About this Journal Submit a Manuscript Table of Contents
Advances in Meteorology
Volume 2013 (2013), Article ID 817870, 10 pages
http://dx.doi.org/10.1155/2013/817870
Research Article

Response of Simulated Surface Air Temperature to the Interannual Variability of Leaf Area Index in Eastern China

Institute of Geographic Sciences and Natural Resources Research (IGSNRR), Chinese Academy of Sciences (CAS), Beijing 100101, China

Received 12 June 2013; Accepted 27 August 2013

Academic Editor: Hongbo Su

Copyright © 2013 Xuezhen Zhang and Qiuhong Tang. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Using the coupled WRF-Noah model, we conducted two experiments to investigate impacts of the interannual variability of leaf area index (LAI) on the surface air temperature (SAT) in eastern China. The Moderate Resolution Imaging Spectroradiometer (MODIS) observed dynamic LAI data from 2002 to 2009 were used in one modeling experiment, and the climatological seasonal cycle of the MODIS LAI was used in the other experiment. The results show that the use of dynamic LAI improves model performance. Compared with the use of climatological LAI, the use of dynamic LAI may reduce the warm (cool) bias in the years with large positive (negative) LAI anomalies. The reduction of the warm bias results from the modeled cooling effect of LAI increase through reducing canopy resistance, promoting transpiration, and decreasing sensible heat flux. Conversely, the reduction of cool bias is a result of the warming effect of negative anomaly of LAI. The use of dynamic LAI can improve model performance in summer and to a lesser extent, spring and autumn. Moreover, the dynamic LAI exerts a detectable influence on SAT in the WRF model when the LAI anomaly is at least 20% of the climatological LAI.

1. Introduction

A large amount of evidence shows that terrestrial vegetation is an important dynamic component of the climate system [14]. Terrestrial vegetation regulates the local/regional weather and climate through modifying the surface energy budget, modifying partition of surface net radiation between latent heat flux and sensible heat flux, modifying surface wind, as reviewed by Notaro et al. [5]. For instance, coupling a dynamic vegetation model could upgrade the ability of the climate model to capture low-frequency variability of precipitation in the Amazon region [6]. The coupled WRF-Noah model could simulate climate warming closer to ground measurements by using a dynamical green vegetation fraction rather than by using a climatological fraction [7]. Therefore, it would be valuable to accurately describe vegetation properties in climate models to improve model performance.

In climate models, the land surface submodel is used to simulate the dynamics of moisture and heat within soil and surface heat and moisture fluxes to the atmosphere. The Noah land surface scheme (Noah LSM) is an intermediate-complexity land surface model that can provide reasonable diurnal and seasonal variations of surface heat fluxes [8]. The Noah LSM is employed in the National Centers for Environmental Prediction (NCEP) operational mesoscale Eta model [9], the Mesoscale Model (MM5) [8], the Weather Research and Forecast (WRF) model, and other mesoscale models. The WRF model is a next-generation mesoscale model designed to serve both atmospheric research and operational forecasting. It includes atmospheric dynamics and parameterizations of mesoscale atmospheric processes that could be comparable to or more comprehensive than those found in most global climate models. The coupled WRF-Noah model has been widely used to simulate regional climate changes [10] and to study surface-atmosphere interactions [7, 11, 12].

There is one canopy layer in the Noah LSM. The canopy properties are represented by a group of parameters including leaf area index (LAI), green vegetation fraction (), albedo, emissivity, and canopy height. The albedo and emissivity regulate the surface radiation balance. The LAI and are the key parameters used to simulate canopy evaporation () and canopy transpiration (). Thus, these parameters regulate the partitioning of surface net radiation between sensible heat flux and latent heat flux [8]. Specifically, while and simulations are largely affected by , simulations are also affected by LAI through the regulation of canopy resistance. Hong et al. [13] reported that evapotranspiration and surface air temperature (SAT) simulations in the coupled WRF-Noah model were highly sensitive to when comparing simulated results from three different datasets to ground measurements. Ge et al. [7] showed that the use of an observed dynamic would better reproduce climate warming in the last two decades of the 20th century than the use of a fixed climatological . These findings suggest that realistic data would be valuable for improving the performance of coupled models. However, little is known about the sensitivity of the coupled WRF-Noah model to LAI.

