Parameterization of Sports Playground Experiments Applying a Hybrid Method to Analyze Microclimate and Outdoor Thermal Comfort

Abstract

Parametric simulation is an effective engineering tool for addressing sustainability challenges, yet small-scale thermal comfort assessment remains limited by plugin-hybridizing complexities and workflow inefficiencies. To address these limitations, here we propose a novel comparative workflow that integrates Lands Design and Dragonfly with the assistance of Ladybug-only (LB) and Honeybee (LB&HB) in the Grasshopper model to predict the Universal Thermal Climate Index (UTCI) as the primary indicator. A playground was selected as a sample site to provide a comprehensive training dataset for the extremely hot summer period. Sensitivity analysis was conducted to assess the impact of input uncertainties on model predictions, and the simulation model’s performance was validated against urban–rural microclimate parameters and the calculated UTCI. Among the microclimate results tested, the wind speed and air temperature predictions achieved the highest accuracy (STDE: 0.10 m/s, 0.20 °C). The UTCI simulation of the LB workflow exhibited a strong correlation between calculated UTCI values (R² = 0.90; p = 0.03). Moreover, the agreement between the LB and LB&HB workflows was strong, with simulated UTCI showing good consistency (R² = 0.70–0.80; r = 0.85–0.88). This framework successfully enables real-time UTCI heatmap analysis in simplified cubic neighborhoods. Additionally, it improves the temporal and spatial resolution of thermal predictions, providing designers with critical insights into the algorithms implemented in new workflows to facilitate urban simulation and parametric sustainability.

1. Introduction

1.1. Outdoor Thermal Comfort (OTC) in Sports Playgrounds

Outdoor thermal comfort (OTC) in campus environments has attracted increasing attention in recent years due to rising temperatures and the growing demand for climate-resilient sports facilities [1,2,3,4,5]. The sensitivity of human thermal comfort is governed by the integrated effects of building envelope design and material selection, together with outdoor environmental configurations [6]. Therefore, the built environment is closely linked to citizens’ physical health, while well-designed sports facilities enhance adolescents’ physical fitness [7]. Moreover, outdoor spaces with thermally comfortable conditions and lower pollutant levels can increase the frequency of physical activity among citizens. Among China’s urban population, university students exceed 41 million [8]. Exposure to extremely hot outdoor conditions reduces outdoor activity, increases sedentary behavior, and threatens physical health [9]. Moreover, rising global temperatures intensify thermal stress in outdoor sports facilities, increasing health risks for individuals engaged in high-intensity physical activity and contributing to higher building energy consumption. Consequently, understanding thermal comfort and climate sensitivity among university students in sports facilities is essential [10]. Previous studies on the Urban Heat Island (UHI) effects of sports facilities have mainly employed hybrid methods to assess thermal environments [5,6,10,11]. These methods combine multiple algorithms or simulation techniques to leverage their strengths, reduce single-method limitations, and address multi-objective optimization in scale determination [12]. Recent studies have emphasized environmental performance research in campus settings, particularly focusing on the quantification and improvement of OTC [1,2,3,5,9,13,14,15,16,17]. In hot–humid and tropical climates, campus outdoor spaces commonly produce thermal discomfort driven by solar radiation, vegetation shading, wind conditions, and surface materials [1,3,5], yet most studies focus on plazas [1,3,14] and large-scale landscapes or building clusters [2,13], with limited validation in small-scale, high-use spaces such as sports and semi-open exercise areas [15,17,18]. Sports ground pavements (e.g., synthetic turf and hardened surfaces) are closely linked to the surrounding thermal environment and contribute to the UHI effect in sports space design, particularly under hot conditions and prolonged solar exposure [17,19,20,21]. Accordingly, enhanced analysis of thermal comfort sensitivity in university sports spaces is crucial for improving urban thermal resilience and reducing heat exposure risks during campus physical activities.

1.2. The Hybrid Method of Parametric Simulation

The hybrid method of parametric simulation enables the closed-loop design–analysis–decision workflow in OTC to achieve parametric sustainability [22,23,24]. Parametric sustainability is an innovative approach to sustainable design and architecture that integrates the principles of parametric design with sustainability objectives [25]. Simultaneously, hybrid simulations are comparatively adopted to improve results consistency and enable more robust assessment of climate-related performance, such as different microclimate variables and thermal indices [26,27,28]. Elwy et al. [26] found that four microclimate parameters showed minor error percentages, with nearly 90% precise Physiological Equivalent Temperature (PET) in a hybrid parametric workflow. López-Cabeza et al. [27] demonstrated that Ladybug tools provided more accurate air temperature (T_a) predictions (3.8–7.5% error) than ENVI-met. Similarly, the hybrid simulation of ENVI-met and Grasshopper plugins found that Ladybug’s output has better accuracy than ENVI-met [29,30]. Kamel et al. [28] explicitly discussed a comprehensive Grasshopper-based OTC workflow (e.g., Ladybug, Honeybee, and Butterfly), demonstrating that changes in materials and urban morphology can lead to an approximately 1 °C difference in the Universal Thermal Climate Index (UTCI) between urban and rural areas. Similarly, Wai et al. [31] used Grasshopper and the Ladybug plugin with the Typical Meteorological Year (TMY) (2017–2021) dataset to perform a parametric analysis of thermal comfort on the hottest and coldest days, showing large summer diurnal variations in UTCI and the Standard Effective Temperature (SET) (approximately 20–40 °C). A simulation performance study [32] generated 200 mixed-parameter cases using Ladybug and Dragonfly and reported that the CatBoost model achieved an R² of 0.93 in predicting the mean radiant temperature (T_mrt). Previous studies have shown that integrating multiple microclimate simulation tools with parametric modeling enables systematic thermal performance assessment and supports adaptive OTC-based public space design. Soflaei et al. [33] evaluated OTC using Ladybug and Honeybee, but specific scripting logic and hybrid workflows were not explicitly described. Hence, parametric simulation errors arising from hybrid plugins and unambiguous explanations of the workflow remain difficult to address in urban microclimate and OTC. Overall, the reviewed studies indicate that current parametric sustainability research still lacks systematic mixed-parameter simulation workflows with sufficient reproducibility and workflow flexibility [28,31].

1.3. Research Gaps and Objectives

Early integration of environmental analysis and building performance simulation can effectively reduce occupant thermal exposure risks and operational energy demand at the campus and urban scales [1,2,3,16]; however, its practical implementation is constrained by the increasing complexity of parametric workflows and persistent challenges in simulation validation, which limit reliability and interpretability [23,25,26,27]. Hybrid simulation approaches integrating multiple Grasshopper plugins are increasingly applied to evaluate urban climate resilience strategies, yet uncertainties related to built-environment representation, meteorological data processing, and workflow connectivity are often overlooked [20,24,28,29,31]. Although previous studies have reported the generally consistent performance of Grasshopper-based tools for urban microclimate simulation, systematic investigation and validation of discrepancies among different simulation workflows are still required to ensure robust and reproducible assessments in Rhino-based urban design [26,27,29,32].

