The effects of small Water Cool(ing) Islands on body temperature

ABSTRACT Water Cool(ing) Island (WCI) literature has paid little attention to smaller waterscapes in developed areas that are most vulnerable to the Urban Heat Island (UHI) effect. To fill this gap, this study investigated participants’ changes in body temperature and stress level associated with the visual, auditory, and thermal properties of small waterscapes. The results of this study demonstrated that smaller waterscapes can provide effective cooling for predominantly hardscaped urban settings prone to the effects of UHI. This approach offers low-income residents without air-conditioning outdoor alternatives to indoor cooling centres in a post-pandemic era with more frequent and severe heat waves.


Macroscale and microscale cooling effects of water bodies on heat adaptation
Urban Heat Islands (UHIs) refer to continuous areas with (as much as 6°C) warmer nighttime temperatures due to the presence of more hardscaped surfaces that trap solar radiation (Zander et al. 2018).A temperature increase of as little as 1.5°C can cause heat stress when the body cannot release excess heat (Elsayed 2012).Urban Cool(ing) Islands (UCIs) allude to continuous areas with cooler temperatures attributed to shades from trees, vegetative evapotranspiration, and evaporative cooling from water bodies (Lee, Oh, and Seo 2016).Among UCIs, Water Cool(ing) Islands (WCIs) denote areas with lower temperatures due to the radiation-protecting and evaporative cooling effects of lowreflectivity water surfaces (Shafaghat et al. 2016).Compared to rivers, lakes tend to have stronger macroscale WCI effects due to lower geometric complexity, higher ratio of blue over grey surfaces, and an edge condition with more trees and vegetation and less hardscape (Coutts et al. 2012).Macroscale WCI effects of large water bodies also vary according to water surface area (Chao, Wu, and Li 2008) and predominant wind direction and speed (Mostofa and Manteghi 2020).While most of the WCI literature has focused on mitigating temperature extremes with large water bodies as heat sinks (Lin et al. 2020), their macroscale WCI effects have been inconsistent and difficult to translate into human-CONTACT Hope Hui Rising hope.rising@tamu.eduscale thermal experience.There is a need to investigate microscale WCI effects from smaller water bodies to help fill the literature gap in the use of water-based thermal comfort for human-scale heat adaptation.
There is a lack of convergeable results from the few studies on measuring the microscale WCI effects from small water features.Mist-based evaporative cooling is more effective in providing thermal comfort than either shade-based cooling or shade-andmist-combined cooling, which is associated with the widest thermal comfort range (Li et al. 2022).For small treeless courtyards sheltered from the wind by buildings, vertical waterscapes with misting effects at a height of at least 1 to 2 m can potentially be more effective evaporative cooling sources than horizontal waterscapes especially when there is not enough space for 1) hosting horizontal waterscapes with a sufficient water surface area of at least 4 m 2 (Imam Syafii et al. 2016;Robitu et al. 2006); 2) providing sufficient fetch and wind for effective evaporative cooling (Du et al. 2016); and 3) enabling convective cooling from air movement between water bodies and air movement (Zeng, Zhou, and Li 2017).Computational Fluid Dynamics (CFD) simulations have been used for the few studies on the microscale WCI effects of small waterscapes in courtyards: A hollow-cone nozzle water spraying system with a 9 × 10 −3 m 3 /s water flow rate and a 3 m height could potentially decrease air temperature by up to 3°C at a height of 1 to 2 m in locations up to 5 m from the spray line (Montazeri et al. 2016).The mists from the leeward area (that is sheltered from the wind) of flowing and dispersing waterscapes can reduce the air temperature by 3°C up to 35 m away (Nishimura et al. 1998) and produce stronger cooling and humidifying effects than the windward (that faces the prevailing wind) and lateral areas in courtyards (Jacobs et al. 2020).
Despite the increasing simulation-based evidence of effective human-scale or microscale WCI effects from individual small waterscapes, no study has investigated their aggregated microscale WCI effects possibly due to two main challenges: 1) most studies employed simulations to investigate the microscale WCI effects of small vertical waterscapes as opposed to measuring human physiological responses to heat stress; 2) most areas investigated for the microscale WCI effects of small horizontal waterscapes contain vegetations, tree shades, and wind exposure as potential confounds.These aforementioned confounds are less significant in courtyards protected from the prevalent wind with limited space to accommodate vegetation and tree shades.There is an urgent need to use physiological measurements to investigate whether human-scale vertical waterscapes can produce greater individual and aggregated microscale WCI effects than horizontal waterscapes in UHI-prone hardscaped courtyards.Such investigations will help inform possible ways to retrofit more hardscape courtyards with small waterscapes to facilitate water-based heat adaptation.

