• Baker, I., 2005: SiB 3.0 prognostic variable and equations set. Atmospheric Science Department, Colorado State University Rep. 19 pp.

  • Baker, I., , Denning A. S. , , Hanan N. , , Prihodko L. , , Uliasz M. , , Vidale P-L. , , Davis K. , , and Bakwin P. , 2003: Simulated and observed fluxes of sensible and latent heat and CO2 at the WLEF-TV tower using SiB2.5. Global Change Biol., 9 , 12621277.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beven, K. J., , and Kirkby M. J. , 1979: A physically based variable contributing area model of basin hydrology. Hydrol. Sci. Bull., 24 , 4369.

  • Bonan, G. B., , Levis S. , , Kergoat L. , , and Oleson K. W. , 2002: Landscapes as patches of plant functional types: An integrated concept for climate and ecosystem models. Global Biogeochem. Cycles, 16 , 10211051.

    • Search Google Scholar
    • Export Citation
  • Chase, T. N., , Pielke R. A. , , Kittel T. G. F. , , Nemani R. , , and Running S. W. , 1996: Sensitivity of a general circulation model to global changes in leaf area index. J. Geophys. Res., 101 , 73937408.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Choudhury, B. J., , and DiGirolamo N. E. , 1998: A biophysical process-based estimate of global land surface evaporation using satellite and ancillary data. I: Model description and comparison with observations. J. Hydrol., 205 , 164185.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Choudhury, B. J., , DiGirolamo N. E. , , Susskind J. , , Darnell W. L. , , Gupta S. K. , , and Asrar G. , 1998: A biophysical process-based estimate of global land surface evaporation using satellite and ancillary data. II. Regional and global patterns of seasonal and annual variations. J. Hydrol., 205 , 186204.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Clapp, R. B., , and Hornberger G. M. , 1978: Empirical equations for some soil hydraulic properties. Water Resour. Res., 14 , 601604.

  • Colello, G. D., , Grivet C. , , Sellers P. J. , , and Berry J. A. , 1998: Modeling of energy, water, and CO2 flux in a temperate grassland ecosystem with SiB2: May–October 1987. J. Atmos. Sci., 55 , 11411169.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collatz, G. J., , Ball J. T. , , Grivet C. , , and Berry J. A. , 1991: Physiological and environmental regulation of stomatal conductance, photosynthesis and transpiration: A model that includes a laminar boundary layer. Agric. For. Meteor., 54 , 107136.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dai, A., , and Trenberth K. E. , 2002: Estimates of freshwater discharge from continents: Latitudinal and seasonal variations. J. Hydrometeor., 3 , 660687.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dai, Y., , Dickinson R. E. , , and Wang Y-P. , 2004: A two-big-leaf model for canopy temperature, photosynthesis, and stomatal conductance. J. Climate, 17 , 22812299.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dickinson, R. E., , Oleson K. W. , , Bonan G. B. , , Hoffman F. , , Thornton P. E. , , Vertenstein M. , , Yang Z. L. , , and Zeng X. , 2006: The community land model and its climate statistics as a component of the community climate system model. J. Climate, 19 , 23022324.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dirmeyer, P. A., , Gao X. , , Zha M. , , Guo Z. , , Oki T. , , and Hanasaki N. , 2005: The second Global Soil Wetness Project (GSWP-2): Multi-model analysis and implications for our perception of the land surface. COLA Tech. Rep. 185, 46 pp.

    • Search Google Scholar
    • Export Citation
  • Feddema, J. J., , Oleson K. W. , , Bonan G. B. , , Mearns L. O. , , Buja L. E. , , Meehl G. A. , , and Washington W. M. , 2005: The importance of land-cover change in simulating future climates. Science, 310 , 16741678.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Garratt, J. R., 1992: The Atmospheric Boundary Layer. Cambridge University Press, 316 pp.

  • Lawrence, D. M., , Thornton P. E. , , Oleson K. W. , , and Bonan G. B. , 2007: The partitioning of evapotranspiration into transpiration, soil evaporation, and canopy evaporation in a GCM: Impacts on land–atmosphere interaction. J. Hydrometeor., 8 , 862880.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lawrence, P. J., , and Chase T. N. , 2007: Representing a MODIS Consistent Land Surface in the Community Land Model (CLM 3.0). J. Geophys. Res., 112 , G01023. doi:10.1029/2006JG000168.

    • Search Google Scholar
    • Export Citation
  • Li, K. Y., , De Jong R. , , and Boisvert J. B. , 2001: An exponential root-water-uptake model with water stress compensation. J. Hydrol., 252 , 189204.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Niu, G-Y., , and Yang Z-L. , 2006: Effects of frozen soil on snowmelt runoff and soil water storage at a continental scale. J. Hydrometeor., 7 , 937952.

  • Niu, G-Y., , Yang Z-L. , , Dickinson R. E. , , and Gulden L. E. , 2005: A simple TOPMODEL-based runoff parameterization (SIMTOP) for use in global climate models. J. Geophys. Res., 110 , D21106. doi:10.1029/2005JD006111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Niu, G-Y., , Yang Z-L. , , Dickinson R. E. , , Gulden L. E. , , and Su H. , 2007: Development of a simple ground water model for use in climate models and evaluation with Gravity Recovery and Climate Experiment data. J. Geophys. Res., 112 , D07103. doi:10.1029/2006JD007522.

    • Search Google Scholar
    • Export Citation
  • Niyogi, D. S., , Xue Y. , , and Raman S. , 2002: Hydrological land surface response in a tropical regime and a midlatitudinal regime. J. Hydrometeor., 3 , 3956.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Oleson, K. W., and Coauthors, 2004: Technical description of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-461+STR, 173 pp.

  • Oleson, K. W., and Coauthors, 2008: Improvements to the Community Land Model and their impact on the hydrological cycle. J. Geophys. Res., 113 , G01021. doi:10.1029/2007JG000563.

    • Search Google Scholar
    • Export Citation
  • Philip, J. R., 1957: Evaporation and moisture and heat fields in the soil. J. Meteor., 14 , 354366.

  • Pielke, R. A., 2001: Influence of the spatial distribution of vegetation and soils on the prediction of cumulus convective rainfall. Rev. Geophys., 39 , 151177.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Qian, T., , Dai A. , , Trenberth K. E. , , and Oleson K. W. , 2006: Simulation of global land surface conditions from 1948 to 2004. Part I: Forcing data and evalutation. J. Hydrometeor., 7 , 953975.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., 1985: Canopy reflectance, photosynthesis and transpiration. Int. J. Remote Sens., 6 , 13351372.

  • Sellers, P. J., 1997: Modeling the exchanges of energy, water, and carbon between continents and the atmosphere. Science, 275 , 502509.

  • Sellers, P. J., , Mintz Y. , , Sud Y. C. , , and Dalcher A. , 1986: A Simple Biosphere Model (SiB) for use within general circulation models. J. Atmos. Sci., 43 , 505531.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., , Shuttleworth W. J. , , Dorman J. L. , , Dalcher A. , , and Roberts J. M. , 1989: Calibrating the simple biosphere model for Amazonian tropical forest using field and remote sensing data. Part I: Average calibration with field data. J. Appl. Meteor., 28 , 727759.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., , Los S. O. , , Tucker C. J. , , Justice C. O. , , Dazlich D. A. , , Collatz G. J. , , and Randall D. A. , 1996a: A revised land surface parameterization (SiB2) for atmospheric GCMs. Part II: The generation of global fields of terrestrial biophysical parameters from satellite data. J. Climate, 9 , 706737.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., and Coauthors, 1996b: A revised land surface parameterization (SiB2) for atmospheric GCMs. Part I: Model formulation. J. Climate, 9 , 676705.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sen, O. L., , Shuttleworth W. J. , , and Yang Z. L. , 2000: Comparative evaluation of BATS2, BATS, and SiB2 with Amazon data. J. Hydrometeor., 1 , 135153.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stöckli, R., , and Vidale P. L. , 2005: Modeling diurnal to seasonal water and heat exchanges at European Fluxnet sites. Theor. Appl. Climatol., 80 , 229243.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stöckli, R., and Coauthors, 2008: Use of FLUXNET in the Community Land Model development. J. Geophys. Res., 113 , G01025. doi:10.1029/2007JG000562.

    • Search Google Scholar
    • Export Citation
  • Thornton, P. E., , and Zimmermann N. E. , 2007: An improved canopy integration scheme for a land surface model with prognostic canopy structure. J. Climate, 20 , 39023923.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • U.S. Geological Survey, cited. 1996: GTOPO30 Global Digital Elevation Model (DEM). [Available online at http://edc.usgs.gov/products/elevation/gtopo30/gtopo30.html.].

    • Search Google Scholar
    • Export Citation
  • von Storch, H., , and Zwiers F. W. , 1999: Statistical Analysis in Climate Research. Cambridge University Press, 484 pp.

  • White, M. A., , Thornton P. E. , , Running S. W. , , and Nemani R. , 2000: Parameterization and sensitivity analysis of the BIOME–BGC terrestrial ecosystem model: Net primary production controls. Earth Interactions, 4 .[Available online at http://EarthInteractions.org.].

    • Search Google Scholar
    • Export Citation
  • Willmott, C. J., , and Matsuura K. , 2000: Terrestrial air temperature and precipitation: Monthly and annual climatologies (Version 2.0.1). [Available online at http://climate.geog.udel.edu/~climate/html_pages/download.html.].

    • Search Google Scholar
    • Export Citation
  • Zeng, X., 2001: Global vegetation root distribution for land modeling. J. Hydrometeor., 2 , 525530.

  • View in gallery

    Comparison of hydrology and sensible heat in the (a) CLM3 and (b) SiB2 land surface parameterizations. Temperature (T) and specific humidity (q) are specified with subscripts for the atmospheric reference height (ATM), the canopy air space (S), at the leaf surface (V), and the ground (G). Total resistance to transfer sensible heat or moisture between the atmospheric reference height, the surface of leaves, and the ground, with the canopy air space are specified as rA, rB, and rD, respectively.

  • View in gallery

    Ensemble mean differences in average seasonal precipitation between Willmott and Matsuura (2000) (a),(b) 1970–99 climatology and current CLM3 hydrology experiment, and with (c),(d) new CLM SiB hydrology experiment, and (e),(f) differences between experiments. Differences with statistical significance >95% are stippled.

  • View in gallery

    Same as Fig. 2, but for ensemble mean differences in average seasonal near-surface air temperature.