To address these limitations, here we develop a sensitivity assessment for key outdoor environmental parameters of school sports playgrounds by integrating microclimatic variables with UTCI into a Rhino–Grasshopper-based workflow. The proposed framework enables the identification of common thermal environmental deficiencies and design shortcomings in school sports spaces of a Cfa climate. These deficiencies are primarily related to the inadequate performance evaluation of parametric hybrid methods and insufficient assessment of ground surface materials, which can exacerbate the UHI and adversely affect students’ thermal comfort, health, and well-being during outdoor physical activities.

This study aims to benchmark the Ladybug-only (LB) and Ladybug&Honeybee (LB&HB) workflows for OTC assessment within a hybrid urban microclimate modelling framework. It is hypothesized that differences in algorithmic structure and radiative modelling between the two workflows lead to simulation discrepancies in thermal indices and spatial heat stress patterns under summer conditions. The findings are intended to support reproducible and transparent OTC assessments in sports facility design, particularly in early-stage decision-making contexts.

2. Materials and Methods

This section addresses a series of methodological issues concerning the algorithms applied by the Grasshopper plugin tools to evaluate thermal comfort in a university playground, discussing and clarifying the structured workflow script as follows: Section 2.1 and Section 2.2 present the district characterization, which identifies the built environment in the LB and LB&HB workflows. Section 2.3 and Section 2.4 describe the methods for evaluating thermal indices and the procedures for handling errors in the output data.

A parametric simulation of the sports playground was implemented on the Grasshopper platform, with the geometric model constructed in Rhino 7. The procedure commenced with the creation of a 3D site model using the Lands Design plugin, based on CAD drawings of the North Field playground at Zhejiang Normal University (ZJNU), a region characterized by a subtropical humid climate. Subsequently, an EPW weather file was acquired from the Climate.OneBuilding.Org database and processed by the Urban Weather Generator (UWG) in the Dragonfly plugin to yield localized hourly microclimate parameters. Finally, these refined parameters served as inputs for the LB and LB&HB plugins to evaluate thermal comfort and microclimate conditions, enabling a comparative analysis of the tools’ outputs and the generation of summer heatmaps (Figure 1).

2.1. Grasshopper Parametric Simulation

Grasshopper, a key parametric platform within Rhino, is central to simulating the thermal, energy, and light performance of built environments [34,35]. The LB and Honeybee (HB) plugins facilitate extensive parametric analyses at the room and building scales with complementary numerical simulations to fulfill complex research goals [28]. Given its novelty, the workflow still requires extensive validation and a combination of simulation mechanisms [36,37].

The LB tool is a parameterized building environment for simulating the thermal environment and an energy consumption plugin based on the Grasshopper platform in Rhino [23,38]. Ladybug imports standard EnergyPlus Weather files (.EPW) in Grasshopper and provides a variety of 2D and 3D designer-friendly interactive graphics to support the decision-making process [23]. Moreover, LB conducts Grasshopper to graphically represent thermal conditions and microclimates [39,40]. HB serves as the primary engine for baseline temperature simulation by parsing EPW files. Dragonfly integrates UWG to transform rural meteorological inputs into urban-scale T_a and relative humidity (RH), accounting for urban canopy effects. LB further computes radiative temperature components and thermal comfort indices, with results visualization and integrated analysis tools. Interoperability among these plugins supports consistent boundary-condition definition and model calibration [35,39,41]. T_mrt and related thermal indices are calculated using LB by integrating radiative fluxes, meteorological variables, and human body parameters via the thermal index module. T_mrt formulation is based on the Man–Environment Heat Exchange model (MENEX_2005) developed by Blazejczyk [20,28]. Given that this tool is open-source and continuously improving, recent updates have incorporated features such as simulated tree and vegetation growth [42]. Due to its novelty, the workflow still requires extensive validation, particularly for hybrid simulation experiments [36,37].

HB, an open-source plugin for Grasshopper developed by Mostapha Sadeghipour Roudsari, facilitates building performance simulation by integrating Rhino–Grasshopper with mature simulation engines (such as EnergyPlus, Radiance, and OpenStudio). The plugin streamlines thermal comfort assessment by leveraging EnergyPlus for energy modeling and Radiance for radiative calculations, allowing for the evaluation of key comfort indices including UTCI, operative temperature, wind speed (WS) and adaptive comfort. This integration provides a robust framework for analyzing human thermal perception in response to architectural and system design parameters. The HB components are used to configure simulations, incorporating factors such as occupancy schedules, material properties, and HVAC system operations. HB identifies key thermal comfort indices and visually evaluates the results to assess compliance with comfort standards in specific environmental conditions.

2.2. Grasshopper Synergistic Plugin Tools

Local UHI effects are governed by the combined influence of urban geometry, morphological characteristics, and anthropogenic activities. Accordingly, the first step of the proposed workflow involves constructing an appropriate geometric representation of the target urban area. Following established modeling practices, building footprints were extruded into solid volumes, while Brep and surface elements were used to generate a fundamentally accurate 3D model. The urban geometric model was developed based on the CityGML standard at Level of Detail 1 (LoD1), which integrates coarse terrain with building masses of uniform height. Core geospatial data were generated using the Lands Design plugin in Rhino 7.0. Given the simple geometry of the sports venue, LoD1 also represented peripheral facilities (e.g., stands, canopies, equipment rooms). The surrounding buildings were modeled through footprint extrusion to provide the urban context required for UWG morphing.

The interaction between the morphological attributes of building clusters and key environmental parameters modulates the intensity of the UHI effect. Surface thermal properties, anthropogenic heat emissions from transportation, and vegetation coverage on building envelopes collectively regulate heat absorption, storage, and release, thereby shaping local microclimates and outdoor thermal comfort. Within the Grasshopper environment, the Dragonfly plugin adapts TMY data from rural weather stations to represent localized urban climate conditions, generating EPW files compatible with building performance simulation tools. In addition to the standard meteorological variables provided in rural TMY datasets, UWG also requires key descriptors of urban morphology to compute parameters such as the average building height, simplified window-to-wall ratio, and façade characteristics. These morphological parameters are derived from the district’s 3D model developed in Rhinoceros. The urban background inputs are generated using the UWG component suite, including modules for model configuration, terrain, traffic, tree cover, and grass cover. The transformation of rural meteorological data into urban climate conditions follows four main steps: (i) defining building geometry and envelope properties (e.g., walls and windows); (ii) specifying urban background parameters; (iii) generating an urban-scale EPW file; (iv) evaluating differences between urban and rural meteorological conditions. UWG accounts for short-wave radiation absorption and long-wave radiation emission within the urban boundary layer, thereby improving the representation of urban climate conditions and reducing uncertainty in the morphed weather data. This workflow enables simulation of local UHI effects using mean urban parameters and supports efficient implementation within the Dragonfly graphical interface in Grasshopper. As shown in Figure 2, urban–rural temperature contrasts indicate consistently higher temperatures in urban areas during the early morning and evening periods, with peak values occurring in July and August. Figure 3 illustrates the weather-data morphing workflow developed in this study, highlighting the interconnections among Dragonfly components.