The goal of this study is to investigate the responses of SAT to the interannual variability of LAI in the coupled WRF-Noah model. The paper is organized into the following sections: Section 2 introduces the experiment design, Section 3 presents the results, and Section 4 presents the concluding remarks and discusses potential future research directions.

2. Methods

2.1. Experiment Design and Data

We conducted two eight-year (2002–2009) simulations using the WRF model with Advanced Research WRF (ARW) dynamic core version 3.4 [14]. The two simulations used the same settings and boundary conditions except for the prescribed LAI data. The control simulation used the Moderate Resolution Imaging Spectroradiometer (MODIS) observed dynamic LAI from 2002 to 2009, which is referred to as the dynamic LAI simulation (DYC). The other simulation used the climatological seasonal cycle of LAI calculated from the dynamic LAI data from 2002 to 2009, which is referred to as the fixed seasonal cycle of LAI simulation (FIX).

The model domain covers eastern China with the central point at 116°E, 36°N. It has a horizontal resolution of 30 km (64 grid cells in west-east and 80 grid cells in south-north; cf. Figure 1) and 28 vertical levels with 50 hPa at the top level of the atmosphere. The simulations were driven by 1.0 × 1.0 degree resolution 6-hourly NCEP FNL reanalysis data (available from the RDA (http://rda.ucar.edu/datasets/ds083.2/) in dataset number ds083.2). The main physical parameterizations used in this study include the WSM 3-class simple ice microphysics scheme [15], the Grell-Devenyi ensemble convective parameterization [16], the Community Atmospheric Model (CAM3) radiation package [17], the Yonsei University planetary boundary layer scheme [18], and the Noah land surface scheme [8].

817870.fig.001
Figure 1: Simulation domain (black-dashed line) and land cover (derived from MODIS land cover products released together with WRF surface static dataset).

The dynamic LAI data used by DYC were taken from MCD15A2, a dataset retrieved by combining observations from both Aqua/MODIS and Terra/MODIS [19]. The global assessment shows that the MODIS LAI has an average uncertainty of 0.17, accounting for 11.5% of the mean LAI. Such uncertainty is much less than that of other prevailing LAI datasets such as GEOV1, GLASS, GLOBMAP, and JRC-TIP [20]. The MCD15A2 dataset is provided with a 1 km resolution and an eight-day interval. We aggregated the 1 km resolution LAI into 30 by 30 km WRF model grid cell resolution by calculating the mean value of the valid pixels within a grid cell. Then, we linearly interpolated the eight-day data to obtain daily LAI. It is worth noting that when more than 80% of the 1 km pixels within one WRF model grid cell were missing or were lower quality (i.e., cloud and cloud shadow, average or high aerosol levels, or snow/ice were detected), the corresponding WRF model grid cell was treated as missing data at that time point. Then, the missing data were estimated by linear interpolation using the LAI values from the last time point and next time point of the same grid cell.

To evaluate the modeling results, the global high-resolution gridded temperature dataset CRU TS3.0 (available at http://badc.nerc.ac.uk/view/badc.nerc.ac.uk__ATOM__dataent_1256223773328276) was used. This dataset has a spatial resolution of 0.5 × 0.5 degrees. It was generated from in situ measurements from a large number of stations worldwide, as described by Mitchell and Jones [21].

2.2. Data Analysis

The data analysis included two parts: evaluating the control (DYC) simulation and diagnosing the effects of dynamic LAI on the SAT simulation. We first evaluated grid cell-based bias of mean SAT of 2002–2009 from DYC. The biases were represented by root mean square error (RMSE; (1)). We also computed the correlations between the time series of simulated SAT from 2002 to 2009 and observations for each grid cell. The correlation coefficients represent the ability of the coupled WRF-Noah model to capture interannual variability of SAT. It is noted that the 30 km simulations were spatially aggregated into the 0.5 degree resolution of CRU grids before evaluation. These evaluations were performed for spring (March to May), summer (June to August), and autumn (September to November). Consider where denotes the RMSE and and represent simulated values and observed values at the year , respectively.

