Quoted from: https://www.mdpi.com/2073-4433/8/8/141/htm
Despite their common origin, there are major differences between the CoLM and the initial version of the CLM. For example, CoLM includes a one-layer, two-big-leaf submodel for photosynthesis, stomatal conductance, leaf temperature, and energy flux, and treats the sunlit and shaded parts of canopy separately [
29]. The sunlit fraction of the canopy is
where kb is the direct beam extinction coefficient and LAI the leaf area index of the canopy. This division is important, as sunlit leaves will receive much more intense light fluxes than shaded leaves under sunny conditions. In this two-big-leaf model, shaded leaves receive diffuse solar radiation only, while sunlit leaves receive both direct and diffuse solar radiation. This type of submodel is adopted in both CLM4.5 and two CoLMs.
Two versions of the CoLM have been released to date: CoLM2005 and CoLM2014. CoLM2014 is radically different from CoLM2005, particularly with respect to global land surface data, pedotransfer functions for soil hydraulic and thermal parameters, the numerical solution of the Richards equation for soil water content, and the groundwater model [
30]. The CaMa-Flood river model [
31] is coupled to CoLM2014, enabling simulations of the global distribution of river discharge.
CoLM2005 derives seven soil hydraulic and thermal parameters (porosity, specific heat capacity soil solids, saturated hydraulic conductivity, thermal conductivity for saturated and dry soil, saturated matrix potential and exponent B defined in Clapp and Hornberger [
32]) from soil sand and clay contents. The FAO (Food and Agriculture Organization) and STATSGO (State Soil Geographic) soil classification data are merged together to provide soil texture data for two soil layers (0–30 cm and 30–100 cm). In CoLM2014, calculation of soil parameters has been improved in three aspects. First, the new BNU global soil texture dataset [
33] is used to provide soil properties at eight levels between the surface and 2.3 m depth. Second, the impact of soil organic matter on soil hydraulic and thermal properties is considered. Third, all seven soil hydraulic and thermal parameters have been optimized. Different results have been generated for each parameter (e.g., 18 results for porosity) using different algorithms based on soil texture data derived from the BNU data set. The median of these results has then been used to define the optimal parameter value.
The ground water model also differs between the two versions of CoLM. In CoLM2005, subsurface runoff is given by
where KD is the saturated soil hydraulic conductivity for the lowermost soil layers contributing to the base flow (KD = 4 × 10−2 mm s−1 in CoLM2005), Ib is a base flow parameter for the saturated fraction of the watershed (Ib = 1 × 10−5 mm s−1 in CoLM2005), is soil wetness weighted by soil layer thickness and hydraulic conductivity in the lowermost five soil layers, and B is the exponent defined by Clapp and Hornberger [
32]. The parameter zw is the mean water table depth calculated as
where fz is a water table depth scale parameter, zbot is the bottom depth of the lowest soil layer, sj is are the soil wetness in soil layer j, and Δzj is the thickness of soil layer j. The liquid soil water content in every soil layer is checked at every time step. If liquid soil water content in any layer exceeds the maximum effective liquid soil water content (porosity minus the volume of ice in the soil layer), then the excess liquid water is added to subsurface runoff (Rexcess in Equation (2)). In CoLM2014, the water table depth changes due to aquifer recharge rate, calculated as
where qo is the flux of water out of the lowest soil layer and Δw is the change in the volume of liquid water contained in the lowest soil layer. A positive value of q corresponds to a rise in the water table. The subsurface runoff is calculated according to the equation
where Im is the ice impedance factor calculated as
The parameters fice and Δzsum represent the total ice content and soil thickness for all soil layers that overlap with the water table, down to the lowest soil layer considered by the model. Two corrections are added to R in CoLM2014. The first of these corrections is equivalent to Rexcess in CoLM2005. The second is intended to limit irreducible wrapping of liquid water in each soil layer (0.01 kg m−2 in CoLM2014). If a soil layer contains insufficient liquid water, this deficiency is compensated by drawing from aquifer water.