Item Type Article
Authors Qin, Yingzuo;Wang, Dashan;Cao, Ye’er;Cai, Xitian;Liang, Shijing;Beck, Hylke;Zeng, Zhenzhong
Citation Qin, Y., Wang, D., Cao, Y., Cai, X., Liang, S., Beck, H. E., & Zeng, Z. (2023). Sub-Grid Representation of Vegetation Cover in Land Surface Schemes Improves the Modeling of How Climate Responds to Deforestation. Geophysical Research Letters, 50(15).
Portico. https://doi.org/10.1029/2023gl104164 Eprint version Publisher's Version/PDF
DOI 10.1029/2023gl104164
Publisher American Geophysical Union (AGU) Journal Geophysical Research Letters
Rights Archived with thanks to Geophysical Research Letters under a Creative Commons license, details at: http://
creativecommons.org/licenses/by/4.0/
Download date 2023-12-20 01:29:09
Item License http://creativecommons.org/licenses/by/4.0/
Link to Item http://hdl.handle.net/10754/693473
1. Introduction
The biogeophysical climate feedbacks of land cover change are critical considerations for land management strat- egies aimed at mitigating global warming (Alkama & Cescatti, 2016; Bonan, 2008; Piao et al., 2019). Regional climate models (RCMs), such as Weather Research and Forecasting (WRF) model, are valuable tools for simu- lating the climate feedback of deforestation and have been widely applied in different regions and research topics (William et al., 2019). However, with increasing intensive and fragmentized global land cover changes, RCMs are becoming highly reliant on spatial resolution (Jiang et al., 2022; Zeng et al., 2018). Most RCMs simulate land surface processes at a relatively coarse resolution, covering approximately millions of square meters, yet substan- tial land cover changes revealed by the advanced satellite observations occur at a fine scale of several hundred square meters. The most effective approach is to represent sub-grid land cover in the land surface scheme (LSS), enabling RCMs to capture the climate response to complex land conversion while avoiding interference owing to the model's relatively coarse grid spacing (Avissar, 1991).
LSSs that adopt the dominant land cover approach consider only the dominant land cover type (LCT) in a grid cell, omitting any other sub-grid LCTs that occupy a smaller portion of the grid cell. In contrast, the “tiling” (or
“mosaic”) approach (Avissar, 1991; Li et al., 2013) divides LCTs into several tiles within a grid cell, providing
Abstract
Understanding the regional climate response to land cover change requires a realistic sub-grid representation of vegetation cover in the land surface scheme (LSS) of climate models. The Community Land Model (CLM) is considered one of the most advanced LSSs; however, when coupled with the Weather Research and Forecasting (WRF) model, the tiling vegetation cover approach was deactivated. Here, we reactivated the theoretical sub-grid vegetation cover representation in WRF-CLM and applied it to assess the impacts of deforestation on regional climate in the Southeast Asian Massif region. We found that CLM-tiling performs more accurate simulations of surface air temperature and precipitation compared to other LSSs using the in situ observations. Importantly, CLM-tiling successfully captures the theoretical sensitivity of evapotranspiration (ET) and temperature to sub-grid deforestation, aligning with Noah-mosaic, and it substantially improves the spatial pattern responses of simulated ET and temperature to regional deforestation.Plain Language Summary
The selection of appropriate land surface schemes (LSSs) is crucial for studying land-atmosphere interactions in the Weather Research and Forecasting (WRF) model. While most LSSs adopt a dominant vegetation cover approach, where each grid cell is assigned a single land cover type, only Noah-mosaic can represent sub-grid vegetation cover changes. However, several studies have mistakenly assumed that the default coupled Community Land Model (CLM) in WRF is capable of simulating such changes, which may have affected their conclusions. In this study, we modified the CLM-default by reactivating the tiling vegetation cover approach (CLM-tiling), which accurately represents sub-grid vegetation cover changes. Our application of the CLM-tiling in the Southeast Asian Massif region led to improved simulations of air temperature and precipitation compared to other LSSs. Moreover, we found that CLM-tiling successfully exhibited theoretical climate sensitivity to sub-grid forest loss, improving the spatial pattern responses of simulated evapotranspiration and temperature to regional deforestation.© 2023. The Authors.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Sub-Grid Representation of Vegetation Cover in Land Surface Schemes Improves the Modeling of How Climate Responds to Deforestation
Yingzuo Qin1, Dashan Wang1 , Ye’er Cao2, Xitian Cai3 , Shijing Liang1 , Hylke E. Beck4, and Zhenzhong Zeng1
1School of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen, China,
2School of Atmospheric Sciences, Sun Yat-sen University, Guangzhou, China, 3Center for Water Resources and Environment, School of Civil Engineering, Sun Yat-sen University, Guangzhou, China, 4King Abdullah University of Science and Technology, Thuwal, Saudi Arabia
Key Points:
• Default WRF-CLM lacked theoretical sub-grid vegetation cover representation, failing to capture climate response to sub-grid deforestation
• CLM-tiling could successfully capture the sensitivity of evapotranspiration and temperature to sub-grid deforestation
• Involving sub-grid vegetation cover representation in WRF-CLM improved the spatial pattern responses of simulated climate to deforestation
Supporting Information:
Supporting Information may be found in the online version of this article.
Correspondence to:
Z. Zeng and D. Wang, [email protected];
Citation:
Qin, Y., Wang, D., Cao, Y., Cai, X., Liang, S., Beck, H. E., & Zeng, Z. (2023).
