2.1. Study area

The Doon Valley, in the western Himalayan region of India within the state of Uttarakhand, forms an elongated intermontane depression bordered by the Siwalik Hills to the south and the Mussoorie Range (Lesser Himalaya) to the north. The domain spans 29.8°N – 31.2°N and 77.2°E – 78.8°E (Fig. 1). The valley’s physiographic configuration results from the tectonic interplay between the Main Boundary Thrust (MBT) and the Himalayan Frontal Fault (HFF), which created a broad synclinal basin filled with Quaternary alluvium and piedmont deposits (Valdiya 1980; Thakur, Rawat 1992). Elevation ranges from about 400 m in the southern plains to over 2000 m along the northern ridges, establishing a steep north-south gradient that strongly influences thermal stratification and wind systems.

Fig. 1.

Terrain of the study domain. The central area is the Dehradun district (Doon Valley area). The two red stars are the locations of IMD observation stations: Dehradun Mokhampur (30.32°N, 78.03°E, ~640 m altitude) and Mussoorie (30.45°N, 78.08°N, ~2005 m altitude). The map was prepared from the data of SRTM V3 product (SRTM Plus) provided by NASA JPL at a resolution of 1 arc-second (Farr et al. 2007).

This complex topography fosters diurnally reversing thermally driven circulations, characterized by daytime upslope (anabatic) and nighttime downslope (katabatic) winds, which play a crucial role in regulating the valley’s temperature and ventilation (Karki et al. 2017). At the basin floor lies Dehradun, the capital of Uttarakhand (~640 m ASL), which has undergone rapid urban expansion in recent decades. The juxtaposition of this highly urbanized valley city with the adjacent Mussoorie hill station (~2005 m ASL) – a relatively undisturbed, forest-dominated region – creates a natural laboratory for studying urban-topographic interactions and the modulation of temperature regimes and urban heat island (UHI) effects in a mountainous environment.

Figure 1 shows the terrain configuration of the study domain derived from the SRTM V3 (1-arc-second) digital elevation model (DEM) provided by NASA JPL (Farr et al. 2007). The Dehradun district (central area) represents the core of the Doon Valley; the two red stars indicate the locations of the India Meteorological Department (IMD) stations at Dehradun Mokhampur (30.32°N, 78.03°E, ~640 m) and Mussoorie (30.45°N, 78.08°E, ~2005 m). The steep elevation contrast between these stations highlights the valley’s orographic control on local meteorology.

Figure 2 presents the LULC distribution derived from the Dynamic World 10 m near-real-time dataset (Brown et al. 2022). Built-up areas (shown in red) are concentrated around the Dehradun municipal region, delineated by the dotted rectangular boundary, whereas surrounding areas are predominantly agricultural or forested. This spatial heterogeneity emphasizes the interplay between urban surfaces and natural terrain, which governs energy exchange, boundary-layer structure, and diurnal thermal gradients.

Fig. 2.

Land use/land cover (LULC) for: the Doon Valley area (left map); the Dehradun district. The region shaded in red shows the built-up area (urban region) (right map). The region inside the dotted rectangular boundary in both (a) and (b) is the Dehradun city municipal area. The LULC map was obtained using a dynamic world dataset, which is a 10 m near-real-time (NRT) LULC dataset derived from Sentinel-2 L1C collection (Brown et al. 2022). The region shown within the dotted line in (a) shows Dehradun city and its surroundings.

Thus, the Doon Valley offers a natural laboratory to study coupled urban-topographic interactions under pre-monsoon conditions, where both steep terrain and rapid urbanization jointly shape temperature distribution, local circulations, and the evolution of UHI phenomena.

2.2. WRF model

The Advanced Research WRF (ARW) modeling system version 4.2 (Skamarock et al. 2019) was used to simulate meteorology over the Doon Valley from 22 December 2020 to 1 January 2022, covering the full calendar year 2021. The first 10 days (22-31 December 2020) were treated as the model spin-up period. The year 2021 was selected because: (1) it represents a recent year with complete availability of high-quality IMD station data at both valley (Dehradun) and hill (Mussoorie) sites, and (2) it exhibited typical pre-monsoon conditions (March-June) with multiple heat events, making it suitable for UHI evaluation.

