M.
2018
Impact of Storage Layer Properties on Infiltration Trench's Retention
.
Wesseloo
J.
,
Visser
A. T.
&
Rust
E.
2009
The stress–strain behaviour of multiple cell geocell packs
.
Geotextiles and Geomembranes
27
(
1
),
31
38
.
Winston
R. J.
&
Hunt
W. F.
2009
Field evaluation of level spreader-vegetative filter strips to improve hydrology and water quality
. In
2009 Reno, Nevada, June 21–June 24, 2009
.
American Society of Agricultural and Biological Engineers
, p.
1
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).