1 Introduction

Rapid urbanization is leading to faster reduction in natural infiltration surface area, which is further intensifying the urban-inundation disaster. The application of green roofs has increasing in China since the declaration of concept of Sponge City Construction (SCC) by the government of People’s Republic of China at the end of 2014 [1]. The first group of 16 cities was selected as the pilot sponge cities (including Nanning, Guangxi province) in 2015, and 1 year later, the pilot program was expanded to another 14 cities. Green roof is one of the measures in SCC and developing fast in the recent years, as reports shown 1,500,000 m2 green roof area in Beijing at 2011 and 1,450,000 m2 in Shanghai at 2012 [2]. Some researches were conducted on the practical green roof in China to explore its hydraulic and thermal performance [3, 4]. The green roofs play a significant role in managing the rainwater by affecting the outflow quantity and quality. As roofs account for considerable areas in cities, the green roofs have become an effective way to alleviate stormwater.

Many studies suggested that biochar is one of the potential soil amendments. Biochar is one of the green materials that has attracted more attention in recent years for its application in green infrastructure (green roofs; bio-filters; green corridors [5,6,7]). It is a stable carbon charcoal produced from the pyrolysis of plant-based or poultry-based biomass. As a soil amendment, it has many advantages such as: (1) enhanced water retention capacity; (2) mitigating nutrient leaching; and (3) reducing the weight of the green roof [6, 7]. In agriculture, biochar also has benefits in improving soil fertility and thus enhances plant growth [8]. The characteristics of the biochar depend on the specific surface area, porosity, and the chemical composition, which may vary with the input feedstock type for pyrolysis and pyrolysis temperatures (300–700 °C) [9]. Furthermore, biochar has much higher stability and resistance to microbial degradation compared to other biomass. Half-life of biochar is expected to be in 100–1000 years [10]. In addition, aging of biochar is likely to enhance water sorption in coarse-textured soils as observed in recent studies [11,12,13]. Thus, even from a long-term perspective, biochar has a potential as a green roof amendment for stormwater management.

Studies have discussed biochar influences on the infiltration rate of the soil. The effect of biochar is not definite in many studies. Most of the results indicated that the biochar decreased the permeability of coarse-grained soil but showed no significant changes or slightly increased in the saturated hydraulic conductivity of fine material (i.e., clayey soil) [14,15,16]. However, according to the studies by Bordoloi et al. [17] and Gopal et al. [18], the biochar-amended soil has developed fewer and smaller cracks than natural soil after drying-wetting cycles and thus showed a lower infiltration rate in the natural environment [18]. The infiltration rate is one of the key parameters to the green roofs, which affects the ability of stormwater management. Most of the above studies were conducted on single layered soil and may not be directly applicable in interpretation of water flow in multi-layered soil (as found in green roofs; [19]) amended with biochar. Beck et al. [20] analyzed saturated water retention and runoff quality and quantity of green roof substrate with and without biochar amendment. Their study found that biochar improved water retention and runoff quality (i.e., decreased in discharge of total nitrogen, phosphate, and organic carbon). Chen et al. [21] found that presence of biochar (10–15%) enhanced water retention and reduced average temperature in green roofs substrate. Kuoppamäki, Lehvävirta [7] found that biochar effects may vary (positive or negative) depending on type of biochar and plant. Their studies cautioned that biochar applications cannot be generalized and require a more customized approach depending on substrate, plant, and climate conditions. Similarly, Olszewski and Eisenman [22] found that effects of biochar may vary depending on type of plant. Their study found that biochar has better effects in peppermint than basil plant. As far as authors are aware, there is lack of systematic study on analyzing water flow (permeability, peak-outflow, storage capacity) in multi-layer substrate with biochar. Furthermore, there is lack any interpretation of unsaturated soil hydraulic parameters that is essential in seepage analysis. Such study may be useful in design of preliminary guidelines for Sponge City Construction (SCC) and Low Impact Development (LID) in Guangxi province of China, where experiences sub-tropical climate similar to those in Singapore, Malaysia, Brazil, Indonesia, and some parts of Asia.