  • View in gallery

    Fig. A1. Differences between CLM3, CLM SiB, and SiB2 soil hydrology parameters for the range of representative soil classes described by Clapp and Hornberger (1978), ordered by (a),(c),(d) sand and (b) clay content: (a) saturated moisture content, (b) wetness exponent, (c) saturated matrix potential, and (d) saturated hydraulic conductivity.

  • View in gallery

    Fig. B1. Idealized comparison of SiB2 and CLM3 bare soil evaporation for (a) sand; (b) loam; (c) clay loam; and (d) clay soils under typical DJF Amazon surface and atmospheric conditions.

  • View in gallery

    Fig. E1. Differences between CLM3, SiB2, and CLM3.5 soil moisture root stress functions for (a) shrub PFTs in sand; (b) grass PFTs in loam; (c) needleleaf-tree PFTs in sandy clay loam; and (d) broadleaf evergreen tree PFTs in sandy clay.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 13 13 9
PDF Downloads 11 11 5

Climate Impacts of Making Evapotranspiration in the Community Land Model (CLM3) Consistent with the Simple Biosphere Model (SiB)

View More View Less
  • 1 Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado
© Get Permissions
Full access

Abstract

In recent climate sensitivity experiments with the Community Climate System Model, version 3 (CCSM3), a wide range of studies have found that the Community Land Model, version 3 (CLM3), simulates mean global evapotranspiration with low contributions from transpiration (15%), and high contributions from soil and canopy evaporation (47% and 38%, respectively). This evapotranspiration partitioning is inconsistent with the consensus of other land surface models used in GCMs. To understand the high soil and canopy evaporation and the low transpiration observed in the CLM3, select individual components of the land surface parameterizations that control transpiration, canopy and soil evaporation, and soil hydrology are compared against the equivalent parameterizations used in the Simple Biosphere Model, versions 2 and 3 (SiB2 and SiB3), and against more recent developments with CLM. The findings of these investigations are used to develop new parameterizations for CLM3 that would reproduce the functional dynamics of land surface processes found in SiB and other alternative land surface parameterizations. Global climate sensitivity experiments are performed with the new land surface parameterizations to assess how the new SiB, consistent CLM land surface parameterizations, influence the surface energy balance, hydrology, and atmospheric fluxes in CLM3, and through that the larger-scale climate modeled in CCSM3. It is found that the new parameterizations enable CLM to simulate evapotranspiration partitioning consistently with the multimodel average of other land surface models used in GCMs, as evaluated by . The changes in surface fluxes also resulted in a number of improvements in the simulation of precipitation and near-surface air temperature in CCSM3. The new model is fully coupled in the CCSM3 framework, allowing a wide range of climate modeling investigations without the surface hydrology issues found in the current CLM3 model. This provides a substantially more robust framework for performing climate modeling experiments investigating the influence of land cover change and surface hydrology in CLM and CCSM than the existing CLM3 parameterizations. The study also shows that changes in land surface hydrology have global scale impacts on model climatology.

Corresponding author address: Peter J. Lawrence, CIRES, University of Colorado, Campus Box 216, Boulder, CO 80309. Email: peter.j.lawrence@colorado.edu

Abstract

In recent climate sensitivity experiments with the Community Climate System Model, version 3 (CCSM3), a wide range of studies have found that the Community Land Model, version 3 (CLM3), simulates mean global evapotranspiration with low contributions from transpiration (15%), and high contributions from soil and canopy evaporation (47% and 38%, respectively). This evapotranspiration partitioning is inconsistent with the consensus of other land surface models used in GCMs. To understand the high soil and canopy evaporation and the low transpiration observed in the CLM3, select individual components of the land surface parameterizations that control transpiration, canopy and soil evaporation, and soil hydrology are compared against the equivalent parameterizations used in the Simple Biosphere Model, versions 2 and 3 (SiB2 and SiB3), and against more recent developments with CLM. The findings of these investigations are used to develop new parameterizations for CLM3 that would reproduce the functional dynamics of land surface processes found in SiB and other alternative land surface parameterizations. Global climate sensitivity experiments are performed with the new land surface parameterizations to assess how the new SiB, consistent CLM land surface parameterizations, influence the surface energy balance, hydrology, and atmospheric fluxes in CLM3, and through that the larger-scale climate modeled in CCSM3. It is found that the new parameterizations enable CLM to simulate evapotranspiration partitioning consistently with the multimodel average of other land surface models used in GCMs, as evaluated by . The changes in surface fluxes also resulted in a number of improvements in the simulation of precipitation and near-surface air temperature in CCSM3. The new model is fully coupled in the CCSM3 framework, allowing a wide range of climate modeling investigations without the surface hydrology issues found in the current CLM3 model. This provides a substantially more robust framework for performing climate modeling experiments investigating the influence of land cover change and surface hydrology in CLM and CCSM than the existing CLM3 parameterizations. The study also shows that changes in land surface hydrology have global scale impacts on model climatology.

Corresponding author address: Peter J. Lawrence, CIRES, University of Colorado, Campus Box 216, Boulder, CO 80309. Email: peter.j.lawrence@colorado.edu

1. Introduction

In our recent paper Lawrence and Chase (2007), we developed new Moderate Resolution Imaging Spectroradiometer (MODIS) consistent current day land surface parameters for the Community Land Model, version 3 (CLM3), to be used with the Community Climate System Model (CCSM3). The average climate simulated in CLM3 and CCSM3 with the new MODIS land surface parameters has year-round reduced precipitation, dryer soils, and reduced evapotranspiration compared to the climate simulated with the current release parameters. This was an unusual result given the new MODIS parameters have increased vegetation through year-round higher average grid cell leaf area index (LAI).

Further investigation in Lawrence and Chase (2007) found the reduced evapotranspiration with the increased vegetation of the MODIS parameters was the product of large decreases in soil evaporation that offset increases in canopy evaporation and transpiration. The absolute magnitudes of transpiration compared to soil and canopy evaporation in both climate simulations regardless of land surface parameters also demonstrated that transpiration in CLM3 contributes a very low fraction (15%) of the average global evapotranspiration, with the dominant contributions coming from evaporation from bare soils (47%) and canopy-intercepted precipitation (38%). This is inconsistent with other land surface models used in global climate models (GCMs), which have average global evapotranspiration dominated by transpiration (47%), with substantially smaller contributions from evaporation from bare soil (36%) and canopy-intercepted precipitation (17%; Dirmeyer et al. 2005).

The low transpiration, and high canopy and soil evaporation found in these experiments are supported by other studies with CLM. In their investigation, Lawrence et al. (2007) found that the evapotranspiration partition of CLM3 was inconsistent with a range of recent global surface hydrology studies, including Dirmeyer et al. (2005), Choudhury and DiGirolamo (1998), and Choudhury et al. (1998). They addressed these evapotranspiration issues through modifying the parameterizations that control canopy interception and evaporation, soil hydrology, soil evaporation, and plant photosynthesis and transpiration in CLM3 to simulate surface hydrology consistently with the global hydrology studies. The changes in parameterizations were highly effective in reproducing the global evapotranspiration partition found in the global studies, with their modified version of CLM being dominated by transpiration (44%), with smaller contributions from evaporation from bare soil (39%) and canopy-intercepted precipitation (17%).

In the most recently released version of CLM3.5, Oleson et al. (2008) also addressed these issues in evapotranspiration partition by incorporating 1) the new land surface datasets of Lawrence and Chase (2007); 2) the improved canopy integration of Thornton and Zimmermann (2007), with new nitrogen limitations; 3) individual plant functional type (PFT) root water stress response functions derived from White et al. (2000); 4) the canopy interception scaling of Lawrence et al. (2007); 5) the simple TOPMODEL-based surface and subsurface runoff of Niu et al. (2005); 6) the groundwater model of Niu et al. (2007); 7) the frozen-soil scheme of Niu and Yang (2006); and 8) SiB2 soil evaporation resistance from Sellers et al. (1996b). The new CLM3.5 parameterizations greatly improved the evapotranspiration partition compared to CLM3 and were nearly as effective in reproducing the global evapotranspiration partition studies as were the modifications of Lawrence et al. (2007). The CLM3.5 parameterizations also improved the performance of CLM surface fluxes of moisture and energy relative to FLUXNET-measured sites, with large improvements in temporal correlation and root-mean-square errors in both latent and sensible heat fluxes for all 15 flux tower sites studied by Stöckli et al. (2008).

The latest release version of CLM3.5, however, is only available as an offline land surface model, which can be run under prescribed atmospheric forcing. For the climate modeling investigations we wished to perform, we required a version of CLM that could simulate realistic evapotranspiration fluxes when fully coupled with the Community Atmospheric Model, version 3 (CAM3). The fully coupled CLM requirement ensured we could use the land model over the range of CCSM climate modeling configurations currently available with CLM3. As discussed later, when coupled to CAM3 the low transpiration, and high soil and canopy evaporation of CLM3 have strong influences on the flux of moisture and energy from the land surface to the atmosphere, and therefore, on the climate simulated in CCSM3.

To satisfy these climate modeling requirements, we developed our own modified version of CLM3 that has evapotranspiration partition consistent with other land surface models used in GCMs studied by Dirmeyer et al. (2005). Our new version of CLM was developed during the same period as the work done by Lawrence et al. (2007) and the subsequent development of the new CLM3.5 model by Oleson et al. (2008). Our new version of CLM has the advantage of being fully coupled into the existing CCSM3 framework addressing our need to use the other climate system models of CCSM. The new version of CLM was developed from a systematic investigation of the individual components that control transpiration, canopy and soil evaporation, and soil hydrology in CLM3. These processes were compared against the equivalent components used in the Simple Biosphere Model, versions 2 and 3 (SiB2 and SiB3), and against alternative component parameterizations from literature and from the recent developments with CLM.

The first level of the investigation studied the behavior of the CLM3, SiB, and alternative parameterizations through idealized offline simulations performed over a range of soil moisture, radiation, and atmospheric conditions. The idealized parameterization simulations were performed as isolated code fragments removed from the rest of the land surface model to prevent interactions between competing processes and the dynamic feedbacks that are found with the complete versions of land surface models. These investigations focused on the similarities and differences of the land surface parameterizations governing surface hydrology and surface fluxes of moisture and energy to the atmosphere.

The findings of the idealized investigations were used to develop modifications of the individual components of CLM3, so that the new parameterizations would reproduce the functional dynamics of land surface processes found in the SiB or the other land surface parameterizations. Each newly modified component of the CLM parameterizations was investigated in offline CLM experiments with prescribed atmospheric forcing to evaluate how those changes influenced the surface hydrology and surface fluxes of the model in the absence of atmospheric feedbacks. Because of the number of parameterizations taken from SiB, hereafter we refer to the new version of CLM3 with all modifications as CLM SiB.