Sub-grid representation of vegetation cover in land surface schemes improves the modeling of how climate responds to deforestation. Geophysical Research Letters, 50, e2023GL104164. https://doi.
org/10.1029/2023GL104164 Received 16 APR 2023 Accepted 5 JUL 2023
10.1029/2023GL104164 Special Section:
Land-atmosphere coupling:
measurement, modelling and analysis
RESEARCH LETTER
a more realistic representation of sub-grid land cover. The tiling approach assumes that the land surface proper- ties related to LCTs, such as water and energy fluxes, are homogenous within each tile in the grid cell and are calculated independently. Finally, the land surface property variables are spatially aggregated to the grid cell level and then conducted to other processes. The Noah-mosaic scheme (Li et al., 2013) in the WRF model is one optional LSS that offers a sub-grid land cover representation. A previous study showed that LSSs with the dominant land cover approach failed to capture (overestimated) the evapotranspiration (ET) response to sub-grid deforestation without (with) dominant land cover changes, whereas only Noah-mosaic with the tiling land cover approach shows reasonable climate sensitivity (Wang et al., 2021). However, as the second-generation land surface model, Noah LSS has not adequately addressed the complex land surface process mechanisms and may provide over-generalized parameterizations (Cai et al., 2014). Therefore, the reasonable model sensitivity to sub-grid level land cover change is essential in capturing the climate response to forest loss in RCMs.
Meanwhile, the advanced community land model (CLM), which is a representation of the next-generation land surface model, has sophisticated parameterizations of land surface energy and water fluxes, making it a more suitable option for studying the ET response of land conversion. The latest version of CLM, CLM5.0 (Lawrence et al., 2019), incorporates a tiling land cover approach that represents five primary sub-grid LCTs and several plant functional types (PFTs), similar to CLM4.0. The utilization of PFTs in the land surface model facilitates more accurate and comprehensive modeling of the spatial and temporal dynamics of vegetation structure and function. It provides a flexible and scalable framework for model validation, parameterization, and widely coupling with other climate models (Bonan et al., 2002), such as Community Earth System Model (CESM) (Lawrence et al., 2019), Regional Climate Model (RegCM) (Steiner et al., 2005) and WRF (Jin & Wen, 2012).
Although the CLM version coupled with the WRF model was CLM4.0, which is now a decade old, recent research has demonstrated the continued efficacy of this coupling in WRF (Chen et al., 2014). However, when CLM was coupled with WRF (CLM-default), the tiling vegetation cover approach was deactivated, retaining only the dominant vegetation cover as one of the five primary LCTs in each grid cell (Wang et al., 2021). Note that this issue is not occurring in the original CLM within CESM and has not been reported in the version of CLM coupled with other RCMs (Jiang et al., 2021). This approach can result in biased estimations of the climate response to land cover change as well as other land-atmosphere interactions, highlighting the need to reactivate the tiling vegetation cover approach in WRF-CLM for a more accurate representation of sub-grid level land cover changes.
In this study, we developed an advanced LSS called CLM-tiling, which is an extension of the CLM-default. By activating the tiling vegetation cover approach in the WRF-CLM code, CLM-tiling is capable of representing sub-grid vegetation fraction and its changes. Specifically, CLM-tiling represents four dominant sub-grid vege- tation cover types in each grid cell. To evaluate the performance of CLM-tiling, we coupled it with the WRF model and tested it on two forest cover scenarios in Southeast Asian Massif (SAM) using Hansen's high-quality remote sensing forest cover product as input (Hansen et al., 2013). In these experiments, our primary objective was to assess the ability of CLM-tiling in capturing the hydrometeorological responses to sub-grid vegetation cover change, specifically in terms of ET and temperature, in comparison to three other LSSs: CLM-default, Noah-mosaic, and Noah-MP.
2. Materials and Methods
2.1. WRF Land Surface Schemes
The WRF model was originally a numerical weather prediction model, widely serving weather forecasting and research on regional climate and environment purposes (William et al., 2019). Several LSSs are coupled in the WRF framework, providing optional LSS parameterizations for different research topics on land-atmosphere interactions. The Noah land surface model (Noah-LSM), the community Noah land surface model with multipa- rameterization options (Noah-MP), and the community land model (CLM4.0) are the three most commonly used LSSs in previous studies. Noah-LSM is the most widely used LSS in WRF (Chen et al., 2010), whose frame- work considers only the dominant LCT in each grid cell. Noah-MP was developed in 2011 based on the Noah- LSM framework and considered more complex land surface processes and parameterizations (Niu et al., 2011), such as vegetation canopy energy balance, layered snowpack, and vegetation phenology, but still following the dominant land cover approach at the grid level. Besides, Li et al. (2013) extended the Noah-LSM framework by introducing the tiling/mosaic land cover approach (Noah-mosaic), which overcomes the limitation of neglecting sub-grid variability of land surface characteristics, yet still following the original simple land surface process and
Geophysical Research Letters
10.1029/2023GL104164parameterization. Noah-mosaic allows users to take homogeneous land cover into specific numbers of tiles (N) in each grid cell; surface water, energy fluxes, and other variables in each tile are calculated independently and spatially aggregated into the value of each grid cell. CLM4.0, a land surface model that was originally developed as part of CESM (Lawrence et al., 2011), has emerged as an important tool for simulating land-atmosphere interactions in various environmental studies. In 2012, CLM4.0 was coupled into the WRF framework (Jin &
Wen, 2012), further expanding its applications in regional climate modeling.