The study aims to evaluate the hydraulic performance of green roofs with biochar and optimize its ability in stormwater management. To achieve these goals, four experimental columns containing local soil and native plant (Zoysia matrella grass species) were built to observe the hydraulic processes (e.g., surface ponding and bottom drainage) under artificial rainfall. Unsaturated hydraulic parameters of soils were derived in four columns by using inverse analysis. Numerical simulation was conducted to explore the mechanism of stormwater management performance of different biochar application modes in the green roof substrate.

2 Experiment and simulation

2.1 Substrate materials including in-house biochar preparation

In accordance with the policy of Sponge City Construction (SCC) and Low Impact Development (LID) in China, full use of local natural soil instead of the cultivated soil or the commercial substrate is promoted. The soil adopted in this study was collected from a natural slope in Nanning city, located in the southern part of China. For improving the uniformity of the soil in column, remolding soil (grinded and sieved through 2 mm) was done before the addition of the biochar and any compaction. The properties of the remolded soil are tabulated in Table 1. According to the Unified Soil Classification System (USCS, ASTM D2487), the tested soil can be classified as Lean clay (CL). Although, it is understood that clay is vulnerable to cracking, however, considering its abundance in South-west Guangxi province make it more feasible economically and in practice to be adopted for green infrastructure. Although, a systematic cost-benefit analysis [23] is required in terms of using such soil including any amendment in green infrastructure considering its engineering, environmental, and economic impact.

Table 1 Basic properties of the soil and biochar
Full size table

As shown in Fig. 1, the biochar was manufactured in-house by pyrolyzing cedar at 750 °C. A large number of grooves (pores) (refer to Fig. 1) implies enhanced specific surface area and provided more porosity for retaining water. The biochar-amended soil was made by mixing the biochar powder (after passing 2 mm sieve) with soil in powder form as shown in Fig. 1. Amendment ratio of 1:10 by weight was adopted as it corresponds to maximum efficiency of biochar in retaining unsaturated water and crack suppression [17].

Fig. 1
figure 1

Apparent feature and microscopic property of biochar

Full size image

For setting the same baseline for the test materials, as well as the same load condition in the design of green roofs, all of the soil columns (as mentioned in next subsection) were compacted at a dry density of 1.2 g/cm3. This value lies within the range specified by Emilsson and Rolf [24]. The degree of compaction (DOC) of natural soil columns (NS) and biochar-amended soil columns (BAS) was 67% and 75%, respectively. It should be noted that this compaction degree was much lower than researches on landfill cover material (compacted usually at 90%); this is because substrate with high DOC may not provide the required healthy balance between water supply and gaseous exchange for ornamental plants in green roofs [25]. Vaseline was applied along the inner wall of the container to avoid the influence of friction and preferential flow between the wall and soil column [26]. Further studies are needed to identify the optimal value of compaction considering the growth of native vegetation and the influence of climatic conditions.

The two vegetated columns, vegetated natural soil column (VNS) and vegetated biochar-amended soil column (VBAS), were fabricated through an identical procedure as bare soil columns. The commercial grass turf (Zoysia matrella) adopted in this study is a perennial grass. It is popularly used as the lawn grass in the southern part of China. It has the advantages of drought-resistant and barren tolerance [27]. The commercial turf was previously grown for eight months in a nursery. The vegetated soil columns were irrigated regularly for 1 month (30 days) before the rainfall infiltration test. The details of the four soil columns are tabulated in Table 2.

