Next Article in Journal
Spatiotemporal Dynamics and Transboundary Differences in Fractional Vegetation Cover in the Red River Basin from 2000 to 2023
Previous Article in Journal
Structure, Mechanisms, and Impacts of Nocturnal Downslope Wind Events in the Taklimakan Desert
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Remote Sensing
Volume 17
Issue 17
10.3390/rs17172985
remotesensing-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Integrating Remote Sensing Data into WRF to Improve 2 M Air Temperature Simulations in the Three-River Source Region of the Tibetan Plateau

by
Yuteng Wang
1,2,3,
Lin Zhao
1,2,*,
Xianhong Meng
1,2,
Lunyu Shang
1,2,
Zhaoguo Li
1,2,
Hao Chen
1,2,
Mingshan Deng
1,2,
Yingying An
1,2 and
Yuanpu Liu
4
1
State Key Laboratory of Cryospheric Science and Frozen Soil Engineering, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou 730000, China
2
Zoige Plateau Wetland Ecosystem Research Station, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou 730000, China
3
University of Chinese Academy of Sciences, Beijing 100049, China
4
Institute of Arid Meteorology, China Meteorological Administration, Lanzhou 730000, China
*
Author to whom correspondence should be addressed.
Remote Sens. 2025, 17(17), 2985; https://doi.org/10.3390/rs17172985
Submission received: 10 May 2025 / Revised: 18 August 2025 / Accepted: 21 August 2025 / Published: 27 August 2025
Browse Figures
Versions Notes

Abstract

The Three-River Source Region (TRSR) of the Tibetan Plateau (TP) is a critical headwater area with complex alpine terrain and significant climate variability. Accurately simulating 2 m air temperature (T2) in this region remains challenging for models such as the Weather Research and Forecasting (WRF) model. This study integrated remote sensing data into the WRF model to improve T2 simulations over the TRSR. Two simulations were conducted for 2020: a control simulation with default static vegetation parameters (EXPcontrol) and a sensitivity simulation with realistic vegetation and associated surface albedo of 2020 from the Global Land Surface Satellite (GLASS) datasets (EXPglass). Results showed that incorporating the GLASS-derived datasets significantly alleviated the cold bias in simulated T2 during winter and spring, while maintaining comparable performance in summer and autumn. The EXPglass run achieved better agreement with observations (R = 0.98, p < 0.01) and reduced root-mean-square error (RMSE) by 36.4% compared to EXPcontrol. Energy balance analysis indicated that the GLASS-derived datasets enhanced solar energy absorption and increased net radiation. Consequently, EXPglass produced greater turbulent heat fluxes and warmer surface skin temperature (TSK) and T2. This study underscores the importance of accurate land surface characterization and highlights the utility of remote sensing data for improving regional climate model performance in high-altitude regions.

Graphical Abstract

1. Introduction

The Three-River Source Region (TRSR) is located in the central-eastern part of the Tibetan Plateau (TP), encompassing the headwaters of the Yangtze, Yellow, and Lancang Rivers (Figure 1). This region, often referred to as “China’s Water Tower,” contributes about 25% of the Yangtze’s flow, 49% of the Yellow’s flow, and 15% of the Lancang’s flow [1]. The TRSR also contains the world’s highest and most extensive alpine wetlands and grasslands, supporting rich biodiversity and ecosystem services [2,3]. However, owing to its fragile ecosystems and strong land atmosphere coupling, the TRSR has been recognized as a key climate change sensitive and initiation area on the TP [4,5,6]. In recent decades, the TRSR has experienced significant climate warming [7,8,9,10]. 2 m air temperature, as a fundamental indicator of surface energy balance and hydrological processes, is therefore essential for reliable climate impact assessments in this fragile highland region.
Multiple approaches exist to obtain 2 m air temperature over the TP, but each presents strengths and limitations. Station observations provide direct and accurate measurements but remain sparsely distributed in the TRSR, leaving large areas ungauged [11]. Global reanalysis products, such as NCEP–DOE Reanalysis 2 (NCEP2), the Modern-Era Retrospective Analysis for Research and Applications version 2 (MERRA-2), and the Japanese 55-year Reanalysis (JRA-55), offer long-term gridded datasets with global coverage. However, their relatively coarse spatial resolution and simplified representation of topography often lead to substantial cold biases and reduced accuracy over the TP’s complex terrain [12,13]. Satellite remote sensing products, such as Moderate Resolution Imaging Spectroradiometer (MODIS) land surface temperature, provide spatially continuous fields but represent surface skin temperature (TSK) rather than 2 m air temperature, and their accuracy is further hampered by cloud contamination and heterogeneous surface conditions [14]. These limitations highlight the need for high-resolution, physically based, dynamical downscaling approaches capable of capturing the TP’s complex land atmosphere interactions.
Regional weather models, such as the Weather Research and Forecasting model (WRF), are widely used to study climate and weather over the TP. However, due to the region’s complex terrain and land surface processes, these models often exhibit systematic errors. In particular, WRF simulations have shown a notable cold bias in T2 across the TP [15]. Several factors contribute to this cold bias, including challenges in representing snow cover, cloud–radiation interactions, and other land surface characteristics [16,17,18,19]. Among these, an important source of uncertainty lies in how vegetation and land surface properties are represented within the model. Specifically, the land surface scheme in WRF (e.g., the Noah land surface model) regulated surface energy fluxes and temperature, but it relied on prescribed vegetation parameters that may be inaccurate for a given region or time period. In the standard WRF configuration, leaf area index (LAI) and green vegetation fraction (GVF) are assigned from static lookup tables [20]. By default, dynamic vegetation phenology was disabled, and each land cover category has a fixed monthly LAI, which does not vary year-to-year. Such simplifications failed to capture the rapid environmental changes and ecological restoration efforts in that region [21]. This misrepresentation of vegetation parameters can lead to errors in simulated surface energy balance, thereby affecting the partitioning of solar radiation and ultimately impacting 2 m temperature.
Improving vegetation representation in models offers a promising way to reduce temperature biases. Remote sensing provides an effective tool to obtain realistic land surface data for model improvement. Incorporating time-varying remote sensing vegetation data and associated surface albedo into WRF has the potential to correct biases arising from the model’s default climatology. For instance, Meng et al. (2018) found that substituting WRF’s default albedo with MODIS surface albedo significantly reduced the cold bias over the TP [17]. Furthermore, integrating remote sensing land surface properties, such as MODIS-derived land surface, such as land use land cover, GVF, albedo, and LAI, has significantly enhanced the simulation of diurnal temperature cycles, improving model accuracy for temperature variability [22]. Similarly, urban climate studies, such as those by Bounoua et al. (2025), have leveraged satellite-based modeling to assess heat patterns in complex urban landscapes [23]. Recent advances in remote sensing, particularly the Global Land Surface Satellite (GLASS) product suite, offer long-term global datasets of vegetation variables derived from satellite observations [24]. Collectively, these developments underscore the importance of incorporating high-resolution remote sensing data to refine land atmosphere interaction modeling, particularly in the challenging, high-elevation environment of the TP.
Parallel efforts to reduce cold biases have also focused on high-resolution reanalysis and parameterization improvements. Datasets such as High Asia Refined analysis version 2 (HARv2), China Meteorological Administration Land Data Assimilation System (CLDAS), and Global Land Data Assimilation System (GLDAS) have been evaluated over the TP, showing better performance than coarse global reanalysis but still exhibiting persistent biases, especially during snow-covered and transitional seasons, similar to findings from Coordinated Regional Downscaling Experiment (CORDEX)—Asia regional simulations [25,26,27,28]. To address these limitations, TP-specific parameterizations have been developed, improving the representation of surface energy balance. These include snow coverage topography relationships, fresh snow albedo schemes [16], cloud macrophysics treatments [29], soil organic matter parameterizations [30], and gravel soil representations [31]. While these developments have advanced model performance, vegetation representation has received comparatively less attention despite its fundamental role in regulating surface energy fluxes.
In this study, we focused on integrating high-resolution GLASS-derived vegetation and associated surface albedo into the WRF model for the TRSR and evaluating its impacts on simulated 2 m temperature and surface energy fluxes. By updating vegetation characteristics (e.g., LAI, FVC, and albedo) with GLASS-derived data, we hypothesize that seasonal variations in vegetation can be more accurately captured, thereby improving WRF’s performance and reducing cold biases. The objectives were (1) to quantify the differences in temperature simulation between the default vegetation case and the GLASS-derived data case and (2) to diagnose the underlying physical mechanisms (in terms of energy balance) responsible for improvements. We conducted year-long simulations for 2020 and compared model results with observational data to assess performance. The findings from this research provide insight into how the integration of remote sensing data can improve regional climate modeling in complex terrains, and underscore the importance of land surface processes in the climate system of the TP.