Outdoor human thermal comfort assessment techniques
Most studies use simulations and models to estimate air temperature and thermal stress indices, including Wet-Bulb Globe Temperature (WBGT), Physiological Equivalent Temperature (PET), Thermal Climate Index (UTCI), and COMfort FormulA (COMFA) (Erell, Bar Kutiel, and Pearlmutter 2020;Thorsson et al. 2021;Zefeng, Liu, and Brown 2020).WBGT provides a direct environmental measure modified by activities while PET, UTCI and COMFA derive heat stress metrics from human energy balance models (Grundstein and Vanos 2020).For a common inside or complicated outside scenario, PET is the air temperature at which the human energy budget is matched without the effect of wind and solar radiation (Jacobs et al. 2020).PET is better at predicting collapses due to heat stress during a half marathon than WBGT (Thorsson et al. 2021).On the other hand, COMFA is more accurate than PET and UTCI in estimating thermal stress (Zefeng, Liu, and Brown 2020).However, these simulations and models have not been able to account for the variation in the subjective perception of thermal comfort to account for individual differences in heat stress responses to the same set of thermal conditions.
UTCI and COMFA use temperature ranges (1.8°C to 23°C as neutral and 35°C to 41°C as hot and very hot) to evaluate perceived levels of heat stress (Matzarakis, Mayer, and Iziomon 1999).These temperature ranges are much greater than 1) the average reductions in simulated air temperature (0.2°C and 0.6°C or 0.36°C and 1.08°C) and 2) PET (0.6°C and 1.9°C or 1.08°C and 3.42°C) due to the cooling effects of small water bodies.If the human energy intake matches the equivalent core and surface temperature, PET becomes interchangeable with the air temperature (without solar and wind effects).However, PET relies on inputs of microclimatic data, which do not have enough variations within less than a minute to adequately measure immediate body temperature reduction due to water-based evaporative cooling.The combined cooling effects of a water mist spray and a fan can easily exceed the thermal load of human bodies to yield nearly instant decreases in body temperature to provide thermal relief on hot days (Farnham, Emura, and Mizuno 2015).This study thus uses direct measurements of body temperature because other simulated heat stress indices do not have sufficient sensitivity levels for investigating the individual and aggregated microscale WCI effects.
In addition to heat stress, body temperature can increase due to cognitive stress associated with hardscape views (Velarde, Fry, and Tveit 2007).Research has demonstrated that the evaporative cooling effects of water could help mitigate the effects of non-heat-related stressors, such as hardscape views (Kendal, Williams, and Williams 2012).