Table 2 Details of the experimental soil columns
Full size table

2.2 Design and development of apparatus

The laboratory infiltration test is one of the effective ways to understand the hydraulic process of green roof substrates [28, 29]. In this study, soil columns (15 cm in height and 30 cm in diameter, refer to Fig. 2) are designed and built for simulating one-dimensional infiltration process in extensive green roofs [30]. Surface ponding was allowed to monitor the saturated infiltration rate during the rain [31]. There were five drain holes and a thin filter plate under the soil column containers to allow free drainage in the bottom boundary of soil samples. As shown in Fig. 2, the artificial rainfall was simulated by using a rainfall system, including a water-replenished Mariotte bottle and a porous disk, which has been proved to be effective in constant-intensity rainfall simulation [32]. The rainfall simulator was fixed at the upper edge of the soil column container to maintain a constant raindrop height (10 cm) during the rainfall process.

Fig. 2
figure 2

Test set-up and sensors

Full size image

A 5-h rainfall with a steady intensity of 0.4 mm/min was applied to simulate a short-term rainstorm. This intensity is equal to a 5-year return period rainstorm intensity [33, 34] in Nanning city. The moisture sensors (EC-5; METER Devices, refer to Fig. 2) were located at 5 cm and 10 cm depths below the surface of soil columns. Monitoring of moisture changes at these depths can provide a clear idea of water flow and storage within the green roof substrate [19]. Readings of volumetric water content (VWC), ponding variation, and the drainage were obtained simultaneously at 3-min interval during the rain. Based on the pan evaporation rate (i.e., 2.6–3.2 mm/day) measured during the test, the amount of evaporation can be assumed to be insignificant in shorter duration of test [35].

Although soil column experiments have been widely used to study the hydraulic characteristics of green roofs, it still has some deficiency. One of the limitations is the vertical preferential flow along the gap between barrel wall and soil column. The influence can be reduced by smearing Vaseline on the barrel walls and increasing the diameter of soil columns. The size of the present experimental pot was 30 cm in diameter, which lies within range of previous studies such as Ni et al. [36] (i.e., 20 cm), Quinn, Dussaillant [29] (i.e., 14 cm), and Leung et al. [37] (i.e., 30 cm). Another limitation is that the rainfall simulators used in this study cannot simulate the varying rainfall intensity as the actual rainfall events. Therefore, this study adopts the method of Yang et al. [4], and further analyzes the runoff quantity of green roof under the actual rainfall through numerical simulation.

2.3 Simulation

The numerical program, HYDRUS-1D, version 4.16 [38], was used to simulate the hydraulic behavior of the substrates in the green roof [39]. The one-dimensional uniform water movement was simulated by solving the modified form of the Richards’ equation:

$$ \frac{\partial \theta }{\partial t}=\frac{\partial }{\partial x}\left[K\left(\frac{\partial h}{\partial x}+\cos \alpha \right)\right]-S $$

The finite-difference method (FDM) was used for discretization of the Richards’ equation. The soil hydraulic prosperity is usually given using the van Genuchten-Mualem (VGM) model:

$$ {S}_e=\frac{\theta -{\theta}_r}{\theta_s-{\theta}_r}=\left\{\begin{array}{cc}\frac{1}{{\left(1+{\left|\alpha h\right|}^n\right)}^m}& h<0\\ {}1& h>0\end{array}\right. $$
$$ K\left({S}_e\right)={k}_s{S_e}^l{\left(1-{\left[1-{S_e}^{1/m}\right]}^m\right)}^2 $$

The unsaturated hydraulic parameters, including residual water content (θr) and saturated water content (θs), model-shape parameters α, n, and m (m = 1 − 1/n), the pore-connectivity parameter l, and saturated hydraulic conductivity ks, were estimated by using the inverse method. The inverse procedure in HYDRUS-1D combines the Levenberg-Marquardt non-linear parameter optimization method [40] with a numerical solution of the Richards' equation.

In this study, soil, boundary and domain (size of 15 cm of column) conditions similar to four soil columns (as mentioned in previous section) were built in HYDRUS-1D. An atmospheric boundary condition with precipitation was imposed on the upper surface. A free drainage boundary condition was applied at the bottom. For the estimation of hydraulic parameters, measured water content before the rainfall was set as the initial condition. The measured hydrological processes data (i.e., ponding and drainage) and water contents during rainfall were inputted for the estimation.