2. Data and Methods

2.1. Model Configuration

In this study, we employed the WRF model version 4.2.1 for all simulations. A single domain was configured, covering the TRSR and surrounding areas with a horizontal resolution of 15 km and 89 × 122 grid points. All simulation outputs were archived at hourly temporal resolution, providing sufficient temporal detail for subsequent evaluation and analysis of atmospheric and surface variables. Within the WRF model, the TRSR exhibited substantial elevation gradients, ranging from approximately 2000 m to over 5000 m, reflecting its complex highland and mountainous terrain (Figure 1b). Grasslands predominantly characterized the default land cover types (Figure 1c). Simulations were conducted for the entire year of 2020, initialized at 00:00 UTC on 1 January 2020 and integrated continuously through 31 December 2020. The model was driven by initial and lateral boundary conditions from the fifth-generation ECMWF reanalysis (ERA5) dataset at 0.25° × 0.25° spatial resolution [32]. Multiple nested domains are often employed in WRF studies to progressively downscale coarse-resolution reanalysis data over complex terrain such as the TP [33,34]. However, since the ERA5 forcing resolution is already close to our simulation resolution (15 km), a single high-resolution domain is a commonly adopted and effective configuration in such cases, allowing for efficient use of computational resources while adequately capturing land atmosphere interactions [29,35].
The WRF model physics options were chosen based on the best performance in simulating summer temperature during preliminary testing (see Figures S1 and S2). Specifically, we evaluated 16 combinations of microphysics and cumulus convection parameterizations (four microphysics schemes × four cumulus schemes) in a one-month test during July 2020, comparing results to observations. The differences in temperature simulation among these trials were minor, and overall, WRF showed strong skill, with a root-mean-square error (RMSE) as low as 0.54 °C and correlation coefficients ranging from 0.87 to 0.92 (Figure S2). Based on these tests, we chose the Thompson cloud microphysics scheme [36] together with the Grell 3D ensemble cumulus parameterization for the final simulations [37].
For other physical processes, we configured parameterizations commonly validated in TP studies. Yonsei University (YSU) planetary boundary layer scheme was selected for its ability to realistically represent entrainment and mixing in high-altitude environments, and it has been shown to outperform the Mellor–Yamada–Janjic (MYJ) scheme over complex terrain in terms of near surface variables, boundary layer structures, and turbulent heat flux instability relationships [38,39]. The Noah land surface model was used to simulate soil moisture, snow cover, and land atmosphere fluxes [40]. This scheme has been extensively evaluated in hydrological and land–atmosphere coupling studies over the TP [41,42]. For radiation, the Rapid Radiative Transfer Model for GCMs (RRTMG) was applied for both shortwave and longwave radiation [43]. These schemes incorporate a correlated-k method and have demonstrated improved performance in simulating surface radiation balance and cloud–radiation interactions [44,45].

2.2. Integration of Remote Sensing Data

To isolate the impact of remote sensing data on meteorological variables, two simulations were designed:
(1) Control experiment (EXPcontrol): This simulation used WRF’s default vegetation parameters. In this simulation, the model adopts its standard land surface properties and parameterizations for vegetation. The Noah land surface model employed a lookup table of monthly LAI and vegetation fraction based on land cover type (derived from climatological MODIS data) and computes surface albedo internally, including adjustments for snow cover. Specifically, the surface albedo in EXPcontrol was determined by Noah’s default parameterization, which used a background (snow-free) albedo value for each land cover and increased it as a function of snow depth and fractional snow cover [40]. Thus, in EXPcontrol, the albedo could vary with snow coverage in the cold season, but the vegetation characteristics (LAI and FVC) are static monthly values not tuned to the specific year 2020.
(2) Sensitivity experiment using GLASS-derived datasets for 2020 (EXPglass): In this simulation, the LAI, FVC, and associated surface albedo (black sky shortwave albedo) were prescribed from the GLASS-derived datasets. Notably, the model’s surface albedo is also directly specified from the GLASS shortwave albedo product, rather than computed by the internal parameterization. In other words, WRF was explicitly forced to use the GLASS albedo as surface albedo. This approach allowed EXPglass to capture the dynamic variations in vegetation greenness and surface albedo, providing a more realistic representation of the TRSR. All other model configurations and forcings remain identical to EXPcontrol, so that the only difference between the two runs was the treatment of vegetation parameters and associated surface albedo.

2.3. Observational and Remote Sensing Datasets

The high-resolution Shuttle Radar Topography Mission (SRTM) 30 m digital elevation data were used for the visualization of regional topographic features in Figure 1, providing geographic context for the broader TP region [46]. For model evaluation, the CN05.1 daily grid dataset (with 0.25° resolution) was employed. This dataset combines station observations across China and has been widely used as a reference for model validation [47]. The GLASS LAI and FVC data provide an 8-day (sub-monthly) varying LAI and FVC at 0.05° resolution. The GLASS surface albedo product provides a similar spatiotemporal resolution. Prior to application, the GLASS-derived datasets were interpolated to match the spatial grid and temporal resolution of the WRF model.

3. Results

3.1. Temperature Simulation Performance

The two simulations revealed notable differences in their simulated T2 when compared to observations. Figure 2 evaluates the spatial, seasonal, and temporal performance of T2 simulations in 2020. Observations (Figure 2a–d) revealed strong seasonal contrasts across the TRSR, with mean T2 values of −3.47 °C in spring, 6.69 °C in summer, −1.00 °C in autumn, and −13.41 °C in winter. EXPcontrol (Figure 2e–h) reproduced the broad spatial distribution but showed systematic cold biases in spring (MB = −2.68 °C; Figure 2m) and winter (MB = −3.10 °C; Figure 2p). Incorporating GLASS-derived datasets (Figure 2i–l) substantially reduced these biases, improving MB to 0.21 °C in spring (Figure 2q) and −0.64 °C in winter (Figure 2t). The strongest improvement occurred in winter, where EXPglass reduced the cold bias by 2.46 °C, underscoring the added value of remote sensing-based vegetation and albedo data in snow-dominated seasons. In contrast, during summer and autumn (Figure 2f,g,j,k), both simulations closely reproduced the observed warm conditions, but EXPcontrol performed slightly better in terms of MB (0.03 °C and −0.05 °C; Figure 2n,o) compared to EXPglass (0.15 °C and 0.60 °C; Figure 2r,s). This suggested that although EXPglass enhanced snow-related processes, uncertainties in the remote sensing algorithm over alpine grasslands may reduce its advantage during the growing season. Nevertheless, EXPglass generally achieved higher R2 (except in autumn), indicating improved consistency with observed temporal variability (Figure 2q–t).
The regional daily time series (Figure 2u) corroborated these seasonal assessments. EXPcontrol exhibited persistent cold biases in winter and spring, yielding an annual RMSE of 2.91 °C. EXPglass reduced this error to 1.85 °C (36.4% improvement) and increased the correlation with observations (R = 0.98, p < 0.01). Overall, these results demonstrated that GLASS derived datasets significantly enhanced model performance during snow-affected seasons by better capturing vegetation–snow–albedo interactions, while their added value was less evident in summer and autumn. This highlighted both the benefits and limitations of current GLASS-derived datasets in improving high-altitude land atmosphere coupling within WRF.

3.2. Land Surface Parameter and Cloud Fraction Differences Between EXPcontrol and EXPglass

To elucidate the mechanisms behind the improved temperature simulation in EXPglass, this section examines the differences in land surface parameters between the two simulations.

3.2.1. LAI Differences Between EXPcontrol and EXPglass

Figure 3a and Figure 3b show the spatial distribution of annual mean LAI in EXPcontrol and EXPglass, respectively, while Figure 3c,d summarize the corresponding regional statistics. Both simulations exhibited the same spatial pattern, with higher LAI values in the southeastern TRSR that gradually decreased toward the northwest. However, the regional average LAI for 2020 in EXPglass was 0.54, whereas it was only around 0.35 in EXPcontrol, indicating that the GLASS data provided an LAI approximately 54.43% higher than the default. The time series of daily LAI (Figure 3d) further revealed that EXPglass LAI was consistently higher than EXPcontrol throughout 2020. Notably, from May onward, the EXPglass LAI increased more rapidly, reaching a higher peak in mid-summer (July–August) before declining in autumn, whereas the EXPcontrol LAI increased more slowly and to a lower maximum. These results suggest that the default WRF vegetation data underestimated the LAI of 2020, especially during the growing season, whereas the GLASS data indicated a rapid green-up in late spring and summer.