The WRF model employs a fully compressible, non-hydrostatic dynamical core with terrain-following vertical coordinates. Three nested domains (1:3 ratio) were used at 18 km, 6 km, and 2 km resolution, respectively, with the innermost domain focused on the Doon Valley (Fig. 3). Boundary conditions were derived from NCEP GDAS/FNL 0.25° global analyses. The following physical parameterizations were applied, based on prior Himalayan studies (Navale, Singh 2020; Singh et al. 2021; 2024b):

Fig. 3.

The three-fold (1:3 ratio) nested domain (D01: 18×18 km², D02: 6×6 km², D03: 2×2 km²) for WRF model set-up in the present study [obtained through WRF-domain wizard (NCAR 2008)]. The yellow circles 1, 2, and 3 depict the centers of domains D01, D02, and D03.

2.3. Urban Canopy Model – Building Effect Parameterization (UCM-BEP)

To represent urban canopy, we used the multilayer Building Effect Parameterization (BEP) scheme (Martilli et al. 2002; Mussetti et al. 2020; Hendricks, Knievel 2022), an option in WRF v4.2. The Urban Canopy Model BEP simulation was run for the pre-monsoon period (21 March – 30 June 2021, considering 21-Mar to 31-Mar as the spin-up period), during summer when UHI effects are strongest.

Unlike the single-layer UCM, BEP discretizes the urban canopy into multiple vertical layers and explicitly treats heat and momentum exchanges between roofs, walls, and roads. The momentum equation in BEP is:

where:

The energy balance in each urban layer is calculated for three distinct surfaces: roof, wall, and road. Each surface has its own surface energy balance:

where:

The vertical profiles of temperature, humidity, and wind are updated using these energy and momentum exchanges at each model level within the urban canopy, thereby improving the representation of the urban boundary layer (UBL) structure (Mussetti et al. 2020).

In this study, the NOAH land surface model (LSM) (Ek et al. 2003) was used in conjunction with the BEP-UCM framework to provide realistic treatment of soil moisture, evapotranspiration, and urban land-use interactions. The NOAH LSM supports the coupling of BEP with surface processes by enabling separate treatment of vegetated and impervious surfaces. The urban classification in the land-use data set (based on MODIS/USGS) triggers the UCM-BEP parameterizations over urban grid cells. Further, the BEP module enables building morphology parameters (e.g., average building height, street width, building fraction) to be user-specified or retrieved from lookup tables, facilitating more accurate representation of urban heterogeneity in grid-scale simulations. The default values used for Dehradun city were: average building height = 7.5 m, building width = 15 m, street width = 20 m, and impervious fraction = 0.9, typical of high-density urban cores (Garuma 2018). By explicitly resolving radiative trapping and dynamic drag in street canyons, the WRF-UCM-BEP system captures the thermal inertia of urban areas, their delayed nighttime cooling, and their suppression of thermally driven flows, which are critical to simulating urban heat island (UHI) effects in mountainous regions like the Doon Valley.

2.4. ERA-5 data

The ERA5 global reanalysis from the European Centre for Medium-Range Weather Forecasts (ECMWF) was used to evaluate large-scale temperature patterns. ERA5 employs the Integrated Forecasting System (IFS) Cycle 41r2 with a four-dimensional variational (4D-Var) data assimilation scheme that incorporates satellite, radiosonde, aircraft, and land observations. It provides hourly data at 0.25° (~31 km) resolution on 137 model levels. For improved near-surface representation, we used the ERA5-Land subset at 0.1°×0.1° (~9 km) resolution, which re-runs the IFS land component (HTESSEL) with enhanced land-surface and boundary-layer physics (Copernicus Climate Data Store 2024). ERA5-Land provides hourly temperature, soil, and flux variables with better land heterogeneity and topographic detail. In this study, ERA5-Land data for January-December 2021 were retrieved from the Copernicus Climate Data Store (CDS) and bilinearly interpolated to the WRF outer domain for consistent comparison with model outputs over the Doon Valley.