Visual-auditory-thermal effects of waterscapes
Auditory-visual stimuli from water features have been found to induce more restorative effects than visual stimuli alone (Deng et al. 2020).Using waterscape recordings and immersive virtual reality, water sound alone has been found to mitigate anxiety and stress to affect body temperature (Yang and Moon 2019).However, not all water sounds are stress-reducing: Waterfalls have been found to generate unpleasant (more percussive) sounds possibly because they are more vertical than most small fountains and channels that produce more pleasant (less percussive) sounds (Patón et al. 2020).Based on monitor presentations of visual stimuli without auditory and thermal effects, urban waterscapes were the most preferred, followed by natural waterfalls and natural ponds with standing water (Polat and Yilmaz 2008).Urban waterscapes that resemble waterfalls have the potential to evoke even more positive valence if they can be designed to have more pleasant sounds than natural waterfalls.Therefore, there is a need to investigate body temperature changes associated with the visual, auditory, and thermal properties of small waterscapes to inform possible ways to retrofit hardscape courtyards using waterscapes with effective cooling effects.
This study sought to fill the knowledge gap pertaining to the aggregated microscale effects of the WCI and to develop effective cooling strategies for hardscaped courtyards that are prone to UHI.To address this issue, participants were randomly assigned to experience different types of waterscapes in a water park as a field experiment.The primary objective of this experiment was to examine the extent to which human-scale vertical waterscapes can generate more substantial individual and aggregated microscale WCI effects compared to horizontal waterscapes.The WCI effect of each waterscape was assessed by measuring the changes in individual body temperature and physiological stress response.

Materials and methods
To investigate the cooling effects of small man-made water features on the UHI effect, Lovejoy Fountain Park in Portland was selected because it is vulnerable to UHI and provides a range of experimental conditions for testing the cooling effects of horizontal and vertical water features.This section covers the study design, including a description of the study site's overall characteristics and thermal conditions, detailed site selection rationale, formulation of hypotheses, experimental procedures, data collection and signal processing, and statistical analysis.

Temporal variations and limitations of macroscale WCI effects
Voelkel, Shandas, and Haggerty ( 2016) used 1-m-resolution temperature data from ground measurements during a day of an extreme heat event on 25 August 2014 in Portland to develop the interactive Urban Heat Index mapping tool used by the study to generate the maps in Figure 1: The macroscale WCI effects of rivers during the summer in Portland, Oregon are limited to the morning hours (6 AM to 7 AM) with the blue to red gradient to illustrate temperatures ranging from 13.0°C to 18.2°C (blue areas in Figure 1 (a)).Three UCIs (locations a through c) during the afternoon (3 PM to 4 PM) and evening hours (7 PM to 8 PM) with the blue to red gradient to illustrate temperatures ranging from 21.7°C to 33.2°C (blue areas in Figure 1(b, c)) are mostly associated with natural areas with tree shades and higher grounds.An UCI associated with the low-lying tree-shaded area around a lake (d) is present in the evening (Figure 1(c)) but negligible in the afternoon (Figure 1(b)).On the other hand, an UCI downtown (e), despite being beyond the 100 m river cooling zone (Syafii et al. 2017) and negligible in the morning, is present when UHI poses the most threats in the afternoon and evening.

Microscale WCI effects of smaller waterscapes
The observed UCI (e) downtown is likely associated with the microscale WCI effects from manmade waterscapes in Keller Fountain Park and Lovejoy Fountain Park, two major outdoor spaces downtown.Lovejoy Fountain Park is a 1.11 ac treeless hardscape courtyard dominated by a variety of human-scaled waterscapes, including a 2.4 m tall and 3.6 m wide source waterfall.Keller Fountain Park is a 0.92 ac city-block with an open parklike setting anchored by one large three-story tall waterfall that holds 285,500 L of water and pumps 49,140 L per minute over the cascade.
Figure 1 shows that macroscale WCI effects are less effective than the aggregated microscale WCI effects from smaller waterscapes in the afternoon and evening for Portland.Yet, the WCI literature has largely focused on comparing the UHI-mitigating effects of large water bodies (Bhargava, Lakmini, and Bhargava 2017).However, few studies investigated the individual and aggregated effects of microscale WCIs in similar treeless hardscaped courtyards surrounded by buildings.These courtyards tend to have little ventilation for convective cooling and insufficient light exposure to support trees for providing shade and vegetation for cooling through evapotranspiration.As a result, they are the most vulnerable to the UHI effect caused by 1) short-wave solar radiation trapped by reflections between building and ground surfaces; 2) less heat transport by a reduction in wind speed; and 3) increased heat storage by surrounding building facades (Kleerekoper, Esch, and Salcedo Rahola 2011).There is a need to investigate the extent to which small man-made water features provide effective cooling in the most UHI-prone locations with the most vulnerable populations, such as the LoveJoy Park that is often used for cooling by the homeless and nearby low-income populations without indoor air conditioning.