3.2.2. FVC Differences Between EXPcontrol and EXPglass

Interestingly, the fractional vegetation cover showed an opposite tendency. Figure 3e–h compare FVC between the two simulations. Although both simulations showed FVC decreasing from southeast to northwest, EXPcontrol generally produced higher fractional cover than EXPglass in many areas. The regional mean FVC in EXPcontrol was about 18.95%, whereas in EXPglass it was about 13.32%, reflecting a 29.68% reduction relative to EXPcontrol. The time series (Figure 3h) showed that both simulations maintained relatively low and stable FVC through winter and early spring. Around May, EXPglass FVC increased more rapidly than EXPcontrol, peaking around August and then declining. Despite this more dynamic seasonal evolution, EXPglass FVC remained lower than that of EXPcontrol for most of the year.
This discrepancy is noteworthy, as both LAI and FVC jointly regulate the land surface’s interaction with incoming radiation, thereby influencing surface energy balance and temperature regulation. Recent evaluations have shown that the GLASS FVC datasets, which drive EXPglass, provide more conservative estimates of vegetation cover in TRSR, particularly in sparsely vegetated regions. For example, in the alpine grasslands of the TRSR, the proportion of pixels with very low FVC values (0–0.1) in GLASS was about 10% higher than in other products [48]. This conservative algorithmic design avoids overestimation in heterogeneous surfaces and has been demonstrated to achieve higher accuracy against ground measurements (R2 = 0.815, RMSE = 0.120). Thus, the relatively lower FVC in EXPglass should not be regarded as a bias but rather as a robust feature of the GLASS dataset.

3.2.3. Surface Albedo Differences Between EXPcontrol and EXPglass

Surface albedo is a critical factor controlling surface energy, and it is strongly affected by both vegetation and snow in these high-altitude regions. Figure 3i–l compared the surface albedo between EXPcontrol and EXPglass. The spatial distributions revealed that EXPglass exhibited substantially lower albedo across most of the TRSR relative to EXPcontrol. This difference was especially pronounced in high-altitude areas affected by seasonal snow. The regional mean albedo in the EXPcontrol run was 0.34, whereas in EXPglass it decreased to approximately 0.22, representing a reduction of 35.61%. The time series of daily albedo (Figure 3l) demonstrated that throughout 2020, EXPglass consistently maintained lower albedo than EXPcontrol, with the largest differences occurring in winter and late spring. Moreover, EXPcontrol’s albedo showed more pronounced day-to-day variability, whereas EXPglass had a more stable seasonal evolution. These results indicated that incorporating GLASS-derived data effectively reduces modeled surface albedo, which could lead to enhanced surface warming through increased net radiation, consistent with the improved cold bias during winter and spring in EXPglass (Figure 2).

3.2.4. Snow Depth Differences Between EXPcontrol and EXPglass

Given the strong coupling between snow depth, surface albedo, and energy balance, a comparison of simulated snow depth between the two experiments was presented in Figure 4a–c. In EXPcontrol, the simulated snow depth was substantial in certain regions, particularly in the southwestern Yellow River source area and the northern Lancang source area. In contrast, EXPglass consistently exhibited lower snow depth across the domain (Figure 4b), primarily because its reduced albedo led to greater net radiation absorption, which enhanced snowmelt. On average, the annual snow depth in EXPglass was 45.56% lower than in EXPcontrol. The daily evolution further confirmed this pattern, with EXPcontrol maintaining higher snow depth throughout the year, and the greatest differences were observed during the spring season (Figure 4c). Moreover, EXPglass reasonably reproduced the observed spatial distribution of snow depth over the TRSR (Figure S3), particularly in high-altitude areas with persistent snow cover, highlighting the added value of realistic vegetation and prescribed albedo inputs.

3.2.5. Cloud Fraction Differences Between EXPcontrol and EXPglass

Cloud fraction, while not a direct land surface parameter, is modulated by surface moisture availability and heat fluxes, which in turn impact the regional radiation balance. Figure 4d–f present the simulated total cloud fraction for both simulations. In EXPcontrol (Figure 4d), a pronounced southeast-to-northwest gradient in cloud fraction was evident across the TRSR, corresponding to spatial variations in moisture availability and elevation. Notably, relatively high cloud fractions were simulated in the eastern Yellow River source and southern Lancang source areas, while lower cloud fractions were found in the northwest of TRSR. The difference map (Figure 4e) revealed that EXPglass enhanced cloud cover in regions already characterized by substantial cloudiness. The regional mean cloud fraction (Figure 4f) was consistently higher in EXPglass, with the most pronounced differences observed during the spring.
This increased cloud fraction in EXPglass could be attributed to several land–atmosphere processes. Specifically, the higher LAI and lower albedo in EXPglass led to increased net radiation absorption, which elevated surface temperatures and promoted enhanced evapotranspiration. Additionally, reduced snow depth in EXPglass exposed more soil and vegetation surfaces earlier in the year, further amplifying moisture fluxes. The resultant increase in the atmospheric humidity facilitated cloud formation. Furthermore, the warmer surface conditions enhanced convective activity, promoting vertical moisture transport and subsequent cloud development.

3.3. Impacts on Surface Radiation Fluxes

To assess the influence of vegetation-related changes on the surface radiation balance, differences in the four primary radiative flux components between EXPcontrol and EXPglass were examined (Figure 5). Both simulations exhibited a pronounced southeast-to-northwest spatial gradient in annual mean downward shortwave radiation (SW↓) across the TRSR, with lower SW↓ in the cloudy southeastern region and higher SW↓ in the clearer northwestern area (Figure 5a,b). However, EXPglass received slightly less SW↓ on average than EXPcontrol, which was mainly influenced by the greater cloud cover mentioned above. The regional mean downward shortwave flux was approximately 234.88 W·m−2 in EXPglass, compared to 244.82 W·m−2 in EXPcontrol, reflecting a reduction of approximately 4.06 % (Figure 5c). In contrast, the upward shortwave radiation (SW↑), which was primarily determined by surface albedo, was substantially lower in EXPglass. The regional mean SW↑ was 83.72 W·m−2, whereas in EXPglass it was reduced by 39.55% (Figure 5g). Thus, EXPglass retained more solar radiation at the surface.
The spatial pattern of downward longwave radiation (LW↓) in the TRSR showed an opposite southeast-to-northwest gradient, with higher values in the southeastern region and lower values in the northwest (Figure 5i,j). Compared to EXPcontrol, EXPglass exhibited a slightly higher LW↓ across most of the region. This increase was primarily attributed to the increased atmospheric moisture and higher cloud fraction, and in EXPglass, as clouds were highly effective in emitting longwave radiation towards the surface. The regional mean LW↓ in EXPcontrol was approximately 217.47 W·m−2, while in EXPglass, it increased to about 222.04 W·m−2, representing a 2.10% increase (Figure 5k). Differences in upward longwave radiation (LW↑) between the two simulations, which directly reflected land surface thermal emission, revealed a clear warming signal in EXPglass. Extensive snow cover and colder land surfaces in EXPcontrol resulted in lower LW↑ in most of TRSR (Figure 4a,b and Figure 5m). Conversely, EXPglass run, with reduced snow depth and generally higher TSK, exhibited a higher LW↑ (Figure 5n). The regional mean LW↑ in EXPcontrol was approximately 290.47 W·m−2, whereas in EXPglass, it was increased by 3.15%, which equates to 299.62 W·m−2 (Figure 5o). This increase in LW↑ was consistent with a notable surface warming observed in EXPglass.
Considering the aforementioned radiation components, EXPglass exhibited higher net shortwave radiation, primarily due to enhanced shortwave absorption despite a slight reduction in SW↓. Moreover, only a minor portion of this additional energy was offset by changes in longwave fluxes. Figure 6a and Figure 6b illustrate the spatial distribution of annual net radiation for EXPcontrol and EXPglass, respectively, while Figure 6c,d present the corresponding statistical analyses. Across the majority of the TRSR, net radiation was notably greater in EXPglass compared to EXPcontrol. On average, the regional mean net radiation in EXPcontrol was approximately 88.10 W·m−2, whereas in EXPglass, it increased to 106.70 W·m−2, representing a 21.11% enhancement relative to the EXPcontrol experiment (Figure 6c). Given that net radiation constitutes the fundamental energy source for turbulent heat fluxes, these fluxes were examined in the subsequent section.