2.5. Surface observation data

Hourly meteorological observations were obtained from India Meteorological Department (IMD) stations at Dehradun Mokhampur (30.32°N, 78.03°E, 640 m) and Mussoorie (30.45°N, 78.08°E, 2005 m) (locations shown in Fig. 1). These served as the reference for validating WRF outputs.

2.6. Evaluation metrics

To evaluate the WRF simulation against the ground-based observations, we used the statistical metrics correlation (r), root mean square error (RMSE), index of agreement (IOA), mean absolute error (MAE), mean absolute percentage error (MAPE), and normalized root mean square error (NRMSE) (Willmott 1981; Karppinen et al. 2000; Borrego et al. 2008).

where S_i and O_i are simulated and observed values, n is the number of samples, and O¯ and S¯ are the means of observed and simulated values.

3.1. High resolution features: WRF simulation at 2 x 2 km resolution vs. ERA-5 Land data at 0.1°×0.1° resolution

The WRF meteorological simulation for 2 m temperature and 2 m relative humidity, and its comparison with ERA-5 Land data at 0.1°×0.1° resolution, are presented in Figures 4 and 5. The WRF simulated at higher resolution, allowing it to capture small-scale temperature variations in greater detail. This feature is evident in the more granular patterns (Jun-2021), especially in the valley and hilly regions. The ERA-5 Land data (right) has a coarser resolution (~10 km), leading to a smoother representation of temperature variations over the region; however, the finer details, such as local temperature variations in valleys and ridges, are not well captured. The WRF model is better at capturing local heating and cooling effects, likely influenced by elevation, land use, and terrain. This enhancement is important for valleys like Doon, where complex topography plays a significant role in temperature distribution. The ERA-5 data set, while useful for large-scale trends, cannot resolve fine-scale topographic influences as effectively as WRF’s higher resolution simulation. The wintertime (January 2021) comparison shows that WRF exhibits stronger cooling in the north (blue shading), corresponding to hill and mountainous areas, while southern valley regions remain warmer, showing the expected winter temperature inversion. Colder regions are less pronounced in ERA-5, suggesting weaker resolution of temperature inversion effects in the valley.

Fig. 4.

Comparison of 2 m temperature from WRF (2×2 km² resolution) simulation (left) and ERA-5 (0.1°×0.1°) resolution (right) for June 2021 (upper panels) and January 2021 (lower panels).

Fig. 5.

Comparison of 2 m relative humidity from WRF (2×2 km² resolution) simulation (left) and ERA-5 (0.1°×0.1°) resolution (right) for July 2021 (monsoon season) (upper panels) and April 2021 (pre-monsoon season) (lower panels).

Figure 5 compares 2 m relative humidity (RH) between the WRF (2×2 km2) simulation and ERA-5 (0.1°×0.1°) data for the monsoon season (July 2021) and the pre-monsoon season (April 2021). In the monsoon season (July 2021), WRF captures significant spatial variability in RH, especially over hilly areas where RH is locally higher. It shows pockets of high RH (>80%) in northern and central regions, likely the effect of orographic lifting causing moisture accumulation. Sharp humidity gradients are evident, reflecting terrain influence on moisture distribution. The ERA-5 results show a smoother distribution of RH with a large-scale gradient from south (~40-50%) to north (~80%). It lacks the fine-scale humidity variations caused by local terrain effects. The WRF’s high resolution simulation captures orographic moisture enhancement due to monsoon winds interacting with the terrain. As seen in WRF, orographic lifting causes higher RH over hills, an effect not well-represented in ERA-5. Evidently, the local

convective systems, which increase humidity locally, are better captured in WRF. In the pre-monsoon season (April 2021), the WRF shows widespread lower RH (<40%), especially in the southern and central parts of the region, while there is higher RH in some northern areas, likely due to local moisture retention in higher-altitude regions.

The ERA-5 depicts a more uniform humidity field, lacking the local variations seen in WRF, showing higher RH (~50-60%) over northern areas, but the distribution of lower RH is smoother.