Site selection rationale
Lovejoy Fountain Park in Portland, Oregon was chosen as a study site because 1) it is within the only UCI (location e in Figure 1) that persists (despite the lack of large vegetated areas, elevated grounds, or large water bodies) during both the afternoon and evening hours most vulnerable to UHI; 2) it is vulnerable to UHI due to being hardscaped and surrounded by buildings; 3) its courtyard setting minimizes confounds associated with direct sun exposure, wind direction, and fetch; 4) it provides a wide range of experimental conditions for testing the cooling effects of horizontal and vertical waterscapes, including a small pond, a large reflecting pool, water jets, bubblers, cascades, and waterfalls (Figure 2, Upper Right); 5) the source waterfall (Figure 2, Bottom Left) has veneer stones cascading down with slight offsets to soften the waterfall sound; 6) potential microscale WCI effects may be produced by the source waterfall, which is about 2.4 m tall and 3.6 m wide (similar to those used for previous simulations); and 7) the source waterfall divides the park into two levels to create potential aggregated microscale WCI effects from the source waterfall, the small waterfall, and the large reflecting pool at the lower level.

Hypothesis A
The stations located at the lower level (S1-S3 coded as 1 for LEVEL) in front of the source waterfall (S1) are associated with significantly lower mean body temperatures (TEMPM), lower stress levels (SCLM), and greater degrees of stress reduction (ΔSCLD) than those at the upper level behind and above (S4-S7 coded as 2 for LEVEL).Validating the hypothesis helps provide evidence for the presence of aggregated microscale WCI effects at the lower level.

Hypothesis B
The source waterfall (S1) induces a significantly greater reduction in body temperature within 15 s (ΔTEMP) than other waterscapes outside of the immediate zone influenced by its evaporative cooling effects.Validating the hypothesis helps substantiate the presence of a microscale WCI about one foot away from the source waterfall (S1) at the lower level Table 1.

Experimental protocol
The study used a within-subject repeated measure design to investigate each participant's changes in body temperature and stress level across experimental conditions (Lok et al. 2019).A total of 17 participants were conveniently sampled from four major routes to the study site during the afternoons of July 7 and July 8 in 2015 when the high temperatures were 32.7°C and 32.2°C at 12 PM and 27.8°C and 32.2°C at 6 PM, respectively, based on the weather report for Portland on timeanddate.com.Both afternoons had cloudy conditions required by the use of outdoor eye-tracking to document participants' visual input.Each participant was randomly assigned to use a clockwise or counterclockwise sequence to visit the seven waterscape stations and to view concrete or water first at each waterscape station.Each participant was instructed to stand at the same location about 0.3 m from the edge of each waterscape for four trials lasting 15 s each: concrete view with natural gaze, water view with natural gaze, concrete view with controlled gaze, and water view with controlled gaze.Controlled gaze referred to holding the gaze at one fixed point while natural gaze shifted with the moving water.

Sample size and statistical power
A total of 28 experimental conditions were generated from four visual stimuli (concrete view with natural gaze, water view with natural gaze, concrete view with controlled gaze, and water view with controlled gaze) at seven waterscape locations.Power analysis using F tests (ANOVA: repeated measures, within factors) in G*Power3.1 suggested that the sample size of 17 participants was adequate with a power of 0.8 with an effect size of 0.25, and an alpha error probability of 0.05.