3.4. Impacts on Surface Turbulent Heat Flux

Figure 6e–h compare the spatial and temporal characteristics of sensible heat flux (SH) between the two simulations. A pronounced increase in SH was observed in EXPglass across most areas of the region (Figure 6f). The regional mean SH in EXPcontrol was 40.71 W·m−2, whereas in EXPglass, it rose to approximately 52.07 W·m−2, representing a 27.90% increase relative to EXPcontrol (Figure 6g). The spatial distribution of this increase corresponded closely with regions characterized by reduced snow depth and lower surface albedo. Notably, the most significant differences in SH between the two simulations were also observed during the spring season (Figure 6h), a period when rapid snowmelt and surface warming were most prominent. The elevated SH in EXPglass represented a critical mechanism for land atmosphere energy exchange, contributing substantially to the observed T2 increases by facilitating more efficient heat transfer from the land surface to the lower atmosphere.
Figure 6i–l present the spatial distribution and temporal evolution of latent heat flux (LH) for both simulations. As expected, both EXPcontrol and EXPglass exhibited a clear spatial gradient, with higher LH in the humid southeastern region and lower values in the arid northwest (Figure 6i,j). However, EXPglass consistently demonstrated higher latent heat flux across nearly the entire domain, with an average increase of approximately 18.92% compared to EXPcontrol (Figure 6k). This enhancement in LH was primarily attributed to the “greening” effect resulting from the incorporation of realistic vegetation data. Specifically, the higher LAI and the earlier seasonal green-up observed in EXPglass promoted increased transpiration, leading to greater water vapor fluxes into the atmosphere. Additionally, enhanced surface warming during spring in EXPglass elevated the vapor pressure deficit, further accelerating evaporation from both soil and canopy surfaces. The increase in LH not only supported greater atmospheric moisture availability but also facilitated cloud formation, as previously discussed.

3.5. Impacts on TSK

The previously described changes in radiative and turbulent heat fluxes ultimately resulted in significant differences in both TSK and T2. The annual mean TSK in EXPcontrol (Figure 7a) indicated that the coldest surfaces were located in high-elevation regions, particularly those with greater snow depth (Figure 4a,b), while relatively warmer temperatures were observed in lower elevations and southern areas. Although EXPglass (Figure 7b) demonstrated a spatial distribution pattern comparable to that of EXPcontrol, due to identical topographic features and climatic forcing, it showed a systematic increase in TSK across the majority of the TRSR. The annual mean TSK in EXPcontrol remained below freezing, while EXPglass raised temperatures to near or above 0 °C (Figure 7c). This enhanced surface warming was consistent with T2 improvements discussed in Section 3.1, indicating that the rise in TSK effectively propagated into the lower atmosphere.

3.6. Possible Mechanism

A comparative analysis of the EXPglass and EXPcontrol simulations revealed a coherent physical mechanism by which the incorporation of GLASS-derived land surface parameters modified land atmosphere interactions and ultimately influenced temperature dynamics across the TRSR (Figure 8). The integration of the GLASS-derived datasets, which included both vegetation parameters (LAI and FVC) and surface albedo, resulted in a substantial increase in the LAI (54.43%), alongside a reduction in FVC (−29.68%), providing a more realistic spatial representation of vegetation conditions in 2020. Simultaneously, the direct application of surface albedo from the GLASS-derived datasets led to a marked decrease in surface albedo (−35.61%) compared to the default model parameterization. This decrease was attributable to the externally prescribed albedo values, rather than vegetation–snow interactions alone. However, the improved land surface representation, including higher LAI, contributed to a significant reduction in snow depth (−45.56%) through enhanced surface energy absorption.
Despite a slight decline in SW↓ (−4.06%) in EXPglass due to increased cloud fraction, the significantly lower albedo allowed much more solar energy to be retained at the surface. As a result, the net absorbed shortwave radiation increased substantially, as evidenced by a 39.55% decrease in SW↑. These radiative changes drove a sharp increase in TSK (97.52%), which in turn amplified turbulent and longwave fluxes, including LW↑ (3.15%), LH (18.92%), LW↓ (2.10%), and SH (27.90%). While the increased LH provided some cooling, the much stronger SH transfer dominated in EXPglass, facilitating more effective heating of the lower atmosphere. These combined land atmosphere interactions ultimately elevated T2 (57.15%) in EXPglass relative to EXPcontrol. These findings underscore the critical role of accurately prescribed land surface characteristics, particularly vegetation structure and surface albedo derived from remote sensing, in modulating the surface energy balance and improving the performance of regional climate simulations over high-altitude regions.

4. Discussion

This study highlighted the critical role of vegetation–snow interactions in regulating the surface energy balance during the cold season over the high-altitude regions. In contrast to the transpiration-driven cooling prevalent in mid- and low-latitude regions, increased vegetation cover in alpine winters reduced surface albedo by masking snow, thereby triggering a snow–albedo positive feedback [49,50,51,52,53]. Our simulations confirmed this mechanism: EXPglass, with higher LAI and lower prescribed albedo, substantially reduced snow depth (−45.56%) and improved winter and spring cold biases by 2–3 °C (Figure 2 and Figure 4). This warming effect significantly outweighed the cooling influence of enhanced evapotranspiration during cold seasons, resulting in net surface warming. By comparison, conventional WRF configurations, which rely on static vegetation parameters, tend to underestimate vegetation cover and neglect interannual albedo variability, while the default Noah snow albedo parameterization further exaggerates snow albedo [54]. Together, these deficiencies prolong snow persistence and lead to systematic cold biases, particularly in winter and spring. These findings underscore the necessity of improving both vegetation and associated surface albedo parameterizations to reduce persistent cold biases in high-altitude climate simulations.
At the same time, the results also showed that the advantages of GLASS-derived datasets were seasonally dependent. While GLASS-derived datasets substantially improved cold biases during snow-dominated seasons, their benefits were less evident during the growing season. In summer and autumn, EXPglass exhibited slightly larger biases than EXPcontrol. This can be partly attributed to the relatively lower FVC in the GLASS-derived dataset, which may weaken the simulated transpiration and reduce the land surface’s ability to buffer temperature extremes. Such inconsistencies highlight the importance of carefully evaluating satellite-derived vegetation products in different seasons and ecosystems.
Despite these advances, several limitations remained. First, uncertainties in satellite-derived land surface parameters, particularly due to cloud contamination in optical remote sensing products, introduced observational noise and potential bias [55]. The GLASS LAI product has been globally evaluated and was reported to achieve an RMSE of 0.86 with an R2 of 0.68 against high-resolution reference datasets, which is considerably better than MODIS (RMSE = 1.22, R2 = 0.47) [56]. For vegetation cover, the GLASS FVC dataset has been systematically validated with 44 Validation of Land European Remote sensing Instruments (VALERI) sites reporting an RMSE of 0.157 and an R2 of 0.809 [24]. Similarly, the GLASS albedo product has been comprehensively assessed against FLUXNET measurements, with reported RMSE less than 0.05 on clear days, although retrievals remain sensitive to cloud and snow contamination [57]. Collectively, these global evaluations support the overall reliability of GLASS-derived datasets for land atmosphere studies, while underscoring the need for careful quality control filtering in complex terrains such as the TRSR. Second, the generalizability of the results is limited by the single-year experimental design, which could not capture interannual variability in vegetation dynamics, snow processes, or atmospheric circulation [58]. Third, feedbacks involving precipitation, which could be influenced by surface moisture and cloud changes, were not explicitly quantified in this study [59]. These factors introduce additional uncertainties into the magnitude of vegetation–snow–climate feedbacks identified here.

