Appendix 6
Critical analysis of the Pasim model for cut and grazed grasslands with particular regard to ammonia exchange.
C. Campbell, M.R. Theobald and M.A. Sutton
Introduction
Very few models have been developed that focus on the NH3 emissions from grazed and cut grassland. From national inventories, it is estimated that the contribution of grazing emissions to total emissions is less than that from livestock housing and land-application of manures and slurries (11% compared with 33% and 33% respectively [Misselbrook et al., 2000]). However, the same inventory estimates that storage losses account for only 5% of the total emissions yet there are several process-based models to simulate them. In addition, it is necessary to be able to model these emissions well, since both emission and deposition occur over grasslands and, due to the large area of grasslands across the country, they have a significant effect on the UK national ammonia budget.
The emission processes are complex because the system contains animal, plant and soil components. Urine and manure excreted by grazing livestock is deposited onto leaf and soil surfaces and the TAN component can either be volatilised or will infiltrate into the soil matrix where it subsequently can be taken-up by the plant roots in the form of NH4+. Once inside the plant, there is potential for the NH4+ to be emitted as NH3 through the stomata. Furthermore, depending on the N status of the plant, atmospheric NH3 can be deposited to the leaf cuticle or taken into the plant through the stomata.
NH3 volatilisation from grazed systems has been modelled by soil-plant process models such as DN-DC (Li, 2000) and CENTURY (Parton et al., 1987). However the NH3 components of these models is quite basic and mainly focus on chemical equilibria within the soils and the resulting volatilisation. Although DNDC does model recapture of NH3 by the canopy (see Appendix 7), neither of the two models simulate the bidirectional exchange of NH3 between the plant and the atmosphere. These models were originally developed to simulate the carbon and nitrogen dynamics of these systems and are therefore more complex than necessary for simulating only ammonia losses. However, by coupling NH3 fluxes to a full treatment of the C-N cycles in grassland, such ecosystem models are uniquely able to show how wider interactions in the grassland system will affect NH3 fluxes.
The only carbon-nitrogen simulation process model that has been modified specifically to model the bidirectional flows of NH3/NH4+ is Pasim (Riedo et al. 2002), which was developed from the Hurley Pasture Model (Thornley, 1998). To allow treatment of ammonia bidirectional exchange, Pasim was modified by coupling it with a resistance model for gaseous NH3 exchange (Sutton et al., 1995; Nemitz et al., 2001). This allows the simulation of the NH3 fluxes shown in Figure 1. Although the model remains a complex way of simulating NH3 losses from grazed systems, this reflects the complexity of the system and is necessary to maintain the process-based approach. The fact that it does include the resistance model for NH3 exchange means that it is a useful tool for investigating NH3 dynamics. It is this model that has been chosen for the critical analysis in this project. It is worth mentioning that there do exist empirical models of NH3 volatilisation from grazed systems. One example is the nitrogen mass balance model NGauge (Brown et al., 2005). NGauge model avoids the complexities of modelling the processes explicitly and instead uses empirical relationships for the N flows.
Figure 1: Schematic representation of the bi-directional ammonia exchange module incorporated into the PaSim model. This canopy-compensation point (c) model formulation derives from the schemes of Sutton et al. (1995) and Nemitz et al. (2001). PaSim generates hourly process based estimates of s and l based on the interaction of plant and soil C-N turnover processes.
Critical analysis
Criteria /PaSim
a)What is the model for? /- Developed for managed productive pastures, incorporating management practices of cutting, fertiliser application and grazing.
- Simulates plant growth dynamics, biomass production, carbon, nitrogen, water and energy fluxes.
- Links two-layer bidirectional ammonia exchange model with a model of grassland dynamics. Replaces empirical estimates of apoplastic and soil surface NH4+ concentrations (e.g. Nemitz et al. 2001) by coupling these concentrations to soil and plant N dynamics.
