The hydrological module (HM) [Zhuang et al., 2002] has been revised to be appropriate for both upland and wetland soils. The revisions include improvements in the simulation of infiltration (IF), evapotranspiration of the vegetation canopy (EV), soil surface evaporation (ES), snowmelt (Smelt), and sublimation (SS) from the snowpack. In addition, soil moisture dynamics are represented in greater detail for upland soils, and algorithms, based on work by Granberg et al. , have been added to simulate water content and water table depth in wetland soils.
D1. Infiltration From the Soil Surface Into the Soil (IF)
 The liquid water from rain throughfall or snowmelt either infiltrates into the soil column or is lost as surface runoff. In the work of Zhuang et al. , all liquid water reaching the soil surface has been assumed to infiltrate into the soil column. In this study, we add algorithms to estimate surface runoff and subtract this estimate from rain throughfall and snowmelt to estimate infiltration (IF). Following Bonan , surface runoff is calculated using the Dunne runoff if the soil surface is saturated or the Horton runoff if the soil surface is not saturated. In the Dunne approach, all the water inputs at the surface (i.e., rain throughfall and snowmelt) are lost as runoff because the soil is already saturated. In the Horton approach, runoff occurs even when the soil is not saturated, but the total water inputs at the surface are greater than the infiltration capacity, which depends on the water content of the surface soil layer relative to the saturated water content of this layer.
D2. Evapotranspiration From the Vegetation Canopy (EV)
 In Zhuang et al. , we simulated evapotranspiration by simulating transpiration and evaporation from the canopy with separate algorithms. In the updated HM, we have replaced these algorithms with those of McNaughton and Jarvis , which are based on the Penman-Monteith approach. Evapotranspiration from the vegetation canopy (EV) is estimated based on short-wave solar radiation absorbed by the vegetation canopy, air temperature, vapor pressure deficit, and canopy conductance. The amount of solar radiation absorbed by the canopy is determined using the incident short-wave solar radiation occurring at the top of the canopy and the leaf area index (LAI) of the vegetation [Zhuang et al., 2002].
 Following Rosenberg et al. , vapor pressure deficit is modeled as
where VP is vapor pressure (kPa) from input data sets. Eadt is saturation vapor pressure (kPa),
where TA is air temperature (°C).
 A simplified equation of Waring and Running  has been adopted to model the canopy water conductance (G),
where gmax is the maximum canopy conductance (mm s−1); f(TA) is a multiplier that describes the effect of air temperature (TA) on the canopy conductance; f(VPD) is a multiplier that describes the effect of the vapor pressure deficit (VPD in mbar) on canopy conductance; and f(ψ) is a multiplier that describes the effect of leaf water potential (lwp in MPa) on canopy conductance. We set gmax to be 3.5, 13.5, and 21.2 mm s−1 for alpine tundra, wet tundra, and boreal forests, respectively. The effects of air temperature on canopy conductance are calculated following Thornton ,
where η is a constant (0.125 °C−1). The effects of vapor pressure deficit on canopy conductance are calculated as
where VPDclose is the vapor pressure deficit at complete conductance reduction and VPDopen is the vapor pressure deficit at the start of canopy conductance reduction. We assume VPDclose is 41.0 mbar and VPDopen is 9.3 mbar for all vegetation types.
 The effects of leaf water potential (lwp) on canopy conductance are calculated in a similar manner,
where ψclose is the leaf water potential at complete conductance reduction and ψopen is the leaf water potential at the start of conductance reduction. We assume that ψclose is −2.3 MPa and ψopen is −0.6 MPa for all vegetation types. As in the work of Zhuang et al. , lwp is calculated as
where WS is mean daily soil water content (millimeters) integrated across the soil profile from the upper boundary to the lower boundary, and SOILCAP is a parameter for soil water capacity (millimeters) of the soils, which is set to 235 [see Zhuang et al., 2002].