To diagnose effects of dynamic LAI on simulated SAT, we first calculated the LAI anomaly for each year and each grid cell. The LAI anomaly was calculated by removing the climatological LAI from the observed dynamic LAI. The years with the maximum (positive) and minimum (negative) LAI anomalies for each grid cell were identified. The grid cell-based spatial correlations between the LAI anomaly and the difference in the DYC- and FIX-simulated SAT (DYC minus FIX) for the corresponding year were computed for the years with maximum and minimum LAI anomalies. Then, we evaluated the grid cell-based bias of the simulated SAT by comparing it to the observations in the maximum/minimum LAI anomaly years for the DYC and FIX simulations. In the following section, we compared the modeled SAT bias of the DYC simulation to that of the FIX simulation. The differences in biases of modeled SAT from the DYC and FIX simulations were revealed.

To investigate the possible physical mechanisms related to the modeled SAT differences in the DYC and FIX simulations, we analyzed the correlations between time series of LAI anomalies and the corresponding differences in the DYC and FIX simulated surface energy budgets (DYC minus FIX) for each grid cell. We performed the analysis in spring (March to May), summer (June to August), and autumn (September to November) because both the LAI anomaly and its potential impacts on SAT may vary with seasons. The study area covers a broad domain from low latitudes to mid-high latitudes (Figure 1). The vegetation growth in the extra-tropical areas, excluding the small tropical area in the low latitudes, ceases in winter (December to February). Therefore, we did not include winter in the analysis. Sparse vegetation and bare land were also excluded.

3. Results

3.1. Evaluation of the Control Simulation

Figure 2 shows the mean and standard deviation of LAI from 2002 to 2009 for spring, summer, and autumn. In spring and autumn, aside from a distinguished low LAI (less than two) area in the northwest with sparse vegetation and bare land, the mean LAI generally decreased from approximately seven at the south end to approximately three at the north end. In summer, the northeast area exhibited a low LAI (less than three). For the other areas, the geographical gradient from south to north is too slight to be detected and the high LAI (greater than six) could be found everywhere. The standard deviations indicate the interannual variability. In spring and autumn, the LAI of southern areas has greater interannual variability (the standard deviation is approximately 0.3) than that in the northern areas. In summer, the interannual variability of LAI is larger in grassland in the north and in evergreen broadleaf forest in the south than that in the other areas. Generally, the grassland in the north had larger interannual variability (the standard deviation is approximately 0.8) than the evergreen broadleaf forest in the south.

fig2
Figure 2: Mean (a) and standard deviation (b) of leaf area index (LAI) from 2002 to 2009 for spring (March to May; MAM), summer (June to August; JJA), and autumn (September to November; SON).

Figure 3 compares the DYC simulated SAT with observations. A general agreement is found between the observations and the DYC simulation, especially with regard to the geographical distribution. Spatial correlations between simulated and observed SAT are high (spring: 0.99; summer: 0.95; autumn: 0.98). The high correlations suggest that the coupled WRF-Noah model with dynamic LAI is capable of reproducing the spatial variability of climatological SAT. The DYC simulation also exhibits systematic biases over certain regions. The spatial distributions of biases are characterized by larger biases along the western edge and smaller biases in the eastern areas of the study domain. The regional means of RMSE have a range of 1.5°C–1.6°C, the negative biases in the southwestern edges reach 5°C–8°C, and the positive biases in the northwestern edges reach approximately 6°C.

fig3
Figure 3: Simulated (a) and observed (b) average air temperature 2 m above ground from 2002 to 2009 and the differences from simulation minus observation (c) (RMSE denotes root mean of square errors between simulation and observation) for spring (March to May; MAM), summer (June to August; JJA), and autumn (September to November; SON).

The DYC simulation is also capable of reproducing the interannual variations of SAT. The correlations between the interannual variations of the observed and simulated SAT are mostly positive (Figure 4). Specially, the correlations in the north of the simulation domain were generally over 0.8 and significant (). For summer and autumn, there are small areas of negative correlations that are mainly found over sparse vegetation.