3.2. Evaluation of WRF simulation with ground observations

The year-long (1 January 2021 through 30 December 2021) validation of WRF-simulated 2 m temperature was compared with observations from two ground-based meteorological stations of the India Meteorological Department (IMD) within the simulated domain. Figure 6 presents the WRF-simulated hourly temperature variations (2 m temperature) versus the ground meteorological observations for 2021 at two locations, Dehradun Mokhampur and Mussoorie. The model was validated against ground observations for hourly temperature, the daily median temperature, and the daily peak 90th percentile temperature. The validation metrics (r, IOA, MAE, and RMSE) for both locations are presented in Table 1. The hourly temperature simulations provide insight into the model’s ability to capture diurnal variations, but daily median temperature is a robust metric better reflecting the central tendency of temperature variations while minimizing the influence of extreme values. The 90th percentile temperature represents peak daytime warming events, which are crucial for assessing extreme heat conditions.

Fig. 6.

The time series of hourly 2 m temperature variations obtained from WRF simulation and the observations from the IMD meteorological stations from 1 January 2021 through 30 December 2021 at Dehradun (left) and Mussoorie (right).

The comparison between WRF-simulated and observed temperatures shows that WRF performs reasonably well in both locations but exhibits greater discrepancies in Dehradun compared to Mussoorie. The correlation coefficient and IOA are relatively high for both locations (Table 1), indicating a general agreement between simulations and observations. The MAE and RMSE are larger for Dehradun than Mussoorie (Table 1), suggesting that WRF is less precise in urban conditions. The validation metric results for daily median temperature show that the correlation and IOA remain high, but slightly better than for hourly temperatures, indicating that WRF can capture the overall temperature trend more accurately. The errors (MAE and RMSE) are still higher for Dehradun, implying that urban dynamics such as heat retention and urban canyon effect are not well represented in the model. The validation metrics for daily peak 90th percentile temperature reveal lower correlation and IOA compared to median temperature, indicating greater difficulty in capturing peak temperature events. The error metrics were larger, especially in Dehradun, where extreme daytime temperatures are more affected by urban morphology.

For Mussoorie, WRF performs better, likely because WRF accurately incorporates the land-atmosphere interactions over complex mountainous terrain. However, for the Dehradun region, the MAE and RMSE are higher, possibly because the city has grown rapidly and the WRF simulation has not included all of the urban features. In the next section, the performance of the WRF simulation, after including urban features, is discussed.

3.3. Impact of urban features: WRF-UCM (BEP) simulation and its evaluation

The WRF-UCM with BEP scheme incorporates urban morphology by considering three-dimensional urban structures, urban canyon effects, and heat exchange between buildings and the atmosphere. The WRF-UCM with BEP scheme was run for the pre-monsoon season (22 Mar 2021 to 20 Jun 2021, where the period 22 Mar 2021 to 31 Mar 2021 is the model’s spin-up period) to assess the urban effects as effectively as possible. The impact of these modifications is assessed using key statistical metrics: r, IOA, MAE, and RMSE for three temperature categories: hourly temperature, daily median temperature, and daily peak 90th percentile temperature (Fig. 7).

Fig. 7.

May 2021 mean of daily peak 90^th percentile temperature of Dehradun by WRF and WRF-UCM (BEP). The upper panel shows all of Dehradun; the lower panel shows the urban core of Dehradun that experiences higher temperatures.