To evaluate how the new CLM SiB land surface parameterization influenced the surface energy balance, hydrology, and atmospheric fluxes when coupled to CAM3 and to assess how those changes influenced larger-scale climate modeled in CCSM3, we performed a series of coupled global climate sensitivity experiments with the new CLM SiB parameterizations relative to CCSM3 with the release CLM3. The climate modeled in the sensitivity experiments were compared with globally observed climate and river runoff data to evaluate the climate effects of new land surface parameters relative to existing climate biases in the model with the current CLM3 parameterizations.

2. Methods

a. Comparing CLM3 to SiB2.0

The current version of the National Center for Atmospheric Research (NCAR) CCSM3 uses the CLM3 as the standard land surface parameterization. The underlying biogeophysics of CLM3 are described in Bonan et al. (2002), Oleson et al. (2004), and Dickinson et al. (2006), with a high-level conceptual model shown in Fig. 1a. The CLM3 model can be run in a number of dynamic and prescribed vegetation modes. In the investigations and experiments of this paper, CLM3 has been run in the prescribed vegetation mode with the distribution of plants and their phenologies described by the new MODIS consistent land surface parameters developed in Lawrence and Chase (2007). This is the same data used to generate the CLM3.5 parameters described in Oleson et al. (2008).

The SiB2 model described in Sellers et al. (1996b) and shown in Fig. 1b is a redevelopment of the SiB1 model described in Sellers et al. (1986) and derived from conceptual work by Sellers (1985). The redevelopment of SiB1 to SiB2 incorporated realistic canopy photosynthesis, and conductance, and satellite data to prescribe vegetation state and phenology. The individual components within the model were derived consistently with observational data collected at the First International Satellite Land Surface Climatology Project Field Experiment (FIFE; Colello et al. 1998) and other field campaigns (Sellers et al. 1989).

We used the SiB parameterization predominantly because the SiB2 model and subsequent versions have been validated and compared to in situ observational data for fluxes of sensible and latent heat and CO2 with good general agreement for surface hydrology and latent heat fluxes. These projects cover a wide range of biospheres including the boreal forest at the WLEF tower (Baker et al. 2003); the Amazon rain forest from the Anglo–Brazilian Amazonian Climate Observation Study (ABRACOS) flux towers (Sen et al. 2000); and many other representative sites with flux tower data from the Large-scale Biosphere–Atmosphere Experiment in Amazonia (LBA), Ameriflux, and Euroflux projects (Stöckli and Vidale 2005).

Both CLM3 and SiB2 are third-generation land surface models as described by Sellers (1997), with photosynthesis and transpiration models derived from Collatz et al. (1991). In both CLM3 and SiB2, the fluxes of sensible heat (H) and water vapor (E) are partitioned into fluxes between the surface of leaves and the ground to the canopy air space, and from the canopy air space to atmospheric reference height, as shown in Figs. 1a and 1b. In both models the air in the canopy air space is assumed to have negligible capacity to store heat or moisture, so all fluxes between the canopy and the ground with the canopy air space must be balanced by the fluxes to the atmospheric reference height by conservation of energy and mass. For simplicity the terminology used in this paper follows Garratt (1992) for flux descriptions in both models; however, these terms are directly equivalent with the terminology used in both Oleson et al. (2004) and Sellers et al. (1996b).

b. CLM parameterization comparison study

The first step of the comparison study was to identify the components of CLM3 that were contributing to the low transpiration, and high soil and canopy evaporation found in the model. To help in this process, we developed a subsystem assessment framework to independently evaluate the model components of CLM3 that simulate 1) soil hydrological properties; 2) soil evaporation; 3) soil infiltration and runoff; 4) deep soil drainage; 5) photosynthesis and transpiration; 6) soil moisture root stress; 7) root zone soil moisture representation; and 8) canopy interception and evaporation. This subsystem framework allowed us to systematically test and evaluate each CLM component against the equivalent subsystems in the SiB model and the other parameterizations from literature.

The subsystems testing framework was developed in Microsoft Excel, with the ability to simulate individual or combinations of subsystems over a range of idealized environmental conditions and periods. The subsystems were developed directly from the mathematical formulations described in Oleson et al. (2004) for CLM3 and in Sellers et al. (1996b) for SiB2. The framework was adapted to include other alternative parameterizations found in literature, allowing direct evaluation against the CLM3 and SiB parameterizations.

In cases where the subsystem comparisons and the idealized simulations identified the CLM3 components were contributing to the low transpiration or high soil and canopy evaporation, we developed new CLM subsystems from either the SiB or the alternative parameterizations that would reduce those influences. We used the criteria that components needed to be changed in CLM when new components were found to 1) simulate behavior more consistently with real world observation; 2) have underlying biogeophysics that were fundamentally more sound; or 3) be closer to the multimodel average of Dirmeyer et al. (2005). A more detailed analysis of each subsystem is provided in appendixes AF.

c. Developing new CLM SiB parameterizations

To address the high soil evaporation and low transpiration found in CLM3, the differences in the soil hydraulic properties of CLM3 were compared to the Clapp and Hornberger (1978) values used in SiB2. Differences between the two prescriptions were addressed with new formulations of saturated volumetric soil moisture (θSAT), soil wetness exponent (B), saturated soil matrix potential (ψSAT), and saturated hydraulic conductivity (KSAT). The new formulations were based on sand and clay content fitted to the experimentally derived values of Clapp and Hornberger (1978) (appendix A). The high soil evaporation rates also were addressed by adding the additional soil resistance term of Sellers et al. (1996b) to the atmospheric resistance for the soil evaporation (appendix B).

Infiltration and runoff were modified to include explicit surface ponding, with the ponding runoff scaled by convective and large scale treatments of precipitation as described in Sellers et al. (1996b). The saturated runoff fraction calculated with the TOPMODEL prognostic water table was removed, with infiltration and surface ponding calculated using the CLM3 infiltration excess scheme as was done in SiB3 (Baker 2005). Evaporation from ponded surface water was explicitly included in soil evaporation using the potential evaporation rate over the pond wetness fraction of the ground, as calculated in SiB2 (Sellers et al. 1996b). Changes in surface runoff and ponding are detailed in appendix C.

The high drainage rates of CLM3 were addressed by removing the lateral drainage components from the lower soil layers and by scaling the gravitational drainage component by topographic slope, as was done in SiB2. The topographic slope was calculated from the Global 30 Arc-Second Elevation Dataset (GTOPO30) 1-km global digital elevation model (DEM; U.S. Geological Survey 1996), with an additional 2 m of topographic fall added for each 1 km to represent the depth of river channels not resolved by the DEM. The soil profile depth was extended to 12 soil layers to increase moisture storage, following field studies with SiB3 by Baker (2005).

To address the low photosynthesis in CLM3, the new canopy integration scheme of Thornton and Zimmermann (2007) with the new nitrogen limitations implemented in CLM3.5 were incorporated into CLM SiB. The new photosynthetic parameterization substantially increased transpiration and photosynthesis for all PFTs (appendix D). Transpiration limitations related to soil moisture stress were addressed by replacing the soil moisture stress function of CLM3 with those derived from White et al. (2000) and used in CLM3.5. The evaluation of the CLM3, SiB2, and CLM3.5 soil moisture stress functions found that the CLM3.5 function had near-identical behavior to the SiB2 scheme, with both schemes allowing much greater extraction of soil water before soil moisture stress was encountered (appendix E).

To allow the plants in CLM to access the whole root zone soil profile in a more extensive manner, the fixed soil layer rooting scheme used in CLM3 was replaced with an exponential root water uptake model that included water stress compensation based on observational and modeling studies from Li et al. (2001). The new scheme dynamically allowed large proportions of transpiration to be met from deeper, more sparsely rooted soil layers when upper layers are depleted of soil moisture (appendix E).

To address the high canopy interception and evaporation in CLM3, we implemented a new canopy interception scheme in CLM SiB that explicitly included large-scale and convective precipitation fractions following the SiB2 methods and constants detailed in Sellers et al. (1996b). We also changed the fraction of the canopy evaporating at the potential evaporation rate to the simple ratio used in Sellers et al. (1996b; appendix F).

d. Experimental design for climate sensitivity to CLM SiB parameterizations

To assess how the various components of the new CLM SiB parameterization contributed to changes in evapotranspiration partition, a series of offline experiments were performed with each component subsystem component. The offline experiments were performed with monthly climatological atmospheric forcing derived from the National Centers for Environmental Prediction (NCEP) reanalysis forcing data of Qian et al. (2006). The incoming solar radiation was scaled to match the climatology of a control run from CCSM3 to address undocumented low incoming radiation biases in the forcing dataset. The precipitation and 2-m air temperature forcing data also were scaled to match long-term climatology values from Willmott and Matsuura (2000) to account for precipitation biases in the NCEP reanalysis data. The offline CLM experiments were performed at the T42 grid resolution for 30 yr each, with the first 10 yr discarded for soil moisture spinup.

To assess how the new CLM SiB land surface parameterization influenced the CLM surface hydrology when coupled to CAM3, we performed a series of climate sensitivity experiments with CCSM3 with the new CLM SiB hydrology compared to a control experiment with the current CLM3 hydrology. The sensitivity experiments were performed at the T42 grid resolution with prescribed monthly climatology sea surface temperatures (SSTs) and sea ice distributions prescribed in the Data Ocean Model (DOCN6) and the Community System Model Sea Ice Model, version 5 (CSIM5), using the CCSM3 standard 1949–2001 monthly climatology data.

To ensure that differences in climate response in CCSM were not dependent on initial conditions, the experiments were run as an ensemble of three realizations for each version of the land surface hydrology, with different initial conditions prescribed for each realization. The differences in initial conditions were set through randomly increasing or decreasing the surface temperature of all land grid cells by ±0.1°C from the original CLM initialization file. Each realization was run for a 40-yr period, with the initial 10 yr discarded for soil moisture spinup. This left a total of 90 yr of climate simulation for each experiment to be used for analysis.