817870.fig.004
Figure 4: Correlations between observed and simulated interannual temperature from 2002 to 2009 for spring (March to May; MAM), summer (June to August; JJA), and autumn (September to November; SON) (shading denotes a confidence level higher than 99%; interval of contour lines is 0.2).
3.2. Effects of Dynamic LAI

Figure 5 illustrates grid cell-based maximum LAI anomalies and the corresponding differences between simulated SAT from DYC and FIX (DYC minus FIX). At the grid-cell scale, the differences in simulated SAT were negatively correlated with the LAI anomaly (; ). Smaller values of simulated TSA in DYC (compared to the simulated SAT in FIX) spatially coexisted with larger positive LAI anomalies. Such negative correlations demonstrate the cooling effects of LAI increases. The cooling effects change seasonally with strong effects in summer and weak effects in spring and autumn. In summer, the SAT can decrease approximately by 0.028°C when LAI increases by 10% of the climatological LAI. In spring and autumn, the cooling effects are small. Moreover, the cooling effects are hard to be detected when the LAI anomaly is less than 20% of the climatological LAI.

817870.fig.005
Figure 5: The maximum (positive) anomalies of LAI from 2002 to 2009 (top panel), the differences of simulated surface air temperature (SAT) by DYC and FIX (DYC minus FIX; middle panel), and the grid cell-based spatial correlations between the LAI anomalies and SAT differences (bottom panel; the red straight line is the best fitting line and the slope as unit of 0.01°C per 1% LAI; the stars represent the confidence level higher than 99.9%) for spring (left panel), summer (middle panel), and autumn (right panel).

Figure 6 shows the minimum LAI anomaly and the corresponding differences between simulated SAT from DYC and FIX (DYC minus FIX). The difference between simulated SAT from DYC and FIX was also negatively correlated with the LAI anomalies (; ). The larger negative LAI anomalies spatially coexisted with higher simulated SAT in DYC (compared to the simulated SAT in FIX). Such negative correlations confirmed the above-mentioned cooling effect of vegetation. Again, the spatial correlations in summer were higher than those in spring and autumn. In summer, SAT increased approximately by 0.06°C when LAI decreased by 10% of the climatological LAI. In spring and autumn, the SAT changes were 25% or less of those changes seen in summer. The effects were generally too small to be detected when the LAI anomaly was less than 20% of the climatological LAI.

817870.fig.006
Figure 6: The minimum (negative) anomalies of LAI from 2002 to 2009 (top panel), the differences of simulated surface air temperature (SAT) by DYC and FIX (DYC minus FIX; middle panel), and the grid cell-based spatial correlations between the LAI anomalies and SAT differences (bottom panel; the red straight line is the best fitting line and the slope as unit of 0.01°C per 1% LAI; the stars represent the confidence level higher than 99.9%) for spring (left panel), summer (middle panel), and autumn (right panel).

Figure 7 shows grid cell-based maximum and minimum LAI anomalies and the corresponding biases of simulated SAT from the DYC and FIX simulations (DYC minus observations and FIX minus observations). Use of dynamic LAI could likely improve the performance of the WRF-Noah model in summer. Compared with the observations, the FIX simulation had a cool bias with a negative LAI anomaly and a warm bias with a positive LAI anomaly. The larger LAI anomaly is associated with the larger SAT bias in the FIX simulation. By using the dynamic LAI, the DYC simulation might reduce the biases. More specifically, the FIX simulation would underestimate SAT at the grids for years with a negative LAI anomaly as a consequence of the prescribed higher LAI value compared with the dynamic LAI. The FIX simulation would also overestimate SAT at the grids for years with a positive LAI anomaly as a consequence of the prescribed lower LAI value compared with the dynamic LAI. It seemed that the DYC simulation with the prescribed dynamic LAI value could reduce the bias. The reductions of the simulated SAT bias were stronger for the grids during a year with large LAI anomalies than those with small LAI anomalies. The bias reduction of simulated SAT was generally detectable when the LAI anomaly was over approximately 20% of the climatological LAI in summer. However, the bias of simulated SAT was generally not reduced in spring and autumn by the use of the dynamic LAI data.

fig7
Figure 7: The maximum and minimum LAI anomalies and the corresponding biases of simulated surface air temperature (SAT) from the DYC (blue circle) and FIX (red circle) simulations (DYC minus observation and FIX minus observation) for spring (a), summer (b), and autumn (c), respectively.
3.3. Physical Mechanism of Regulation of LAI on SAT

