The role of land surface dynamics in glacial inception: a study with the UVic Earth System Model

Abstract

The first results of the UVic Earth System Model coupled to a land surface scheme and a dynamic global vegetation model are presented in this study. In the first part the present day climate simulation is discussed and compared to observations. We then compare a simulation of an ice age inception (forced with 116 ka BP orbital parameters and an atmospheric CO₂ concentration of 240 ppm) with a preindustrial run (present day orbital parameters, atmospheric [CO₂] = 280 ppm). Emphasis is placed on the vegetation’s response to the combined changes in solar radiation and atmospheric CO₂ level. A southward shift of the northern treeline as well as a global decrease in vegetation carbon is observed in the ice age inception run. In tropical regions, up to 88% of broadleaf trees are replaced by shrubs and C₄ grasses. These changes in vegetation cover have a remarkable effect on the global climate: land related feedbacks double the atmospheric cooling during the ice age inception as well as the reduction of the meridional overturning in the North Atlantic. The introduction of vegetation related feedbacks also increases the surface area with perennial snow significantly.

Appendices

Appendix 1: Land Surface Scheme

The land surface scheme consists of a simplified version of ‘MOSES’ (Met Office Surface Exchange Scheme) described in Cox et al. (1999a). The scheme used here defines the state of the land surface in terms of lying snow, skin temperature, soil temperature and moisture content, using a single soil layer. It recognizes the five TRIFFID vegetation types as well as bare soil (see Table 3 for values of constant parameters).

Table 3. Constant parameters of the land surface scheme

1.1 Soil moisture and snow mass

The soil moisture is updated according to:

$$ \frac{\partial M} {\partial t} = P_R + S_M - E_M - Y $$

(1)

where M is the soil moisture in the top 1 m of soil, P _R the precipitation rate (rain), S _M the snow melt, E _M the evaporation rate and Y the continental runoff. The evaporation rate E _M is made up of transpiration by vegetation E _v and bare soil evaporation E _b (E _M = E _v + E _b ). Transpiration by vegetation E _v is calculated separately for each plant functional type (PFT) i $(E_v = \sum {E_{v,i}})$ and occurs from the vegetated fraction $\nu (\nu = {\sum {\nu _{i} } })$ :

$$E_{v,i} = \nu _i \frac{\rho } {r_a + r_{c,i}} (q_{sat} (T^\ast_i) - q_1) $$

(2)

where ρ is the surface air density, r _a the aerodynamic resistance (depending on roughness length and windspeed), r _c,i the canopy resistance of PFT i, q _sat (T* _i ) the saturated specific humidity at the skin temperature T* _i and q ₁ the atmospheric specific humidity (Cox et al. 1999a). The surface air density ρ is calculated with the constant surface pressure P* and the surface air temperature T ₁ (ρ = P*/(RT ₁)). The canopy resistance r _c is the inverse of the canopy conductance, g _c , which is calculated within the canopy conductance and primary productivity module described in Cox et al. (1999a). Bare soil evaporation E _b is defined as follows:

$$E_{b} = (1 - \nu )\frac{\rho }{{r_{a} + r_{{ss}} }}(q_{{sat}} (T^{*}_{b} ) - q_{1} )$$

(3)

where T* _b is the bare soil skin temperature and r _ss is the soil surface resistance scaling inversely with the soil moisture availability factor μ(Θ):

$$r_{{ss}} = \min {\left( {10^{6} ,\frac{{100}}{\mu }} \right)}{\text{sm}}^{{ - 1}} $$

(4)

$$ \mu (\Theta ) = \left\{ \begin{array}{*{20}l} {1: \hfill} & {\Theta > \Theta _{c} \hfill} \\ \frac{\Theta - \Theta _w} {\Theta _c - \Theta _w: \hfill} & {\Theta _w < \Theta \leqslant \Theta _c \hfill} \\ {0: \hfill} & {\Theta \leqslant \Theta _w \hfill} \\ \end{array} \right. $$

(5)