OTC was assessed using Grasshopper-based plugins, with UTCI calculated from key environmental and physiological variables, including RH, T_mrt, and metabolic parameters, via algorithms implemented in LB (LB v1.8.0) and HB (HB v1.8.0) (Figure 4). This study proposes a methodological framework to benchmark the LB and LB&HB workflows by quantifying discrepancies in simulated microclimatic variables and thermal comfort indices. A comparative analysis shows that UTCI results differ between the LB tool (“LB UTCI Comfort” component) and the combined LB and HB tool (“HB UTCI Comfort Map” component), owing to their distinct computational methods. The “LB UTCI Comfort” component integrates several environmental parameters (such as T_a, RH, T_mrt, and WS) to calculate UTCI. Proper execution of this simulation requires specific input connections, particularly the _air_temp and _rel_humid inputs, which must be supplied for the component to operate correctly. The LB&HB workflow employs the “HB UTCI Comfort Map” component, which uses EPW data to calculate the surface temperature and humidity. In this study, outdoor meteorological conditions (T_a, RH, and WS) are obtained directly from the EPW file processed via UWG. Spatial comfort assessment is performed using the Radiance sensor grids defined in the model. To initiate the simulation, both the _model and _epw inputs need to be connected. The resulting UTCI values were classified according to established thermal perception categories (e.g., extreme heat stress, strong heat stress, moderate heat stress, no thermal stress) as shown in Table 1, enabling quantitative evaluation of outdoor comfort conditions and supporting evidence-based design optimization for the study area.

Figure 5 and Figure 6 present the simulation workflow developed and applied in this study, illustrating the interactions among the underlying computational engines. The LB-based UTCI workflow comprises five sequential steps: (i) calculation of T_mrt; (ii) determination of solar trajectory; (iii) definition of sky–human geometric interactions; (iv) specification of the analysis period; (v) computation and output of UTCI values (Figure 5). Accurate estimation of T_mrt in urban public spaces remains challenging due to the complex interactions between short-wave and long-wave radiative fluxes. Reliable simulation requires high-resolution surface property data and robust radiative transfer algorithms [28,43]. The simulation procedure begins with the computation of T_mrt from surface temperatures using the “LB Outdoor Solar MRT” component, supplied with processed EPW data. Subsequently, the “LB Sun Path” component is employed to generate sun vectors for solar and shading analyses. The sun path is calculated for the study period to determine solar incidence angles and exposure durations, which are critical for accurately estimating solar gains and T_mrt. Using Ladybug, T_mrt is further calculated by decomposing radiative exchange, combining surface temperatures obtained from EPW with view factors derived from ray-tracing analyses. Long-wave radiation exchange with the sky is represented using a human–environment heat exchange model, while the human body is assigned default parameters for short-wave absorption and long-wave emission. The methodological framework involves specific definitions and procedures: the ground geometry within the study domain is modeled as an adiabatic virtual thermal zone, whereas discrete objects are treated as shading surfaces, excluding long-wave radiation exchange and evapotranspiration processes, particularly for vegetation. Accurate thermal comfort modeling in the UTCI framework requires a precise representation of the interaction between the sky vault and human body geometry to evaluate radiant heat gains and losses on the human surface. The “LB Human to Sky Relation” component is used to define simulated urban geometry and human physiological characteristics. The LB workflow calculates the key parameters governing the relationship between human geometry and the sky vault based on the specified human position and the surrounding environmental context. The outputs of this component can be integrated into either the “LB Outdoor Solar MRT” or the “LB Indoor Solar MRT” components to account for context shading around a human subject in T_mrt calculations. A grid of sensor points within the thermal zone, generated using the “LB Generate Point Grid” component, is employed to capture the simulation period and subsequently calculate UTCI.

Compared to the LB-only workflow, LB&HB simulations involve greater complexity due to their detailed representation of building geometry and material characteristics. Accurate parameterization of surface properties, including ground, walls, and vegetation, has a significant influence on thermal zone conditions and presents challenges for precise UTCI prediction. The UTCI simulation workflow in LB&HB comprises five sequential steps: (i) definition of building geometry and material properties, including walls, ground, shading structures, and vegetation; (ii) placement of sensor grids within the model; (iii) specification of the simulation periods; (iv) import of EPW weather data and generation of UTCI comfort maps; (v) extraction and output of results. In the HB plugin, the “HB Opaque Material” component defines the thermal and optical properties of opaque building elements (e.g., walls, roofs, and floors) (Figure 6). Users can specify key material parameters, including thickness, thermal conductivity, density, specific heat, roughness, and solar, thermal, and visible absorptance (ranging from 0 to 1), which collectively govern the material’s thermal performance and response to solar radiation. The following HB tool is employed for this purpose: the “LB Generate Point Grid” component creates a structured array of points within a defined two- or three-dimensional domain, serving as a foundational input for subsequent analyses and sensor placement. This component requires key inputs, including geometry, grid dimensions, and spacing, which collectively determine the distribution and resolution of the generated point grid. It enables systematic evaluation of environmental performance across a surface or volume, facilitates data interpolation, and prepares inputs for downstream tools such as “HB Surface Mapping” within the LB and HB workflow. The “LB Analysis Period” component is employed to define the simulation run period for a specified duration. The model and analysis period are linked via the “HB UTCI Comfort Map” component, which computes the thermal conditions for buildings adjacent to the playground. This component calculates OTC using UTCI based on biometeorological inputs. When applied over a defined point grid or surface, it classifies thermal stress levels and facilitates the visualization and optimization of pedestrian-level thermal comfort in urban environments.

2.3. Universal Thermal Climate Index (UTCI)

Quantitative evaluation of thermal environments is primarily based on thermal indices, with more than 100 indices currently reported in the literature. However, only a limited subset has been rigorously validated for their relevance to human thermal perception and pedestrian-level thermal comfort [44]. UTCI was developed to indicate the T_a in a reference environment that represented a human physiological sensitivity response close to the actual environment [26,36,44]. Among commonly used thermal indices, UTCI provides a physiologically based representation of human thermal stress across multiple spatial and temporal climate scales. Moreover, UTCI applies to many complex climatic backgrounds and geographic regions, and UTCI has been investigated and identified to be applicable for the assessment of humid subtropical climate (Cfa) [27]. Previous studies on campus thermal environment assessment have widely applied UTCI to evaluate and optimize thermal environment design strategies for campus-type settings [45,46]. UTCI is particularly sensitive to human physiological responses, as it comprehensively integrates variations in key microclimate parameters. Additionally, the specific thresholds to quantify thermal stress conditions are assessed with reference to the Python 3.7 script for UTCI in battery packs within Grasshopper, as referenced in Table 1. UTCI can also be calculated through the following simplified Equation (1) [28,35,47]:

UTCI = 3.21 + (0.872 × T_a) + (0.2459 × T_mrt) − (2.5078 × WS) − (0.0176 × RH) (°C)

(1)

where T_a is the air temperature (°C), T_mrt is the mean radiant temperature (°C), WS is the wind speed at 10 m above ground (m/s), and RH is the relative humidity of air (%).

Table 1. The UTCI range defines specific thresholds for thermal stress.

UTCI Range	UTCI Interval Value (°C)	Thermal Press Level Classification
−5	UTCI < −40	Extreme cold stress
−4	−40 ≤ UTCI < −27	Very strong cold stress
−3	−27 ≤ UTCI < 13	Strong cold stress
−2	−12 ≤ UTCI < 0	Moderate cold stress
−1	0 ≤ UTCI < 9	Slight cold stress
0	9 ≤ UTCI < 26	No thermal stress
+1	26 ≤ UTCI < 28	Slight heat stress
+2	28 ≤ UTCI < 32	Moderate heat stress
+3	32 ≤ UTCI < 38	Strong heat stress
+4	38 ≤ UTCI < 46	Very strong heat stress
+5	46 < UTCI	Extreme heat stress

Table 1. The UTCI range defines specific thresholds for thermal stress.

