**2.2 Baseflow regulation**

*Advanced Evapotranspiration Methods and Applications*

at least 0.11 m3

effect [50].

m<sup>−</sup><sup>3</sup>

green roofs by enhancing ET [39]. Under well-watered conditions, the nighttime air above green roof can be even colder than the cool roof, though the reverse may be found during the daytime [42, 44]. With unrestricted irrigation, green roof has a comparable cooling potential as the white roof, but green roof becomes less effective when only sustainable irrigation (harvested roof runoff) or no irrigation is available [45]. During dry summer, mean daytime Bowen ratio (sensible heat flux/latent heat flux) above a green roof could reach 3, as a typical value for the urban environment; while during wet periods, mean daytime Bowen ratio can be as low as 0.3 [46]. The substrate volumetric water content is recommended to be

ratio < 1) [46]. In a study in Australia, the daytime Bowen ratio on top of a green roof reduced from above four during dry conditions to less than one after irrigation; however, the sensible heat flux on the green roof was still larger than that on the cool roof [47]. A downside of applying irrigation is that the increased moisture content may build a notable heat sink, which partly offsets the cooling effect; therefore, finer soil mix with fewer mesopores and minimized moisture storage was recommended to reduce the heat-sink effect [36]. Apart from supporting active cooling, irrigation is necessary for establishment, survival, and success of green roof plants in semi-arid and arid climates [48]. Deficit watering strategy (adapting to the vegetation requirement) and alternative sources (gray water, harvested rainwater, or condensed water from air conditioning) can be tested for controlling irrigation demand [48, 49]. So far, the role of irrigated GI for cooling urban areas is still not fully examined yet, while less is known regarding how the optimum type, amount, and arrangement of GI units influence the overall cooling

The choice of plant species also affects the cooling effect of a green roof. Sedum, though proposed as the default green roof species, often comes with incomplete plant cover, sluggish transpiration, and limited substrate moisture storage, which altogether result in a weak ET cooling effect or even a downward heat transmission toward indoor space that raises the cooling load [36]. Sedum provided no significant cooling potential over a soil substrate roof alone, so adding a thin cover of white gravel or stones on top of the green roof is recommended to increase the albedo [47]. Furthermore, sedum is also difficult to maintain and subject to the widespread decline caused by high temperature and humidity [36, 49]. Plants with higher transpiration rates and denser foliage have better cooling effect and create a blanket on top of substrate and roof to block heat transmission [36]. A promising option is woodland vegetation, which, with a 1-m substrate, can filter 90% of incoming short-wave radiation during daytime [51]. Although a deeper substrate (>10 cm) was often preferred because of the larger moisture storage [48], shallow-

Urban greening in the street canyon level includes mesic lawns and shade trees. Their cooling effect, limited by the vegetation abundance and moisture content as well, tends to be more effective over desert/xeric than over mesic/oasis landscapes [42]. At a city scale, increasing the ground vegetation has a stronger impact than implementing green roofs on reducing street temperature; whereas green roofs are more cost-effective to reduce a building's energy consumption [52]. Turfgrass was observed to represent the largest contribution to annual ET in recreational and residential land types (87 and 64%, respectively), followed by trees (10 and 31%, respectively) [53]. Urban ET amount overall relates to the urban forest coverage. Following the increasing ET gradient (464.43–1000.47 mm) through the conterminous United States, urban forest cover and forest volume correspondingly had a doubled and a threefold increase, respectively [7]. Under the shade of tree canopies, the cooling effect of the added lawn will be significantly restrained [42]. Of all

rooted plants like sedum may not able to take this advantage [49].

to maintain a favorable green roof energy partitioning (Bowen

**112**

Another major ecosystem service provided by evapotranspiration from green infrastructure is to regulate the regime of urban baseflow in terms of its peak discharge, lag time, recession coefficient, and water yield [46, 55]. Runoff and infiltration determine the upper limit in the volume of surface and subsurface return flows to streams, respectively; while ET, as a sink/loss term in the water balance, determines the lower limit in the volume of the return flow.

The goal of regulating baseflow is ambiguous to define and dependent on each case. Urbanization tends to elevate imperviousness percentage and leads to excessive surface runoff in the postdevelopment condition, which raises flooding risk and causes the urban stream syndrome at the downstream [22]. Reducing the volume of surface runoff is often set as a common goal of all GI applications [6, 10], since GI creates the extra sink near the source of rainfall and effectively reduces the volume of surface runoff traveling downstream [6, 56, 57]. In this case, the ET-focused GI (green roof, lined bioretention) would be recommended, which would transform portions of recharge and baseflow into ET [35, 58–60].

On the other hand, regulating baseflow can also mean to strengthen the percolation, when the aquifer is heavily tapped by the urban basin [61, 62]. In such case, the percolation-focused GI would be recommended such as drywell, unlined bioretention (sometimes referred as bioinfiltration), retention pond, and permeable pavement, which would transform portions of ET into recharge and eventually baseflow [63]. However, the influence of percolated water on ET is not clearly understood. Conventionally, percolation is assumed to recharge groundwater and contribute to baseflow through subsurface hidden paths [60]. Yet, lateral seepage from the bioretention is not negligible, and it can be comparable to ET amount [64] or even a much more dominant term than both ET and vertical percolation [65]. The fate of the lateral seepage has not been extensively studied yet, which could end up being intercepted by downstream rooting systems and eventually released into the air by ET again, instead of reaching the channels as baseflow. Further, water from shallow water table (<2.5 m deep) can move upwards to the root zone as capillary flow; for example, 1-m capillary upward groundwater can supply 41% of ET [66]. The knowledge gaps regarding the fate of percolation water as well as occasional capillary flow prevents the accurate appraisal of the GI influence on the local or broader scale water balance. The contributing areas to the baseflow of an urban watershed should be identified, and building GI at such locations would be cost-effective.

Connection to storm drainage network is another factor affecting the ratio of rainfall redistribution. Employment of an underdrain underneath bioretention can bypass most infiltration through the drainage network and lead to minimal ET and percolation [67, 68]. From the volume reduction perspective, underdrains make GI more resemble a conventional storm pipeline. Without connecting to a drainage network, GI can manage infiltrated water more through ET or percolation.

Choosing the percolation-focused GI in the urban areas with limited aquifer extraction and ecosystem water demand (humid climates) may overcompensate the groundwater and increase the volume of return flow to the downstream channels due to the increased baseflow. Further, the percolation-focused GI, only designed

for managing impervious surfaces, may also drain extra stormwater from pervious surfaces and then unintendedly result in a larger baseflow than the predevelopment condition [60]. Overcompensating groundwater recharge can lead to deleterious effects on downstream waters and ecosystem like in arid regions with intermittent and ephemeral streams [24]. Moreover, excessive recharge from GI may cause groundwater mounds, which, taking a long time to dissipate [69], endanger the foundations of other infrastructures and compromise drought resilience by promoting shallow-rooted plant systems that do not extract water from deep soil [70]. Therefore, determining the appropriate ET amount for an urban watershed is complicated and requires an overview of the complete water budget. This discussion goes beyond the viewpoint of baseflow restoration and gives rise to the emerging trend of using GI to reestablish the urban water budget.