5. Conclusions

In this study, the integration of remote sensing-derived vegetation parameters from the GLASS dataset (LAI, FVC, and surface albedo) into the WRF model (EXPglass) led to notable seasonal improvements in the simulation of T2 relative to the default configuration (EXPcontrol). In particular, the persistent cold bias observed during winter and spring in EXPcontrol was significantly alleviated. The EXPglass experiment produced higher surface and T2 that closely matched observational data, achieving a daily correlation exceeding 0.98 and reducing RMSE to approximately 1.85 °C. However, the improvements were not uniform across all seasons. During the growing season (summer and autumn), EXPglass performed slightly worse than EXPcontrol. This finding emphasizes that the benefits of satellite-based datasets are most pronounced during snow-dominated seasons, whereas their performance during the active vegetation period requires further evaluation.
Overall, the results underscore the importance of prescribing realistic land surface conditions in enhancing model performance, especially in high-elevation, cold regions such as the TP. The use of satellite-based vegetation parameters from GLASS in the WRF modeling framework represents an effective and practical approach for improving temperature simulations, but the seasonal dependence and dataset-specific uncertainties must be carefully considered in future applications.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/rs17172985/s1, Figure S1. Spatial distribution of 2-m air temperature (T2, °C) biases relative to observations under different combinations of microphysics (MP) and cumulus (CU) parameterization schemes in WRF. (a–d) MP2 + (CU1–CU6), (e–h) MP6 + (CU1–CU6), (i–l) MP8 + (CU1–CU6), and (m–p) MP10 + (CU1–CU6). The microphysics schemes include MP2 (Lin et al. scheme) [60], MP6 (WSM 6-class graupel scheme) [61], MP8 (Thompson scheme) [36], and MP10 (Morrison double-moment scheme) [62]. The cumulus schemes include CU1 (Kain–Fritsch scheme) [63], CU2 (Betts–Miller–Janjic scheme) [64], CU5 (Grell 3D ensemble scheme) [37], and CU6 (Modified Tiedtke scheme) [65]. Blue shading indicates cold biases, while red shading indicates warm biases. T2 Observations are derived from the CN05.1 dataset. Figure S2. Taylor diagram of 2 m air temperature (T2, °C) simulated by WRF under different combinations of microphysics (MP) and cumulus (CU) parameterization schemes. The microphysics schemes are MP2 (Lin et al. scheme) [60], MP6 (WSM 6-class graupel scheme), MP8 (Thompson scheme), and MP10 (Morrison double-moment scheme). The cumulus schemes are CU1 (Kain–Fritsch scheme), CU2 (Betts–Miller–Janjic scheme), CU5 (Grell 3D ensemble scheme), and CU6 (Modified Tiedtke scheme). T2 Observations are derived from the CN05.1 dataset. Figure S3. Spatial distribution of snow depth (cm) over the Three-River Source Region. (a) Observations in 2020 from the long term snow depth dataset of China provided by the National Tibetan Plateau Data Center at https://www.tpdc.ac.cn/zh-hans/data/df40346a-0202-4ed2-bb07-b65dfcda9368 (accessed on 28 October 2024) [66,67], and (b) WRF simulation using GLASS-derived datasets (EXPglass). Refs. [60,61,62,63,64,65,66,67] have been cited in the Supplementary Materials.

Author Contributions

Conceptualization, Y.W., L.Z. and X.M.; methodology, Y.W. and L.Z.; software, Y.W., H.C.; validation, L.Z., Z.L. and Y.A.; formal analysis, Y.W., L.Z. and X.M.; investigation, Y.W., L.Z. and M.D.; resources, H.C.; data curation, Y.W. and Y.A.; writing—original draft preparation, Y.W. and L.Z.; writing—review and editing, Y.W., L.Z., X.M., L.S., Z.L., H.C., M.D., Y.L. and Y.A.; visualization, Y.W., L.S. and M.D.; supervision, L.Z. and X.M.; project administration, L.Z. and Z.L.; funding acquisition, L.Z. and Z.L. All authors have read and agreed to the published version of the manuscript.

Funding

This research was supported by the National Natural Science Foundation of China (42325502, 42275045, 42475105), the West Light Foundation for Western Cross Team of the Chinese Academy of Sciences (xbzg-zdsys-202215), and the West Light Foundation of Chinese Academy of Sciences (E2290302).