2.2. WRF CLM-Tiling Framework
CLM4.0 incorporates advanced parameterizations of surface characteristics that enable more sophisticated simu- lations of land surface water vapor and energy fluxes. The tiling approach is utilized to represent the land cover in each grid cell, which is divided into five primary sub-grid LCTs, including vegetated, urban, lake, wetland, and glacier. In the original CLM4.0 implementation of CESM, the vegetated land cover was further divided into 16 sub-grid PFTs.
However, when coupled with the WRF framework, the CLM-default implementation replaces the tiling vegetation cover approach with the dominant vegetation cover approach (Wang et al., 2021). Consequently, the CLM-default only considers the dominant vegetation cover type in each grid cell, while sub-grid vegetation type and fraction information are omitted. As a result, the model fails to capture the anticipated climate response to changes in fractional forest cover (Figure 1a). In the model grid scale, CLM-default only considers the dominant vegetation cover type, with its fraction set to 100% (Figure 1b), leading to underestimation in grids without dominant land cover change but overestimation in grids with dominant land cover change (Figures 1a and 1b; Wang et al., 2021).
Figure 1. Diagram of activating the mosaic land cover method in WRF-CLM. Different performance of climate response to dominant land cover change and sur-grid forest cover change in WRF CLM-default and CLM-tiling (a). Differences of dominant and tiling land cover approach between WRF CLM-default and CLM-tiling in model grid samples 1 and 2 (b).
Forest cover decreased by 30% both in grids 1 and 2, but the dominant land cover did not change in grid 1 (underestimated);
dominant land cover changed from forests to croplands in grid 2 (overestimated), while the tiling approach (N = 4) represent such sub-grid deforestation successfully. Brief conclusion of changes we made in WRF/CLM-tiling (Detailed codes see Text S1 in Supporting Information S1) based on the main program of WRF/CLM-default (c).
19448007, 2023, 15, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023GL104164 by Cochrane Saudi Arabia, Wiley Online Library on [06/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
In this study, we propose a novel LSS, CLM-tiling, which enhances the default model's ability in simulating historical climate and capturing the climate response to sub-grid deforestation intensity. We have reintroduced the tiling approach in WRF-CLM, with a bubble sort method that identifies the top four dominant land covers and their fractions (Figure 1c; Text S1 in Supporting Information S1). These fractions are translated into PFTs and their respective fractions. The properties and variables of each PFT are calculated separately and then weighted and averaged to the grid level before being passed on to other processes in WRF-CLM (Figure 1c).
CLM-tiling represents an improvement over the CLM-default as it reconstructs the original tiling vegetation cover approach and enhances the model's ability to capture the climate sensitivity to vegetation cover changes. It provides more comprehensive vegetation cover information for each vegetated grid cell, accounting for the top four dominant vegetation cover types and their fractions. Our reactivation of CLM’s multi-PFT functionality at the sub-grid scale has improved the model's ability to represent the complex vegetation cover change in reality, especially when the dominant cover type remains unchanged but sub-grid vegetation cover fractions are altered (Figure 1b). More details about the codes for CLM-tiling can be found in Supporting Information S1 (Text S1).
2.3. Study Areas and Experimental Design
The SAM has been subjected to rapid and patchy deforestation induced by agricultural activities, particularly in mountainous regions, over the past two decades, making it an appropriate experimental site for this study.
We designed two one-way nested domains in Southeast Asia, as depicted in Supporting Information S1 (Figure S1). The outer domain, with a grid spacing of 25 km, covered the entire Southeast Asian region to capture the background climate features. The inner domain focused on deforestation in the SAM and had a higher resolu- tion of 5 km. To account for large-scale forest loss in Southeast Asia during the early twenty-first century, we did not directly apply the recent Moderate Resolution Imaging Spectroradiometer (MODIS) land cover map as the deforestation scenario. Instead, we utilized Hansen's high-resolution forest cover gain/loss product (Hansen et al., 2013) to generate forest cover change information in each model grid of SAM. Our study specifically focused on the period 2000–2014, consistent with the experimental designs of our previous works (Wang et al., 2021;
Zeng et al., 2021). We designed two forest cover scenarios in SAM: a control scenario (“CTL”) that represents the local vegetation cover in 2000; and a deforestation scenario (“S2014”) that was generated by superimposing the high-quality satellite-derived forest cover change during 2000–2014 onto the CTL scenario. For grid cells that experienced net forest losses, the percentages of non-forest categories (e.g., bare ground, grassland, shrubland, and cropland) were increased accordingly, with the percentages for each category increasing proportionally from the original percentages in the CTL scenario; while the percentages of forest categories were decreased propor- tionally. In contrast, for grid cells that experienced net forest gain, the percentages of forest categories were increased proportionally while the percentages of non-forest categories were decreased. Thus, the S2014 scenario maintained the same relative proportion among the five MODIS forest cover types (e.g., evergreen needleleaf forest, evergreen broadleaf forest, deciduous needleleaf forest, deciduous broadleaf forest, mixed forests) as the CTL scenario, while varying with total forest cover percentage. Note that for WRF simulations coupled with the LSS using the dominant land cover approach, although they were initially run using forest cover change as input, the results solely reflect the change signals within the dominant LCT (Figure 1b).