The ensemble mean differences between the CCSM3 coupled climate experiments with the new CLM SiB parameterizations and the current CLM3 parameterizations were assessed relative to the total model climate biases of CCSM3 with prescribed climatology SSTs and sea ice. Precipitation and air temperature from the CCSM experiments were compared to observed climatology data from Willmott and Matsuura (2000). The land surface hydrology of the CLM SiB and CLM3 parameterizations were assessed relative to the global hydrology found with other GCM land surfaces in the multimodel experiments of Dirmeyer et al. (2005). The multimodel average of Dirmeyer et al. (2005) was used as a guide to evapotranspiration, as it was consistent with a range of recent global flux partition studies as reported in Lawrence et al. (2007). The net influence of the CLM SiB parameterizations on the surface hydrological balance was assessed for both the offline and coupled experiments for major river basins using the annual river discharge of Dai and Trenberth (2002).

Overall changes in ensemble mean climate variables with the CLM SiB hydrology were tabulated for all land and for the three analysis regions of Amazon, boreal, and Sahara and Arabia for all seasons, with the statistical strengths of these relationships assessed through multiple statistical tests. The statistical testing involved a Student’s t test and a Wilcoxon signed rank test on the entire ensemble time series to identify statistically significant differences in the experiments for all land and for the regions. The Wilcoxon signed rank test was used following von Storch and Zwiers (1999), as it is nonparametric and provides insurance against moderate departures from the distribution assumptions required for calculating t-test significance.

In addition to the all land and regional analysis, the ensemble mean seasonal differences in a range of representative climate variables were mapped globally between the CLM SiB and CLM3 experiments and against the observed climatology data from Willmott and Matsuura (2000). The global maps showed differences in the mean climatological values for each season for each grid cell and for larger regions. Time series t tests were performed on each grid cell to identify statistically significant changes for the grid cell and for larger regions between the ensembles.

3. Results

a. Global differences in surface hydrology

The influence on global annual surface hydrology for each component of the new CLM SiB parameterization were investigated through the series of offline CLM experiments, shown in Table 1. In this series of experiments, each component was incrementally added to the CLM model, as described in appendixes AF, and then assessed for changes in global surface hydrology. The initial offline experiment shows the average global surface hydrology of the release CLM3 model. The surface hydrology reflects the issues discussed in the introduction, with low transpiration contribution and high soil and canopy evaporation relative to the multimodel average of Dirmeyer et al. (2005).

The second experiment investigated the influence of changing the soil hydrological properties to better match the experimentally derived values used in SiB2 and described in appendix A. The net influence of the new soil properties was increased evapotranspiration through increased transpiration, which was partially offset by reduced soil evaporation. The two main drivers for this change were the lower soil matrix potential, which allowed greater moisture to be extracted by plant roots, and the increased hydraulic conductivity, which allowed moisture to infiltrate to the plant root zone more rapidly. The increase in evapotranspiration was offset by reduced surface runoff and drainage.

The third experiment investigated the influence of adding the SiB2 soil resistance term to the soil evaporation equation, as described in appendix B. The additional soil resistance resulted in reduced evapotranspiration through reduced soil evaporation, with the lower soil evaporation offset by increased transpiration, canopy evaporation, surface runoff, and drainage. The forth experiment investigated the influence of adding the SiB2 surface ponding, runoff, and drainage schemes, described in appendix C. The SiB2 schemes had the combined influence of increasing evapotranspiration through increased transpiration and soil evaporation. The SiB2 schemes also reduce surface runoff and increased deep soil drainage.

The fifth experiment investigated the influence of the new canopy integration scheme of Thornton and Zimmermann (2007), described in appendix D. The higher photosynthesis of the scheme resulted in increased evapotranspiration through higher transpiration. The higher transpiration was offset by reduced soil evaporation and deep soil drainage. The sixth experiment investigated the new root zone soil moisture stress scheme, described in appendix E. The increased evapotranspiration resulted from increased transpiration with the increased soil moisture availability. The increase in transpiration was offset by decreased deep soil drainage.

The final experiment investigated the influence of including the SiB2 canopy interception and evaporation scheme in CLM SiB, as described in appendix F. This experiment had all of the components of the new CLM SiB parameterization. The influence of the new canopy interception and evaporation scheme was to reduce evapotranspiration back below the initial CLM3 parameterization. The decrease in evapotranspiration was the product of greatly reduced canopy evaporation, which was partially offset by increases in transpiration and soil evaporation. The net reduction in evapotranspiration was offset by increases in both the surface runoff and the deep soil drainage.

The final CLM SiB experiment demonstrated that the new surface hydrology parameterizations were successful in addressing the very low transpiration, and high soil and canopy evaporation of CLM3. The CLM SiB experiment was substantially closer to the average transpiration contribution found in the multimodel experiments (44% compared to 47%) of Dirmeyer et al. (2005) and had the same absolute transpiration flux in terms of millimeters per day. Canopy evaporation was much lower than CLM3 but still higher than the multimodel average in both contribution and millimeters per day. Soil evaporation also was lower than the multimodel average in both terms. Surface runoff was the same as the multimodel average, however, deep soil drainage was considerably lower. The CLM SiB hydrology also compared well with the CLM3.5 results of Oleson et al. (2008).

The influence of the new CLM SiB land surface parameterizations on the ensemble of coupled CCSM experiments is shown in Table 2 for global annual average land surface hydrology relative to the multimodel average values of Dirmeyer et al. (2005). The changes in the global average surface hydrology in the coupled experiments reflect the findings of the offline experiments, with similar changes in terms of both evapotranspiration partition and for changes in individual moisture fluxes in millimeters per day. The coupled CCSM experiments also had the same changes in global surface runoff and drainage as were found in the offline experiments.

b. Differences in precipitation compared to observations in coupled CLM experiments

As discussed in the previous section, the changes in surface hydrology with the new CLM SiB parameterizations had a limited influence on the global annual average land precipitation in the coupled experiments. There were, however, larger seasonal and regional changes in precipitation in CCSM. The ensemble mean differences in precipitation between the CLM SiB and the CLM3 control experiments are shown in Table 3 relative to differences with the observed climatology data of Willmott and Matsuura (2000). The seasonal differences in precipitation between each experiment and the observed climatology data of Willmott and Matsuura (2000) also are mapped globally in Fig. 2.

Table 3 shows that globally there were significant changes in average land precipitation with the CLM SiB parameterizations for all seasons, with the overall influence of the new parameterization reducing the wet bias of CCSM3 by 0.02 mm day−1 annually. The decrease in wet bias was largest for December–February (DJF), March–May (MAM), and September–November (SON), which was partially offset by a decrease in dry bias in June–August (JJA). In the Amazon region, there was a significant decrease in dry bias in MAM that offset significant increases in the dry bias of JJA and the wet bias of SON. The boreal region had significant decreases in the wet biases of MAM and SON that were offset by the new wet bias of JJA, resulting in an overall increase in the wet bias annually. The Sahara and Arabia region had significant increases in the dry bias of DJF and the wet bias of JJA that were partially offset by the decrease in the wet bias of SON.

While there were mixed changes in regional precipitation with the CLM SiB parameterizations, the global mapping of Fig. 2 shows there were some significant improvements in precipitation simulation in CCSM3 with the CLM SiB parameterizations compared to the current CLM3 parameterization. In DJF there were decreases in the wet biases of northern Australia, southern Africa, and sub-Saharan Africa. The decreases in these African wet biases, however, were offset by increases in wet biases in central Africa. The CLM SiB parameterizations also resulted in reductions in both wet and dry DJF biases over the Amazon, with localized redistribution of precipitation but limited net regional change.

In JJA there were decreases in the dry biases of North America, northern Europe, and Russia with the CLM SiB parameterizations. These decreases in dry biases corresponded with the boreal forests. There were significant increases in JJA precipitation in the central and eastern Amazon; however, these were offset by larger decreases in the northwest. There also were large decreases in JJA precipitation over the Sahara desert, Arabian Peninsula, and Indian subcontinent, which reduced some of the largest wet biases in the CCSM.

Despite these improvements in precipitation, large biases remained with the CLM SiB hydrology for both DJF and JJA, with these biases having the same spatial pattern as found with the original CLM3 hydrology. One interesting note on the differences in precipitation between the two experiments was that the influence of changes in land surface hydrology is not limited to the land surface. The significant changes in precipitation over the oceans indicates there are large-scale circulation changes in the coupled model associated with changes in land surface fluxes of moisture and energy, consistent with studies by Feddema et al. (2005), Chase et al. (1996), Pielke (2001), Niyogi et al. (2002), and others.

c. Differences in air temperature compared to observations in coupled CLM experiments

The changes in surface hydrology with the CLM SiB parameterizations had large influences on global and regional land 2-m air temperature as well. Again, the differences in air temperature between the CLM SiB parameterizations and the CLM3 control experiments were assessed relative to differences with the observed climatology data of Willmott and Matsuura (2000). The results of these analyses are shown in Table 4, with the ensemble mean seasonal differences in air temperature between the experiments and the observed climatology data of Willmott and Matsuura (2000) mapped globally in Fig. 3.

Table 4 shows that the CLM SiB2 parameterizations increased the global average warm bias over land from the CLM3 hydrology for all seasons except JJA. The regional analysis showed that the warmer temperatures were not uniform, with the near-surface air temperature cooler in the Amazon region for all seasons except DJF. For the boreal region, the CLM SiB hydrology increased the warm biases of CLM3 for all seasons and annually. In the Sahara and Arabia region, the CLM SiB parameterization reduced the cool biases in JJA and SON, resulting in a decrease in the cool bias annually.

The global mapping of Fig. 3 shows that although there was an overall increase in global warm biases with the CLM SiB hydrology, changes between the experiments were again more detailed than could be captured in a global or large regional analysis. The differences mapped between the experiments and the observed air temperature climatology of Willmott and Matsuura (2000) show that like precipitation, there were some significant improvements in CCSM3 air temperature with the CLM SiB parameterization.

In DJF the CLM SiB hydrology resulted in warming over Australia, reducing a cool bias in northern Australia but increasing a warm bias in southern Australia. The CLM SiB hydrology also resulted in warming over central and southern Africa, reducing the cool biases in these areas found with the CLM3 hydrology. The CLM SiB hydrology also resulted in warming over eastern Europe, southern Russia, central Asia, China, and North America, which increased existing warm biases with the CLM3 hydrology.

In JJA the CLM SiB hydrology resulted in widespread cooling over South America, and central and southern Africa. The cooling in these areas reduced warm biases found with the CLM3 hydrology. The CLM SiB hydrology also resulted in warming from the Arabian Peninsula through India into Asia, reducing cool biases in these areas. Like precipitation, the overall influence of the CLM SiB hydrology on the near surface was mixed, with large biases remaining with the CLM SiB hydrology for both DJF and JJA, and again these biases have the same spatial pattern as was found with the original CLM3 hydrology.

d. Differences in river discharge compared to observations

Further assessment of the influence of the new CLM SiB parameterizations on surface hydrology was performed to compare annual river discharge from both offline and coupled CCSM experiments against the river discharge data of Dai and Trenberth (2002) for major river basins of the world. The annual river discharge values are shown in Table 5.

