**4. Discussion**

#### *4.1. The ET Characteristics of Urban Hedges*

The ET rates of two common urban hedges were estimated in this study, and both were found to be relatively high. The ET rates of the *H. littoralis* hedge were 0.04, 0.38, 0.13, and 0.18 mm h−<sup>1</sup> during the four typical sunny days from four seasons. The ET rates of the *L. quihoui* hedge were slightly lower: on the four days selected, they were 0.03, 0.33, 0.09, and 0.17 mm <sup>h</sup>−1, respectively. The ET rates of the two hedges have also been studied in other cities. A study conducted in Hubei, China showed that the ET rate of *H. littoralis* (0.04 mm <sup>h</sup>−1) was higher than that of six other plants in summer [60]. It also found that the *H. littoralis* had the highest light utilization efficiency and the third highest water use efficiency. Another study conducted in Changsha, China showed that the *L. quihoui* could transport 2576.52 <sup>g</sup>·m<sup>−</sup>2·d−<sup>1</sup> (approximately 0.11 mm <sup>h</sup>−1) of water into the air through ET in August [61]. It was the third highest out of the 13 studied shrubs. However, the ET rates in the two studies above are much lower than our results, as the two previous cities ge<sup>t</sup> less solar radiation compared with our study sites.

The winter in Shenzhen is warm enough to sustain plant growth, so almost all local plants are evergreen [62]. As a result, the ET rate of the *H. littoralis* hedge was still high in the winter day. In addition, with its high light and water utilization efficiency, its ET rate might be slightly higher than that of the *L. quihoui* hedge. The ET rates of the two hedges were both higher than the ET rates of the lawn. LAI might be the predominant reason [63].

#### *4.2. Cooling Effect of the Urban Hedges*

Three techniques were used to describe the cooling effects of the urban hedges in this study. Among them, the cooling effects of plants through ET alone was calculated using a reference method. For 10 cubic meters of air, this *H. littoralis* hedge could generate cooling at rates of 0.12 ◦C min−<sup>1</sup> m<sup>−</sup>2, 1.29 ◦C min−<sup>1</sup> m<sup>−</sup>2, 0.42 ◦C min−<sup>1</sup> m<sup>−</sup>2, and 0.61 ◦C min−<sup>1</sup> m<sup>−</sup><sup>2</sup> on the four studied days. Meanwhile, the cooling rates of the *L. quihoui* hedge were 0.10 ◦C min−<sup>1</sup> m<sup>−</sup>2, 1.13 ◦C min−<sup>1</sup> m<sup>−</sup>2, 0.30 ◦C min−<sup>1</sup> m<sup>−</sup>2, and 0.56 ◦C min−<sup>1</sup> m<sup>−</sup>2. In our reference research, a 2885-m<sup>2</sup> *S. superba* forest in Guangzhou, another subtropical megacity near Shenzhen, could cool a 10-m<sup>3</sup> air column at rates of 0.15 ◦C min−<sup>1</sup> m<sup>−</sup><sup>2</sup> and 0.13 ◦C min−<sup>1</sup> m<sup>−</sup><sup>2</sup> in July in 2007 and 2008 [58]. Its cooling rates by per unit area vegetation were lower than our results, as the ET during the day and night in all kinds of weather were included in their study. The daily UHIIs around our study sites over four seasons were approximately 0.76, 1.06, 1.04, and 0.80 ◦C [64]. For the whole city, the yearly average UHII of Shenzhen was 2.6 ◦C [65]. Therefore, the urban hedges showed grea<sup>t</sup> cooling potential in the mitigation of UHI.