All numerical simulations shared the same physic parameterization schemes except for the differences in land surface model schemes (Table 1) (Dudhia, 1989; Hong et al., 2006; Jiménez et al., 2012; Kain, 2004;
Mlawer et al., 1997). To better isolate the climate feedback induced by deforestation from the influence of
Physics options Parameterization scheme References
Microphysics WSM 6-class scheme Hong et al. (2006)
Cumulus Kain-Fritsch scheme Kain (2004)
Planetary boundary layer YSU scheme Hong et al. (2006)
Longwave radiation RRTM scheme Mlawer et al. (1997)
Shortwave radiation Dudhia scheme Dudhia (1989)
Surface layer Revised MM5 Monin-Obukhov scheme Jiménez et al. (2012)
Table 1
The Physical Schemes Used in This Study
Geophysical Research Letters
10.1029/2023GL104164large-scale background climate systems, four paired WRF simulations coupled with different LSSs (CLM-tiling, Noah-mosaic, CLM-default, and Noah-MP) were conducted during the local dry season from 15 November 2014, to 15 February 2015. The local dry season was chosen because the atmosphere tends to be more stable and the climate noise resulting from external forcing was relatively small, minimizing its impact on the analysis (Wang et al., 2021; Zeng et al., 2021). The first 16 days of each simulation are referred to as the spin-up period, which is not included in the validations and analyses. The initial and lateral boundary conditions for the four paired simulations were obtained from the fifth-generation reanalysis (ERA5) data provided by the European Centre for Medium-Range Weather Forecasts. The ERA5 data has a spatial resolution of 0.25° × 0.25° and hourly tempo- ral resolution. The model contained 30 vertical levels, with the upper boundary extended to 100 hPa. The sea surface temperature was updated every 6 hr from the ERA5 data. Specifically, Leaf Area Index (LAI) parameters for different land cover types in the ecosystems of SAM were updated using the Advanced Himawari Imager LAI product according to the setting in the previous study (Chen et al., 2019; Wang et al., 2021).
3. Results
3.1. Accurate Representation of Historical Climate by CLM-tiling
We validate the model simulations of the S2014 scenario with daily in situ observed daily surface air temper- ature and precipitation from 125 Global Surface Summary of the Day (GSOD) stations (Figures 2a and 2d) during the study period. Overall, the results show distinct variations in bias and root mean square error (RMSE) among different simulations, indicating that various land-atmosphere interaction processes and parameterizations in different LSSs can result in distinct model performances in capturing the historical climate in the tropics. Our results show that CLM-tiling performs well in representing historical climate, outperforming Noah-mosaic and Noah-MP and having a comparative performance with CLM-default. Specifically, CLM-tiling and CLM-default Figure 2. Validation of simulated surface air temperature and precipitation in CLM-tiling compared with CLM-default, Noah-MP, and Noah-mosaic. Mean daily surface air temperature (a) and precipitation (d) of CLM-tiling overlayed with the 125 observed data of Global Surface Summary of the Day (GSOD) (triangles) from 1 December to 28 February. Biases of daily mean temperature and precipitation between the S2014 simulations and GSOD data during the study period (b and e). Error bars show the 95% confidence interval of the bias. Normalized root mean square error of daily mean temperature and precipitation between the S2014 simulations and GSOD data during the study period (c and f).
19448007, 2023, 15, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023GL104164 by Cochrane Saudi Arabia, Wiley Online Library on [06/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
exhibit a lower mean bias and normalized RMSE for surface air temperature than Noah-MP and Noah-mosaic.
Moreover, for daily precipitation, different LSSs exhibit similar normalized RMSE, but CLM-tiling yields the lowest mean bias, indicating that it offers the best simulation of water flux among the selected LSSs in the dry season.
Our analysis reveals that the Noah-mosaic LSS exhibits a considerable mean bias (−1.76°C) and normalized RMSE (0.08), indicating that it inadequately represents the historical climate due to its basal land surface parameteriza- tion derived from Noah-LSM. Conversely, the Noah-MP, which features improvements in multi-parameterization based on the Noah-LSM, demonstrates better performance in representing the historical climate. Additionally, our modified version of the Community Land Model (CLM-tiling) performs comparably to the CLM-default in representing the historical surface air temperature, while slightly improving the representation of precipitation.
These findings suggest that our modification preserves the original performance of the model in realistically representing the historical climate, while also enhancing its ability to capture water vapor fluxes.
3.2. Correcting WRF-CLM's ET Response to Deforestation
Figure 1a shows the dominant LCT change and sub-grid forest loss fraction at a model resolution of 5 km in SAM. Our study revealed that the dominant land cover change is relatively limited (Figure S2a in Supporting Information S1), whereas the sub-grid forest cover fraction changes extensively (Figure S2b in Supporting Infor- mation S1). Deforestation was primarily occurring in the north of Thailand and Laos, with Cambodia displaying the most intensive forest loss fraction (Figure S2b in Supporting Information S1). Interestingly, some grid cells in the northeast of Thailand showed a signal of afforestation.
By activating the tiling vegetation cover approach, CLM-tiling demonstrates a spatial pattern of ET that aligns well with the distribution of forest loss fraction (Figure 3a). In contrast, CLM-default, which lacks the tiling vege- tation cover approach, only presents significant ET change in those grid cells with changes in the dominant LCT, while no systematic changes are observed in other grid cells with sub-grid forest cover change (Figure 3b). This suggests that the CLM-default only considers the dominant LCT and does not adequately account for sub-grid vegetation cover variations. Upon activating the tiling vegetation cover approach, CLM-tiling shows a strong sensitivity of ET to the percentage of forest loss, demonstrating a linear relationship between ET and forest loss fraction (0.15 mm day −1 per 10% forest loss) (Figure 3e). Conversely, the ET responses in the CLM-default are Figure 3. Simulated regional evapotranspiration (ET) response to forest loss fraction (S2014 minus CTL). Spatial pattern of ET response to regional deforestation in CLM-tiling (a), CLM-default (b), Noah-mosaic (c), and Noah-MP (d). ET sensitivity to forest loss fraction in CLM-tiling (e), CLM-default (f), Noah-mosaic (g), and Noah-MP (h). Black dots in figures (e–h) represent grids without dominant land cover change, while red dots represent grids with dominant land cover change.