Mobile eye-tracker for recording visual and auditory stimuli and gaze patterns
Each participant wore Tobii Pro Glasses 2, a mobile eye tracker during the experiment.For each trial, the investigator used a tablet to wirelessly access the real-time and in-situ video recording from Tobii Pro Glasses 2 to ensure the correct gaze (natural or focused) was used for viewing the correct visual stimulus (water or concrete) and to document the auditory stimulus each participant experienced from each waterscape.

Mobile sensor for recording physiological data
Empatica E4, a multimodal wearable sensor, has been proven to provide 1) consistent and accurate stress detection from continuous recordings of physiological signals (Indikawati and Winiari 2020); and 2) stress response markers from episodic changes in the signals (Birenboim et al. 2019).The sensor recorded each participant's body temperature (TEMP) and electrodermal activity (EDA) at a sampling rate of 4 Hz.EDA was decomposed into tonic and phasic components of skin conductance.Skin conductance level (SCL), or the tonic component of EDA, is a stress index based on the sympathetic nervous system's activity (Horstick, Siebers, and Backhaus 2018).Skin conductance response (SCR) is the phasic increase in the conductance of the skin (EDA) that can last between 10 and 20 s before returning to the tonic or baseline level of skin conductance (SCL); SCR as the eventrelated autonomic response can occur 1-4 s after each exposure to a discrete or discontinuous stimulus as a clip of a visual scene or a sound as short as 7 s duration (Khalfa et al. 2002).

Epoch analysis for physiological data
SCL is an index for the sympathetic nervous system's activity.The standard protocols in AcqKnowledge 4.4.were used to decompose EDA data into SCL and SCR in addition to: 1) removing movement artefacts; 2) normalizing within each recording; 3) standardizing across recordings for each participant; 4) standardizing across participants; and 5) performing epoch analysis of 15 s segments to generate response variables including change in SCL (ΔSCL), change in SCR (ΔSCR), change in body temperature (ΔTEMP), as well as the mean SCL (SCLM), mean SCR (SCRM), and mean body temperature (TEMPM).

Data coding
As shown in Table 2, five independent variables in this study were coded 1 or 2 to account for: 1) concrete or water view for the CONCRETE/WATER (C/W) variable; 2) natural or controlled gaze for the FOCUS/NOT (F/N) variable; 3) the source waterfall or other waterscapes for the WATERFALL/NOT (W/N) variable; 4) lower or upper level of the source waterfall for the LEVEL (L) variable; and 5) the waterscapes were coded as 1, 2, 3, 4, 5, 6, or 7 for the STATION (S), respectively.

Normality testing for residuals in ANOVA
As shown in Table 3, the normality distributional assumption for residuals in the Analysis of Variance (ANOVA) was violated for all response variables.Therefore, in lieu of ANOVA, this study employed linear mixed-effects models (a form of multivariate analysis of variance) to allow correct estimates across a wide range of variance structures to be robust despite violations of normality distributional assumptions (Vasey and Thayer 1987).

Multicollinearity diagnostics
A Chi-squared correlation analysis was conducted to assess potential multicollinearity among the five categorical variables coded to indicate participants' waterscape experiences in Table 2.The coefficient of Cramer's V ranges from 0 to 1, with a higher value indicating a stronger correlation between the two variables.The results from Table 4 show that CONCRETE/WATER was highly related to FOCUS/NOT (Cramer's V = 0.12, p < 0.05), whereas STATION was highly correlated with WATERFALL/NOT (Cramer's V = 1.00, p < 0.01) and LEVEL (Cramer's V = 0.47, p < 0.01), respectively.