- The key point to allow simulation of foliar NH3 fluxes is that plant N is separated into structural N, symplastic substrate N and apoplastic substrate N. The apoplastic substrate N is linked to ammonia exchange. Soil surface exchange of ammonia is linked to soil surface ammonium concentration, with a multi-layer soil model. The stomatal compensation point for ammonia exchange is linked to plant internal N dynamics via the concentration of NH4+ in the apoplast.
- Apoplastic substrate N is a function of root N uptake, biological N fixation, passive transport of NH3 from symplast, active transport of N to the symplast, exchange with the atmosphere via stomatal and loss through grazing and cutting.
- Nitrogen balance components modelled are soil ammonium, soil nitrate, nitrate leaching, ammonia exchange, nitrogen immobilisation/mineralisation, nitrogen uptake by roots, fractional N in roots and shoot, nitrification, denitrification and biological N fixation. Ammonium and nitrate input in fertiliser and atmospheric NH3 concentration are driving variables, atmospheric NO3- deposition and NHx deposition other than NH3 are site-specific model parameters.
- Carbon balance components modelled are canopy photosynthesis, plant respiration (associated with N uptake, maintenance and growth), microbial respiraton, animal respiration, methane loss, C in excreta and milk, C in roots and shoot, mineralisation in the soil and root exudation.
- Energy fluxes modelled are sensible, latent and soil heat fluxes.
b)Common processes / Not relevant, as examining only one model.
c)Parameters that need measuring or estimating / Input driving variables needed on an hourly timestep are:
- Global radiation (W m-2)
- Precipitation rate (mm d-1)
- Air temperature (K)
- Water vapour pressure (kPa)
- Wind speed (m s-1)
- Atmospheric CO2 concentration (ppm)
- Atmospheric NH3 concentration (g NH3 m-3)
- Cutting - cutting dates (day of year)
- Fertilisation
- Day of year of each fertiliser application
- NH4 in each mineral fertiliser application
- NO3 in each mineral fertiliser application
- N in applied liquid manure
- N in applied slurry
- N in applied manure
- Grazing
- Stocking density in livestock units
- Start of grazing (day of year)
- Duration of grazing
Latitude (rad)
Time of the heighest position of the sun (h.min)
Slope (rad)
Aspect (rad)
Height above sea level (m)
Micrometeorological reference height above soil surface (m)
NH3 reference height above soil surface (m)
Number of soil layers
Depth of soil layers (mm)
Depth of lower soil boundary layer (mm)
Maximal canopy height (m)
Canopy height parameter
Clover fraction (kg/kg)
Relative root dry matter in different soil layers (-)
Main rooting depth (m)
Bulk density (kg/l)
Volume fraction of quartz in soil (m3/m3)
Clay fraction of texture (-)
Silt fraction of texture (-)
Saturated soil water content (m3/m3)
Saturated soil water content of lower soil boundary layer (m3/m3)
Air entry potential (mm)
Air entry potential of lower soil boundary layer (mm)
Parameter b in ks(Psi) (-)
Parameter b of lower soil boundary layer (-)
Saturated hydraulic conductivity (mm/d)
Saturated hydraulic conductivity of lower soil boundary layer (mm/d)
Parameter for determining field capacity (iO)
Soil pH
Parameter a for soil NH4+ partitioning
Parameter b for NH4+ partitioning
Shoot dry matter after cutting (kg/m2)
LAI after cutting (m2 leaf/m2)
Capillary rise from lower soil boundary layer possible or not
Water content of lower soil boundary layer in spring (m3/m3)
Water content of lower soil boundary layer in automn (m3/m3)
Average temperature of lower soil boundary layer (K)
Amplitude of temperature of lower soil boundary layer (K)
Phase of temperature of lower soil boudary layer (-)
NH4+ deposition other than gaseous NH3 (kg N m-2 d-1)
NO3- deposition (kg N m-2 d-1)
Potential eating rate of lactating cows (kg/(GVE*m2))
Weight of lactating cows (kg)
Type of animal
Initial Conditions (can be generated by model “spin up”):
plant C substrate concentration (kg C/kg)
plant N substrate concentration (kg N/kg)
N concentration of structural plant dry matter (kg N/kg)
C in structural dead plant material (kg C/m2)
C in metabolic dead plant material (kg C/m2)
C in active soil organic matter (kg C/m2)
C in slow soil organic matter (kg C/m2)
C in passive soil organic matter (kg C/m2)
N in metabolic dead plant material (kg C/m2)
N in active soil organic matter (kg C/m2)
N in slow soil organic matter (kg C/m2)
N in passive soil organic matter (kg C/m2)
soil ammonium (kg N/m2)
soil nitrate (kg N/m2)
soil water content (m3/m3)
root dry matter (kg/m2)
LAI (m2 leaf/m2)
shoot dry matter (kg/m2)
d)Limits in application and extrapolation / Previous range of application
- Used for continuous 24 year simulations covering the full annual cycle.