According to the existing research [41], the residual water content θr and the pore-connectivity parameter l in the model were least sensitive to the bottom drainage and the water content. Residual water content and pore-connectivity parameter were fixed respectively as 0.01 and 0.5 as suggested by Mualem [42]. The saturated hydraulic conductivity of bare soil columns (refer to Table 1) was set to the measured values. In total, 4 parameters of the VGM model, including saturated moisture content θs, saturated hydraulic conductivity ks (only estimated in VNS and VBAS), and two shape parameters α and n, were estimated based on methods similar to previous works [31]. The goodness of the estimation can be reflected by the R2 value for regression of the observed values versus fitted values. R2 value of 1 indicates a perfect correlation between the fitted and observed data.

3 Result and discussion

3.1 Experimental result

During the rain simulation experiment, column BAS generated a visible surface ponding (refer to Fig. 3) at the 90th min from the start of the rain. Ponding was found to increase at a steady rate of 0.1 mm/min. The ponding phenomenon was not observed in the other three soil columns, which indicated that column BAS possessed a lowest infiltration rate than the other soil columns. The surface water level progressively reached to a maximum height of 20 mm at the end of the rain. The ponding water permeated into the soil column with a steady infiltration rate of 0.2 mm/min and dissipated at approximately, 400th min.

Fig. 3
figure 3

The measured results and simulation curve of surface ponding above column BAS

Full size image

Figure 4 compares the drainage (drainage volume divided by the cross-sectional area of the soil column) among soil columns. Drainage was generated progressively from these four tested soil columns. As shown in Fig. 4, column VNS generated drainage at the 16th min, followed by NS and VBAS. BAS shows the latest in drainage occurrence with a 182 min delay after the rain beginning.

Fig. 4
figure 4

The measured results and simulated curve of bottom drainage

Full size image

The drainage from the four soil columns increased linearly after the occurrence, which indicated a steady infiltration rate in the soil. As shown in Fig. 5, the drainage rate of column NS, VNS, and VBAS stabilized at 0.4 mm/min, which is close to the rainfall intensity. In comparison, BAS showed a slower drainage rate, with the velocity at about 0.2 mm/min. The steady drainage rate of BAS was approximately equal to the saturated hydraulic conductivity (represented by the dotted lines in Fig. 5). The rainfall intensity exceeded the saturated infiltration rate of BAS, and thus caused a part of the rainwater to accumulate on the surface of the soil column. However, the drainage rates of NS, VNS, and VBAS were close to the rainfall intensity, because of their higher saturated hydraulic conductivities compared to rainfall intensity.

Fig. 5
figure 5

Drainage rate of soil columns

Full size image

3.2 Unsaturated hydraulic parameters of treated and untreated soil

The optimized unsaturated hydraulic parameters of the four soil columns are summarized in Table 3. Figures 3 and 4 show the measured data (in color marks) and calculated results (in lines). The high agreement between the measured data and simulation results, as well as the R2 value (refer to Table 3), indicated that the numerical simulations were able to capture the bottom drainage and surface ponding of these four soil columns. The optimized parameters were able to reflect the hydraulic characteristics of these substrate materials.

Table 3 Estimated values of the van Genuchten hydraulic parameters obtained by inverse solution in HYDRUS-1D
Full size table

As shown in Table 3, the saturated water content of biochar-amended soil columns (BAS and VBAS) was higher than those of natural soil columns (NS and VNS). It can be found a 15% increase of the saturated water content from VBAS to VNS and a 16% increase from BAS to NS. These results have a similar trend as some previous experimental reports [43, 44], even though the results obtained in this study are slightly higher than Garg et al. [43] (i.e., 13%) and Ni et al. [44] (i.e., 5%). The embedded effect caused by the particle size distribution of clay [45] and the different types of biochar may probably be responsible for the higher saturated water content. On the other hand, the saturated water content of vegetated soil columns was higher than the bare soil columns, with a 4% increase from VNS to NS and a 3% increase from VBAS to BAS. Leung et al. [46] reported a similar amount of increase in saturated water content (i.e., 4.5%), and pointed out that the plant roots could change the soil structure and thus affect the soil-water retention chrematistics.