Θ _w is the volumetric soil moisture concentration below which stomata close, Θ _c is the volumetric soil moisture concentration above which stomata are not sensitive to soil water and Θ the volumetric soil moisture. Finally, the runoff Y is given by:

$$Y = K_{s} {\left( {\frac{M}{{M_{{SAT}} }}} \right)}^{{(2b + 3)}} $$

(6)

where K _S is the saturated hydraulic conductivity, M _SAT is the saturated soil moisture and b is the Clapp-Hornberger exponent. River drainage basins are described in Weaver et al. (2001). The snow mass is updated according to:

$$\frac{{\partial S}}{{\partial t}} = P_{S} - S_{M} - E_{S} $$

(7)

where S is the snow mass, P _S the rate of precipitation (snow), and E _S the rate of sublimation. The evaporative demand E is met by any lying snow first (E = E _S if there is lying snow in the grid cell, E = E _M otherwise).

1.2 Energy balance

The energy balance for each PFT and for bare soil (i indicates a PFT or bare soil) consists of the net radiation $R_{{N_{i} }}$, the latent heat due to evaporation or sublimation LE _i , the sensible heat H _i and the soil heat flux G _i :

$$R_{{N_{i} }} = LE_{i} + H_{i} + G_{i} $$

(8)

The net radiation is defined as follows:

$$R_{{N_{i} }} = R^{{down}}_{{SW}} (1 - \alpha _{{s,i}} ) + R^{{down}}_{{LW}} - \sigma T^{\ast 4}_{i} $$

(9)

where R _SW ^down is the incoming short wave radiation at the surface, α _s,i is the surface albedo of PFT i, R _LW ^down is the downward atmospheric longwave radiation and σ is the Stefan Boltzmann constant. The latent heat due to evaporation or sublimation LE _i can be written as:

$$LE_{i} = L_{c} E_{{M,i}} + (L_{c} + L_{f} )E_{S} $$

(10)

where L _c and L _f are the latent heat of condensation and fusion, respectively. The sensible heat is a function of the skin temperature T* _i and the atmospheric temperature T ₁:

$$H = \frac{{\rho c_{p} (T^{*}_{i} - T_{1} )}}{{r_{a} }}$$

(11)

where c _p is the specific heat capacity of the air. Finally, the the soil heat flux G _i is defined as follows:

$$G = \frac{{2\lambda }}{{\Delta z_{1} }}(T^{*}_{i} - T_{S} )$$

(12)

where λ is the soil conductivity, Δz ₁ the thickness of the soil layer and T _S is the soil temperature. The soil temperature T _S is calculated as a function of the ground heat flux and the energy required to melt snow:

$$\frac{{\partial T_{S} }}{{\partial t}} = \frac{G}{{C_{S} \Delta z_{1} }} - L_{f} S_{M} $$

(13)

where C _S is the soil heat capacity. The surface energy balance Eq. (8) is solved using equations 9 to 12 to eliminate the skin surface temperature, which yields an extended Penman-Monteith equation for each surface type (Essery et al. (2003)).

Appendix 2: IGBP data set

In Sect. 3, the simulated vegetation cover is compared to the IGBP data set (Loveland and Belward 1997). The IGBP biomes have been converted to PFTs by Mike Dunderdale (see Dunderdale et al. 1999 for a detailed description on how this conversion has been carried out). As the photosynthetic pathways (C₃ versus C₄) are not distinguished in the IGBP dataset, the ratio has been estimated based on the ratio of “long grass” to “short grass” (Dunderdale et al. 1999; Wilson and Henderson-Sellers 1985).

To calculate the correlation between the areal coverage of simulated vegetation and IGBP data, the correlation for each PFT in each grid cell has been calculated as follows:

$$ {\text{corr}}({\text{PFT}}_{{\text{i}}} ) = 1 - {\left| {{\text{areal coverage}}({\text{PFT}}_{{\text{i}}} ,{\text{IGBP}}) - {\text{areal coverage}}({\text{PFT}}_{{\text{i}}} ,{\text{model}})} \right|} $$

(14)

The total correlation represented in Fig. 2 is the minimum of each correlation in each grid cell:

$${\text{total}}\_{\text{corr}} = \min [{\text{corr}}({\text{PFT}}_{{\text{i}}} )]_{{{\text{i}} = 1,5}} $$

(15)