To explain the above-mentioned findings, we analyzed the role of LAI in the Noah LSM. The Noah LSM includes parameterization of canopy resistance () following (2) [22]. Lower has the potential to result in more latent heat flux and less sensible heat flux and vice versa for higher : where where where , , , and are subject to 0 and 1 as lower and upper bounds and represent the effects of photosynthetically active radiation, vapor pressure deficit, air temperature, and soil moisture. , , , and are coefficients depending fully on vegetation types without regard to the environment. Based on (2), LAI regulates through two pathways: (1) direct reverse linear regulation and (2) the modifying factor of photosynthetically active radiation. In the second pathway, high LAI results in more photosynthetically active radiation and thus lower . Overall, high LAI has the potential to reduce and produce more transpiration. Additionally, a warm and dry atmosphere and moist soil are good for low .

Figure 8 confirms the role of LAI prescribed by (2). We found widespread positive correlations between the LAI anomaly and differences in latent heat flux from DYC and FIX (DYC minus FIX) and negative correlations between the LAI anomaly and differences in sensible heat flux over a large area. Additionally, there are widespread negative correlations between the LAI anomaly and differences in surface Bowen ratio. The correlations do not always have a consistent sign (positive or negative) over the study area, suggesting that the LAI interannual variability might not only affect local fluxes but also have remote influence through its impact on atmospheric processes. Nonetheless, the widespread correlations with the same sign suggested that high LAI could result in more latent heat and less sensible heat at the grid scale. Significant correlations () could be found in the northern part of the grassland and in the evergreen broadleaf forest where the LAI interannual variability is large.

fig8
Figure 8: Correlations between LAI anomaly and differences in the simulated latent heat flux (a), sensible heat flux (b), and surface Bowen ratio (c) from DYC simulation and FIX simulation (DYC minus FIX) for spring (March to May; MAM), summer (June to August; JJA), and autumn (September to November; SON) (white lines denote a confidence level higher than 95%).

These findings explain the above-mentioned cooling effect of LAI increases. In the Noah LSM, the increased LAI would reduce the canopy resistance () and thus have the potential to produce more transpiration. Therefore, more energy would be used for transpiration and less sensible heat flux would be available to heat the lower atmosphere.

Equation (2) also helps explaining why the climatic effects of LAI anomalies were larger in summer than in spring and autumn. It is well known that in this study area, spring and autumn generally have a cooler atmosphere and drier soil than summer. According to (2), low temperature and soil moisture would lead to high . Under these conditions, would remain high and experience sensitivity to temperature and soil moisture, whereas the would not be sensitive to the changes in LAI. As a result, the anomaly of LAI only weakly regulates SAT. Summer is characterized by a warm atmosphere and moist soil. The heat and moisture can adequately lead to low . Under these conditions, LAI would largely determine the changes. Therefore, the simulated Bowen ratio and SAT are more sensitive to LAI anomaly in summer than in spring and autumn.

4. Conclusion and Discussion

In this study, we investigated the responses of SAT to the interannual variability of LAI using the coupled WRF-Noah model. The results show that high LAI implies low canopy resistance () in the model and may produce more latent heat flux and less sensible heat flux. Therefore, high LAI would have a cooling effect and low LAI would have a warming effect in the model. As a result, use of the fixed seasonal cycle of LAI could produce different SAT from the use of the dynamic LAI. Our results also show that the WRF-Noah model with the prescribed dynamic LAI value could reduce the bias of simulated SAT compared with the prescribed fixed seasonal cycle of LAI in summer. However, the simulated SAT bias was generally not reduced in spring and autumn using the dynamic LAI data. Our findings indicate that use of the dynamic LAI dataset in the coupled WRF-Noah model may improve the model performance in simulating SAT in summer. Furthermore, these findings support the value of dynamical vegetation for understanding climate changes. Including a component of dynamical vegetation in the integrated climate system model is valuable, as pointed out by Wang et al. [23].