Comparison between the vegetated soils and bare soils implied that the permeability of vegetated soil has increased in both columns with and without biochar mixture. As discussed in previous studies [26, 47], the preferential water flow channel formed by the growth of plant roots might contribute to the change of saturated hydraulic conductivity. Besides, the development of cracks during the daily irrigation process in vegetated soil may have a significant influence on the hydraulic conductivity [48].

It can be found that the biochar-amended soils have lower saturated hydraulic conductivity. One of the possible reasons is the higher degree of compaction (DOC) in biochar-amended soil samples (i.e., 75% for BAS and 67% for NS). The higher DOC might result in less volume of inter-pores among the clay aggregates. Meanwhile, biochar particles acted as a filling material among the clay particles and subsequently impeded the flow of water [49].

In conclusion, the vegetated soils have higher saturated hydraulic conductivity values and higher saturated water content, while the addition of biochar decreased the saturated hydraulic conductivity and increased the saturated water content of the soil. The combination of vegetation and biochar (i.e., VBAS) shows the highest saturated water content among the four tested soil columns. The saturated hydraulic conductivity of VBAS was higher than NS and BAS but lower than that of VNS. This might be a comprehensive result of biochar and vegetation, which influenced by the growth condition of plant and biochar type. Considering that biochar may have benefits in plant growth, further studies are needed to understand the coupling effect of biochar and vegetation on the long-term hydraulic properties.

3.3 Stormwater management in biochar-amended substrate

To understand the rainstorm management of biochar-amended substrate, a set of measured rainstorm data was input in the HYDRUS-1D for simulation. The selected rainfall happened on May 22, 2020, and the rainfall data was recorded by a meteorological station on the campus of Guangxi University. The rain lasted for 10 h, with a maximum instantaneous rainfall intensity of 0.21 cm/min. The cumulative precipitation was up to 100 mm. The recorded rainfall data was set as the time-variable boundary conditions in the upper surface of the simulated model. The lower boundary condition was set as free drainage, which was able to reproduce the drainage condition of green roof substrates. For comparative purpose, all soil columns were set at the same initial water content (i.e., 0.3 m3/m3). The hydraulic parameters obtained from the inverse solution were input in the simulation (refer to Table 2). With these initial and boundary conditions, six types of green roof substrates were simulated, as shown in Table 4. The stormwater management (e.g., peak flow delay and reduction effect) of green roofs was compared and discussed between different biochar addition modes. These biochar addition modes can be classified as (1) single-layer mode with and without biochar-amended soil (substrate A and B), (2) two types of dual-layer modes: biochar mixed in the upper layer (substrate F), and biochar mixed in the bottom layer (substrate C, D, and E).

Table 4 Outflow reduction and delay effect of different substrates
Full size table

Figure 6a–f shows the flux of surface runoff and drainage during the rainstorm. The positive value represents the surface runoff, and the negative value denotes the bottom drainage. It can be found that there was no surface runoff from substrate A during the rainfall, and the other substrates generated both surface runoff and bottom drainage. During the light rain (0–100 min), there was no rainwater outflowed from each of the tested substrates. This is because of the sufficient storage capacity in the substrate before saturation, and the infiltration rate of the substrate was higher than the rain intensity. Along with an increase in rainfall intensity (100–145 min), the outflow from substrate A generated the earliest in the 116 min after rain beginning, and the substrate B showed the longest delay (i.e., 142 min) in the occurrence of rainwater outflow (refer to Table 5). For substrate A, no surface runoff was generated during the rain, and the outflow only came from the bottom drainage because of the higher saturated hydraulic conductivity than the rain intensity, while the substrate B generated a small quantity of surface runoff because its saturated hydraulic conductivity (i.e., 0.16 cm/min) was slightly lower than the maximum rainfall intensity (i.e., 0.21 cm/min). All the dual-layer substrates (i.e., substrate C, D, E, and F) generated both surface runoff and bottom drainage, and generally, the runoff occurred earlier than bottom drainage.