Mixed-effects model
To prevent potential multicollinearity, five independent variables were split into three models for mixed-effect analysis   and skin temperature (TEMP) collected using the Empatica E4 sensor.To investigate the effects of small waterscapes on cognitive stress and heat stress, ΔSCL, SCLM, ΔSCR, SCRM, ΔTEMP, and TEMPM were used as the response variables for running each of the three mixedeffects models in SPSS Statistics 26 while controlling for the random effects from variability in participants' psychophysiological baselines and the sequence in which participants experienced waterscapes.To parse out the effects of cognitive stress from viewing hardscape from those of heat stress on body temperature, the study monitors each participant's view to hardscaped versus water scenes as a control variable CONCRETE/WATER (C/W).For each condition, two trials were also conducted with unfocused and focused gazes to control the effects of gaze type FOCUS/NOT (F/N).To investigate the aggregated microscale WCI effects, each model uses another variable, LEVEL, to indicate waterscape locations at the upper level behind versus the lower level in front of the source waterfall.In contrast, each model uses WATERFALL/NOT (W/N) to detect the presence of potential individual microscale WCI due to the source waterfall (S1) based on whether there is a greater reduction in body temperature.

Data analysis
A Chi-squared correlation analysis and linear mixed-effect model analysis were performed in SPSS Statistics 26.

Models using ΔSCL and SCLM as response variables
The effects of STATION ) at the lower level significantly reduced the mean SCL (SCLM) with the lowest SLCM at the source waterfall (S1: as indicated by a significant negative fixed-effect on SCLM; p < .01),followed by the small waterfall (S2: as indicated by a significant negative fixedeffect on SCLM; p < .01),and whirlpool (S3: as indicated by a significant negative fixedeffect on SCLM; p < .01).The findings suggest that the evaporative and/or auditory effects of each of the three lower-level waterscapes not only helped mitigate the stress response from viewing concrete but also reduced participants' level of arousal.The effects of STATION on SCLM had marginal or no significance for the upper-level waterscapes, including the top bubbler (S4: .05< p < .1), the lower bubbler (S5: .05< p < .1)and the Upper Cascade (S6: p > .1).The marginally significant effects of the top bubbler and the lower bubbler were similar in magnitude with slightly less arousal by the lower bubbler.

The effects of LEVEL
As shown in Compared to the upper level, participants experienced a lower level of arousal (SCLM) at the lower level (as indicated by a significant negative fixed-effect on SCLM; p < .01).The findings suggest that the evaporative and/or auditory effects of the lower-level waterscapes not only helped mitigate the stress response from viewing concrete but also reduced participants' level of arousal significantly.

The effects of WATERFALL/NOT
As shown in Table 5, despite no significant effects on ΔSCL (AIC = −145.10;p > .05),WATERFALL/NOT significantly influenced SCLM (AIC = 1504.72;p < .01)with a lower SLCM average by the source waterfall (S1: as indicated by a significant negative fixed-effect on SCLM; p < .01)than other waterfalls while CONCRETE/WATER (p > .05)and FOCUS/NOT (p > .05)did not.The finding suggests that participants experienced a significantly lower level of arousal by the source waterfall.

Models using ΔSCR and SCRM as response variables
The effects of STATION  continuous visual, auditory, and thermal stimulus without detectable event-related responses by SCR.

The effects of LEVEL
As shown in Table 5, LEVEL did not significantly influence ΔSCR (AIC = −619.99;p > .05)or SCRM (AIC = −1558.64;p > .05).The lack of event-related responses by SCR showed the aggregated visual, auditory, and thermal effects of waterscapes were continuous.

The effects of WATERFALL/NOT
As shown in Table 5, WATERFALL/NOT did not significantly influence ΔSCR (AIC = −619.88;p > .05)or SCRM (AIC = −1557.52;p > .05).The finding indicated that the visual, auditory and thermal effects from the source waterfall were continuous without detectable eventrelated responses by SCR.