Geophysical Research Letters
10.1029/2023GL104164generally separated into two clusters, which could be affected by the two different types of grid cells (dominant LCT changed and unchanged) (Figures 3e and 3f). The negative slope between ET and forest loss in CLM-default may be attributed to an overestimation of ET response in those grid cells with dominant land cover change (Figure 3f). On the other hand, Noah-mosaic shows improved ET sensitivity to deforestation compared to Noah-MP, owing to the inclusion of the sub-grid land cover approach (Figures 3c, 3d, 3g, and 3h).
3.3. CLM-tiling Yields the Theoretical Temperature Response to Deforestation
Similar to ET, the approaches of using LSSs with dominant land cover failed to capture an appropriate temper- ature response to forest cover change at the sub-grid level, and only produced patterns that align with dominant land cover change (Figures 4b and 4d). This approach amplified the signals of forest cover change in grids with dominant land cover change while ignoring them in grids without land cover change. Consequently, the tempera- ture sensitivity to forest cover change was overestimated (underestimated) in grids with (without) dominant land cover change (Figures 4f and 4h).
By incorporating the mosaic approach, CLM-tiling successfully yields the theoretical temperature response to tropical deforestation aligning with Noah-mosaic (Figure 4), which is the most widely used LSS that can capture the sub-grid vegetation cover change in WRF. CLM-tiling and Noah-mosaic credibly perform the spatial patterns of simulated temperature responses to regional deforestation, particularly in Laos, where extensive sub-grid deforestation occurs without dominant land cover change. In contrast, CLM-default and Noah-MP fail to capture this signal.
Moreover, the simulated temperature responses in CLM-tiling better align with the spatial pattern of forest loss fraction compared to Noah-mosaic. For instance, Figure S2b in Supporting Information S1 shows that the most intensive deforestation occurs in Cambodia, but Noah-mosaic shows a weak temperature response (Figure 4c).
This discrepancy is primarily due to Noah-mosaic overestimating the albedo changes resulting from the conver- sion of forests to croplands, thereby suppressing the warming effects induced by deforestation over most areas.
By contrast, CLM-tiling produces substantial temperature increases in Cambodia (Figure 4a), which is more reasonable considering the high-intensity local deforestation. Additionally, while Noah-mosaic fails to capture the negative temperature response to afforestation in the northeast of Thailand, CLM-tiling presents a significant cooling effect.
Figure 4. Simulated regional surface air temperature response to forest loss fraction (S2014 minus CTL). Spatial pattern of the surface temperature response to regional deforestation in CLM-tiling (a), CLM-default (b), Noah-mosaic (c), and Noah-MP (d). Temperature sensitivity to forest loss fraction in CLM-tiling (e), CLM-default (f), Noah-mosaic (g), and Noah-MP (h). Black dots in figures (e–h) represent grids without dominant land cover change, while red dots represent grids with dominant land cover change.
19448007, 2023, 15, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023GL104164 by Cochrane Saudi Arabia, Wiley Online Library on [06/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4. Discussion and Conclusion
The current LSS coupled in WRF, including RUC (Smirnova et al., 2016), Noah-LSM (Chen et al., 2010; Ek et al., 2003), Noah-MP (Niu et al., 2011), Pleim-Xiu (Pleim & Xiu, 1995), and SSiB (Xue et al., 1991), are all developed using the dominant land cover approach. Among the optional LSSs in WRF, only Noah mosaic (Li et al., 2013) and CLM4.0 (Jin & Wen, 2012) incorporate sub-grid land cover information and are widely used in land-atmosphere interaction research (Cao et al., 2015; Ge et al., 2020; Glotfelty et al., 2021). However, previous research has shown that relying on the CLM-default to represent the sub-grid level PFTs can be misleading, as it is only a matrix that transfers MODIS land cover types into PFTs in the CLM structure (Lu & Kueppers, 2012).
Although the CLM-default shows superior performance in representing historical climate compared with other LSSs (Chen et al., 2014), its application may lead to underestimations or overestimations in simulating climate responses to vegetation change due to the deactivated tiling vegetation cover approach (Wang et al., 2021). For example, underestimations can occur in grids with sub-grid vegetation cover fraction change but no dominant vegetation cover type change, while overestimations would happen in grid cells with dominant vegetation cover change but with fractional sub-grid vegetation cover change (Wang et al., 2021).
Alkama and Cescatti (2016) reported regional warming of 0.16°C per 10% forest loss fraction during the dry season based on satellite observation. Here we revealed regional warming of 0.31°C per 10% forest loss using CLM-tiling, while Noah-mosaic shows a much lower surface air temperature sensitivity of 0.07°C per 10% forest loss. It’s worth noting that Alkama and Cescatti’s study quantifies the local effects of deforestation (Alkama &
Cescatti, 2016), specifically examining changes in the surface energy budget and their first-order interactions with the boundary layer. In contrast, our model simulations encompass both local and non-local effects, includ- ing atmospheric circulation. Therefore, the estimated warming effect of tropical deforestation in Alkama and Cescatti’s study may be underestimated due to the non-negligible non-local effects (Winckler et al., 2019). While a direct comparison between our simulation results and the satellite-based results may not be entirely fair due to differences in scale and methodology, it serves as an important reference that highlights the improved capability of CLM-tiling in representing the observed deforestation-induced climate feedback.