UTCI Range	UTCI Interval Value (°C)	Thermal Press Level Classification
−5	UTCI < −40	Extreme cold stress
−4	−40 ≤ UTCI < −27	Very strong cold stress
−3	−27 ≤ UTCI < 13	Strong cold stress
−2	−12 ≤ UTCI < 0	Moderate cold stress
−1	0 ≤ UTCI < 9	Slight cold stress
0	9 ≤ UTCI < 26	No thermal stress
+1	26 ≤ UTCI < 28	Slight heat stress
+2	28 ≤ UTCI < 32	Moderate heat stress
+3	32 ≤ UTCI < 38	Strong heat stress
+4	38 ≤ UTCI < 46	Very strong heat stress
+5	46 < UTCI	Extreme heat stress

2.4. Validation of the Microclimate Parameters and Thermal Indices

The microclimate parameters were selected as key parameters to assess the simulation’s accuracy in the sports playground, morphing urban and rural weather data from the historical climate database. Table 2 and Figure 2 provide details of these weather station data, along with their corresponding local physical parameters. The microclimate parameters are assessed using statistical metrics in Section 4.1, including the standard deviation of errors (STDE), mean absolute error (MAE), and mean bias error (MBE), following Equations (2), (3), and (4), respectively [47].

$$\mathrm{S}\mathrm{T}\mathrm{D}\mathrm{E}=\sqrt{{\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{N}}{\left(\left({\mathrm{S}}_{\mathrm{i}}-{\mathrm{U}}_{\mathrm{i}}\right)-\mathrm{M}\mathrm{B}\mathrm{E}\right)}^2}{\mathrm{N}}}}$$

(2)

$$\mathrm{M}\mathrm{A}\mathrm{E}={\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{N}}\left|{\mathrm{S}}_{\mathrm{i}}-{\mathrm{U}}_{\mathrm{i}}\right|}{\mathrm{N}}}$$

(3)

$$\mathrm{M}\mathrm{B}\mathrm{E}={\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{N}}({\mathrm{F}\mathrm{S}}_{\mathrm{i}}-{\mathrm{U}}_{\mathrm{i}})}{\mathrm{N}}}$$

(4)

where S is the simulated data, U is the calculated data, i corresponds to the time step, and N is the total number of time steps.

To compare the LB and LB&HB workflows, correlation and regression analyses were conducted between simulated UTCI and calculated UTCI. Statistical significance was evaluated using p-values, with thresholds of p < 0.05 (*). Linear associations between UTCI outputs from the two workflows were quantified using the Pearson correlation coefficient (r), where values approaching ±1 indicate stronger linear relationships. Model performance was further evaluated using the coefficient of determination (R²) to assess goodness of fit. Before regression analysis, thermal indices were screened and normalized to remove outliers and minimize the influence of period-dependent variations in minimum and maximum UTCI values, as described in Section 4.3.

3. Case Study and Experimental Settings

3.1. Study Site and Simulation Objectives

The study area spans 344 m × 408 m and is centered on a university playground at Zhejiang Normal University, Jinhua City (29°08′08″ N, 119°38′27″ E), located in a subtropical humid climate typical of the Yangtze River Delta. The playground is enclosed by campus buildings, including student dormitories, academic blocks, and supporting sports and service facilities. Surrounding structures are primarily mid- to high-rise (2–7 stories) with an average floor height of 3 m. As shown in Figure 7, the playground comprises spectator stands, a running track, an outdoor football field, and horizontal bar facilities, representing a typical spatial configuration of a university sports playground. The site is intensively used by students and residents during the afternoon and early evening (14:00–20:00), when pedestrian thermal exposure is most pronounced. Typical urban microclimate challenges involve high surface temperatures in public spaces, limited vegetation coverage, and insufficient shade, leading to rising air and radiation temperatures during hot seasons. As a result, this site serves as a representative case for parametric design and environmental simulation studies. Specifically, the sports playground’s frequent use on weekdays and holidays further amplifies its suitability as a representative healthy public space.

The building masses were extruded in Rhino 7 to generate simplified geometric forms, enabling efficient extraction of urban morphological parameters for the Dragonfly plugin in UWG to perform meteorological data morphing. Terrain elevation was constructed using the Lands Design plugin, employing an urban background model integrating trees (modeled with Brep surfaces), buildings, and ground textures to generate the urban EPW file. The simulated core area covers approximately 244,490 m², incorporating sports facilities, synthetic running tracks, asphalt surfaces, artificial turf, and scattered tree cover. The north–south-oriented sports field forms the central zone of the simulation domain.

3.2. Experimental Settings and Analysis Variables

The experimental parameters required for the UWG morphing procedure are defined in Table 2. Buildings were categorized by functional use (e.g., classrooms, dormitories, sports facilities, campus shops) and modeled in Rhino 7.0. In Figure 3, the urban EPW was generated using Honeybee, Ladybug, and Dragonfly, and simulation zones were assigned building types via the “LB Building Programs” component, consistent with the classifications in Table 2. Parameter settings for the morphing were produced with UWG of the Dragonfly plugin. The “DF Building from Solid” component in Dragonfly was employed to generate the building and playground geometry required for urban EPW data processing. According to ASHRAE climate zoning, the study area is categorized as “Mixed-Humid,” corresponding to the “Mixed-Mode” Honeybee thermal zone [48]. The window-to-wall ratio (WWR) for all building orientations was limited to 0.7 in accordance with Article 4.2.4 of the Chinese national standard GB 50189-2005 [49]. Based on this regulation, a uniform simple window ratio of 0.6 was applied to all campus public buildings. Accordingly, tree and turf coverage ratios were extracted from AutoCAD 2018 site plans. Traffic-related anthropogenic heat was specified using the “DF Traffic Parameters” component, with peak intensities of 10 W/m² and daily averages of 5.5 W/m² on weekdays and 4 W/m² on weekends, reflecting emissions from the adjacent commercial district [34,47]. The additional parameters in Table 2 (e.g., average building height, site coverage ratio, façade-to-site ratio) were computed using the “DF Assign Model UWG Properties” component, as illustrated in Figure 3.

Using the UWG-morphed EPW dataset, OTC was independently simulated for the evaluation period from 06:00 to 18:00 on days 6–18 of July and August by the LB and LB&HB workflows. The LB configuration employs fixed parameter settings to ensure model reproducibility and reinforce parameter control with a fixed northward angle of 350°, an analysis grid scale of 5, a solar trajectory scale factor of 2, and a sky-dome radius of 5. These parameters were selected to ensure consistent solar exposure and view-factor resolution across simulations, thereby reducing inter-workflow variability and facilitating a more reliable comparison of radiative sensitivity between the two workflows. A sensitive point grid of building areas with a grid size of 8 is generated using the “LB Generate Point Grid” component, providing sufficient resolution for pedestrian-level thermal assessment. A planar analysis layer was placed 1.1 m above the playground surface with a 6 m horizontal offset to capture near-ground thermal conditions and align with key circulation areas. Moreover, the sky exposure configuration for the human geometry was defined using the “LB Human to Sky Relationship” component, which governs the radiative exchange assumptions and directly affects the sensitivity of T_mrt and UTCI calculations, as seen in Figure 5.