At Dehradun, for hourly temperature simulation, the WRF-UCM (BEP) scheme demonstrates a marginal improvement in IOA (0.72 to 0.78) and r (0.80 to 0.83) over the standard WRF model (Table 2). The noticeable reduction in MAE (4.9 to 4.1) and RMSE (5.7 to 4.9) for the WRF-UCM (BEP) simulation suggests a more accurate representation of diurnal temperature variations. This improvement is the result of including urban heat storage and canyon radiative trapping in the WRF through UCM (BEP scheme). At Mussoorie, the improvement is less or negligible, indicating that BEP modifications primarily benefit urban regions, where the representation of building-energy interactions is crucial. The performance remains comparable to WRF, as the model is already capable of simulating natural terrain with limited urban influence. For daily median temperature at Dehradun, noticeable improvement is observed with WRF-UCM (BEP), as reflected in higher IOA values (0.53 to 0.58). The reduction in MAE (5.3 to 4.3) and RMSE (5.7 to 4.7) indicates a more precise simulation of overall daily temperature trends. The standard WRF model underestimates urban heat retention, leading to biases that BEP corrects. At Mussoorie, the results remain relatively unchanged compared to WRF, reinforcing that the urban modifications in BEP are less relevant in high-altitude, rural settings. For daily peak 90th percentile temperature performance at Dehradun, the inclusion of BEP significantly improves high-temperature peak prediction. The urban heat island (UHI) effect amplifies peak temperatures in cities, which WRF-UCM (BEP) captures better than the standard WRF model. The reduction in MAE and RMSE highlights the importance of including urban morphology in WRF for accurate extreme temperature simulation. At Mussoorie, the differences between WRF and WRF-UCM (BEP) are minor, reaffirming that urban parameterization has little influence at a hill station where urbanization is less, and therefore, the urban effect is less pronounced. The BEP scheme improves WRF’s performance significantly in Dehradun by incorporating urban physics, whereas its impact in Mussoorie is minimal. Hourly and peak temperatures benefit the most from urban modifications, as they are more sensitive to urban heat storage and canyon effects.

Figure 7 illustrates the spatial distribution of the daily peak 90th percentile temperature over Dehradun averaged for May 2021, as simulated by the WRF and WRF-UCM (BEP). The results highlight the temperature variations across the region, emphasizing the impact of urbanization on local thermal conditions. The upper panels of Figure 7 compare the temperature distribution from WRF (left) and WRF-UCM (BEP scheme) (right) for the entire Dehradun region. Both simulations depict a clear spatial gradient, with elevated temperatures in the southern urban areas and cooler temperatures in the northern high-altitude regions. However, the inclusion of the urban canopy model (BEP scheme) leads to a more pronounced urban morphology effect, with higher temperatures over the city center compared to the standard WRF simulation.

The lower panel of Figure 7 focuses on the section of Dehradun that includes the urban core of Dehradun, where the temperature differences between the WRF and WRF-UCM (BEP) simulations become more apparent. The WRF-UCM simulation shows systematically higher temperatures in built-up areas due to better representation of urban geometry, thermal properties of buildings, and reduced evaporative cooling. In contrast, the standard WRF model underestimates these effects, leading to relatively lower temperatures. The enhanced warming in the WRF-UCM simulation is consistent with the urban heat island phenomenon, which is driven by increased sensible heat flux and a lower fraction of latent heat flux in urban environments.

3.4. Assessment of UHI effect

The UHI effect was evaluated in two complementary ways: (1) UHI based on near-surface air temperature (2-m temperature (T2)) to assess the atmospheric manifestation of the phenomenon, and (2) Surface UHI based on skin temperature (TSK) and satellite-derived MODIS LST to examine the radiative surface imprint of urban warming. Both approaches provide a comprehensive understanding of nocturnal UHI dynamics. The UHI effect is evaluated for the Dehradun city areas (municipal localities) where build-up is predominant and administratively regulated by urban municipal corporations.

3.4.1. UHI (air temperature-based)

During April, the WRF-BEP simulation produced higher nocturnal UHI intensities compared to the control WRF run, with maximum ΔT exceeding 6 K at 00:00 UTC (05:30 IST) (Fig. 8). The enhancement is attributable to the representation of urban canopy processes in BEP, which trap heat more effectively during nighttime. The UHI then diminished rapidly after sunrise, reaching near-neutral or slightly negative values between 04:00 – 10:00 UTC (09:30 – 15:30 IST), before re-emerging in the late afternoon. Spatial maps for April indicate a clear warming core over central Dehradun, more pronounced in WRF-BEP, highlighting the importance of urban morphology in shaping near-surface temperature fields (Fig. 8).

Fig. 8.