In the offline experiments, the new CLM SiB hydrology improved annual river discharge relative to the data of Dai and Trenberth (2002) for seven of the rivers studied. The CLM SiB hydrology resulted in increases in the low flow of the Amazon, Yangtze, Orinoco, Brahmaputra, St. Lawrence, Danube, and Sepik Rivers and with no effective influence on the low flow of the Mississippi River. The CLM SiB hydrology, however, did degrade the river discharge of six of the rivers studied, with decreases in the low flow of the Congo, Mekong, and Amur Rivers and increases in the high flow of the Ganges, Yukon, and Columbia Rivers.

In the coupled experiments, the precipitation biases of CAM had additional influence on surface hydrology, further complicating the river discharge assessment. Despite these feedbacks the CLM SiB hydrology improved the annual river discharge of six of the rivers studied. The reasonable performance of the CLM SiB surface hydrology for river discharge was a necessary requirement for modeling experiments where CLM is coupled to the Parallel Ocean Program (POP) where freshwater is discharged from the river network into the oceans.

4. Discussion

The results show that new CLM SiB parameterizations effectively changed the partition of average global evapotranspiration in CLM3 in both offline and coupled climate simulations from being completely inconsistent with the multimodel average found in GCM land surface models by Dirmeyer et al. (2005) to a partition that is now highly consistent with the average. The changes in evapotranspiration partition included a substantial increase in transpiration (15%–42%), a substantial reduction in canopy evaporation (38%–22%), and a substantial decrease in soil evaporation (47%–36%). The results also show that changes in evapotranspiration have a strong influence on precipitation distribution and near-surface air temperature.

Comparing the changes in evapotranspiration in the CCSM sensitivity experiments with the idealized offline parameterization investigations provides insight into how the changes in the CLM parameterizations influenced each component of evapotranspiration. The increase in transpiration can be traced to the changes in the canopy photosynthesis and transpiration model with the new two-leaf canopy model of Thornton and Zimmermann (2007). The new two-leaf model had higher photosynthetic rates, which combined with the new root zone soil moisture stress function and the root stress compensation model of Li et al. (2001) to enable plants in the new hydrology to more readily access soil moisture for transpiration. The new soil infiltration combined with the new soil hydrological parameters also contributed to the storage of moisture in the plant root zone.

The reduced canopy evaporation can be directly traced to the SiB2 explicit representation of large-scale and convective precipitation intercepting the canopy. This combined with the simpler SiB2 canopy wetness fraction resulted in less precipitation being intercepted by the canopy and slower recycling of the intercepted precipitation back to the atmosphere. The largest change for soil evaporation with the new CLM parameterizations was the addition of the SiB2 soil resistance to the atmospheric resistance to transport ground specific humidity to the atmosphere. This combined with the explicit surface ponding and the new soil hydrological properties that allowed soils to drain into the soil profile greatly reduced soil evaporation.

The changes in surface fluxes resulted in a number of improvements in precipitation and near-surface air temperature, although large biases remain with the new CLM SiB parameterizations. The largest improvements in precipitation biases occurred in areas with large areas of bare soil and in the boreal forests. In Australia the large decrease in DJF wet bias was associated with large decreases in soil evaporation. This also was the case for the DJF wet biases in southern Africa and sub-Saharan Africa. The decreases in JJA dry biases over the boreal forests were associated with increases in transpiration for these forests. The changes in near-surface air temperature reflected the changes in precipitation, with decreases in the DJF cool biases of northern Australia and southern Africa corresponding to the decreases in precipitation in these areas, and the decrease in warm biases over Boreal forests corresponding with the increase in precipitation in these areas.

The underlying motivation for this work was to develop climate modeling framework to perform historical land-cover-change experiments. With the current release version of CLM3, land-cover-change experiments had land surface flux changes that were opposite of those expected. Reflecting the results shown here and the new MODIS parameters experiments of Lawrence and Chase (2007), the deforestation experiments we performed with the existing version of CLM3 had much larger increases in soil evaporation than decreases in transpiration and canopy evaporation. The new CLM parameterizations, however, enable us to perform these experiments with the changes in vegetation having the expected results. The CLM model also does not include irrigation as a component of the surface hydrology. Given the magnitude of the climate changes found with changes in surface hydrology, we expect that adding irrigation to CLM will have large surface changes with similar climate changes as those found with the new hydrology.

5. Conclusions

We have developed new parameterizations for CLM3 that address the low transpiration, and high canopy and soil evaporation found with the current parameterizations in Lawrence and Chase (2007) and other studies. The new parameterizations are derived from SiB2 and other literature sources to consistently reproduce the land surface processes modeled in SiB2. The new parameterizations enable CLM to simulate evapotranspiration partitioning consistently with the multimodel average of other land surface models used in GCMs, as evaluated by Dirmeyer et al. (2005). The changes in surface fluxes resulted in a number of improvements in the simulation of precipitation and near-surface air temperature in CCSM3, although large biases remain. The improvements in evapotranspiration partition, however, do provide a substantially more robust framework for performing land-cover-change experiments in CLM and CCSM than the existing CLM3 parameterizations.

Acknowledgments

The use of the computing time for the model experiments was supplied through a grant from the National Center for Atmospheric Research’s Community Climate System Model (CCSM) Land Working Group, which is sponsored by the National Science Foundation. Funding for this research also was supported by National Science Foundation Grants ATM 0639838, ATM0001476, and ATM0437538.

REFERENCES

  • Baker, I., 2005: SiB 3.0 prognostic variable and equations set. Atmospheric Science Department, Colorado State University Rep. 19 pp.

  • Baker, I., , Denning A. S. , , Hanan N. , , Prihodko L. , , Uliasz M. , , Vidale P-L. , , Davis K. , , and Bakwin P. , 2003: Simulated and observed fluxes of sensible and latent heat and CO2 at the WLEF-TV tower using SiB2.5. Global Change Biol., 9 , 12621277.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beven, K. J., , and Kirkby M. J. , 1979: A physically based variable contributing area model of basin hydrology. Hydrol. Sci. Bull., 24 , 4369.

  • Bonan, G. B., , Levis S. , , Kergoat L. , , and Oleson K. W. , 2002: Landscapes as patches of plant functional types: An integrated concept for climate and ecosystem models. Global Biogeochem. Cycles, 16 , 10211051.

    • Search Google Scholar
    • Export Citation
  • Chase, T. N., , Pielke R. A. , , Kittel T. G. F. , , Nemani R. , , and Running S. W. , 1996: Sensitivity of a general circulation model to global changes in leaf area index. J. Geophys. Res., 101 , 73937408.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Choudhury, B. J., , and DiGirolamo N. E. , 1998: A biophysical process-based estimate of global land surface evaporation using satellite and ancillary data. I: Model description and comparison with observations. J. Hydrol., 205 , 164185.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Choudhury, B. J., , DiGirolamo N. E. , , Susskind J. , , Darnell W. L. , , Gupta S. K. , , and Asrar G. , 1998: A biophysical process-based estimate of global land surface evaporation using satellite and ancillary data. II. Regional and global patterns of seasonal and annual variations. J. Hydrol., 205 , 186204.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Clapp, R. B., , and Hornberger G. M. , 1978: Empirical equations for some soil hydraulic properties. Water Resour. Res., 14 , 601604.

  • Colello, G. D., , Grivet C. , , Sellers P. J. , , and Berry J. A. , 1998: Modeling of energy, water, and CO2 flux in a temperate grassland ecosystem with SiB2: May–October 1987. J. Atmos. Sci., 55 , 11411169.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collatz, G. J., , Ball J. T. , , Grivet C. , , and Berry J. A. , 1991: Physiological and environmental regulation of stomatal conductance, photosynthesis and transpiration: A model that includes a laminar boundary layer. Agric. For. Meteor., 54 , 107136.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dai, A., , and Trenberth K. E. , 2002: Estimates of freshwater discharge from continents: Latitudinal and seasonal variations. J. Hydrometeor., 3 , 660687.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dai, Y., , Dickinson R. E. , , and Wang Y-P. , 2004: A two-big-leaf model for canopy temperature, photosynthesis, and stomatal conductance. J. Climate, 17 , 22812299.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dickinson, R. E., , Oleson K. W. , , Bonan G. B. , , Hoffman F. , , Thornton P. E. , , Vertenstein M. , , Yang Z. L. , , and Zeng X. , 2006: The community land model and its climate statistics as a component of the community climate system model. J. Climate, 19 , 23022324.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dirmeyer, P. A., , Gao X. , , Zha M. , , Guo Z. , , Oki T. , , and Hanasaki N. , 2005: The second Global Soil Wetness Project (GSWP-2): Multi-model analysis and implications for our perception of the land surface. COLA Tech. Rep. 185, 46 pp.

    • Search Google Scholar
    • Export Citation
  • Feddema, J. J., , Oleson K. W. , , Bonan G. B. , , Mearns L. O. , , Buja L. E. , , Meehl G. A. , , and Washington W. M. , 2005: The importance of land-cover change in simulating future climates. Science, 310 , 16741678.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Garratt, J. R., 1992: The Atmospheric Boundary Layer. Cambridge University Press, 316 pp.

  • Lawrence, D. M., , Thornton P. E. , , Oleson K. W. , , and Bonan G. B. , 2007: The partitioning of evapotranspiration into transpiration, soil evaporation, and canopy evaporation in a GCM: Impacts on land–atmosphere interaction. J. Hydrometeor., 8 , 862880.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lawrence, P. J., , and Chase T. N. , 2007: Representing a MODIS Consistent Land Surface in the Community Land Model (CLM 3.0). J. Geophys. Res., 112 , G01023. doi:10.1029/2006JG000168.

    • Search Google Scholar
    • Export Citation
  • Li, K. Y., , De Jong R. , , and Boisvert J. B. , 2001: An exponential root-water-uptake model with water stress compensation. J. Hydrol., 252 , 189204.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Niu, G-Y., , and Yang Z-L. , 2006: Effects of frozen soil on snowmelt runoff and soil water storage at a continental scale. J. Hydrometeor., 7 , 937952.