The thermal and optical properties of opaque materials (e.g., walls, ground, canopies, vegetation) were assigned in the LB&HB workflow using the “HB Opaque Material” component (Figure 6). The core playground was modeled as an enclosed layout, with trees deliberately excluded according to the configuration shown in Figure 3, to isolate the effects of built surfaces on thermal conditions. The point grid settings were aligned with those used in the LB workflow to ensure consistency and comparability in sensitivity analyses. The material properties are summarized in Table 3. The wall construction was defined as a composite via the “HB Construction Types” component, with a thermal conductivity of 0.93 W/m·K and a density of 1800 kg/m³, corresponding to typical concrete brick masonry [50]. These parameters were carefully selected to control the thermal response of building surfaces and maintain reproducibility across simulations. The thermal properties of ground materials (such as rubber, sand, and artificial turf) were obtained from the U-value database, with rubber and artificial turf modeled on polyurethane (PU) and polyethylene (PE), respectively [51]. Material parameters for the steel canopy, concrete pavement, and vegetation (turf and shrubs) were sourced from the ENVI-met 5.1.1 database [52]. The absorption properties (thermal, solar, and visible) of opaque materials used the default HB setting, and the soil-related parameters in vegetation materials were left unchanged. The canopy defined using the “HB Shade” component had a transmittance of 0, reflectance of 0.8, and emissivity of 0.1, consistent with the wall/roof material settings in the database (Table 4).

Table 2. Parameter settings for the morphing were produced with UWG of the Dragonfly plugin.

UWG of Dragonfly	Urban Context Parameter	Setting Values
User’s input and default	Building typology	Midrise apartment; small office; warehouse; strip mall, college; retail
	Climate zones	Mixed
	Simple window ratio	0.6
	Terrain albedo	0.1
	Terrain conductivity (W/m⋅k)	1
	Traffic parameters (W/m2)	10
Calculated by Dragonfly	Average building height (m)	11.1
	Site coverage ratio	0.16
	Façade-to-site ratio	0.28
	Tree coverage ratio	0.5
	Grass coverage ratio	0.3

Table 2. Parameter settings for the morphing were produced with UWG of the Dragonfly plugin.

UWG of Dragonfly	Urban Context Parameter	Setting Values
User’s input and default	Building typology	Midrise apartment; small office; warehouse; strip mall, college; retail
	Climate zones	Mixed
	Simple window ratio	0.6
	Terrain albedo	0.1
	Terrain conductivity (W/m⋅k)	1
	Traffic parameters (W/m2)	10
Calculated by Dragonfly	Average building height (m)	11.1
	Site coverage ratio	0.16
	Façade-to-site ratio	0.28
	Tree coverage ratio	0.5
	Grass coverage ratio	0.3

Table 3. Parameter settings for the morphing were produced with two simulation processes.

Workflow Types	Urban Context Parameter	Setting Values
Calculated by Ladybug (LB) simulation processing	Compass direction (°)	350
	Scale of compass	5
	Sphere radius	5
	Scale of sun’s path	2
	Grid size	8
	Offset distance	6
Calculated by Ladybug and Honeybee (LB&HB) simulation processing	Material types	Grass; shrub; wall; plastic track; canopy; concrete pavement; sand; artificial turf
	Grid size	8
	Offset distance	6
Simulation Period	Start of simulation time	6:00 AM on 6 July; 6:00 AM on 6 August
	End of simulation time	18:00 PM on 18 July; 18:00 PM on 18 August

Table 3. Parameter settings for the morphing were produced with two simulation processes.

Workflow Types	Urban Context Parameter	Setting Values
Calculated by Ladybug (LB) simulation processing	Compass direction (°)	350
	Scale of compass	5
	Sphere radius	5
	Scale of sun’s path	2
	Grid size	8
	Offset distance	6
Calculated by Ladybug and Honeybee (LB&HB) simulation processing	Material types	Grass; shrub; wall; plastic track; canopy; concrete pavement; sand; artificial turf
	Grid size	8
	Offset distance	6
Simulation Period	Start of simulation time	6:00 AM on 6 July; 6:00 AM on 6 August
	End of simulation time	18:00 PM on 18 July; 18:00 PM on 18 August

Table 4. Parameter settings in LB&HB were produced with urban context processing.

	Urban Context Parameter	Wall	Plastic Track (Rubber)	Canopy	Concrete Pavement	Sand	Artificial Turf
Opaque materials/Shade material	HB construction types	Mass	-	-	-	-	-
	Material thickness (m)	0.24	0.6	0.02	0.01	0.15	0.06
	Material conductivity (W/m⋅k)	0.93	0.25	45	1.6	0.7	0.5
	Material density (kg/m3)	1800	1200	800	2220	835	980
	Specific heat (J/(kg⋅k))	1080	1800	4800	850	1500	1800
	Material roughness	Medium rough	Very rough	Very smooth	Medium rough	Medium rough	Very smooth
	Thermal absorption /Transmittance	0.9	0.9	0	0.5	0.9	0.9
	Solar absorption /Reflectance	0.7	0.7	0.8	0.7	0.7	0.7
	Visible absorption /Emission	1	1	0.1	1	1	1
Vegetation material	Vegetation parameter	Grass	Shrub	-	-	-	-
	Plant height (m)	0.25	1	-	-	-	-
	Leaf area index	0.075	2.5	-	-	-	-
	Leaf reflect	0.2	0.2	-	-	-	-
	Leaf emissivity	0.97	0.97	-	-	-	-
	Soil reflective	0.2	0.2	-	-	-	-
	Soil emission	0.9	0.9	-	-	-	-
	Stomatal resistance	180	180	-	-	-	-
	Soil thickness	0.1 m	0.1 m	-	-	-	-
	Soil conductivity	0.35	0.35	-	-	-	-
	Soil density	1100	1100	-	-	-	-
	Specific heat of soil	1200	1200	-	-	-	-

Table 4. Parameter settings in LB&HB were produced with urban context processing.

	Urban Context Parameter	Wall	Plastic Track (Rubber)	Canopy	Concrete Pavement	Sand	Artificial Turf
Opaque materials/Shade material	HB construction types	Mass	-	-	-	-	-
	Material thickness (m)	0.24	0.6	0.02	0.01	0.15	0.06
	Material conductivity (W/m⋅k)	0.93	0.25	45	1.6	0.7	0.5
	Material density (kg/m3)	1800	1200	800	2220	835	980
	Specific heat (J/(kg⋅k))	1080	1800	4800	850	1500	1800
	Material roughness	Medium rough	Very rough	Very smooth	Medium rough	Medium rough	Very smooth
	Thermal absorption /Transmittance	0.9	0.9	0	0.5	0.9	0.9
	Solar absorption /Reflectance	0.7	0.7	0.8	0.7	0.7	0.7
	Visible absorption /Emission	1	1	0.1	1	1	1
Vegetation material	Vegetation parameter	Grass	Shrub	-	-	-	-
	Plant height (m)	0.25	1	-	-	-	-
	Leaf area index	0.075	2.5	-	-	-	-
	Leaf reflect	0.2	0.2	-	-	-	-
	Leaf emissivity	0.97	0.97	-	-	-	-
	Soil reflective	0.2	0.2	-	-	-	-
	Soil emission	0.9	0.9	-	-	-	-
	Stomatal resistance	180	180	-	-	-	-
	Soil thickness	0.1 m	0.1 m	-	-	-	-
	Soil conductivity	0.35	0.35	-	-	-	-
	Soil density	1100	1100	-	-	-	-
	Specific heat of soil	1200	1200	-	-	-	-

4. Results and Discussion

This section presents the results of the microclimate and thermal comfort assessments conducted in this study. Section 4.1 reports the simulated microclimatic parameters across the sports playground on summer days. Section 4.2 presents the corresponding UTCI heatmaps, while Section 4.3 provides a comparative analysis of UTCI results derived from the LB and LB&HB workflows, including simulation validation and sensitivity assessment. Accordingly, Section 4.4 addresses the limitations of the hybrid simulation in this study and compares the two parametric hybrid simulation methods.