The effects of STATION
As shown in Table 5, STATION had significant effects on both ΔTEMP (AIC = 776.12;p < .01)and TEMPM (AIC = 821.66;p < .01)while CONCRETE/WATER (p > .05)and FOCUS/NOT (p > .05)did not.The source waterfall was associated with the highest temperature reduction (ΔTEMP) (S1: as indicated by a significant positive fixed-effect on ΔTEMP; p < .01)followed by the small waterfall (S2: as indicated by a significant positive fixedeffect on ΔTEMP; p < .01).Other waterscapes did not have significant effects on temperature reduction.On the other hand, the TEMPM was the highest for the source waterfall (S1: as indicated by a significant positive fixed-effect on TEMPM; p < .01),followed by the small waterfall (S2: as indicated by a significant positive fixed-effect on TEMPM; p < .01), the whirlpool (S3: as indicated by a significant positive fixed-effect on TEMPM; p < .01), the top bubbler (S4: as indicated by a significant positive fixed-effect on TEMPM; .p< .01), the lower bubbler (S5: as indicated by a significant positive fixed-effect on TEMPM; p < .01)and the Upper Cascade (S6: .05< p < .1).The findings suggest that participants had significantly higher body temperature at three lower-level waterscapes and two bubblers, while the evaporative and/or auditory effects of the source waterfall and the small waterfall significantly helped reduce participants' body temperature.

The effects of LEVEL
As shown in Compared to the upper level, the lower level was associated with significantly higher reduction in body temperature (ΔTEMP) (as indicated by a significant positive fixed-effect on ΔTEMP; p < .01)and higher mean body temperature (TEMPM) (as indicated by a significant positive fixed-effect on TEMPM; p < .01).The findings suggest that participants had significantly higher body temperature at the lowerlevel waterscapes, while the evaporative and/or auditory effects of the lower-level waterscapes significantly contributed to a reduction in participants' body temperature.

The effects of WATERFALL/NOT
As shown in (as indicated by a significant positive fixed-effect on ΔTEMP; p < .01)and higher mean body temperature (TEMPM) (as indicated by a significant positive fixed-effect on TEMPM; p < .01).The findings suggest that participants had significantly higher body temperature at the source waterfall, while the evaporative and/or auditory effects of the source waterscapes helped reduce participants' body temperature significantly.

Models for testing the effects of LEVEL
When the fixed effects of CONCRETE/WATER and FOCUS/NOT were included as control variables, the TEMPM model shows significantly higher mean body temperature for the lower-level waterscapes.The higher heat stress is possibly due to greater solar radiation from the vertical concrete surfaces by the two waterfalls.The ΔTEMP model shows greater body temperature reduction for the lower level to substantiate aggregated microscale WCI effects at the lower level for heat adaptation.No significant effect was found for the ΔSCL or ΔSCR model.This may be because significant reductions in stress levels indicated by ΔSCL and by ΔSCR are likely to be attributed to potential compounding effects from CONCRETE/WATER or FOCUS and STATION.

Models for testing the effects of the source waterfall (WATERFALL/NOT)
The overall results of mixed-effect model analysis for ΔTEMP validated Hypothesis B to substantiate the presence of a microscale WCI within a zone of more effective instantaneous cooling effects that can be detected as a greater reduction in body temperatures within 15 s: the source waterfall (S1) is significantly more effective in decreasing body temperatures within 15 s.In addition, the results of the effect of STATION suggest that the cooling effect of the source waterfall can be equally felt by the small waterfall (S2) about 3.0 m to the right of the source waterfall, as evidenced by the significant reduction in participants' body temperature at S2.