Noah-mosaic has been applied in several pieces of research on land-atmosphere interactions owing to the excel- lent improvement of adding the mosaic land cover approach into default Noah LSS and improved the simulation of urban climate feedback (Ramamurthy & Bou-Zeid, 2017; Zeng et al., 2021). However, the lack of more complex land surface process parameterizations limits Noah-mosaic's application to more cases of climate feed- back of land cover change (Wang et al., 2021). In contrast, CLM-tiling incorporates sophisticated parameteriza- tions of land surface energy and water vapor flux, resulting in powerful simulation abilities to represent historical climates. The original CLM4.0 (Lawrence et al., 2011) has three major sophisticated components, namely, biogeophysics, biogeochemistry, and hydrology, which enable excellent regional and global scale simulations.
Furthermore, the CLM4.0 scheme includes detailed descriptions of vegetation interception and transpiration, a multilayer snowpack considering snow accumulation and melt, a 10-layer model of soil water with an explicit treatment of liquid water and ice, and a runoff production module (Beven & Kirkby, 1979; Bonan et al., 2002), leading to improved hydrological cycle simulation.
This study has identified different climate responses to deforestation between Noah mosaic and CLM-tiling, which mainly result from the distinct ET and albedo calculations. Owing to tropical deforestation-induced warm- ing effects being dominated by the decreased ET, improved ET calculation helps represent better climate simu- lations, while the influence of albedo is limited (Zeng et al., 2021). Noah LSM employs the Penman equation to calculate the total ET, which includes bare soil evaporation and canopy transpiration (Chen & Dudhia, 2001), while the CLM uses a complex vertical canopy structure method that comprises three ET sources: non-vegetated surface evaporation, vegetation interception and transpiration, and ground evaporation of vegetated surface (Subin et al., 2011). The disparity in the ET calculation methods causes different hydrological cycle simulations and leads to variations in the ET response to the same deforestation scenario. As for the albedo calculations, Noah LSM uses albedo from lookup table values according to land cover types, whereas CLM calculates albedo with the two-stream approximation of Dickinson (1983) and Sellers (1985), which is highly related to soil albedo. The different albedo calculation methods lead to diverse performances among LSSs.
In this study, we developed an advanced CLM-tiling LSS by activating the tiling vegetation cover approach in the CLM-default, aiming to enhance the accuracy and reasonability in capturing the climate response to sub-grid
Geophysical Research Letters
10.1029/2023GL104164vegetation cover change. Our results showed that CLM-tiling is comparable in its ability to represent the histor- ical climate with default WRF CLM (Jin & Wen, 2012) and is reasonably effective in capturing the climate sensitivity to sub-grid deforestation similar to Noah-mosaic (Li et al., 2013). We highly recommend that users apply CLM-tiling in more cases of land-atmosphere interaction, correcting previous mistaken understanding and helping validate and improve model performance.
Conflict of Interest
The authors declare no conflicts of interest relevant to this study.
Data Availability Statement
The authors have filed the paperwork and provided the documentation and codes to incorporate the CLM-tiling parameterization in future official releases of WRF. All analysis scripts are available at https://doi.org/10.6084/
m9.figshare.22598050. ERA 5 reanalysis data is available at https://www.ecmwf.int/en/forecasts/dataset/
ecmwf-reanalysis-v5. WRF depository is available at https://github.com/wrf-model/WRF. Global high-resolution 21st-century forest cover change maps are available at http://earthenginepartners.appspot.com/science-2013- global-forest. GSOD daily surface air temperature and precipitation data are available at https://www.ncei.noaa.
gov/data/global-summary-of-the-day/.
References
Alkama, R., & Cescatti, A. (2016). Biophysical climate impacts of recent changes in global forest cover. Science, 351(6273), 600–604. https://
doi.org/10.1126/science.aac8083
Avissar, R. (1991). A statistical-dynamical approach to parameterize subgrid-scale land-surface heterogeneity in climate models. Surveys in Geophysics, 12(1–3), 155–178. https://doi.org/10.1007/bf01903417
Beven, K. J., & Kirkby, M. J. (1979). A physically based, variable contributing area model of basin hydrology. Hydrological Sciences Bulletin, 24(1), 43–69. https://doi.org/10.1080/02626667909491834
Bonan, G. B. (2008). Forests and climate change: Forcings, feedbacks, and the climate benefits of forests. Science, 320(5882), 1444–1449. https://
doi.org/10.1126/science.1155121
Bonan, G. B., Oleson, K. W., Vertenstein, M., Levis, S., Zeng, X., Dai, Y., et al. (2002). The land surface climatology of the community land model coupled to the NCAR community climate model. Journal of Climate, 15(22), 3123–3149. https://doi.org/10.1175/1520-0442(2002)015<3123:tlsc ot>2.0.co;2
Cai, X., Yang, Z.-L., Xia, Y., Huang, M., Wei, H., Leung, L. R., & Ek, M. B. (2014). Assessment of simulated water balance from Noah, Noah-MP, CLM, and VIC over CONUS using the NLDAS test bed. Journal of Geophysical Research: Atmospheres, 119(24), 13751–13770.