4.1. Simulated Microclimatic Parameters Comparison

The UWG-morphed urban microclimate (T_a, RH, WS, and T_mrt) was compared with rural reference data using TMY weather files. The simulation defined a core area with a point grid resolution size of 8 and a horizontal offset of 6 m across both workflows, ensuring controlled comparison and isolating model sensitivity (Figure 7 and Table 3) [28,29,35]. The temporal evolution of microclimate parameters was consistent between July and August, reflecting the typical summer heatwave pattern (Figure 8) [1,5,10]. Urban zones showed higher T_a, T_mrt, and WS, while RH was lower compared to rural references, consistent with typical UHI effects observed in previous studies [3,4,53]. The peak urban T_a occurred in the afternoon, reaching 35.7 °C in July and 34.7 °C in August, with more pronounced diurnal fluctuations than in rural areas, corroborating findings from other university campus studies in hot-humid climates [1,3,5]. The mean T_mrt difference between July and August was 0.6 °C in urban zones and 0.7 °C in rural zones. The mean RH increased by 7.4–8.4% from July to August, maintaining a persistent urban–rural contrast. The maximum WS ranged 6–8 m/s for both months, whereas the mean WS in August decreased by 1.4 m/s relative to July, indicating seasonal variability in wind exposure (

Table 5. Microclimate conditions evaluated at urban and rural areas using the Grasshopper plugin tool.

Simulation Parameter	Ta (°C)			Tmrt (°C)			RH (%)			WS (m/s)
	Max	Min	Mean	Max	Min	Mean	Max	Min	Mean	Max	Min	Mean
Urban weather in July	35.7	25.5	30.1	68.7	28.8	52.4	93	49	72	6	1	2.4
Urban weather in August	34.7	23.5	29.7	66.9	25.4	51.8	92	43	64.6	8	1	3.8
Rural weather in July	35.7	24.2	29.7	67.9	27.2	51.1	99.0	48.0	73.8	6.0	0.0	2.3
Rural weather in August	34.9	23.2	29.5	67.0	25.1	51.8	92.0	43.0	65.4	8.0	0.0	3.7

) [3,10].

The quantitative evaluation of LB and LB&HB models reveals a distinct microclimatic performance across meteorological parameters under comparisons of urban and rural conditions (

Table 6. Assessment of LB and LB&HB model accuracy metrics for air temperature (T_a), mean radiation temperature (T_mrt), wind speed (WS), and relative humidity (RH) predictions including standard deviation of errors (STDE), mean absolute error (MAE), and mean bias error (MBE) evaluated for urban and rural weather during a playground heat wave event. Lower values indicate better performance for STDE and MAE metrics, while MBE values closer to zero indicate less systematic bias.

Statistical Metrics	Ta (°C)			Tmrt (°C)			RH (%)			WS (m/s)
	STDE	MAE	MBE	STDE	MAE	STDE	MAE	MBE	STDE	MAE	MBE	STDE
July	0.22	2.49	0.26	0.87	7.30	−0.64	1.03	11.91	8.40	0.11	2.02	−1.44
August	0.20	2.50	0.46	0.85	7.40	0.56	0.93	11.14	7.44	0.10	1.87	−1.30

). For T_a predictions on hot summer days, August exhibits a lower STDE (0.2 °C) than July (0.22 °C), indicating more consistent alignment with urban and rural conditions [23,40,46]. The MAE (2.49–2.50 °C) of T_a remains low across both July and August, while urban areas show minimal MBE bias (0.26–0.46 °C) relative to rural areas. WS output performs particularly well in urban areas, with the lowest STDE and MBE (0.10–0.11 m/s, −1.30–1.44 m/s), reflecting the model’s ability to capture urban airflow patterns [3,5,10,29]. T_mrt maintains relatively low errors (STDE: 0.85–0.87 °C; MBE: −0.64 to 0.56 °C) despite diurnal variations in solar radiation and solar altitude, consistent with prior parametric microclimate simulation [32,43]. RH exhibits the largest discrepancies among parameters, with lower STDE values (0.93–1.03%) but higher MAE and MBE (11.14–11.91%, 7.44–8.40%), reflecting a pronounced negative bias. This aligns with previous studies showing RH is sensitive to surface–atmosphere interactions during hot periods. These patterns highlight the complementary strengths of each model across different microclimatic conditions, with performance variations likely resulting from their distinct parameterization schemes and handling of surface–atmosphere interactions during extreme heat events.

4.2. UTCI Maps Evaluation in the Sports Playground

By accounting for both microclimatic conditions and human thermophysiological responses, UTCI effectively captures the heat stress experienced by active users in outdoor sports spaces [5,10,20,31], offering clear advantages for heat-safety-oriented assessments in open and radiation-dominated campus environments [1,4,18]. To assess the performance of the thermal comfort simulation, the modeled UTCI values were benchmarked against outputs from the LB and LB&HB workflows during the heatwave periods (06:00–18:00, 6–18 July and 6–18 August). This section presents representative workflow results, including spatial and temporal UTCI distributions, with a focus on peak conditions at 14:00. The findings provide practical insights into pedestrian thermal exposure and enable the identification of thermally critical zones within the sports playground across the simulated periods, thereby informing subsequent engineering analyses and design interventions (Figure 9 and Figure 10).

A hybrid parametric simulation coupled with multi-objective evaluation was employed to map the spatial distribution of UTCI-based thermal sensitivity across the sports playground. Heatmaps depict the UTCI profile along a transect across the sports playground, comparing the geometric model with material-based information following the line indicated in Figure 7. The observed UTCI values of 42 °C (indicating very strong heat stress) around the artificial turf pavement at 14:00 are primarily attributed to intense direct solar radiation absorbed by the pavement materials and the facades of adjacent sports facilities and classroom buildings (Figure 10). Singh et al. [19] reported that synthetic turf significantly increases local thermal load compared to natural grass, with higher surface temperatures (range: 9.4–33.7 °C) and ambient temperatures (range: 0.5–1.2 °C), potentially exacerbating local UHI effects. Consistently, studies on playground materials indicate that synthetic grass and rubber surfaces lead to higher heat exposure risks compared to natural vegetative materials [17,54,55]. However, areas under shaded outdoor canopy and vegetation exhibited substantially lower UTCI due to reduced solar radiation and cooling from plant transpiration (Figure 9 and Figure 10), consistent with previous findings that shaded corridors and vegetation substantially improve thermal comfort in hot–humid campuses [1,5,9,18,22]. The LB&HB simulations indicated significant spatial variation in UTCI (Figure 10), closely associated with different pavement materials [1,2,3,4]. The materials were ranked from the most thermally comfortable to the least as follows: vegetation (grass/shrubs) < concrete < rubber track < artificial turf [17,18,19]. Vegetation contributes to cooling primarily via evapotranspiration and shading, which reduces the radiant heat load [3,5]. In contrast, artificial turf composed predominantly of synthetic fibers and rubber infill absorbs substantial solar radiation, leading to markedly higher surface temperatures [19,55]. This thermal discomfort likely results in reduced frequency of use during the summer months.