  • Niu, G-Y., , Yang Z-L. , , Dickinson R. E. , , and Gulden L. E. , 2005: A simple TOPMODEL-based runoff parameterization (SIMTOP) for use in global climate models. J. Geophys. Res., 110 , D21106. doi:10.1029/2005JD006111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Niu, G-Y., , Yang Z-L. , , Dickinson R. E. , , Gulden L. E. , , and Su H. , 2007: Development of a simple ground water model for use in climate models and evaluation with Gravity Recovery and Climate Experiment data. J. Geophys. Res., 112 , D07103. doi:10.1029/2006JD007522.

    • Search Google Scholar
    • Export Citation
  • Niyogi, D. S., , Xue Y. , , and Raman S. , 2002: Hydrological land surface response in a tropical regime and a midlatitudinal regime. J. Hydrometeor., 3 , 3956.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Oleson, K. W., and Coauthors, 2004: Technical description of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-461+STR, 173 pp.

  • Oleson, K. W., and Coauthors, 2008: Improvements to the Community Land Model and their impact on the hydrological cycle. J. Geophys. Res., 113 , G01021. doi:10.1029/2007JG000563.

    • Search Google Scholar
    • Export Citation
  • Philip, J. R., 1957: Evaporation and moisture and heat fields in the soil. J. Meteor., 14 , 354366.

  • Pielke, R. A., 2001: Influence of the spatial distribution of vegetation and soils on the prediction of cumulus convective rainfall. Rev. Geophys., 39 , 151177.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Qian, T., , Dai A. , , Trenberth K. E. , , and Oleson K. W. , 2006: Simulation of global land surface conditions from 1948 to 2004. Part I: Forcing data and evalutation. J. Hydrometeor., 7 , 953975.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., 1985: Canopy reflectance, photosynthesis and transpiration. Int. J. Remote Sens., 6 , 13351372.

  • Sellers, P. J., 1997: Modeling the exchanges of energy, water, and carbon between continents and the atmosphere. Science, 275 , 502509.

  • Sellers, P. J., , Mintz Y. , , Sud Y. C. , , and Dalcher A. , 1986: A Simple Biosphere Model (SiB) for use within general circulation models. J. Atmos. Sci., 43 , 505531.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., , Shuttleworth W. J. , , Dorman J. L. , , Dalcher A. , , and Roberts J. M. , 1989: Calibrating the simple biosphere model for Amazonian tropical forest using field and remote sensing data. Part I: Average calibration with field data. J. Appl. Meteor., 28 , 727759.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., , Los S. O. , , Tucker C. J. , , Justice C. O. , , Dazlich D. A. , , Collatz G. J. , , and Randall D. A. , 1996a: A revised land surface parameterization (SiB2) for atmospheric GCMs. Part II: The generation of global fields of terrestrial biophysical parameters from satellite data. J. Climate, 9 , 706737.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., and Coauthors, 1996b: A revised land surface parameterization (SiB2) for atmospheric GCMs. Part I: Model formulation. J. Climate, 9 , 676705.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sen, O. L., , Shuttleworth W. J. , , and Yang Z. L. , 2000: Comparative evaluation of BATS2, BATS, and SiB2 with Amazon data. J. Hydrometeor., 1 , 135153.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stöckli, R., , and Vidale P. L. , 2005: Modeling diurnal to seasonal water and heat exchanges at European Fluxnet sites. Theor. Appl. Climatol., 80 , 229243.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stöckli, R., and Coauthors, 2008: Use of FLUXNET in the Community Land Model development. J. Geophys. Res., 113 , G01025. doi:10.1029/2007JG000562.

    • Search Google Scholar
    • Export Citation
  • Thornton, P. E., , and Zimmermann N. E. , 2007: An improved canopy integration scheme for a land surface model with prognostic canopy structure. J. Climate, 20 , 39023923.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • U.S. Geological Survey, cited. 1996: GTOPO30 Global Digital Elevation Model (DEM). [Available online at http://edc.usgs.gov/products/elevation/gtopo30/gtopo30.html.].

    • Search Google Scholar
    • Export Citation
  • von Storch, H., , and Zwiers F. W. , 1999: Statistical Analysis in Climate Research. Cambridge University Press, 484 pp.

  • White, M. A., , Thornton P. E. , , Running S. W. , , and Nemani R. , 2000: Parameterization and sensitivity analysis of the BIOME–BGC terrestrial ecosystem model: Net primary production controls. Earth Interactions, 4 .[Available online at http://EarthInteractions.org.].

    • Search Google Scholar
    • Export Citation
  • Willmott, C. J., , and Matsuura K. , 2000: Terrestrial air temperature and precipitation: Monthly and annual climatologies (Version 2.0.1). [Available online at http://climate.geog.udel.edu/~climate/html_pages/download.html.].

    • Search Google Scholar
    • Export Citation
  • Zeng, X., 2001: Global vegetation root distribution for land modeling. J. Hydrometeor., 2 , 525530.

APPENDIX A

CLM3 Soil Hydrological Properties Assessment

The differences in soil hydrological properties between CLM3 and SiB 2 were investigated, as they strongly influence the simulation of soil hydrology. The assessment compared the effective values of saturated volumetric soil moisture (θSAT), soil wetness exponent (B), saturated soil matrix potential (ψSAT), and saturated hydraulic conductivity (KSAT) used in CLM3 to those used in SiB2 across a range of soil types.

In CLM3 the soil hydrological properties are calculated directly from the sand and clay fraction of each soil layer, where as in SiB2 these properties are prescribed directly from a soil class and the experimentally derived values provided by Clapp and Hornberger (1978). The differences in soil properties calculated in CLM3 are compared to the experimentally derived values used in SiB2 in Fig. A1 across the range of representative soil types investigated by Clapp and Hornberger (1978). The differences in soil hydrological properties have a large influence on water holding capacity, infiltration rates, soil moisture drainage and recharge, and water availability for plant roots.

To reduce the differences found between the Clapp and Hornberger (1978) values used in SiB2 and the values used in CLM3, we developed new formulations of the soil hydrological properties for CLM SiB based on the sand and clay texture of the soil [Eqs. (A1)(A4)]. The new formulations were developed to more faithfully reproduce the values derived experimentally by Clapp and Hornberger (1978) while retaining the soil texture parameters used in CLM3. The new formulations are plotted against the CLM3 and Clapp and Hornberger (1978) values in Fig. A1, ordered by sand or clay composition to reflect the original CLM3 formulation dependence of those properties.

The original CLM3 formulation of saturated soil moisture content [Eq. (A1)] has a strong linear dependence on sand content. This relationship is supported by the experimentally derived values of Clapp and Hornberger (1978), with saturated soil moisture content linearly increasing with decreasing sand content (Fig. A1a). The CLM3 values, however, are consistently lower than the experimentally derived values. To reduce the differences between the CLM3 formulation and those used in SiB2, we modified the CLM SiB formulation to have fractionally higher saturated soil moisture content with sand content, as shown in the second part of Eq. (A1). The new formulation had overall better agreement with SiB2 values, with half of the root-mean-square error of the current CLM3 formulation (Fig. A1a).

The old and new CLM formulations of the saturated soil moisture content (θSAT) are
i1525-7541-10-2-374-ea1
i1525-7541-10-2-374-eqa1
The CLM3 soil wetness exponent [Eq. (A2)] has a strong linear dependence on clay content. Again, this relationship is supported by the experimentally derived values, which have the wetness exponent increasing linearly with increasing clay content (Fig. A1b). The CLM3 formulation, however, has a too-steep slope on the exponent-to-clay relationship, with low clay content soils being lower and high clay content soils being higher than the values derived experimentally. The new CLM SiB formulation was modified to reduce the slope of the exponent-to-clay relationship and then offset with a higher y-intercept value, as shown in the second part of Eq. (A2). The new formulation had overall better agreement with SiB2 values, with the reduced root-mean-square error compared to the current CLM3 formulation (Fig. A1b).
The old and new CLM formulations of the soil wetness exponent (B) are
i1525-7541-10-2-374-ea2
i1525-7541-10-2-374-eqa2
The CLM3 formulation of saturated soil matrix potential [Eq. (A3)] was solely dependent on sand content. This relationship is not supported by the experimentally derived values, which has increasing saturated soil matrix potential with decreasing sand, spiking at silt loam before decreasing rapidly for soils with the lowest sand content (Fig. A1c). The CLM3 formulation also has saturated soil matrix potential substantially higher than the experimentally derived values for all but the sand and silt loam soils. To capture the behavior of the experimentally derived saturated soil matrix potential, the CLM SiB formulation was modified to include a clay content term with new weighting terms for both the sand and clay components, as shown in the second part of Eq. (A3). The new CLM SiB formulation captured the nonlinear nature of the experimental values used in SiB2 and reduced the root-mean-square error to less than a half of the current CLM3 formulation (Fig. A1c).
The old and new CLM formulations of saturated soil matrix potential (ψSAT) are
i1525-7541-10-2-374-ea3
i1525-7541-10-2-374-eqa3
The CLM3 formulation of saturated hydraulic conductivity [Eq. (A4)] also was solely dependent on sand content and like saturated soil, matrix potential is not supported by the experimentally derived values (Fig. A1d). The CLM3 formulation has saturated hydraulic conductivity values that are an order of magnitude smaller than the experimentally derived values for sandy soils but are similar for soils with less sand and more clay. To capture the behavior of the experimentally derived saturated hydraulic conductivity values, the CLM SiB formulation was modified to include a clay content term with new weighting terms, as shown in the second part of Eq. (A4). The new CLM SiB formulation greatly increased saturated hydraulic conductivity for soils with high sand content while maintaining low values for soils with higher clay content. The new formulation reduced the root-mean-square error to a quarter of the current CLM3 formulation (Fig. A1d).
The old and new CLM formulations of saturated hydraulic conductivity (KSAT) are
i1525-7541-10-2-374-ea4
i1525-7541-10-2-374-eqa4
In addition to the differences shown in hydraulic conductivity with the CLM3 formulation, CLM3 also has exponentially decreasing hydraulic conductivity with depth following the TOPMODEl parameterization of Beven and Kirkby (1979). SiB2, however, consistently specifies the saturated hydraulic conductivity from the soil type through the entire soil profile. This further increases the differences between CLM3 and SiB2 values, which already have sandy soils an order of magnitude lower with the CLM3 formulation. To address this difference in CLM SiB, we set saturated hydraulic conductivity directly as a function of soil texture, as is done in SiB2.