The cooling rate above is the temperature reduction by ET without heat input. The LE/Rn reflects the proportions of the net radiation that ET dissipates. The greater the proportions of net radiation that were consumed by latent heat, the smaller proportion of net radiation could heat the environment through sensible heat. The *H. littoralis* hedge could consume 37.27%, 68.44%, 56.10%, and 65.71% of the net radiation as latent heat over the four days, while for the *L. quihoui* hedge the ratios were 35.72%, 60.81%, 41.45%, and 61.58%. These ratios were significantly higher than artificial underlying surfaces. It was found that only 123 Wh m<sup>−</sup><sup>2</sup> out of the 1949 Wh m<sup>−</sup><sup>2</sup> net radiation reaching to the asphalt roof was consumed by latent heat [66]. Grimmond et al. reported 23% of the LE/Rn in Marseille, where the area fraction of vegetation and water was 10–20%. Meanwhile, areas like Me93 and Vl92 that contained less vegetation had a lower LE/Rn [67]. In Kansas City, the LE/Rn could reach 46–58% in an exurban residential neighborhood, where the vegetation accounted for 58% of the total area [68]. This phenomenon was also demonstrated in a study conducted in Kugahara, Tokyo, where the LE/Rn in the daytime was always larger in hot months and smaller in cooler months [69]. According to LE/Rn, the *H. littoralis* hedge had better cooling effects than the *L. quihoui* hedge. The LE/Rn was larger when the radiation was stronger, which means the cooling effects of ET might be stronger in hotter days.

The albedo differences may result in the surface temperature differences between hedges and asphalt pavements [70,71]. Moreover, the hedges could consume much more heat through ET than artificial underlying surfaces [72]. Compared to the asphalt pavement, the surface temperature of the studied *H. littoralis* hedge were 4.22, 19.94, 8.57, and 7.00 ◦C lower on the four days. The surface temperature of the *L. quihoui* hedge were 2.80, 19.17, 7.80, and 5.92 ◦C lower at the same time. The forested land could also cool the surface more than 10 ◦C in November compared with developed land [73]. Leuzinger et al. found that tree canopies in Basel were 19 ◦C cooler than roofs in July, and different trees had different canopy temperatures [74]. The land surface temperature differences between land use types of transportation and green spaces in Shenzhen were approximately 4.8 ◦C in daytime in October [75]. In comparison with the studies above, the cooling effects of the urban hedges in our research are remarkable. The amplitudes of the hedges' surface temperature were also smaller than those of the asphalt road in the daytime, which means the thermal environment was more stable in the urban hedge area (Table 4). Similar results have been found in previous studies [76]. They found that the maximum daily variation of surface temperature was no more than 3 ◦C, and the maximum surface temperature was only 26.5 ◦C for *Raphis palm*, while for the hard surface, they were 30 ◦C and 57 ◦C, respectively. Latent heat could significantly reduce the maximum surface temperature in a day but showed minimal effects on the minimum temperature [77].

**Table 4.** Standard deviations of the surface temperatures on the four days (◦C).


#### *4.3. Applicability of the '3T + IR' Method for the ET Estimation of Urban Hedges*

In this study, a new method based on '3T + IR' was applied to accurately estimate urban ET. The applicability of this method on urban vegetation has been verified in this study by comparison with the BREB method. The results showed grea<sup>t</sup> reliability in this new method. With this method, the ET rates of the hedges can be calculated by surface temperature, which could be easily obtained by thermal images. Therefore, this method will not be limited by the complexity of urban underlying surfaces, which is the main obstacle for traditional methods. It also has a higher temporal and spatial resolution than traditional satellite remote sensing. Based on accurate ET rates, the specific cooling effects of ET could be obtained.

In this study, to simplify the calculation and measurement, the emissivity of the hedges and the lawn were defined as 0.98 and the net radiation was estimated based on the solar radiation and temperatures. Therefore, there might be some bias of the results. Besides, the leaf with the highest surface temperature was selected to be the reference leaf. The ET of this leaf actually is larger than zero, therefore will leading a little overestimation of the actually ET of the vegetation based on three-temperature model. Though it has been applied and validated in various field experiments out of the city, this is the first application of this new method in the study of urban hedges. Therefore, more research is needed. For example, an idealized reference leaf is still hard to select. The shape, emissivity and albedo of the reference leaf could affect the results. Besides, this method could not be used in continuous measurement at a high frequency, as we could only photo the thermal images and measure the net radiation by hand. Automatic imaging technique of infrared remote sensing and net radiation measurement will be helpful in the future.