Figure 9 and Figure 10 show the UTCI maps produced by the LB and LB&HB workflows for two heatwave periods. Compared with LB-only, LB&HB simulations yield systematically higher UTCI values and more spatially uniform heat accumulation across the playground, reflecting enhanced representation of surface heat storage and Radiance-derived T_mrt calculations [20,23,24]. The daytime mean UTCI peaks at 36.1–37.8 °C in July, forming a broad and persistent hotspot at the playground center, while cooling is limited to narrow shadow zones along building edges. In August, the mean UTCI slightly decreases to 35.1–37.2 °C, yet the spatial hotspot pattern persists, indicating that shading geometry exerts stronger control than seasonal variation [17,19,55]. Thermal stress intensifies around 14:00, reaching 41.0–42.1 °C in July and 38.4–41.8 °C in August, with most of the playground falling within strong to very strong heat-stress categories, and this period is not recommended for continuous exercise when UTCI exceeds 38.2 °C [10]. Overall, LB&HB simulations underscore the dominant influence of solar exposure, surface thermal properties, and shading morphology on micro-scale thermal stress, particularly during midday peak conditions [35,43,45].

4.3. Comparison of UTCI with LB and LB&HB

Figure 11 presents boxplots of daily UTCI values during heatwave periods obtained from the LB workflow (Group A), the coupled LB&HB workflow (Group B), and the reference-equation-based calculations (Group C). Distinct and systematic differences are observed among the three datasets in terms of both central tendency and dispersion, indicating that the choice of simulation workflow substantially influences the magnitude and variability of estimated outdoor thermal stress [26,35,45].

During the simulation periods, Group B consistently exhibits higher median UTCI values and wider interquartile ranges than Group A, indicating more intense and variable thermal stress when surface temperature dynamics and detailed radiative exchange are explicitly resolved [19,29,43]. In contrast, Group A shows lower median UTCI values and more compact distributions, reflecting the simplified treatment of surface thermal behavior and T_mrt in the LB-only workflow [45,46]. Group C generally lies between Groups A and B, with median values more closely aligned with Group A on most days, suggesting that the reference-equation-based approach captures overall thermal conditions but underrepresents variability associated with complex surface–radiation interactions [27,43]. Despite differences in absolute magnitude, all three groups show similar day-to-day variability, with UTCI peaks occurring during the heatwave periods in mid-July and mid-August. Median UTCI values are generally lower in July (approximately 30–40 °C) and increase substantially in August, frequently exceeding 40 °C for Groups B and C, consistent with hotter August conditions. These elevated UTCI levels indicate markedly reduced OTC during peak heatwave days [1,10]. As UTCI integrates T_a, RH, WS, and T_mrt, elevated values indicate deteriorated outdoor thermal comfort. The coupled LB&HB workflow explicitly resolves surface heat storage and short- and long-wave radiative exchange through EnergyPlus and Radiance, thereby enhancing sensitivity to solar loading and shading configuration and resulting in higher and more variable UTCI estimates [35]. In contrast, the LB workflow relies on simplified surface temperature representations, which dampen radiative feedback and yield more stable UTCI distributions.

The calculated UTCI of Group C serves as an independent reference and shows closer agreement with the LB results, suggesting that simplified workflows may better approximate site-scale thermal conditions under clear-sky heatwave scenarios. Nevertheless, the LB&HB workflow remains valuable for identifying conservative, worst-case thermal exposure and for assessing the effectiveness of shading and surface-based mitigation strategies. These findings highlight the importance of selecting modeling approaches according to study objectives, balancing physical detail against output stability and computational efficiency.

The correlation analysis between model predictions and calculations provides valuable insights into the characteristics of both the LB and LB&HB workflows (Figure 12a). For the simulation days, both workflows demonstrated a high goodness of fit with the reference UTCI calculations, confirming their capability to capture the overall temporal evolution of outdoor thermal stress during heatwave conditions [27,43]. The LB shows robust coefficients (R² = 0.9), whereas the LB&HB model demonstrates moderate fitness (R²= 0.6–0.7), reflecting the increased physical complexity introduced by explicit surface heat storage and detailed radiative exchange modeling [26,35,45]. In Figure 12b, both workflows effectively reproduce UTCI variability in July and August, with strong correlations (R² = 0.7–0.8), indicating consistent representation of diurnal and inter-day thermal stress patterns [29,46]. For UTCI predictions in LB comparing LB&HB, the correlations were strongly positive, with two workflows achieving a correlation of r = 0.88 in July and a mean of r = 0.85 in August. The correlation strength varies when using TMY data generated by Dragonfly. The UTCI predictions from the LB workflow show statistically significant agreement with the reference calculations for both July (p = 0.03) and August (p = 0.04), whereas the LB&HB workflow exhibits weaker and less consistent significance (July: p = 0.16; August: p = 0.04). This aligns with previous studies indicating that simplified models better match reference calculations under clear-sky heatwaves, while detailed radiative modeling increases variability through enhanced sensitivity to solar exposure and shading [27,45]. Overall, LB provides conservative, reference-consistent estimates suitable for rapid assessment, whereas LB&HB offers a more physically detailed approach for evaluating worst-case thermal exposure and mitigation performance under extreme heat.

4.4. Limitations of the Hybrid Workflows

The hybrid workflows developed in this study are conducted to evaluate OTC within a sports playground, utilizing an integrated simulation environment that couples the Lands Design and UWG plugins with both LB and LB&HB workflows. This integrated framework proves particularly valuable for assessing the effectiveness of various mitigation strategies in urban sports facilities from a broad perspective. However, such approaches demonstrate practical utility that relies on certain simplified assumptions and algorithms in order to maintain computational efficiency and burden. For instance, the simulation of the thermal field incorporates refinements to balance accuracy and computational load, and the characterization of decorative materials along spatiotemporal variations is grounded in widely validated simulation engines.

More specifically, the validation exercise detailed in Section 4.1 demonstrates that the proposed workflow effectively predicts the outdoor microclimate under extreme summer conditions. The model exhibits strong accuracy for WS and T_a predictions, with STDE values of 0.10 m/s and 0.20 °C, respectively. However, a significant overestimation was observed for RH, as indicated by an MAE of 11.14–11.91% and an MBE of 7.44–8.40%, which substantially exceeds the STDE (0.93–1.03%) and confirms a consistent negative bias. A similar overestimation was identified for T_mrt, with an MAE of approximately 7.30–7.40 °C. This discrepancy aligns with findings from previous studies, which have reported that the integration within the workflow of the Ladybug engine tends to overestimate T_mrt by approximately 10% [37]. Similarly, a study of LB&HB indicates that a significant difference exists between the T_mrt in points exposed to direct sunlight and shadow conditions, with a 6.2 °C discrepancy from measurements [56]. These specific biases exist and have been acknowledged as effective for early design analysis, and the comprehensive workflow confirms the practicality of a robust tool for simulating microclimate conditions in urban environments [37].