APPENDIX B

CLM3 Soil Evaporation Assessment

To investigate the high soil evaporation rates in CLM3, we compared the soil evaporation parameterization of CLM3 against those used in SiB2. Both CLM3 and SiB2 use the closed soil pore space relative humidity calculation developed by Philip (1957) to limit soil moisture availability for soil evaporation from the top soil layer. In CLM3 the total resistance to transfer moisture from the soil to the reference height is simply the atmospheric resistance (rdE) calculated from the Monin–Obukhov similarity theory.

In SiB2, however, a soil resistance term (rSOIL) is added to the calculated atmospheric resistance term (rdE) to substantially increase the total resistance (rDE) to transfer soil moisture to the atmosphere. This term is added so that bare soil evaporation rates simulated in SiB2 more closely match observational values collected in the FIFE field campaign. The soil resistance term in SiB2 is calculated as
i1525-7541-10-2-374-eb1
The soil resistance term represents the real physical process of diffusively transferring soil moisture from the closed soil pore space to the soil surface roughness height, prior to it being available for atmospheric exchange to the reference height.

To assess the influence of the soil resistance term on soil evaporation, we performed idealized simulations of bare soil evaporation for various soils types over a range of soil moisture, wind, air temperature, and ground temperature values. The results of these simulations are shown in Fig. B1 for a range of soil types for average DJF Amazon conditions. The plots show that soil evaporation with soil resistance is substantially lower than the CLM3 formulation for all soil types. The influence of the soil resistance term also is similar for all soil types, with the underlying soil parameters influencing the range of soil evaporation terms but the resistance term acting to reduce the evaporation curve with soil wetness. To make the soil evaporation formulation in CLM SiB consistent with SiB2, we modified the CLM3 parameterization to add soil resistance to the atmospheric resistance for the soil evaporation component of the ground to the canopy air space or atmospheric reference height moisture flux.

APPENDIX C

Infiltration, Runoff, and Drainage Assessment

The differences in the idealized soil evaporation shown in Fig. B1 do not include precipitation and snowmelt that are ponded explicitly at the soil surface in SiB2. The water ponded at the surface evaporates at the potential evaporation rate, reducing some of the large differences shown between the model schemes. In SiB2 the contribution from pond evaporation to total ground evaporation is set by the pond wetness fraction. Sellers et al. (1996b) defines this as the simple ratio of the pond depth relative to a maximum pond depth of 0.2 mm. To make CLM SiB consistent with SiB2, we added the explicit surface ponding and allowed ponded water to evaporate at the potential rate over the pond wetness fraction of the ground, as calculated in SiB2.

Differences in soil infiltration and runoff parameterizations between CLM3 and SiB2 were further investigated to assess how these processes influenced soil moisture availability for transpiration and soil evaporation. In CLM3 runoff follows the TOPMODEL parameterization, with infiltration set as the residual of the precipitation and snowmelt once surface runoff and evaporation have been removed. The surface runoff is calculated based on the combined Dunne runoff from the saturated fraction of the grid cell based on a prognostic water table, and Hortonian runoff based on infiltration excess due to mean soil wetness of the top three soil layers. In CLM3, surface ponding is implicitly represented as an additional 10 mm of soil moisture storage in the top soil layer over saturation. The water table depth is calculated based on soil wetness weighted by soil layer thickness over the complete soil profile relative to the total soil depth.

SiB2 uses a significantly simpler soil representation of precipitation and snowmelt infiltration and runoff. Surface runoff is calculated only as Hortonian runoff based on the infiltration rate of the top soil layer, with water in excess of infiltration explicitly ponded to a depth of 0.2 mm and the remainder returned to the river system through surface runoff. In SiB2 Sellers et al. (1996b) specifies the infiltration rate of the top soil layer as the hydraulic conductivity of the top soil layer (KINF). This specification results in very low infiltration rates and, therefore, very high Hortonian runoff rates for dry soils. In the most recent Colorado State University (CSU) implementation of SiB 3.0, the infiltration rate of the top soil layer is set as the residual of the CLM3 infiltration excess calculated from the mean soil wetness of the top three soil layers (Baker 2005).

To account for the differences in precipitation intensity due to the spatial distribution of different type of rainfall events, SiB2 divides the runoff behavior of precipitation into convective and large-scale components with different representations for both. Large-scale precipitation is uniformly distributed across the grid cell and added to the surface ponding evenly. Convective precipitation, however, is nonuniformly distributed, with a fraction of the surface ponding represented as experiencing intense convective precipitation with the remainder free from convective precipitation. The fraction of the surface experiencing the intense convective precipitation in SiB2 is more likely to have ponding rates that exceed the remaining available surface ponding depth compared with the uniform treatment of large-scale precipitation.

To make CLM SiB consistent with SiB2, we modified the surface runoff and infiltration schemes to have the same convective and large-scale treatment of precipitation as SiB2, with explicit surface ponding to 0.2 mm. We also removed the saturated runoff fraction calculated with the TOPMODEL prognostic water table currently used in CLM3. The infiltration rate from surface ponding was set using the residual of the CLM3 infiltration excess, as was done in SiB3. The new surface runoff scheme also explicitly included the large-scale and convective precipitation fractions represented in SiB2, following the methods and constants detailed in Sellers et al. (1996b).

Further investigation found that there also were differences in the parameterizations of soil drainage between CLM3 and SiB2 that influenced soil moisture availability for soil evaporation and transpiration. In CLM3 drainage follows the TOPMODEL parameterization, with lateral drainage from the saturated and unsaturated fractions of the third and second bottom soil layers and gravitational flows from the bottom soil layer. The lateral drainage terms are determined by the prognostic water table depth and the soil wetness of the two layers. The gravitation drainage term is determined from soil conductivity of the bottom soil layer, and the rate of change of the soil hydraulic conductivity with respect to soil moisture multiplied by the change in soil moisture of the bottom soil layer.

The SiB2 parameterization for soil moisture drainage is much simpler than CLM3, with all drainage simulated through gravitational drainage formulated from the soil hydraulic conductivity of the bottom soil layer scaled by the sine of the average topographic slope of the grid cell. However SiB2 with no lateral drainage terms, Sellers et al. (1996b) includes a basin-scale drainage component to allow for drainage in flat terrain from the bottom soil layer.

To make CLM SiB consistent with SiB2 for deep soil drainage, we removed the lateral drainage components from soil layers 8 and 9 and scaled the gravitational drainage component from the bottom soil layer by the sine of the average topographic slope of the CLM grid cell. The topographic slope was calculated from the GTOPO30 1-km global digital elevation model (U.S. Geological Survey 1996). We performed CCSM model investigations with the basin-scale drainage of Sellers et al. (1996b) and found this resulted in excessive drainage with dry soil over much of the planet. As an alternative solution to capturing the large-scale base flow in areas of flat terrain, we added an additional 2 m of topographic fall for each 1-km DEM grid cell in the topographic slope calculation. This value was essentially a tuning parameter but can be rationalized to represent the depth of river channels not resolved by the DEM. The soil profile depth also was extended to 12 soil layers to increase moisture storage following field studies with SiB3 undertaken by Baker (2005).

APPENDIX D

CLM3 Photosynthesis and Transpiration Assessment

To investigate the low transpiration in CLM3, we investigated the photosynthesis and transpiration parameterizations in CLM3 against those used in SiB2. In both models photosynthesis and transpiration parameterizations are derived from Collatz et al. (1991), where the relationship between stomatal conductance and photosynthesis is based on the rate of carbon dioxide assimilation (AN) relative to the ambient concentration of carbon dioxide (CS) and relative humidity (hS) at the surface of the leaf. The stomatal conductance (gS)–resistance (rS) expression used in both models is written as
i1525-7541-10-2-374-ed1
With gS in μmol m−2 s−1, p is atmospheric pressure in Pa, and m and b are experimentally derived vegetation specific coefficients.

In both models transpiration is highly dependent on the canopy resistance term, which is itself an integrated component of stomatal resistance over the canopy leaf area. In this regard Thornton and Zimmermann (2007) have recently found that CLM3 has an unorthodox implementation of the big-leaf model of photosynthesis as well as large systematic biases in the calculation of absorbed photosynthetically active radiation (PAR) through the canopy. In their research they found these two large biases in the CLM3 photosynthesis and transpiration components resulted in unusual model behavior compared to the results of previous canopy integration studies.

To remove this behavior, Thornton and Zimmermann (2007) have developed an alternative two-leaf canopy model based on the work of Dai et al. (2004) for use in CLM. The new scheme maintains the same basic CLM3 transpiration model; however, the sunlit and shaded photosynthesis and stomatal resistance terms are explicitly calculated based on PAR absorption in the canopy. In addition the behavior of carbon dioxide assimilation (AN) in the new scheme is specified as a dynamic function dependent on leaf concentrations of Rubisco and enzyme activity. The new scheme can be used with dynamically calculated canopy leaf distributions and leaf concentrations of Rubisco, or prescriptively with monthly LAI values and literature values for PFT leaf concentrations of Rubisco and nitrogen limitation supplied as parameters.

Thornton and Zimmermann (2007) found that compared to CLM3, the new scheme greatly increased the absorption of PAR, which has corresponding large increases in photosynthesis and transpiration. In their study they found that the new scheme helped to address the low photosynthetic and transpiration rates found in CLM3 in regions that are not highly moisture limited. They do, however, suggest that additional modifications are required to soil hydrology to improve photosynthetic rates in drought prone environments. To capture the improvements in CLM photosynthesis and transpiration, we included the new scheme of Thornton and Zimmermann (2007) in the new CLM SiB parameterization with the new PFT nitrogen limitations specified by Oleson et al. (2008).

APPENDIX E

CLM3 Soil Moisture Root Stress Assessment

In both CLM3 and SiB2 models, the catalytic capacity of Rubisco (Vm), and therefore transpiration and photosynthesis, are both strongly limited by soil water availability through root zone soil moisture stress functions. To assess how the differences in the soil moisture stress contribute to the low transpiration in CLM3 experiments, we performed further investigation into this component of both models. As part of this investigation, we also compared these two models to the root zone moisture stress function derived from White et al. (2000) and used in CLM3.5.

In CLM3 the soil moisture stress function describes the response of the plant to the difference of the root zone soil matrix potential (ψRi) relative to a wilting point soil matrix potential (ψMAX) and the saturated soil matrix potential (ψSATi). The soil moisture stress function of CLM3 is calculated for each soil layer independently and then combined as a root fraction weighted sum for all soil layers. The CLM3 individual soil layer root stress parameterization is calculated as
i1525-7541-10-2-374-ee1
where ψMAX = −1.5 × 105 mm for all PFTs.
The soil moisture stress function of SiB2 uses an inverse exponential function that describes the response of the plant to differences in the average soil matrix potential in the root zone relative to a half inhibition water soil matrix potential value (ψCRIT). The SiB2 parameterization of soil moisture stress is calculated as
i1525-7541-10-2-374-ee2
where ψCRIT = −2.0 × 105 mm from Sellers et al. (1996a).