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Wang, Z.; Song, K.; Hu, L. China’s largest scale ecological migration in the three-river headwater region. Ambio 2010, 39, 443–446. [Google Scholar] [CrossRef]
  2. Yin, T.; Feng, Y.; Liang, S.; Wang, G.; Yuan, N.; Yan, D.; Wu, P.; Kuang, X.; Wan, L. A suitable method for alpine wetland delineation: An example for the headwater area of the yellow river, Tibetan Plateau. Front. Environ. Sci. 2022, 10, 1062954. [Google Scholar] [CrossRef]
  3. Xinliang, X.; Jiyuan, L.; Quanqin, S.; Jiangwen, F. The dynamic changes of ecosystem spatial pattern and structure in the Three-River Headwaters region in Qinghai Province during recent 30 years. Geogr. Res. 2008, 27, 829–839. [Google Scholar]
  4. Xu, M.; Kang, S.; Chen, X.; Wu, H.; Wang, X.; Su, Z. Detection of hydrological variations and their impacts on vegetation from multiple satellite observations in the Three-River Source Region of the Tibetan Plateau. Sci. Total Environ. 2018, 639, 1220–1232. [Google Scholar] [CrossRef]
  5. Jiang, C.; Zhang, L. Climate change and its impact on the eco-environment of the Three-Rivers Headwater Region on the Tibetan Plateau, China. Int. J. Environ. Res. Public Health 2015, 12, 12057–12081. [Google Scholar] [CrossRef]
  6. Zhuang, J.; Li, Y.; Bai, P.; Chen, L.; Guo, X.; Xing, Y.; Feng, A.; Yu, W.; Huang, M. Changed evapotranspiration and its components induced by greening vegetation in the Three Rivers Source of the Tibetan Plateau. J. Hydrol. 2024, 633, 130970. [Google Scholar] [CrossRef]
  7. Liu, X.; Yang, M.-X.; Dong, N.-P.; Xie, W.-N.; Li, H.-G.; Guan, Z.; Wen, F.; Wang, H.; Chen, D. Increased southerly and easterly water vapor transport contributed to the dry-to-wet transition of summer precipitation over the Three-River Headwaters in the Tibetan Plateau. Adv. Clim. Change Res. 2023, 14, 502–510. [Google Scholar] [CrossRef]
  8. Wang, Y.; Cheng, C.; Xie, Y.; Liu, B.; Yin, S.; Liu, Y.; Hao, Y. Increasing trends in rainfall-runoff erosivity in the Source Region of the Three Rivers, 1961–2012. Sci. Total Environ. 2017, 592, 639–648. [Google Scholar] [CrossRef] [PubMed]
  9. Guo, B.; Zhou, Y.; Zhu, J.; Liu, W.; Wang, F.; Wang, L.; Yan, F.; Wang, F.; Yang, G.; Luo, W. Spatial patterns of ecosystem vulnerability changes during 2001–2011 in the three-river source region of the Qinghai-Tibetan Plateau, China. J. Arid Land 2016, 8, 23–35. [Google Scholar] [CrossRef]
  10. Luo, S.; Fang, X.; Lyu, S.; Ma, D.; Chang, Y.; Song, M.; Chen, H. Frozen ground temperature trends associated with climate change in the Tibetan Plateau Three River Source Region from 1980 to 2014. Clim. Res. 2016, 67, 241–255. [Google Scholar] [CrossRef]
  11. Jiao, Y.; Tseten, Y.; Yang, Q.; Bai, L. Spatiotemporal variability of near-surface air temperature lapse rates in the Qinghai–Tibet plateau using high-density meteorological observations. Sci. Rep. 2025, 15, 20702. [Google Scholar] [CrossRef]
  12. Peng, Y.; Duan, A.; Zhang, C.; Tang, B.; Zhao, X. Evaluation of the surface air temperature over the Tibetan Plateau among different reanalysis datasets. Front. Environ. Sci. 2023, 11, 1152129. [Google Scholar] [CrossRef]
  13. Decker, M.; Brunke, M.A.; Wang, Z.; Sakaguchi, K.; Zeng, X.; Bosilovich, M.G. Evaluation of the reanalysis products from GSFC, NCEP, and ECMWF using flux tower observations. J. Clim. 2012, 25, 1916–1944. [Google Scholar] [CrossRef]
  14. Phan, T.N.; Kappas, M. Application of MODIS land surface temperature data: A systematic literature review and analysis. J. Appl. Remote Sens. 2018, 12, 041501. [Google Scholar] [CrossRef]
  15. Miao, X.; Guo, W.; Hu, X.; Wang, L.; Li, Y.; Li, W. Divergent transformation of wet to cold bias on the Tibetan Plateau in climate models during snow season. Geophys. Res. Lett. 2024, 51, e2024GL110762. [Google Scholar] [CrossRef]
  16. Zhou, X.; Ding, B.; Yang, K.; Pan, J.; Ma, X.; Zhao, L.; Li, X.; Shi, J. Reducing the cold Bias of the WRF model over the Tibetan plateau by implementing a snow coverage-topography relationship and a fresh snow albedo scheme. J. Adv. Model. Earth Syst. 2023, 15, e2023MS003626. [Google Scholar] [CrossRef]
  17. Meng, X.; Lyu, S.; Zhang, T.; Zhao, L.; Li, Z.; Han, B.; Li, S.; Ma, D.; Chen, H.; Ao, Y. Simulated cold bias being improved by using MODIS time-varying albedo in the Tibetan Plateau in WRF model. Environ. Res. Lett. 2018, 13, 044028. [Google Scholar] [CrossRef]
  18. Li, Y.; Gao, Y.; Chen, G.; Wang, G.; Zhang, M. Decomposition and reduction of WRF-modeled wintertime cold biases over the Tibetan Plateau. Clim. Dyn. 2024, 62, 4189–4203. [Google Scholar] [CrossRef]
  19. Yue, S.; Yang, K.; Lu, H.; Zhou, X.; Chen, D.; Guo, W. Representation of stony surface-atmosphere interactions in WRF reduces cold and wet biases for the Southern Tibetan Plateau. J. Geophys. Res. Atmos. 2021, 126, e2021JD035291. [Google Scholar] [CrossRef]
  20. Glotfelty, T.; Ramírez-Mejía, D.; Bowden, J.; Ghilardi, A.; West, J.J. Limitations of WRF land surface models for simulating land use and land cover change in Sub-Saharan Africa and development of an improved model (CLM-AF v. 1.0). Geosci. Model Dev. 2021, 14, 3215–3249. [Google Scholar] [CrossRef]
  21. Lü, F.; Yan, X. The three rivers source region alpine grassland ecosystem was a weak carbon sink based on BEPS model analysis. Remote Sens. 2022, 14, 4795. [Google Scholar] [CrossRef]
  22. Chen, W.; Zhou, Y.; Passe, U.; Zhang, T.; Wang, C.; Asrar, G.R.; Li, Q.; Li, H. Improving estimation of diurnal land surface temperatures by integrating weather modeling with satellite observations. Remote Sens. Environ. 2024, 315, 114393. [Google Scholar] [CrossRef]
  23. Bounoua, L.; Zhang, T.; Thome, K.J.; Ed-dahmany, N.; Lachkham, M.A.; Bahi, H.; Yacoubi Khebiza, M.; Messouli, M. The Urban Archipelago Effect: A Case Study in Morocco. Urban Sci. 2025, 9, 97. [Google Scholar] [CrossRef]
  24. Liang, S.; Cheng, J.; Jia, K.; Jiang, B.; Liu, Q.; Xiao, Z.; Yao, Y.; Yuan, W.; Zhang, X.; Zhao, X. The global land surface satellite (GLASS) product suite. Bull. Am. Meteorol. Soc. 2021, 102, E323–E337. [Google Scholar] [CrossRef]
  25. Maussion, F.; Scherer, D.; Mölg, T.; Collier, E.; Curio, J.; Finkelnburg, R. Precipitation seasonality and variability over the Tibetan Plateau as resolved by the High Asia Reanalysis. J. Clim. 2014, 27, 1910–1927. [Google Scholar] [CrossRef]
  26. Wang, X.; Tolksdorf, V.; Otto, M.; Scherer, D. WRF-based dynamical downscaling of ERA5 reanalysis data for High Mountain Asia: Towards a new version of the High Asia Refined analysis. Int. J. Climatol. 2021, 41, 743–762. [Google Scholar] [CrossRef]
  27. Han, S.; Liu, B.; Shi, C.; Liu, Y.; Qiu, M.; Sun, S. Evaluation of CLDAS and GLDAS datasets for near-surface air temperature over major land areas of China. Sustainability 2020, 12, 4311. [Google Scholar] [CrossRef]
  28. Niu, X.; Tang, J.; Chen, D.; Wang, S.; Ou, T.; Fu, C. The performance of CORDEX-EA-II simulations in simulating seasonal temperature and elevation-dependent warming over the Tibetan Plateau. Clim. Dyn. 2021, 57, 1135–1153. [Google Scholar] [CrossRef]
  29. Zhao, D.; Lin, Y.; Dong, W.; Qin, Y.; Chu, W.; Yang, K.; Letu, H.; Huang, L. Alleviated WRF summer wet bias over the Tibetan Plateau using a new cloud macrophysics scheme. J. Adv. Model. Earth Syst. 2023, 15, e2023MS003616. [Google Scholar] [CrossRef]
  30. Sun, J.; Chen, Y.; Yang, K.; Lu, H.; Zhao, L.; Zheng, D. Influence of organic matter on soil hydrothermal processes in the Tibetan Plateau: Observation and parameterization. J. Hydrometeorol. 2021, 22, 2659–2674. [Google Scholar] [CrossRef]
  31. Xu, Y.; Ma, Y.; Lyu, S.; Zuo, H.; Hu, W. Coupling soil gravel parameterization into WRF: A case study of the Tibetan Plateau vortex. J. Adv. Model. Earth Syst. 2025, 17, e2024MS004923. [Google Scholar] [CrossRef]
  32. Hersbach, H.; Bell, B.; Berrisford, P.; Hirahara, S.; Horányi, A.; Muñoz-Sabater, J.; Nicolas, J.; Peubey, C.; Radu, R.; Schepers, D. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc. 2020, 146, 1999–2049. [Google Scholar] [CrossRef]
  33. Maussion, F.; Scherer, D.; Finkelnburg, R.; Richters, J.; Yang, W.; Yao, T. WRF simulation of a precipitation event over the Tibetan Plateau, China–an assessment using remote sensing and ground observations. Hydrol. Earth Syst. Sci. 2011, 15, 1795–1817. [Google Scholar] [CrossRef]