Although using the dynamic LAI in the summer demonstrated detectable improvement, the reduction of the simulated SAT bias was found only when the LAI anomaly was over 20%. Furthermore, the bias reduction was relatively small (approximately 0.03°C when the LAI anomaly was 10%) compared with the model bias. The sensitivity of simulated SAT to LAI variability seems to be much less than that of variability, as shown by Ge et al. [7].

This study improves our understanding of the sensitivity of SAT to LAI interannual variability in the coupled WRF-Noah model in eastern China. However, several caveats should be noted. First, the satellite-based LAI product is not an exact match for the Noah LSM. The satellite-based LAI data describe the gridded LAI, which includes the effects of both the vegetated and nonvegetated land, whereas the LAI in the Noah LSM refers to the value over the vegetated portion of the grid. Such discrepancies might cause large uncertainties of the prescribed LAI values in the model. Second, we have only analyzed the local effects of LAI variability in this study. The LAI variability could also remotely affect atmospheric processes through its impact on heat and water fluxes between the land surface and the atmosphere. The remote influence should be investigated in future research. Third, this study considered all types of vegetation together to investigate the effects of LAI anomalies on SAT; however, the intensity of such effects might vary with vegetation type. In future research, it is worthwhile to study the effects of each vegetation type individually. Finally, it is well known that the occurrence of rainfall is also associated with surface heat and moisture fluxes. The effects of land use change on rainfall have been well-documented [24]; however, we know little about the effects of interannual variability of LAI on rainfall. Therefore, the effects of dynamical LAI on rainfall would also be a potential research direction.

Acknowledgments

This research is jointly financed by China Global Change Research Program (2010CB950903; 2010CB950102) from the Ministry of Science and Technology of China, National Nature Science Foundation (no. 41001122), and the Hundred Talent Program of the Chinese Academy of Sciences, China.