Monthly variation of evening-time ambient urban heat-island (UHI) intensity (ΔT = Turban − Trural) derived from WRF and WRF-UCM-BEP simulations for the pre-monsoon months of April, May, and June 2021. The top panel represents the monthly averaged diurnal cycles of ambient ΔT, showing stronger daytime and evening UHI development in the WRF-UCM-BEP run than in the standard WRF simulation. The lower panels include sets of maps illustrating spatial distributions of UHI intensity at 14:00 UTC (≈ 19:30 IST) for the same months, enabling consistent inter-month comparisons under peak-heating conditions. Warm anomalies exceeding 4-6°C are concentrated over the Dehradun urban core, while weaker contrasts occur in June due to enhanced convective mixing. The WRF-UCM-BEP configuration produces a more compact and realistic UHI footprint, emphasizing the role of urban morphology in regulating late-afternoon ambient temperature.

In May, both WRF and WRF-BEP runs reproduced consistent nocturnal UHI peaks, but the magnitude was slightly lower than in April, with maximum ΔT values of about 4-4.5 K. Interestingly, the WRF simulation exhibited somewhat larger UHI magnitudes than WRF-BEP during late evening hours. This reversal suggests seasonal modulation, where stronger background heating in May reduces the relative contribution of urban canopy effects. Spatial plots show localized hot-spots over the city core during afternoon hours, although the areal extent of elevated temperatures remains broader in WRF compared to the more spatially confined but sharper warming simulated by WRF-BEP.

By June, the nocturnal UHI weakened further, with peak intensities generally below 3.5 K. The WRF control runs simulated higher UHI magnitudes than WRF-BEP, particularly during late evening (18:00 – 22:00 UTC). The weaker UHI during June can be attributed to enhanced convective activity and stronger mixing, which diminished the urban–rural thermal contrast. Spatial fields for June confirm this, showing more diffuse warming signatures with reduced intensity compared to April and May. The observational analyses in an earlier study (Dhankar et al. 2024) report that increased cloud cover and moisture during the monsoon transition dampen UHI signatures in Dehradun, matching the attenuation we diagnose in June.

Across all months, the UHI was strongest during the late evening and early nighttime hours, while it weakened substantially during the forenoon and was nearly absent around sunrise.

These results indicate a strong seasonal (monthly) modulation of UHI intensity, with maximum nocturnal intensities in April, a moderate decline in May, and a further reduction in June. The inclusion of the BEP scheme enhances the simulation of nocturnal warming during April, when stable atmospheric conditions prevail, but produces weaker signals during late pre-monsoon months when convective mixing dominates. This highlights the sensitivity of UHI representation to both urban canopy processes and background meteorological conditions. Prior Dehradun-focused UHI works (Mishra, Arya 2024) emphasize the role of built-up growth and vegetation gradients in setting the magnitude and timing of thermal contrasts; our ambient UHI diagnostics reproduce those space-time features at high resolution and attribute the April peak to stronger stability and reduced evaporative cooling over the urban core.

3.5. Role of mountain-breeze circulation and the dominance of urban morphology in regulating valley temperature

The Doon Valley’s north-south relief, bounded by the Siwalik Hills and Mussoorie Range, supports strong diurnal thermally driven winds, i.e., daytime upslope (anabatic) and nighttime downslope (katabatic) flows, that naturally ventilate the basin (Karki et al. 2017; Singh et al. 2021). Under clear pre-monsoon conditions, these circulations promote nocturnal cooling and maintain valley-slope temperature contrasts.

In the standard WRF simulation, the nocturnal pattern at 01:00 UTC on 15 May 2021 (Fig. 10a) clearly depicts cooler air pooled within the valley and a temperature gradient toward the northern slopes, indicating organized katabatic drainage. The WRF-UCM (BEP) run (Fig. 10b) alters this structure substantially: valley temperatures remain elevated, horizontal gradients flatten, and the difference map (Fig. 10c) reveals localized warming of +1.5-2°C across the urbanized Doon-Rispana corridor. Cooler anomalies confined to forested ridges imply that radiative trapping and heat storage within built surfaces suppress nocturnal cold-air formation.

Fig. 9.