https://doi.org/10.1002/2014jd022113
Cao, Q., Yu, D., Georgescu, M., Han, Z., & Wu, J. (2015). Impacts of land use and land cover change on regional climate: A case study in the agro-pastoral transitional zone of China. Environmental Research Letters, 10(12), 124025. https://doi.org/10.1088/1748-9326/10/12/124025 Chen, F., & Dudhia, J. (2001). Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model
implementation and sensitivity. Monthly Weather Review, 129(4), 569–585. https://doi.org/10.1175/1520-0493(2001)129<0569:caalsh>2.0.co;2 Chen, F., Liu, C., Dudhia, J., & Chen, M. (2014). A sensitivity study of high-resolution regional climate simulations to three land surface models over the western United States. Journal of Geophysical Research: Atmospheres, 119(12), 7271–7291. https://doi.org/10.1002/2014jd021827 Chen, Y., Sun, K., Chen, C., Bai, T., Park, T., Wang, W., et al. (2019). Generation and evaluation of LAI and FPAR products from Himawari-8
advanced Himawari imager (AHI) data. Remote Sensing, 11(13), 1517. https://doi.org/10.3390/rs11131517
Chen, Y., Yang, K., Zhou, D., Qin, J., & Guo, X. (2010). Improving the Noah land surface model in arid regions with an appropriate parameteri- zation of the thermal roughness length. Journal of Hydrometeorology, 11(4), 995–1006. https://doi.org/10.1175/2010jhm1185.1
Dickinson, R. E. (1983). Land surface processes and climate-surface albedos and energy balance. Advances in Geophysics, 25, 305–353. https://
doi.org/10.1016/s0065-2687(08)60176-4
Dudhia, J. (1989). Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model.
Journal of the Atmospheric Sciences, 46(20), 3077–3107. https://doi.org/10.1175/1520-0469(1989)046<3077:nsocod>2.0.co;2
Ek, M. B., Mitchell, K. E., Lin, Y., Rogers, E., Grunmann, P., Koren, V., et al. (2003). Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. Journal of Geophysical Research, 108(D22), 8851. https://
doi.org/10.1029/2002jd003296
Ge, J., Pitman, A. J., Guo, W., Zan, B., & Fu, C. (2020). Impact of revegetation of the Loess Plateau of China on the regional growing season water balance. Hydrology and Earth System Sciences, 24(2), 515–533. https://doi.org/10.5194/hess-24-515-2020
Glotfelty, T., Ramírez-Mejía, D., Bowden, J., Ghilardi, A., & West, J. J. (2021). 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). Geoscientific Model Development, 14(6), 3215–3249. https://doi.org/10.5194/gmd-14-3215-2021
Hansen, M. C., Potapov, P. V., Moore, R., Hancher, M., Turubanova, S. A., Tyukavina, A., et al. (2013). High-resolution global maps of 21st-century forest cover change. Science, 342(6160), 850–853. https://doi.org/10.1126/science.1244693
Hong, S.-Y., Kim, J., Lim, J., & Dudhia, J. (2006). The WRF single-moment 6-class microphysics scheme (WSM6). Asia-Pacific Journal of Atmospheric Sciences, 42, 129–151.
Jiang, X., Ziegler, A. D., Liang, S., Wang, D., & Zeng, Z. (2022). Forest restoration potential in China: Implications for carbon capture. Journal of Remote Sensing, 2022, 0006. https://doi.org/10.34133/remotesensing.0006
Acknowledgments
This study was supported by the National Natural Science Foundation of China (42071022, 42001321) and the startup fund provided by the Southern University of Science and Technology (29/Y01296002, 29/Y01296122, 29/
Y01296222). We thank the Center for Computational Science and Engineering at the Southern University of Science and Technology for providing computing resources. We acknowledge the anon- ymous reviewers for their detailed and helpful comments on the original manu- script. We thank the editor Guiling Wang for her valuable guidance and input, which greatly improved this manuscript.
19448007, 2023, 15, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023GL104164 by Cochrane Saudi Arabia, Wiley Online Library on [06/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Jiang, Y., Wang, G., Liu, W., Erfanian, A., Peng, Q., & Fu, R. (2021). Modeled response of South American climate to three decades of deforest- ation. Journal of Climate, 34(6), 2189–2203. https://doi.org/10.1175/jcli-d-20-0380.1
Jiménez, P. A., Dudhia, J., González-Rouco, J. F., Navarro, J., Montávez, J. P., & García-Bustamante, E. (2012). A revised scheme for the WRF surface layer formulation. Monthly Weather Review, 140(3), 898–918. https://doi.org/10.1175/mwr-d-11-00056.1
Jin, J., & Wen, L. (2012). Evaluation of snowmelt simulation in the Weather Research and Forecasting model. Journal of Geophysical Research, 117(D10), D10110. https://doi.org/10.1029/2011jd016980
Kain, J. S. (2004). The Kain–Fritsch convective parameterization: An update. Journal of Applied Meteorology, 43(1), 170–181. https://doi.