Fig. 6
figure 6

The flux of surface runoff and bottom drainage from different substrates

Full size image
Table 5 Outflow reduction and delay effect of different substrates
Full size table

The rainstorm reached the maximum intensity of 0.21 cm/min at the 145 min, followed by the peak outflow from each substrate. As shown in Table 5, substrate F showed a longest peak-outflow delay (i.e., 6 min), and then substrate A, B, and C, while substrate E shows no peak-outflow delay. As shown in Fig. 7, the rainwater divided into three parts in most substrates (i.e., runoff, storage, and drainage) according to the flow direction. The occurrence of surface runoff and bottom drainage depended on the infiltration rate and storage capacity of the substrates. This might result in different delay effect and reduction in peak outflow. Among the tested substrates, substrate C showed the highest reduction in peak outflow, followed by substrate F, while substrate D and substrate E show no reduction. This was mainly due to the off-peak effect in the occurrence of runoff and drainage in these two substrates. Even though the surface runoff came out earlier in these two substrates, the runoff flux reduced the infiltration flux and assumed the distributary role. Therefore, the occurrence of bottom drainage was further delay. In addition, the higher storage volume of substrate C in comparison to other substrates (refer to Fig. 7) suggests a higher detention capacity and provides a better delay and reduction effect, whereas substrate D and E had generated the runoff and drainage before the maximum rainfall intensity, which indicated that the substrate has reached saturation. The precipitation divided into two parts (i.e., runoff and drainage) after the substrate reached its storage capacity; thus, the peak flow delay and reduction effect have been decreased.

Fig. 7
figure 7

The volume of rainwater outflow

Full size image

3.4 Prospect of the biochar-amended green roofs

The benefits of biochar as a soil amendment in green roofs have been investigated extensively. Studies indicated that different application modes have resulted in varying runoff quality, quantity [50], and nitrogen dynamics, which is important for plant growth [8]. Compared to the situation in slope and landfill cover, it is much more convenient to mix biochar particles in layers for the construction of green roof. In present study, a new biochar application mode with dual-layer substrate was proposed. Substrates with similar structure have been proved to be effective in retaining rainfall and pollution reduction [51]. Simulation results show that different substrate structures (i.e., upper layer with natural soil and bottom layer with biochar-amended soil) and the height of biochar-amended layer can cause different peak flow delay and reduction effects in green roofs. An optimum structure of biochar-amended substrate can be designed based on the local climate conditions. However, the following factors should be considered before using biochar in substrate:

3.4.1 Pyrolysis effect

Feedstock type and pyrolysis temperature may significantly change the physio-chemical characteristics of biochar. Studies showed that the treatment process of biochar can influence the pyrogenic nano-pores and hydrophobic surfaces [52], and therefore affect the hydraulic properties such as soil water retention and plant available water [53, 54]. It was suggested that higher pyrolysis temperature is likely to lead to higher hydrophobicity of biochar. Furthermore, volatilization of organic compounds in the feedstock under higher pyrolysis temperature would enhance surface area and pores formation [55]. Apart from this, application rate, soil type, and particle size of biochar may also influence hydraulic properties of biochar amended soil [56]. The hydraulic performance of biochar-amended substrate can only be determined after obtaining the hydraulic parameters of the specific biochar type and soil mixture.