The UWG tool has been widely employed in real-city contexts to predict the UHI effect, with several studies validating simulation performance against experimental measurements. These efforts confirm UWG’s capability to capture the average trend of urban T_a [35,41]. As an integrated extension module within the Ladybug environment, UWG employs simplified physical models to simulate urban T_mrt. Grasshopper plugins potentially realize higher precision, but often lack public availability or involve substantial computational load, which limits practical application within design-oriented workflows. Notably, many studies rely on detailed on-site measurements to perform multi-dimensional comparisons of simulation errors against ENVI-met [26,27]. However, this hybrid simulation increases the complexity of model validation and discrepancies arising from the update simulation engines embedded within Grasshopper’s internal plugins. In this study, an attempt was made in the Grasshopper–Rhino platform to avoid the accumulation of simulation errors from multiple models. The rural TMY data were used to generate urban microclimate conditions, and material parameters were set by the U-value database and ENVI-met model to evaluate the thermal load. Reduction in microclimatic deviations subsequently influences the outdoor thermal environment by altering the convective heat exchange between building façades and the ambient air. These results significantly affect UTCI analysis, as shown in Section 4.2 and Section 4.3, where the LB and LB&HB simulation workflows have a consistent simulation trend (R² = 0.7–0.8, r = 0.85–0.88) in albedo variation for the surface materials and fixed building geometries. The accuracy of UTCI was evaluated using a linear regression between calculations and simulations in July and August. The precision and statistical significance of the comparison results were assessed using R² values and p-values, confirming that the LB simulation process (R² = 0.9; p = 0.03) and computational results exhibit good accuracy (Figure 12). Compared to previous studies validating the accuracy of UTCI using the LB tool and the ENVI-met 4.4.0 model with R² = 0.96 [57], the simulated UTCI results in this study also demonstrate a substantial level of agreement compared to the calculated UTCI, specifically with the LB tool.

The LB workflow is limited by its inability to simulate evapotranspiration from vegetation and water surfaces [27]. Comparisons between LB and LB&HB reveal clear methodological trade-offs for evaluating outdoor sports environments. The UWG-based Dragonfly integrated LB workflow efficiently computes canopy-layer climate parameters, making it well-suited for early design stages and urban-scale analyses where many alternatives need to be explored [28,34,57,58]. In contrast, the Honeybee-integrated workflow offers more detailed calculations of radiative exchange, surface temperature, and T_mrt, which are essential in later design phases. Accurate T_mrt estimation has been shown to improve the reliability of thermal comfort metrics such as UTCI [35,43,53,59], a critical factor in sports contexts where heat stress affects athletic performance. Additionally, the Dragonfly plugin benefits from Grasshopper’s active open-source community, supporting rapid improvements in functionality and speed [27].

Several methodological limitations introduce inaccuracies into the simulations. The primary issue lies in the simplification of wind field representation: both LB and LB&HB rely on assumptions that do not capture the localized ventilation patterns characteristic of playground environments. Two factors contribute to errors in wind and UHI modeling. First, EPW-based meteorological data fails to reflect real-world wind variability, thereby influencing predictions of surface heat transfer. Second, the UWG model in Dragonfly employs averaged urban geometry and material properties, which provide only a partial characterization of the local microclimate [20]. Consequently, the transformation of boundary conditions often results in noticeable deviations in RH, particularly over turf surfaces or under humid conditions, reducing the accuracy of UTCI estimates (Table 5). This also explains why LB exhibits better agreement with reference UTCI values, whereas LB&HB is more sensitive to uncertainties related to vegetation-associated material properties.

Both workflows can estimate UTCI, but model accuracy depends heavily on the input boundary conditions. The LB model captures temporal variations in T_a and WS, and the LB&HB offers higher spatial resolution via detailed ground properties and shading elements. Nevertheless, neither approach inherently resolves airflow complexity as a key limitation for thermal comfort prediction in semi-enclosed or geometrically complex sports venues. The following methodological limitations should be noted:

• Both workflows rely on simplified assumptions and cannot resolve local ventilation effects or accurately model wind fields, which are particularly significant in stadium environments [60].
• UWG introduces deviations in relative humidity and evapotranspiration effects, particularly when irrigating lawns or under high humidity conditions, leading to bias in UTCI calculations [44].
• The quality and representativeness of EPW data are still limiting factors, as discrepancies between meteorological station data and field conditions can affect the reliability of the output results [26].
• HB improves radiation prediction accuracy; its results depend heavily on the precise optical properties of surfaces and shading geometry, parameters often uncertain in early design stages [27,36]. In addition, Honeybee’s higher computational cost limits its applicability in large-scale parametric analyses [46].
• Without calibration against field measurements, the simulation results in this case should be interpreted as relative scenario comparisons rather than absolute predictions [29].

5. Conclusions

A new hybrid workflow for the outdoor playground spaces and thermal-sensitive assessment has been developed within Grasshopper for integrating parametric sustainability. Grasshopper is one of the free plugins applied to integrate algorithmic commands into digital form to connect various plugins, such as Lands Design, Honeybee, Ladybug, Dragonfly, etc. Each of these plugins is used for the previously described workflow, but how to use the Lands Design tool linking to terrain and landscape design is uninterpreted. Microclimate parameters and UTCI are used to compare the thermal environment for pedestrians in a sports playground, while UWG is used to morph the available rural datasets into an urban weather EPW dataset file. Based on the minimum-error results obtained after validating the urban–rural data transformation, the simulations show relatively small errors for T_a (STDE: 0.20 °C, MAE: 2.49 °C, MBE: 0.26 °C), WS (STDE: 0.10 m/s, MAE: 1.87 m/s, MBE: –1.30 m/s), and T_mrt (STDE: 0.85 °C, MAE: 7.30 °C, MBE: 0.56 °C). In contrast, RH exhibits a noticeably larger deviation (STDE: 0.93%, MAE: 11.15%, MBE: 7.44%). Within the simulation workflows, LB performs well and shows agreement between the simulated and calculated UTCI with a strong correlation (R² = 0.90, p < 0.05). The agreement between the LB and LB&HB workflows is also strong, with simulated UTCI showing good consistency (R² = 0.70–0.80, r < 1).

Overall, the LB and LB&HB workflows achieved satisfactory performance when coupled with Lands Design and Dragonfly. The LB workflow demonstrated faster computational efficiency and stronger agreement with reference UTCI values, whereas the slower performance of LB&HB reflects its more detailed handling of urban climate and building materials. In this study, we developed a new Grasshopper-based workflow and confirmed the plausibility of the discrepancies observed between the two simulation workflows. The results show that both workflows are capable of supporting climate-resilient design in complex terrains. Notably, this work is the first to incorporate ENVI-met parameter settings, such as opaque, shading, and vegetation material definitions, to characterize the thermal zone of sports fields. Related findings provide a methodological foundation for future studies that integrate Grasshopper plugins with ENVI-met for coupled-model validation and the development of effective urban heat-mitigation strategies.