Monthly variation of nighttime surface UHI (ΔT_skin = TSKurban − TSKrural) from WRF and WRF-UCMBEP simulations compared with MODIS 1 km land-surface-temperature (LST) for April and May 2021. The top panel shows the monthly average diurnal cycles of surface ΔT_skin, demonstrating stronger nocturnal UHI intensity and slower post-sunset cooling in the WRF-UCM-BEP simulation. The subsequent maps depict spatial distributions of nighttime surface UHI at 01:00 UTC (≈ 06:30 IST) for each month. MODIS LST (Aqua night 1:30 to 1:50 local overpass time) maps (top row of each month) serve as observational references; WRF and WRF-UCM-BEP maps illustrate modeled skin-temperature and UHI intensity fields. The WRF-UCM-BEP configuration captures a compact, warmer surface core centered over Dehradun in close agreement with MODIS observations, whereas the control WRF run exhibits diffuse heating extending into rural surroundings. These results confirm the improved capability of the BEP scheme to reproduce nocturnal surface-UHI magnitude, structure, and monthly variability under pre-monsoon conditions.

Fig. 10.

Simulated 2m temperature (WRF) at 01:00 UTC on 15 May (a); Simulated 2m temperature (WRF-BEP) at 01:00 UTC on 15 May 2025 (b); Difference in 2m temperature (WRF-BEP − WRF) (c).

Vertical cross-sections of temperature and potential temperature (Fig. 11) reinforce these contrasts. The control WRF simulation shows a shallow inversion around 700-800 m ASL and steep potential-temperature gradients typical of stable stratification. In WRF-BEP, this inversion nearly vanishes, and vertical gradients weaken, producing a more mixed boundary layer. The added urban heat flux erodes surface stability and limits nocturnal cooling.

Fig. 11.

(a) Vertical cross-section of temperature (from WRF run); (b) temperature vs. height (from surface) and θ (potential temperature) vs. height (from surface) profile (from WRF run); (c) vertical cross-section of temperature (from WRF-UCM-BEP run); (d) temperature vs. height (from surface) and θ (potential temperature) vs. height profile (from surface) (from WRF-UCM-BEP). The temperature inversion is suppressed in the WRF-UCM-BEP run.

The wind cross-sections (Fig. 12) depict a similar transition. In WRF, downslope winds of ~1-2 m s^-1 descend from the Mussoorie slopes toward the valley floor, forming a coherent mountain-breeze cell. In WRF-BEP, these flows become disorganized and weaker, occasionally reversing near the surface above the city core. Difference plots (Figs. 13-14) quantify this effect: warming of +1-2°C and wind-speed reductions up to 1.8 m s^-1 within 0-2 km AGL show that urban heat alters both thermal and dynamical structures. Surface warming enhances near-surface stability, weakening katabatic momentum and reducing natural nighttime ventilation.

Fig. 12.

(a) Vertical cross-section of wind speed and vectors (WRF), (b) Vertical cross-section of wind speed and vectors (WRF-BEP).

Fig. 13.

Vertical cross-section of temperature difference (WRF-BEP − WRF).

Fig. 14.

Vertical cross-section of wind speed difference (WRF-BEP − WRF).

Collectively, these diagnostics demonstrate that urban morphology dominates over terrain-induced thermodynamics in controlling nocturnal conditions within the Doon Valley. The multilayer BEP scheme captures how building geometry, high heat capacity, and canyon radiation trapping disrupt cold-air drainage and maintain persistent nighttime warmth. Such suppression of mountain-breeze circulation implies reduced ventilation efficiency, elevated heat stress, and potential deterioration of nighttime air quality.

Comparable results have been reported for other mountainous cities – Lisbon (Silva et al. 2021), Manila (Bilang et al. 2022), and Athens (Roukounakis et al. 2023) – but this study extends the evidence to the western Himalayas, demonstrating how rapid urbanization can override natural cooling processes in intermontane basins. The findings highlight the necessity of incorporating realistic urban morphology in mesoscale modeling to accurately represent thermal dynamics and inform climate-responsive urban planning for Himalayan valleys such as Dehradun.

The analysis revealed that the WRF-BEP framework provides a more realistic representation of urban– topography interactions in the Doon Valley, highlighting the dominance of built environment characteristics in controlling nocturnal temperature and wind patterns. These findings have critical implications for urban planning, nighttime heat stress management, and air quality regulation, particularly in rapidly urbanizing intermontane basins like Dehradun.