D3. Evaporation From the Soil Surface (ES)
 The evaporation rate from the soil surface is modeled using the Penman approach [Zhuang et al., 2002], which uses air temperature, vapor pressure deficit, short-wave solar radiation at the soil surface, and the throughfall of rain from the overlying vegetation canopy. In the work of Zhuang et al. , a mean daily rate of potential evaporation is estimated for a month and a monthly rate is determined by multiplying this mean daily rate by the number of days per month (MD). In this study, we use the daily potential evaporation (PES) estimates directly (i.e., MD = 1.0) when calculating daily evaporation from the soil surface (ES). If the daily throughfall of rain (RTH) is greater than or equal to PES, then ES is assumed to equal PES; otherwise, ES is equal to RTH.
D4. Snowmelt (Smelt) and Snow Sublimation (SS)
 The rate of snowmelt has been modeled by Zhuang et al. , using monthly shortwave solar radiation, throughfall of snow from the overlying vegetation canopy, snow albedo, and the number of days per month. The potential snowmelt rate (PSmelt in millimeters) now uses a daily time step, which depends on daily air temperature and solar radiation [Brubaker et al., 1996; Edward Rastetter, personal communication, 2002],
where mq is a constant (2.99 kg MJ−1), Rn is the incident short-wave solar radiation to the snowpack (J cm−2 d−1), AR is a constant (2.0 mm °C−1 d−1), and TA is the daily air temperature (°C). If the daily throughfall of snow is greater than PSmelt, then Smelt is equal to PSmelt; otherwise Smelt is equal to the daily throughfall of snow.
 The rate of snow sublimation has also been modeled by Zhuang et al.  based on monthly short-wave solar radiation and throughfall of snow from the overlying vegetation canopy. In the work of Zhuang et al. , a mean potential sublimation rate is determined and multiplied by the number of days per month (MD) to obtain a monthly rate. In this study, we use the potential daily sublimation rate (PSS) directly (i.e., MD = 1.0) based on daily shortwave solar radiation. If the PSS is greater than water equivalent of the snowpack, then SS is assumed to equal the water equivalent of the snowpack; otherwise SS is assumed to equal PSS.
D5. Upland Soils
 In the work of Zhuang et al. , the soil profile has been represented with three soil layers: a moss or litter layer, an organic soil layer, and a mineral soil layer. Changes to the water content of the whole soil profile (WS in millimeters) have depended on infiltration (IF), evapotranspiration from the vegetation canopy (EV), evaporation from the soil surface (ES), and drainage from the deep mineral layer (DR),
Within each soil layer, changes in water content have been determined using a water balance approach similar to that described in equation (D9). The terms IF and DR are replaced by percolation into and out of a soil layer, respectively, and ES is assumed to occur only from the top moss or litter layer. Only the organic soil and mineral soil layers are assumed to contribute to EV, and this flux has been partitioned between the two layers based upon the relative soil water content of the two layers. Soil moisture has been assumed to be uniformly distributed within each of the three soil layers.
 To improve our simulation of water dynamics in upland soils in high-latitude ecosystems, we now represent the soil profile with six layers with different hydrologic characteristics: a 10-cm-thick moss or litter layer, a 20-cm-thick upper organic soil layer, a 40-cm-thick lower organic soil layer, an 80-cm-thick upper mineral soil layer, a 160-cm-thick lower mineral soil layer, and a 320-cm-thick deep mineral soil layer. We assume that all upland soils have the same soil profile structure for our soil water dynamics due to a lack of spatially explicit data sets for each grid cell. Changes to the water content of the entire soil profile are still influenced by the factors given in equation (D9). However, soil moistures within each of the six layers are now assumed to vary as described by the Richards equation [Hillel, 1980; Celia et al., 1990],
where WC is the volumetric water content (mm3 mm−3); k is the hydraulic conductivity (mm s−1); and ψs is the soil matrix potential (millimeters), which varies as a function of WC and soil texture [Clapp and Hornberger, 1978]. The soil water content (WC) for the midpoint of each of the different layers of the unsaturated soils is obtained simultaneously through solving equations (D9) and (D10) numerically with a tridiagonal system of equations [see Press et al., 1990]. Infiltration (IF) into the first soil layer sets the upper boundary condition for the numerical solution and the drainage (DR) of the deep mineral soil layer, which is equal to the hydraulic conductivity of this layer, sets the lower boundary condition. Although the Richards equation could be used to estimate soil moistures at each 1 cm depth in the profile, large amounts of computation time would be required to extrapolate this approach across the Pan-Arctic. Instead, we use the soil moisture contents at the midpoints of each of the six soil layers to interpolate soil moistures at each 1 cm depth across the soil profile.
D6. Wetland Soils
 Because Zhuang et al.  only considered water dynamics in unsaturated soils, new algorithms needed to be developed to estimate the proportion of the soil profile that becomes saturated, the depth of the resulting water table, and the influence of the water table on soil moisture in the unsaturated portion of the soil profile. We assume that wetland soils are always saturated below 30 cm, which represents the maximum water table depth (zb). Thus changes in water content (WS) of the top 30 cm of the soil profile can be calculated with a water balance model that considers the water input and outputs at the daily time step,
where IF is infiltration, EV is evapotranspiration of the vegetation canopy, ES is evaporation from the soil surface, and QDR is the saturated flow drainage below zb. Calculation of the IF, EV, and ES terms for wetlands use the same algorithms that have been described in the previous sections of Appendix D. Similar to Walter et al. [2001a], QDR is calculated as
where QDRMAX is the maximum drainage rate of 20 mm d−1; and f(coarse) is the relative volume of coarse pores in the soil. The calculation of f(coarse) is described in equation (C3).
 Instead of the six layers used to simulate upland soils, we assume that water dynamics in wetland soils can be represented by two functional layers or “zones”: an upper oxygenated, unsaturated zone; and a lower anoxic, saturated zone. The water table represents the boundary between these two zones, and its depth is allowed to change over time with changes in soil moisture. The maximum thickness of the upper unsaturated layer is represented by the maximum water table depth (zb), which is assumed to be 30 cm [Frolking et al., 1996; Granberg et al., 1999]. The minimum thickness of the lower saturated layer is the difference between the depth of the lower boundary (LB) and 30 cm. The total volume of water in the top 30 cm of the soil profile (VTOT in centimeters) is represented by
where ϕ is the soil porosity, WT is the actual water table depth (centimeters), and θus(z) is the volumetric water content in the unsaturated zone at depth z. We assume ϕ is equal to 0.9 cm3 cm−3 [Frolking and Crill, 1994] for the entire soil profile. If WS is greater than zb × ϕ, the water table will be above the soil surface and the height of water above the soil surface is determined by the difference of WS and zb × ϕ. Otherwise, VTOT is equal to WS. After setting VTOT to equal WS, equation (D13) can be integrated and inverted to solve for the water table depth (WT) following Granberg et al. ,
where az is the gradient in soil moisture resulting from evaporation at the soil surface and is calculated as the ratio of ϕ-θs,min to zθs,min; θs,min is the minimum volumetric water content at the soil surface; and zθs,min is the maximum depth where evaporation influences soil moisture. We assume θs,min is 0.25 and zθs,min is 10 cm for all wetland soils. A negative value of the water table depth indicates that the water table is above the soil surface, whereas a positive value indicates that the water table is below the soil surface.
 After determining the water table depth, the volumetric water content at each 1 cm depth can then be estimated. If depth z is in the saturated zone, the volumetric water content is assumed to be equal to ϕ. If depth z is in the unsaturated zone, the volumetric water content (θus(z)) is estimated following Granberg et al. ,
where θs is the volumetric water content at the soil surface and is calculated as