org/10.1175/1520-0450(2004)043<0170:tkcpau>2.0.co;2
Lawrence, D. M., Fisher, R. A., Koven, C. D., Oleson, K. W., Swenson, S. C., Bonan, G., et al. (2019). The community land model version 5: Description of new features, benchmarking, and impact of forcing uncertainty. Journal of Advances in Modeling Earth Systems, 11(12), 4245–4287. https://doi.org/10.1029/2018ms001583
Lawrence, D. M., Oleson, K. W., Flanner, M. G., Thornton, P. E., Swenson, S. C., Lawrence, P. J., et al. (2011). Parameterization improvements and functional and structural advances in version 4 of the community land model. Journal of Advances in Modeling Earth Systems, 3(1).
https://doi.org/10.1029/2011ms00045
Li, D., Bou-Zeid, E., Barlage, M., Chen, F., & Smith, J. A. (2013). Development and evaluation of a mosaic approach in the WRF-Noah frame- work. Journal of Geophysical Research: Atmospheres, 118(21), 11918–11935. https://doi.org/10.1002/2013jd020657
Lu, Y., & Kueppers, L. M. (2012). Surface energy partitioning over four dominant vegetation types across the United States in a coupled regional climate model (Weather Research and Forecasting Model 3-Community Land Model 3.5). Journal of Geophysical Research, 117(D6), D06111. https://doi.org/10.1029/2011jd016991
Mlawer, E. J., Taubman, S. J., Brown, P. D., Iacono, M. J., & Clough, S. A. (1997). Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. Journal of Geophysical Research, 102(D14), 16663–16682. https://doi.org/10.1029/97jd00237 Niu, G.-Y., Yang, Z.-L., Mitchell, K. E., Chen, F., Ek, M. B., Barlage, M., et al. (2011). The community Noah land surface model with multipa- rameterization options (Noah-MP): 1. Model description and evaluation with local-scale measurements. Journal of Geophysical Research, 116(D12), D12109. https://doi.org/10.1029/2010jd015139
Piao, S., Wang, X., Park, T., Chen, C., Lian, X., He, Y., et al. (2019). Characteristics, drivers and feedbacks of global greening. Nature Reviews Earth & Environment, 1(1), 14–27. https://doi.org/10.1038/s43017-019-0001-x
Pleim, J. E., & Xiu, A. (1995). Development and testing of a surface flux and planetary boundary layer model for application in mesoscale models.
Journal of Applied Meteorology, 34(1), 16–32. https://doi.org/10.1175/1520-0450-34.1.16
Ramamurthy, P., & Bou-Zeid, E. (2017). Heatwaves and urban heat islands: A comparative analysis of multiple cities. Journal of Geophysical Research: Atmospheres, 122(1), 168–178. https://doi.org/10.1002/2016jd025357
Sellers, P. J. (1985). Canopy reflectance, photosynthesis and transpiration. International Journal of Remote Sensing, 6(8), 1335–1372. https://
doi.org/10.1080/01431168508948283
Smirnova, T. G., Brown, J. M., Benjamin, S. G., & Kenyon, J. S. (2016). Modifications to the rapid update cycle land surface model (RUC LSM) available in the Weather Research and Forecasting (WRF) model. Monthly Weather Review, 144(5), 1851–1865. https://doi.org/10.1175/
mwr-d-15-0198.1
Steiner, A. L., Pal, J. S., Giorgi, F., Dickinson, R. E., & Chameides, W. L. (2005). The coupling of the Common Land Model (CLM0) to a regional climate model (RegCM). Theoretical and Applied Climatology, 82(3–4), 225–243. https://doi.org/10.1007/s00704-005-0132-5
Subin, Z. M., Riley, W. J., Jin, J., Christianson, D. S., Torn, M. S., & Kueppers, L. M. (2011). Ecosystem feedbacks to climate change in Cali- fornia: Development, testing, and analysis using a coupled regional atmosphere and land surface model (WRF3-CLM3.5). Earth Interactions, 15(15), 1–38. https://doi.org/10.1175/2010ei331.1
Wang, D., Wu, J., Huang, M., Li, L. Z. X., Wang, D., Lin, T., et al. (2021). The critical effect of subgrid-scale scheme on simulating the climate impacts of deforestation. Journal of Geophysical Research: Atmospheres, 126(17), e2021JD035133. https://doi.org/10.1029/2021jd035133 William, S. C., Klemp, J. B., Dudhia, J., & Jimy, K. (2019). A description of the advanced research WRF model version 4 (Vol. 145, p. 145).
National Center for Atmospheric Research. Retrieved from https://opensky.ucar.edu/islandora/object/technotes:576/
Winckler, J., Lejeune, Q., Reick, C. H., & Pongratz, J. (2019). Nonlocal effects dominate the global mean surface temperature response to the biogeophysical effects of deforestation. Geophysical Research Letters, 46(2), 745–755. https://doi.org/10.1029/2018gl080211
Xue, Y., Sellers, P. J., Kinter, J. L., & Shukla, J. (1991). A simplified biosphere model for global climate studies. Journal of Climate, 4(3), 345–364. https://doi.org/10.1175/1520-0442(1991)004<0345:asbmfg>2.0.co;2
Zeng, Z., Estes, L., Ziegler, A. D., Chen, A., Searchinger, T., Hua, F., et al. (2018). Highland cropland expansion and forest loss in Southeast Asia in the twenty-first century. Nature Geoscience, 11(8), 556–562. https://doi.org/10.1038/s41561-018-0166-9
Zeng, Z., Wang, D., Yang, L., Wu, J., Ziegler, A. D., Liu, M., et al. (2021). Deforestation-induced warming over tropical mountain regions regu- lated by elevation. Nature Geoscience, 14(1), 23–29. https://doi.org/10.1038/s41561-020-00666-0