  34. Gupta, P.; Verma, S.; Mukhopadhyay, P.; Bhatla, R.; Payra, S. Fidelity of WRF model in simulating heat wave events over India. Sci. Rep. 2024, 14, 2693. [Google Scholar] [CrossRef]
  35. Du, S.; Zhuo, L.; Kendon, E.J.; Han, D. Exploring suitable domain size for high-resolution urban rainfall simulation. Urban Clim. 2025, 61, 102489. [Google Scholar] [CrossRef]
  36. Thompson, G.; Field, P.R.; Rasmussen, R.M.; Hall, W.D. Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part II: Implementation of a new snow parameterization. Mon. Weather Rev. 2008, 136, 5095–5115. [Google Scholar] [CrossRef]
  37. Grell, G.A.; Dévényi, D. A generalized approach to parameterizing convection combining ensemble and data assimilation techniques. Geophys. Res. Lett. 2002, 29, 38-31–38-34. [Google Scholar] [CrossRef]
  38. Hong, S.-Y.; Noh, Y.; Dudhia, J. A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Weather Rev. 2006, 134, 2318–2341. [Google Scholar] [CrossRef]
  39. Ma, Y.-F.; Wang, Y.; Xian, T.; Tian, G.; Lu, C.; Mao, X.; Wang, L.-P. Impact of PBL schemes on multiscale WRF modeling over complex terrain, Part I: Mesoscale simulations. Atmos. Res. 2024, 297, 107117. [Google Scholar] [CrossRef]
  40. Chen, F.; Dudhia, J. Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Weather Rev. 2001, 129, 569–585. [Google Scholar] [CrossRef]
  41. Sun, G.; Hu, Z.; Ma, Y.; Xie, Z.; Yang, S.; Wang, J. Analysis of local land-atmosphere coupling in rainy season over a typical underlying surface in Tibetan Plateau based on field measurements and ERA5. Atmos. Res. 2020, 243, 105025. [Google Scholar] [CrossRef]
  42. Ma, S.; Zhou, L.; Li, F.; Zhu, J. Evaluation of WRF land surface schemes in land-atmosphere exchange simulations over grassland in Southeast Tibet. Atmos. Res. 2020, 234, 104739. [Google Scholar] [CrossRef]
  43. Iacono, M.J.; Delamere, J.S.; Mlawer, E.J.; Shephard, M.W.; Clough, S.A.; Collins, W.D. Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res. Atmos. 2008, 113, D13103. [Google Scholar] [CrossRef]
  44. Liu, L.; Zhang, X.; Han, C.; Ma, Y. A performance evaluation of various physics schemes on the predictions of precipitation and temperature over the Tibet Autonomous Region of China. Atmos. Res. 2023, 292, 106878. [Google Scholar] [CrossRef]
  45. Ji, P.; Yuan, X.; Li, D. Atmospheric radiative processes accelerate ground surface warming over the southeastern Tibetan Plateau during 1998–2013. J. Clim. 2020, 33, 1881–1895. [Google Scholar] [CrossRef]
  46. Farr, T.G.; Rosen, P.A.; Caro, E.; Crippen, R.; Duren, R.; Hensley, S.; Kobrick, M.; Paller, M.; Rodriguez, E.; Roth, L. The shuttle radar topography mission. Rev. Geophys. 2007, 45, RG2004. [Google Scholar] [CrossRef]
  47. Wu, J.; GAO, X.-J. A gridded daily observation dataset over China region and comparison with the other datasets. Chin. J. Geophys. 2013, 56, 1102–1111. [Google Scholar]
  48. Huang, R.; Chen, J.; Feng, Z.; Yang, Y.; You, H.; Han, X. Fitness for purpose of several fractional vegetation cover products on monitoring vegetation cover dynamic change—A case study of an alpine grassland ecosystem. Remote Sens. 2023, 15, 1312. [Google Scholar] [CrossRef]
  49. Jin, Z.; Liang, W.; Yang, Y.; Zhang, W.; Yan, J.; Chen, X.; Li, S.; Mo, X. Separating vegetation greening and climate change controls on evapotranspiration trend over the Loess Plateau. Sci. Rep. 2017, 7, 8191. [Google Scholar] [CrossRef]
  50. Yang, Y.; Roderick, M.L.; Guo, H.; Miralles, D.G.; Zhang, L.; Fatichi, S.; Luo, X.; Zhang, Y.; McVicar, T.R.; Tu, Z. Evapotranspiration on a greening Earth. Nat. Rev. Earth Environ. 2023, 4, 626–641. [Google Scholar] [CrossRef]
  51. Shen, X.; Liu, Y.; Wu, L.; Ma, R.; Wang, Y.; Zhang, J.; Wang, L.; Liu, B.; Lu, X.; Jiang, M. Grassland greening impacts on global land surface temperature. Sci. Total Environ. 2022, 838, 155851. [Google Scholar] [CrossRef]
  52. Qi, Q.; Yang, K.; Li, H.; Ai, L.; Wang, C.; Wu, T. Negative impacts of the withered grass stems on winter snow cover over the Tibetan Plateau. Agric. For. Meteorol. 2024, 352, 110053. [Google Scholar] [CrossRef]
  53. Wu, F.; You, Q.; Pepin, N.; Zhai, P.; Kang, S. Quantifying processes of winter daytime and nighttime warming over the Tibetan Plateau. Clim. Dyn. 2025, 63, 9. [Google Scholar] [CrossRef]
  54. Liu, L.; Ma, Y.; Menenti, M.; Su, R.; Yao, N.; Ma, W. Improved parameterization of snow albedo in Noah coupled with Weather Research and Forecasting: Applicability to snow estimates for the Tibetan Plateau. Hydrol. Earth Syst. Sci. 2021, 25, 4967–4981. [Google Scholar] [CrossRef]
  55. Liu, Y.; Xiao, J.; Ju, W.; Zhu, G.; Wu, X.; Fan, W.; Li, D.; Zhou, Y. Satellite-derived LAI products exhibit large discrepancies and can lead to substantial uncertainty in simulated carbon and water fluxes. Remote Sens. Environ. 2018, 206, 174–188. [Google Scholar] [CrossRef]
  56. Li, J.; Xiao, Z. Evaluation of the version 5.0 global land surface satellite (GLASS) leaf area index product derived from MODIS data. Int. J. Remote Sens. 2020, 41, 9140–9160. [Google Scholar] [CrossRef]
  57. Liu, Q.; Wang, L.; Qu, Y.; Liu, N.; Liu, S.; Tang, H.; Liang, S. Preliminary evaluation of the long-term GLASS albedo product. Int. J. Digit. Earth 2013, 6, 69–95. [Google Scholar] [CrossRef]
  58. Jia-Wen, Z.; Xiao-Dong, Z. Comparison of the influence of interannual vegetation variability between offline and online simulations. Atmos. Ocean. Sci. Lett. 2014, 7, 453–457. [Google Scholar] [CrossRef]
  59. Li, Q.; Xue, Y. Simulated impacts of land cover change on summer climate in the Tibetan Plateau. Environ. Res. Lett. 2010, 5, 015102. [Google Scholar] [CrossRef]
  60. Lin, Y.-L.; Farley, R.D.; Orville, H.D. Bulk parameterization of the snow field in a cloud model. J. Appl. Meteorol. Climatol. 1983, 22, 1065–1092. [Google Scholar] [CrossRef]
  61. Hong, S.-Y.; Lim, J.-O.J. The WRF single-mo ment 6-class microphysics scheme (WSM6). Asia-Pac. J. Atmos. Sci. 2006, 42, 129–151. [Google Scholar]
  62. Morrison, H.; Thompson, G.; Tatarskii, V. Impact of cloud microphysics on the development of trailing stratiform precipitation in a simulated squall line: Comparison of one-and two-moment schemes. Mon. Weather Rev. 2009, 137, 991–1007. [Google Scholar] [CrossRef]
  63. Kain, J.S. The Kain–Fritsch convective parameterization: An update. J. Appl. Meteorol. 2004, 43, 170–181. [Google Scholar] [CrossRef]
  64. Janjić, Z.I. The step-mountain eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Weather Rev. 1994, 122, 927–945. [Google Scholar] [CrossRef]
  65. Zhang, C.; Wang, Y.; Hamilton, K. Improved representation of boundary layer clouds over the southeast Pacific in ARW-WRF using a modified Tiedtke cumulus parameterization scheme. Mon. Weather Rev. 2011, 139, 3489–3513. [Google Scholar] [CrossRef]
  66. Che, T.; Li, X.; Jin, R.; Armstrong, R.; Zhang, T. Snow depth derived from passive microwave remote-sensing data in China. Ann. Glaciol. 2008, 49, 145–154. [Google Scholar] [CrossRef]
  67. Dai, L.; Che, T.; Wang, J.; Zhang, P. Snow depth and snow water equivalent estimation from AMSR-E data based on a priori snow characteristics in Xinjiang, China. Remote Sens. Environ. 2012, 127, 14–29. [Google Scholar] [CrossRef]
Remotesensing 17 02985 g001
Figure 1. Study area: (a) location of the TRSR within the TP, based on SRTM; (b) topography of the TRSR derived from WRF model data; (c) land use types in the TRSR, obtained from WRF model datasets.
Figure 1. Study area: (a) location of the TRSR within the TP, based on SRTM; (b) topography of the TRSR derived from WRF model data; (c) land use types in the TRSR, obtained from WRF model datasets.
Remotesensing 17 02985 g001
Remotesensing 17 02985 g002
Figure 2. Seasonal and daily assessment of daily T2 (°C) in 2020: (a–d) observed seasonal mean T2 for spring, summer, autumn, and winter, respectively; (e–l) corresponding seasonal simulations from EXPcontrol and EXPglass; (m–t) scatterplots of seasonal mean T2 from EXPcontrol and EXPglass against observations, with mean bias (MB) and determination coefficients (R2) indicated; (u) the time series of average daily T2 in 2020. The black line represents the observed data, the red line represents EXPcontrol, and the blue line represents EXPglass. Correlation coefficient (Corr), significance level (p, assessed using Student’s t-test), and RMSE are also provided. Note that for winter, the temperature was simulated using data from December 2020, as well as January and February 2020, due to the one-year simulation period.