3.4.2 Biochar aging

The aging or weathering of biochar with time in a field is likely to cause changes in physiochemical properties and disintegration of biochar into smaller particles [57]. The change in porosity due to aging alters the soil water retention characteristics (SWRC). Paetsch et al. [58] found a higher water retention in a 3-year-aged biochar-amended soil compared to the soil with fresh biochar from same feedstock and pyrolysis process. Aller et al. [11] found that the effect of aged and fresh biochar on water retention varied with the soil type, and pointed out that aged biochar is more hydrophilic then fresh biochar. Based on literatures, the influence of fresh and aged biochar on soil hydraulic properties is highly variable, which may depend on biochar type and soil type.

3.4.3 Cost-benefit analysis

Unlike other natural material commonly used for green roofs (e.g., coir, peat, sewage sludge, and farmyard waste), biochar is a stable, carbon-rich charcoal that results from pyrolysis of biomass materials, accompany with gas and oils [59]. The manufacturing process of biochar provided some additional benefits such as potential for C sequestration, reductions in greenhouse gas emission, renewable energy generation, and economic viability [60]. The feedstock collection and pyrolysis process are the main cost of the biochar application in green roofs. Waste biomass such as water hyacinth may have the greatest potential to be economically viable while still being net energy positive [17].

In conclusion, we suggested that the feedstock collection should be obtained from the local waste material (i.e., near testing site) to minimize the cost of biochar production, transportation, and site application. Systematic studies including effects of feedstock selection, pyrolysis effect, biochar aging, nutrient leaching, and the biochar application mode (i.e., layer height of biochar-amended substrate) on water dynamics under local climate conditions are needed for preliminary design of biochar-amended green roofs.

4 Conclusion

This study aims to evaluate the hydraulic performance of green roofs with biochar and optimize its ability in stormwater management. To achieve these goals, four experimental soil columns were built to observe the hydraulic processes under artificial rainfall. Unsaturated hydraulic parameters of soils were derived in four column tests using inverse analysis. Numerical simulations were conducted to explore the mechanism of stormwater management performance in dual-layer biochar-amended substrates. Following can be concluded from study:

  • Unsaturated hydraulic parameters were derived for four experimental columns using inverse analysis. The simulation results show high agreement with the measured results. The derived parameters were able to reflect the hydraulic characteristics and the models were able to capture the surface runoff and bottom drainage of these substrate materials.

  • The vegetated soil has higher saturated water content and saturated hydraulic conductivity, whereas biochar-amended soil has higher saturated water content but shows a lower permeability than natural soil. The vegetated biochar-amended soil (VBAS) has the highest saturated water content among the four tested substrate materials, but the saturated hydraulic conductivity of VBAS was lower than vegetated natural soil.

  • For utilizing higher water retention capacity of biochar-amended soil and permeability of natural soil, dual-layer biochar-amended substrate is proposed for green roofs. The different height ratio of the layers showed variation in peak-outflow delay and reduction. The case study simulated under measured rainfall data showed that under the given substrate thickness (i.e., 15 cm), dual-layer substrate with 3 cm vegetated natural soil in the upper layer and 12 cm biochar-amended soil possessed a longer delay and higher reduction in peak outflow and thus provide the highest rainwater storage capacity.

  • Besides the hydraulic performance, the economic viability and environmental sustainability should be taken into consideration. Comprehensive studies including feedstock selection, pyrolysis effect, biochar aging, nutrient leaching, and the biochar application mode based on the local climate are needed before the biochar used in the green roofs substrate.

The dual-layer substrate with biochar-amended soil is a flexible utilization of biochar in green roofs. Biochar, being a more stable material (expected half-life greater than 100 years [10]), is more feasible amendment than other types of biomass (coir, peat, sewage sludge and farmyard waste, etc.). This study also can help in preliminary design of green roof by considering native vegetation, local soil, and climatic conditions. Further studies are needed to combine the probabilistic analysis and provide a design guideline for optimizing the rainstorm management of green roofs.