Discussion
The results of linear mixed-effects models demonstrated the feasibility of using the mean body temperature and change in body temperature within 15 s to investigate aggregated microscale and individual microscale WCI effects, respectively.This study first validated Hypothesis A and supports the presence of an aggregated microscale WCI formed by the collecting effects of the source waterfall (S1), the small waterfall (S2), and the whirlpool (S3) at the lower level.The finding from the STATION model, suggests significant microscale stress reduction associated with each of the three waterscapes at the lower level, suggesting the potential of microscale effect aggregated from the effects of the three waterscapes at the lower level.As the small waterfall was less than 0.30 m wide, it had limited evaporative cooling and auditory effects.The whirlpool generated neither evaporative cooling nor auditory effects on its own.The findings suggest that the source waterfall's microscale evaporative cooling and/or auditory effects significantly reduced participants' arousal in an aggregated microscale zone that extended at least as far as the small waterfall and whirlpool stations and potentially beyond.Therefore, the evaporative cooling effect of the source waterfall could have been the main contributor to the aggregated microscale WCI.
In addition, this study also validated Hypothesis B and suggests that the presence of a source waterfall that generates microscopic WCI can significantly reduce body temperature within 15 s.The mixed-effects model analyses for TEMPM and ΔTEMP revealed that people exhibited higher mean body temperatures at the source waterfall or the lowerlevel waterscapes, where more significant reductions in body temperature were observed.The observed decrease in body temperature may be attributed to the greater evaporative cooling effect of the waterscapes.But there can be another potential explanation: people with higher body temperatures may be more sensitive to and benefit more from the WCI effect, leading to a substantial decrease in their body temperature.These findings provide new insights into the potential role of water features in regulating body temperature, particularly for those who are more vulnerable to hot weather, and have implications for public health interventions aimed at mitigating the adverse effects of extreme temperatures.
The lack of significant correlations between SCRM or ΔSCR (for measuring event-related responses) and LEVEL, TEMPM, or ΔTEMP may be related to the lack of instantaneous stressors due to the continuous nature of the visual, auditory, and thermal effects of each station.As cognitive and thermal stress have additive effects on TEMP, a significant reduction in TEMP without significant changes in either SCL or SCR (changes in cognitive stress) suggests the presence of cooling effects beyond cooling effects from stress reduction.

Conclusion
This study offers preliminary evidence that, compared to other horizontal waterscapes and smaller vertical waterscapes, a cascading waterfall about 0.91 m tall and 1.22 m wide is more effective in reducing the mean body temperature within 15 s within 0.30 m from its edge to create a microscale WCI.The cascading waterfall is also effective in creating an aggregated microscale WCI as a zone of lower temperatures when combined with a reflecting pool extending at least 3.0 m from the waterfall.
During the pandemic and the deadly heat dome in the summer of 2021, rather than using the official indoor cooling centres for heat relief, many Portlanders without airconditioning flocked to various fountains downtown like those in Lovejoy Fountain Park.The park is known to have homeless populations resting by its waterscapes, particularly during the summer.The design parameters of the cascading waterfall and the lower reflecting pool provide a point of departure for developing form-based codes to mandate the integration of waterscapes in a way that provides effective cooling for predominantly hardscaped urban settings that are prone to the effects of UHI to offer heat relief to

Figure 1 .
Figure 1.UHIs (in red) and UCIs (in blue) in Portland, Oregon.Letters in the figures account for: (a) Rocky Butte natural area (UCI natural areas with tree shades and higher grounds); (b) Mt.Tabor Park (UCI natural areas with water reservoirs, tree shades, and higher grounds); (c) Powell Butte nature park and happy valley nature park (UCI natural areas with tree shades and higher grounds); (d) Crystal Springs Rhododendron garden and Westmoreland Park (UCI around a lake); and (e) UCI downtown.(a) During the morning hours (6 AM to 7 AM) (b) During the afternoon hours (3 PM to 4 PM) (c) During the evening hours (7 PM to 8 PM) Source: Generated by the interactive mapping tool from Sustaining Urban Places Research Lab (SUPRL), 2019, labelled by the author.https://climatecope.research.pdx.edu/UHI/#11/45.5504/-122.6545

Table 1 .
Seven waterscape stations characterized by water features.

Table 2 .
Coding scheme for independent variables.

Table 3 .
Test of normality.
a Lilliefors Significance Correction.

Table 5 .
Mixed-effect model of fixed effects for response variables (N = 476).