The soil moisture stress function used in CLM3.5 is very similar to the CLM3 but with the generic wilting point soil matrix potential (ψMAX) and the saturated soil matrix potential (ψSATi) values replaced with PFT specific stomatal closure soil matrix potential (ψCLOSE) and full stomatal open soil matrix potential (ψOPEN) values. The PFT values for the closed and open soil matrix potential values were derived from the work of White et al. (2000). The differences in the CLM3, SiB2, and CLM 3.5 soil moisture stress functions were assessed through idealized simulations of each parameterization over a range of soil moisture values, soil classes, and PFTs, with representative results shown in Fig. E1.

The plots show that for all PFTs and all soil types, the SiB2 and CLM3.5 parameterizations behave almost identically, with both models having soil moisture stress functions that have complete stomatal closure (value of 0) and complete stomatal opening (value of 1) at much lower soil wetness values than CLM3. The early shutdown in photosynthesis and transpiration in the CLM3 parameterization has major influences on the transpiration fluxes in CLM3 compared to the SiB2 and CLM 3.5 models. To reduce the transpiration limitations related to soil moisture root stress and to make CLM SiB consistent with SiB2 and CLM3.5, we replaced the soil moisture stress function with the one used in CLM3.5.

The availability of soil moisture to plant roots in SiB2 also is influenced by the way the model represents the root zone with a single soil layer. This is functionally very different to CLM3, as the fixed weighted average of soil moisture stress from soil layers that a plant has roots in is very different from the average soil moisture properties of the root zone soil layer calculated in SiB2. There are other differences in the soil root zone representation between the models as well, with SiB2 having a thin surface layer above the root zone to simulate soil surface processes and a deep soil layer that stores soil water below the root zone.

In CLM3, the total soil moisture stress is calculated for each PFT using fixed root density allocation (ri) and the individual soil layer stress values fR(Wi) from each soil layer. The CLM3 total plant soil moisture stress is calculated as
i1525-7541-10-2-374-ee3
where the root fraction distribution for soil layers range from 0 to 1, following Zeng (2001).

In respect to the fixed soil layer rooting scheme used in CLM3, Li et al. (2001) provides numerous studies that show that such schemes have substantial difficulties in representing real world plant water stresses, where plants compensate water stress in one part of the root zone by enhancing water uptake from other more moist areas of the root zone. To address these limitation, Li et al. (2001) developed an exponential root water uptake model that includes water stress compensation based on observational and modeling studies. The new scheme dynamically allows large proportions of transpiration to be met from deeper, more sparsely rooted soil layers when upper layers are depleted of soil moisture. Compared to observed plant transpiration studies, the new model substantially improves the simulation of water stress in areas where soil water in the upper soil layers is exhausted during the growing season, relative to fixed soil layer contribution schemes.

The new water stress compensation model redistributes soil layer weightings for plant root water stress and water extraction based on the root density profile and the water stress in each soil layer. The variable weightings of the soil layers are calculated with a weighted stress index (σi) that dynamically ranges from 0 to 1 for each soil layer. For n soil layers, each soil layer weighted stress index is formulated as
i1525-7541-10-2-374-ee4
where λ = 0.5 from field observation (Li et al. 2001).
Although the water stress compensation scheme is not a component of the SiB2 model, it does provide a mechanism to allow CLM SiB to simulate soil moisture availability in the root zone in a conceptually similar manner to SiB2. This is done so that the soil moisture stress is calculated over the whole root zone rather than with the fixed allocation scheme in CLM3. To incorporate the root zone compensation scheme in CLM, the fixed root fraction (ri) soil layer weighting values used in Eq. (E3) are replaced with the weighted stress index (σi) values, dynamically calculated for each soil layer based on the root fraction and the soil wetness stress of each PFT at each time step. The total water stress through the root zone is calculated as
i1525-7541-10-2-374-ee5

APPENDIX F

CLM Canopy Interception and Evaporation Assessment

To investigate the large contribution to evapotranspiration from canopy evaporation in CLM3, we compared the interception and evaporation parameterizations of the CLM3 with SiB2 to identify major differences and similarities in the schemes. CLM3 and SiB2 both model the canopy interception of precipitation and dew using an exponential extinction formulation, with the precipitation conceptually represented as vertical radiation falling through the canopy without reflection following Beer’s law. Like surface runoff, SiB2 divides precipitation into nonuniformly distributed convective, which is concentrated into a fraction of the grid cell and large-scale precipitation that is uniformly distributed across the grid cell.

The fraction of the canopy experiencing the intense convective precipitation in SiB2 is more likely to have interception rates that exceed the remaining canopy moisture store compared to the grid cell average interception rate of CLM3. The parameterization in SiB2, therefore, results in substantially higher amounts of canopy-intercepted water being added to the canopy through fall, reducing the total grid cell canopy interception for the convective precipitation component. The resulting lower canopy moisture store thereby reduces the moisture available for canopy evaporation in SiB2 compared to CLM3. In both cases the maximum canopy store is set to 0.1-mm leaf water depth, with the total canopy store calculated as the leaf water depth multiplied by the leaf and stem area.

The rate of canopy evaporation from the canopy water store is also treated differently in CLM3 than in SiB2. In CLM3 the fraction of the canopy that is covered in water and evaporates at the potential rate is calculated with a 2/3 power function of the fraction of the maximum canopy store. In SiB2, however, the canopy evaporation is represented with a simple fraction of the canopy store to the maximum canopy store multiplied by the potential rate. The 2/3 power function of CLM3 results in a much higher fraction of the canopy evaporating at the potential rate than the simple ratio of SiB2. The largest differences between the schemes occur for small fractions of the canopy water storage where CLM3 can be more than twice the SiB2 value.

To remove the differences in canopy interception and evaporation between CLM3 and SiB2, we implemented a new canopy interception scheme in CLM SiB that explicitly included large-scale and convective precipitation fractions, following the methods and constants detailed in Sellers et al. (1996b). We also changed the fraction of the canopy evaporating at the potential evaporation rate to the simple ratio used in Sellers et al. (1996b).

Fig. 1.
Fig. 1.

Comparison of hydrology and sensible heat in the (a) CLM3 and (b) SiB2 land surface parameterizations. Temperature (T) and specific humidity (q) are specified with subscripts for the atmospheric reference height (ATM), the canopy air space (S), at the leaf surface (V), and the ground (G). Total resistance to transfer sensible heat or moisture between the atmospheric reference height, the surface of leaves, and the ground, with the canopy air space are specified as rA, rB, and rD, respectively.

Citation: Journal of Hydrometeorology 10, 2; 10.1175/2008JHM987.1

Fig. 2.
Fig. 2.

Ensemble mean differences in average seasonal precipitation between Willmott and Matsuura (2000) (a),(b) 1970–99 climatology and current CLM3 hydrology experiment, and with (c),(d) new CLM SiB hydrology experiment, and (e),(f) differences between experiments. Differences with statistical significance >95% are stippled.

Citation: Journal of Hydrometeorology 10, 2; 10.1175/2008JHM987.1

Fig. 3.
Fig. 3.

Same as Fig. 2, but for ensemble mean differences in average seasonal near-surface air temperature.

Citation: Journal of Hydrometeorology 10, 2; 10.1175/2008JHM987.1

i1525-7541-10-2-374-fa01

Fig. A1. Differences between CLM3, CLM SiB, and SiB2 soil hydrology parameters for the range of representative soil classes described by Clapp and Hornberger (1978), ordered by (a),(c),(d) sand and (b) clay content: (a) saturated moisture content, (b) wetness exponent, (c) saturated matrix potential, and (d) saturated hydraulic conductivity.

Citation: Journal of Hydrometeorology 10, 2; 10.1175/2008JHM987.1

i1525-7541-10-2-374-fb01

Fig. B1. Idealized comparison of SiB2 and CLM3 bare soil evaporation for (a) sand; (b) loam; (c) clay loam; and (d) clay soils under typical DJF Amazon surface and atmospheric conditions.

Citation: Journal of Hydrometeorology 10, 2; 10.1175/2008JHM987.1

i1525-7541-10-2-374-fe01

Fig. E1. Differences between CLM3, SiB2, and CLM3.5 soil moisture root stress functions for (a) shrub PFTs in sand; (b) grass PFTs in loam; (c) needleleaf-tree PFTs in sandy clay loam; and (d) broadleaf evergreen tree PFTs in sandy clay.

Citation: Journal of Hydrometeorology 10, 2; 10.1175/2008JHM987.1

Table 1.

Global annual average surface hydrology comparison from offline experiments. Incremental changes from CLM3 hydrology to CLM SiB hydrology (CSiB), compared to the multiple model average of Dirmeyer et al. (2005) and CLM3.5 hydrology from Oleson et al. (2008). All values are in mm day−1. Percentage contribution to total ET is shown in parenthesis. Incremental changes refer to adding new CSiB components described in appendices AF. Arrows indicate direction of flux change with parameterization.

Table 1.
Table 2.

Global annual average surface hydrology from coupled experiments: CLM3 hydrology, CSiB, and the multiple model average of Dirmeyer et al. (2005). All values are in mm day−1. Percentage contribution to total ET is shown in parenthesis.

Table 2.
Table 3.

Average global and regional land precipitation (mm day−1) for CLM3 hydrology (CLM), and with the CSiB in coupled CCSM experiments. Differences between the experiments and 1970–99 observed climatological values from Willmott and Matsuura (2000) are shown in parenthesis, with “—” signifying no difference. Statistical significance in differences between current CLM and CSiB hydrology experiments are shown as n = neither test has significance ≥0.95; t = Student’s t test has significance ≥0.95; w = Wilcoxon signed rank test has significance ≥0.95; and b = both tests have significance ≥0.95.

Table 3.
Table 4.

Same as Table 3 but for average global and regional land 2-m air temperature (°C).

Table 4.
Table 5.

Annual average river discharge volumes and percentage of river discharge database of Dai and Trenberth (2002; D&T) for offline CLM3 hydrology; offline CLM SiB hydrology; coupled CLM3 hydrology; and coupled CLM SiB hydrology. All values are in 103 m3 sec−1.

Table 5.
Save