Figure 2. Seasonal and daily assessment of daily T2 (°C) in 2020: (a–d) observed seasonal mean T2 for spring, summer, autumn, and winter, respectively; (e–l) corresponding seasonal simulations from EXPcontrol and EXPglass; (m–t) scatterplots of seasonal mean T2 from EXPcontrol and EXPglass against observations, with mean bias (MB) and determination coefficients (R2) indicated; (u) the time series of average daily T2 in 2020. The black line represents the observed data, the red line represents EXPcontrol, and the blue line represents EXPglass. Correlation coefficient (Corr), significance level (p, assessed using Student’s t-test), and RMSE are also provided. Note that for winter, the temperature was simulated using data from December 2020, as well as January and February 2020, due to the one-year simulation period.
Remotesensing 17 02985 g002
Remotesensing 17 02985 g003
Figure 3. (a) Spatial distribution of LAI in EXPcontrol; (b) spatial distribution of LAI in EXPglass; (c) Boxplot of LAI for EXPcontrol and EXPglass (red dots indicate mean values); (d) daily variation in regional mean LAI, with light blue bars representing the relative difference between EXPglass and EXPcontrol (calculated as (EXPglass − EXPcontrol)/EXPcontrol × 100%). Panels (e–h) present corresponding results for FVC, and panels (i–l) for surface albedo.
Figure 3. (a) Spatial distribution of LAI in EXPcontrol; (b) spatial distribution of LAI in EXPglass; (c) Boxplot of LAI for EXPcontrol and EXPglass (red dots indicate mean values); (d) daily variation in regional mean LAI, with light blue bars representing the relative difference between EXPglass and EXPcontrol (calculated as (EXPglass − EXPcontrol)/EXPcontrol × 100%). Panels (e–h) present corresponding results for FVC, and panels (i–l) for surface albedo.
Remotesensing 17 02985 g003
Remotesensing 17 02985 g004
Figure 4. (a) Spatial distribution of snow depth in EXPcontrol; (b) difference in snow depth (mm) between EXPglass and EXPcontrol (EXPglass − EXPcontrol); (c) daily variation in regional mean snow depth, with the red line denoting EXPcontrol and the blue line denoting EXPglass; (d) spatial distribution of cloud fraction in EXPcontrol; (e) differences in cloud fraction between EXPglass and EXPcontrol (EXPglass − EXPcontrol); (f) daily variation in regional mean cloud fraction, with the red line denoting EXPcontrol and the black line denoting EXPglass − EXPcontrol.
Figure 4. (a) Spatial distribution of snow depth in EXPcontrol; (b) difference in snow depth (mm) between EXPglass and EXPcontrol (EXPglass − EXPcontrol); (c) daily variation in regional mean snow depth, with the red line denoting EXPcontrol and the blue line denoting EXPglass; (d) spatial distribution of cloud fraction in EXPcontrol; (e) differences in cloud fraction between EXPglass and EXPcontrol (EXPglass − EXPcontrol); (f) daily variation in regional mean cloud fraction, with the red line denoting EXPcontrol and the black line denoting EXPglass − EXPcontrol.
Remotesensing 17 02985 g004
Remotesensing 17 02985 g005
Figure 5. (a,b) Spatial distribution of SW↓ (W·m−2) in EXPcontrol and EXPglass; (c) boxplot of SW↓ for both EXPcontrol and EXPglass (red dots indicate means); (d) daily variation in regional mean SW↓, with the red line representing EXPcontrol and the blue line representing EXPglass. Panels (e–h) present corresponding results for SW↑ (W·m−2). Panels (i–l) for LW↓ (W·m−2), and panels (m–p) for LW↑ (W·m−2).
Figure 5. (a,b) Spatial distribution of SW↓ (W·m−2) in EXPcontrol and EXPglass; (c) boxplot of SW↓ for both EXPcontrol and EXPglass (red dots indicate means); (d) daily variation in regional mean SW↓, with the red line representing EXPcontrol and the blue line representing EXPglass. Panels (e–h) present corresponding results for SW↑ (W·m−2). Panels (i–l) for LW↓ (W·m−2), and panels (m–p) for LW↑ (W·m−2).
Remotesensing 17 02985 g005
Remotesensing 17 02985 g006
Figure 6. (a,b) Spatial distribution of net radiation (W·m−2) for EXPcontrol and EXPglass, respectively; (c) boxplot of net radiation for both experiments (red dots indicate mean values); (d) daily variation in regional mean net radiation, with the red line representing EXPcontrol and the blue line representing EXPglass. Panels (e–h) present corresponding results for SH (W·m−2), and panels (i–l) for LH (W·m−2).
Figure 6. (a,b) Spatial distribution of net radiation (W·m−2) for EXPcontrol and EXPglass, respectively; (c) boxplot of net radiation for both experiments (red dots indicate mean values); (d) daily variation in regional mean net radiation, with the red line representing EXPcontrol and the blue line representing EXPglass. Panels (e–h) present corresponding results for SH (W·m−2), and panels (i–l) for LH (W·m−2).
Remotesensing 17 02985 g006
Remotesensing 17 02985 g007
Figure 7. (a,b) Spatial distribution of TSK (°C) for EXPcontrol and EXPglass, respectively; (c) boxplot of TSK for both experiments (red dots indicate mean values); (d) daily variation in regional mean TSK, with the red line representing EXPcontrol and the blue line representing EXPglass.
Figure 7. (a,b) Spatial distribution of TSK (°C) for EXPcontrol and EXPglass, respectively; (c) boxplot of TSK for both experiments (red dots indicate mean values); (d) daily variation in regional mean TSK, with the red line representing EXPcontrol and the blue line representing EXPglass.
Remotesensing 17 02985 g007
Remotesensing 17 02985 g008
Figure 8. Schematic diagram illustrating variations in the surface energy balance over the TRSR. Albedo—surface albedo; LAI—leaf area index; FVC—fractional vegetation cover; Snow—snow depth (mm); SW↓—downward shortwave radiation (W·m−2); SW↑—upward shortwave radiation (W·m−2); LW↓—downward longwave radiation (W·m−2); LW↑—upward longwave radiation (W·m−2); LH—latent heat flux (W·m−2); SH—sensible heat flux (W·m−2); T2—2 m air temperature (°C); TSK—surface skin temperature (°C). The numbers in the figure represent the value of EXPcontrol, and the percentages represent the percentage change of EXPglass relative to EXPcontrol. Red text: EXPcontrol (relative increase to EXPcontrol); Black text: EXPcontrol (relative reduction to EXPcontrol).
Figure 8. Schematic diagram illustrating variations in the surface energy balance over the TRSR. Albedo—surface albedo; LAI—leaf area index; FVC—fractional vegetation cover; Snow—snow depth (mm); SW↓—downward shortwave radiation (W·m−2); SW↑—upward shortwave radiation (W·m−2); LW↓—downward longwave radiation (W·m−2); LW↑—upward longwave radiation (W·m−2); LH—latent heat flux (W·m−2); SH—sensible heat flux (W·m−2); T2—2 m air temperature (°C); TSK—surface skin temperature (°C). The numbers in the figure represent the value of EXPcontrol, and the percentages represent the percentage change of EXPglass relative to EXPcontrol. Red text: EXPcontrol (relative increase to EXPcontrol); Black text: EXPcontrol (relative reduction to EXPcontrol).
Remotesensing 17 02985 g008
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Wang, Y.; Zhao, L.; Meng, X.; Shang, L.; Li, Z.; Chen, H.; Deng, M.; An, Y.; Liu, Y. Integrating Remote Sensing Data into WRF to Improve 2 M Air Temperature Simulations in the Three-River Source Region of the Tibetan Plateau. Remote Sens. 2025, 17, 2985. https://doi.org/10.3390/rs17172985

AMA Style

Wang Y, Zhao L, Meng X, Shang L, Li Z, Chen H, Deng M, An Y, Liu Y. Integrating Remote Sensing Data into WRF to Improve 2 M Air Temperature Simulations in the Three-River Source Region of the Tibetan Plateau. Remote Sensing. 2025; 17(17):2985. https://doi.org/10.3390/rs17172985

Chicago/Turabian Style

Wang, Yuteng, Lin Zhao, Xianhong Meng, Lunyu Shang, Zhaoguo Li, Hao Chen, Mingshan Deng, Yingying An, and Yuanpu Liu. 2025. "Integrating Remote Sensing Data into WRF to Improve 2 M Air Temperature Simulations in the Three-River Source Region of the Tibetan Plateau" Remote Sensing 17, no. 17: 2985. https://doi.org/10.3390/rs17172985

APA Style

Wang, Y., Zhao, L., Meng, X., Shang, L., Li, Z., Chen, H., Deng, M., An, Y., & Liu, Y. (2025). Integrating Remote Sensing Data into WRF to Improve 2 M Air Temperature Simulations in the Three-River Source Region of the Tibetan Plateau. Remote Sensing, 17(17), 2985. https://doi.org/10.3390/rs17172985

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop