An Evaluation of Soil Processes in Eastern Canada Using Daycent, DNDC and STICS

Supplementary Materials for,

An evaluation of soil processes in eastern Canada using DayCent, DNDC and STICS

Authors: G. Guest[1], R. Kröbel[2], B. Grant3, W. Smith4, J. Sansoulet5, E. Pattey6, R. Desjardins7, G. Jégo8, N. Tremblay9, G. Tremblay10

S1. Methods Supporting Information

S.1.1 DayCent

The DayCent model (Parton et al. 1998) was derived from the CENTURY model (Parton et al. 1994), and simulates changes in soil organic matter (SOM), plant productivity, nutrient availability, and N gas emissions at a daily time step (Del Grosso et al. 2008b). Accordingly, submodels for calculating plant growth, decomposition, soil water and temperature dynamics, as well as N gas fluxes, are included in the model (Del Grosso et al. 2008b).

S.1.1.1 Soil water

The DayCent model combines a cascade modeling approach with a Darcy’s law function to simulate the soil moisture profile, and combines it with estimates of runoff, snow dynamics, and the effect of soil freezing on saturated flow (Del Grosso et al. 2001) (Table 1). Rainfall interception (canopy) and evaporation (from canopy and surface litter) are calculated. The Penman equation is used to estimate PET (Parton et al. 1998). Under freezing temperatures, all precipitation is treated as snowfall and accumulates in a snowpack, subject to sublimation and melting (Del Grosso et al. 2001). Non-intercepted rainfall, irrigation and melting snow infiltrate the soil surface after surface runoff is subtracted (Parton et al. 1998).

Water infiltrating into the soil induces a 4-hr period of saturated flow (unidirectional downward flow), where water only moves on to the next layer once the prior layer is filled (Del Grosso et al. 2001) – an approach commonly used in cascade models. Remaining surplus water (e.g., above field capacity) is continuously drained to lower layers after the 4-hr period is finished. In any other case, unsaturated water flow is calculated with Darcy’s law using hydraulic conductivity and hydraulic potential of the soil layers (Del Grosso et al. 2001) – a basic principle of the Richard’s equation approach used in more complex water models to simulate soil hydrology. Actual evaporation is partitioned amongst two individual processes. The evaporation of water intercepted by plant foliage is lost at the rate of PET followed by a separate algorithm derived from Hillel (1977) for water evaporated from the soil profile. Actual transpiration is calculated next and is determined as a function of PET, leaf biomass, and soil water potential (wettest layer in the top 30 cm or the weighted average of the rooting zone) and the root biomass (Parton et al. 1998).

S.1.1.2 Soil Nitrogen

In DayCent, N is supplied to the mineral nutrient pools through N mineralization, N fertilization, N deposition, and N2 fixation. While NO3- concentrations are calculated for all layers in the soil profile, the calculations for NH4+ are limited to the top 15 cm (Parton et al., 2001). Nitrate is subject to runoff and leaching through the profile as a function of sand content and saturated flow. Soil mineral N can also be immobilized by microbial populations via decomposing plant residues with high C:N ratios. Nutrient pools are subject to nitrification and denitrification processes that lead to N gas losses (Parton et al., 2001). N gas losses from nitrification are proportional to the nitrification rates that are controlled by the NH4+ concentration, water content, temperature, pH, and texture. Denitrification rates (and N gas loss due to denitrification) are determined by the availability of labile C, soil NO3- and O2 (Del Grosso et al., 2001). A gas loss ratio determines the contribution to either N2 or N2O, which is regulated by the ratio of NO3- to the labile C. Nitrogen gas losses are calculated on a daily time step for each simulated soil layer, and an N balance calculation ensures that N gas losses cannot be higher than the availability of NO3- and NH4+ (Del Grosso et al., 2001). Soil organic carbon inputs come from dead plant material (subdivided into fast and slow turn-over pools), and their decomposition re-supplies the NH4+ pools (and thus plant growth and microbial processes) and transfers carbon from fast to slow turn-over pools (Del Grosso et al., 2008b; Parton et al., 2001).

S.1.2 DNDC

The DNDC model (Li et al. 1992a, 1992b, 1994) is a process-based model with focus on estimating greenhouse gas emissions from agricultural production systems. For this purpose, environmental driving factors (e.g. climate and management) are combined with system properties (e.g. soil properties) to estimate soil temperature and moisture, soil nitrogen concentrations, and soil carbon compounds in the soil profile in addition to the simulation of crop biomass accumulation.

S.1.2.1 Soil water

In the submodel for thermal-hydraulic flow, soil moisture is calculated using a cascade modelling approach in a layered water budget on an hourly time step. The simulated soil is subdivided into a series of horizontal layers. Water inputs (precipitation, and irrigation) are added at midnight at constant intensity, and therefore are of variable duration (Li et al. 1992a). At the beginning of each hour, water inputs saturate the soil layer by layer, with field capacity being the upper limit of water holding capacity that determines the amount of water infiltrating to the next layer at the following time step (Li et al. 1992a). Surplus water in the lowest layer is leached from the soil profile. If air temperatures drop below 0°C, precipitation is assumed as snow and accumulates in a snow pack. Water will pond on the soil surface as long as the top layer of the soil is frozen, and therefore be susceptible to runoff if not present as snow. Water losses from the soil occur by evapotranspiration or drainage. Potential evapotranspiration is calculated by an adapted Thornthwaite equation (Li et al. 1992a), which includes a multiplier for wind speed. The calculated T is based on crop water requirement (g water g-1 biomass), thus depends on the simulated crop biomass, and crop water is taken up from the root profile. Potential evaporation is estimated as the difference between PET and T. Actual evaporation depends on available soil moisture in the upper 15 cm of the soil, as well as on the presence of surface litter.

S.1.2.2 Soil Nitrogen

Similar to DayCent, N pools are supplied by N fertilization, N deposition, N mineralization, and N2 fixation. Ammonia losses, hydrolysis and/or nitrification are calculated on a daily time step (Li et al., 1994). Nitrate travels through the profile in solution, and is leached from the deepest soil layer. Nitrifier-bacteria growth and activity are dependent on temperature, soil moisture and NH4+ availability, and N2O emissions are a function of temperature and the quantity of N being nitrified (Li et al., 1992a, 2000). The mass of denitrifier-bacteria is estimated using a multi-nutrient dependent (Michaelis-Menten) growth function (Li et al., 1992a), dependent on temperature, moisture, soil redox potential (Eh) and soil pH (Li et al., 1994, 2000). Denitrification is calculated as a stepwise transformation process, depending on microbial activity and soil pH, so that the amount of the N2O emitted is the difference between the N2O formed in the soil and the N2O transformed further into N2.

The submodels for nitrification and denitrification are regulated through the “anaerobic balloon”, where nitrification occurs in the aerobic outside and denitrification in the anaerobic inside. Depending on the soil redox potential (Eh), the size of the balloon increases or shrinks and thus determines the daily contribution to N-turnover by the two submodels. The Nernst equation is used to estimate the soil redox potential, by charging the concentration of all oxidising species (e.g., O2, NO3-, SO42-, Fe3+, Mn5+, CO2) against the product of all reducing species (e.g., dissolved organic carbon, H2S, Fe2+, Mn3+, H2, etc.) in the liquid phase (Li et al., 2000). Nitrogen is released from decomposing carbon pools (first-order kinetics) in proportion to the C/N ratio of the different carbon pool, of which there are: plant residues (or litter), microbial biomass, humads (or active humus) and passive humus (Li et al., 1992a; Li et al., 1994). The decomposed N is either immobilized into another SOC pool along with the carbon (depending on the C/N ratio), or it is released into the nutrient pools as NH4+ (Li et al., 1994), from where it re-enters the N-cycle.

S.1.3 STICS

The STICS model is a dynamic soil-crop model that calculates crop biomass accumulation, as well as water, nitrogen and carbon budgets on a daily time step, using climate and soil properties as input (Brisson et al. 1998). Crop parameters are predefined and cultivar parameters need to be regionally adjusted by calibration and verification prior to its use, as was done for eastern Canada by Jégo et al. (2010).

S.1.3.1 Soil water

STICS has several submodels that describe plant growth and development (phenology, shoot growth, and yield formation), soil-crop interaction (root growth, water balance, nitrogen balance, and soil water transfer using the Cascade model approach), as well as submodels for crop management and microclimate in the canopy (Brisson et al. 2008). The soil is subdivided in five user-defined soil horizons (Brisson et al. 1998). The STICS model can be used on successive crop sequences without re-initialization every year. The current version of STICS accounts for snowfall and soil freezing through a separate model, called the snowMAUS model (Trnka et al. 2010), that alters the climate data file to account for snow formation. As the coupled STICS-snowMAUS module was not tested for Canadian conditions, particularly for overwintering, the STICS model was employed in this study to simulate individual snow-free seasons only. Nitrogen and water in the soil profiles was re-initialized at the beginning of each growing season for a given location.

Soil water input from precipitation is reduced by water intercepted by the canopy. Similarly to DNDC, STICS also employs a cascade modelling approach. Water flow down the profile is calculated as water fills successively 1-cm layers, assuming that the upper limit of each reservoir corresponds to the soil layer’s field capacity defined in the soil horizons (Brisson et al. 1998). Run-off is not accounted for in this study. In this study, the resistance approach was used to calculate the crop water requirements (Brisson et al. 2008). The saturation deficit within the vegetation is calculated using the Shuttleworth and Wallace daily time-step model (Brisson et al. 1998), wherein each component (soil, plant, mulch) is defined with a resistance parameter. The fraction of potentially active radiation is used to evaluate the distribution of the available energy between the soil and the plant. Actual transpiration is a function of the available water in the root zone. Actual evaporation is equal to PE until a threshold (defined through soil properties) is reached and then is reduced depending on the weather (mainly wind speed), soil moisture at field capacity and soil clay content. The contribution of each 1-cm soil layer to E decreases with depth.

S.1.3.2 Soil Nitrogen

Nitrogen enters the simulated system as either synthetic fertilizer or as deposition from precipitation and irrigation. An efficiency factor is assigned to each fertilizer type, the non-efficient part contributing to 66% to humus build up and to 34% to N gas losses through volatilization and denitrification. Nitrate transport in the soil profile is a function of N available in the layer and the rate of water transport in the profile (Brisson et al., 1998). Nitrification occurs in layers as a function of biological activity, and is dependent on NH4+ availability, soil temperature, soil moisture, and soil pH; while denitrification is estimated as a constant potential rate proportional to the NO3- content and limited by temperature, and soil moisture status (Brisson et al., 2008). Nitrous oxide emissions from nitrification and denitrification occur as a constant fraction of the NH4+ being nitrified and the NO3- being denitrified, respectively (Brisson et al., 2008). The mineralization of nitrogen from soil organic matter is calculated for biologically active layers (e.g., ploughing depth), and the decomposition rate of the “active” carbon pool is calculated using first-order kinetics, while the “stable” remains inert (Brisson et al., 1998, 2008).

S2. Results Supporting Information

S.2. Nitrogen

S.2.3. Crop N Uptake

Liao et al. (2004) indicated that wheat N uptake is increased with increased rooting depth and linear relationships between root density and N uptake have been found (Kristensen and Thorup-Kristensen 2004). However, nitrogen-deficient plants will direct their root growth towards areas of available N (Drew and Saker 1975; Brady et al. 1995). This behaviour has also been found with respect to soil water uptake (Hamblin and Tennant 1987). Each of the investigated models has a rather simplified root depth estimation (and no estimate of root density), so that abovementioned processes were not properly accounted for in the model simulations. This is, however, partially due to the lack of appropriate datasets needed to develop and test more complex root growth models.

S.2.4. Gaseous N Loss

Ruser et al. (2001) found that N2O emissions are closely related to the mean soil nitrate content in the Ap horizon, a finding that was also reported by Zebarth et al. (2008a). Such a correlation was found in the DayCent simulations, however, in the DNDC simulations correlation analysis showed a linkage between the surface ammonium content and N2O emissions instead, thus rather suggesting nitrification as a major contributor. Zebarth et al. (2008a) reported that they were not able to tie the timing of N2O emissions to fertilization events, a finding that was repeated by Zebarth et al. (2008b). Burton et al. (2008) found steadily low N2O emissions interrupted by a few higher emission events, which are correlated to spring thaw, fertilization and/ precipitation. This was more clearly represented in the simulations of DNDC, where steady small N2O emissions contributed roughly as much to the total as the precipitation driven peaks, than it was in DayCent, where extended N2O peaks tended to dominate in fall when both moisture and nitrogen availability coincided (not shown).