References

  1. J. A. Foley, R. DeFries, G. P. Asner et al., “Global consequences of land use,” Science, vol. 309, no. 5734, pp. 570–574, 2005. View at Publisher · View at Google Scholar · View at Scopus
  2. G. B. Bonan, “Forests and climate change: forcings, feedbacks, and the climate benefits of forests,” Science, vol. 320, no. 5882, pp. 1444–1449, 2008. View at Publisher · View at Google Scholar · View at Scopus
  3. P. A. Dirmeyer, D. Niyogi, N. de Noblet-Ducoudré, R. E. Dickinson, and P. K. Snyder, “Impacts of land use change on climate,” International Journal of Climatology, vol. 30, no. 13, pp. 1905–1907, 2010. View at Publisher · View at Google Scholar · View at Scopus
  4. R. Mahmood, R. A. Pielke Sr., K. G. Hubbard et al., “Impacts of land use/land cover change on climate and future research priorities,” Bulletin of the American Meteorological Society, vol. 91, no. 1, pp. 37–46, 2010. View at Publisher · View at Google Scholar · View at Scopus
  5. M. Notaro, Z. Liu, and J. W. Williams, “Observed vegetation–climate feedbacks in the United States,” Journal of Climate, vol. 19, no. 5, pp. 763–786, 2006. View at Publisher · View at Google Scholar · View at Scopus
  6. G. Wang, S. Sun, and R. Mei, “Vegetation dynamics contributes to the multi-decadal variability of precipitation in the Amazon region,” Geophysical Research Letters, vol. 38, no. 19, Article ID L19703, 2011. View at Publisher · View at Google Scholar · View at Scopus
  7. Q. Ge, X. Zhang, and J. Zheng, “Simulated effects of vegetation increase/decrease on temperature changes from 1982 to 2000 across the Eastern China,” Internal Journal of Climatology, 2013. View at Publisher · View at Google Scholar
  8. F. Chen and J. Dudhia, “Coupling and advanced land surface-hydrology model with the Penn State-NCAR MM5 modeling system—part I: model implementation and sensitivity,” Monthly Weather Review, vol. 129, no. 4, pp. 569–585, 2001. View at Scopus
  9. M. B. Ek, K. E. Mitchell, Y. Lin et al., “Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model,” Journal of Geophysical Research D, vol. 108, 16 pages, 2003. View at Publisher · View at Google Scholar · View at Scopus
  10. L. R. Leung, Y.-H. Kuo, and J. Tribbia, “Research needs and directions of regional climate modeling using WRF and CCSM,” Bulletin of the American Meteorological Society, vol. 87, no. 12, pp. 1747–1751, 2006. View at Publisher · View at Google Scholar · View at Scopus
  11. J. Zhang, W.-C. Wang, and L. R. Leung, “Contribution of land-atmosphere coupling to summer climate variability over the contiguous United States,” Journal of Geophysical Research D, vol. 113, no. 22, Article ID D22109, 2008. View at Publisher · View at Google Scholar · View at Scopus
  12. J. Zhang, L. Wu, and W. Dong, “Land-atmosphere coupling and summer climate variability over East Asia,” Journal of Geophysical Research-Atmosphere, vol. 116, no. 15, Article ID D05117, 2011. View at Publisher · View at Google Scholar
  13. S. Hong, V. Lakshmi, E. E. Small, F. Chen, M. Tewari, and K. W. Manning, “Effects of vegetation and soil moisture on the simulated land surface processes from the coupled WRF/Noah model,” Journal of Geophysical Research D, vol. 114, no. 18, Article ID D18118, 2009. View at Publisher · View at Google Scholar · View at Scopus
  14. W. C. Skamarock, J. B. Klemo, J. Dudhia et al., “A Description of the advanced research WRF version 3,” NCAR Technical Note NCAR/TN-475+STR, 2008. View at Publisher · View at Google Scholar
  15. S.-Y. Hong, J. Dudhia, and S.-H. Chen, “A revised approach to ice microphysical processes for the bulk parameterization of clouds and precipitation,” Monthly Weather Review, vol. 132, no. 1, pp. 103–120, 2004. View at Scopus
  16. G. A. Grell and D. Dévényi, “A generalized approach to parameterizing convection combining ensemble and data assimilation techniques,” Geophysical Research Letters, vol. 29, no. 14, pp. 38-1–38-4, 2002. View at Publisher · View at Google Scholar · View at Scopus
  17. W. D. Collins, P. J. Rasch, B. A. Boville et al., “Description of the NCAR Community Atmosphere Model (CAM 3. 0),” NCAR Technical Note NCAR/TN-464+STR, 2004.
  18. S.-Y. Hong and H.-L. Pan, “Nonlocal boundary layer vertical diffusion in a medium-range forecast model,” Monthly Weather Review, vol. 124, no. 10, pp. 2322–2339, 1996. View at Scopus
  19. R. B. Myneni, S. Hoffman, Y. Knyazikhin et al., “Global products of vegetation leaf area and fraction absorbed PAR from year one of MODIS data,” Remote Sensing of Environment, vol. 83, no. 1-2, pp. 214–231, 2002. View at Publisher · View at Google Scholar · View at Scopus
  20. H. Fang, C. Jiang, W. Li et al., “Characterization and intercomparison of global moderate resolution leaf area index (LAI) products: analysis of climatologies and theoretical uncertainties,” Journal of Geophysical Research, vol. 118, no. 2, pp. 529–548, 2013. View at Publisher · View at Google Scholar
  21. T. D. Mitchell and P. D. Jones, “An improved method of constructing a database of monthly climate observations and associated high-resolution grids,” International Journal of Climatology, vol. 25, no. 6, pp. 693–712, 2005. View at Publisher · View at Google Scholar · View at Scopus
  22. B. Jacquemin and J. Noilhan, “Sensitivity study and validation of a land surface parameterization using the HAPEX-MOBILHY data set,” Boundary-Layer Meteorology, vol. 52, no. 1-2, pp. 93–134, 1990. View at Publisher · View at Google Scholar · View at Scopus
  23. G. Wang, S. Sun, and R. Mei, “Vegetation dynamics contributes to the multi-decadal variability of precipitation in the Amazon region,” Geophysical Research Letters, vol. 38, no. 19, Article ID L19703, 2011. View at Publisher · View at Google Scholar · View at Scopus
  24. R. A. Pielke Sr., J. Adegoke, A. Beltrán-Przekurat et al., “An overview of regional land-use and land-cover impacts on rainfall,” Tellus B, vol. 59, no. 3, pp. 587–601, 2007. View at Publisher · View at Google Scholar · View at Scopus