- Variation in environmental drivers:
- doubled CO2
- air temperature increased by 2 and 4C
- 20% increase in precipitation amount and 10% increase in probability of precipitation.
- Can deal with snowfall, and surface coverage by snow
- Accounts for management by cutting or grazing.
- PaSim cannot be applied to rotational grassland.
- Tested for up to 270 kg N ha-1 yr-1 fertiliser application
- Pasim is coded for grazing by lactating cows such that other animals must be scaled by their relative effect compared to lactating cows.
- Grazing animals do not grow in the model, so no C or N is retained by animals.
This suggests that if Pasim was used to provide simple relationships between changes in management or climate and ammonia exchange, relationships would have to be derived for many combinations of climate, management and site-specifc parameters.
e)Strengths of parts or whole / Pasim reproduces qualititative changes in NH3 exchange after cutting and fertilisation. The ammonia submodel uniquely simulates bi-directional exchange of ammonia between the canopy, soil and atmosphere based on controlling mechanisms. Soil NH3 emission is now based on the NHx pool at the soil surface, and NHx in the soil is partitioned between soil layers. The stomatal compensation point for NH3 emission/deposition is modelled as directly related to the plant nitrogen status, as simulated by the ecosystem grassland model. The coupling of these models allows the effects of changes in climate and management to be applied to the ammonia flux model, and allows the ammonia exchange to reflect the nitrogen dynamics of the canopy and soil.
f)Weaknesses of parts or whole (e.g. the multiple linear regression equation developed by Zhang et al., 1994 can suggest very misleading relationships between air speed and temperature that were otherwise mechanistically sound) / An increase in soil NH4+ after cutting has been detected in measurements but not in Pasim simulations, although PaSim did reproduce the observed increase in foliar ammonium levels (Loubet et al. 2002).
Pasim does not distinguish between the form of nitrogen taken up by the roots. In reality NH3 emissions increase more when NH4+ is absorbed relative to when NO3- is absorbed (Sutton et al. 2001).
NHx released from decomposing leaf litter on the soil surface is not accounted for.
NH3 emission is also sensitive to NH3 permeability of the plasmalemma, which was fitted with the assumption that the rate of passive diffusion of NH3 through the plasmalemma was an order of magnitude smaller than that of active transport.
The model is sensitive to initial N concentration of structural dry matter, and to N fertilisation. It is also sensitive to the initial conditions of C/N ratio of plant residue and soil organic matter, plus amount of C in the litter and soil. The metabolic turnover of C between pools is important, but is modelled as a collection of scaling factors.
g)How mechanistic and empirical it is / Leaf cuticular resistance (determining NH3 deposition to the cuticles) is an empirical function of relative humidity, based on measurementderived parameters. This is an essential simplification as a fully process based approach (Sutton et al. 1998; Flechard et al. 1999) requires a major increase in model complexity and run-time.
The stomatal compensation point is modelled mechanistically, based on the apoplast pH and ammonium concentration. The ammonium concentration is modelled as dependent on the apoplastic N concentration and water content. The form of the relationship between ammonium concentration in the apoplast and apoplast N uses six parameters. These fixed parameters include the ratio of fresh to dry weight of the whole plant, and the fraction of leaf water in the apoplast. Apoplast N is modelled incorporating fluxes of N within the plant due to root uptake, biological fixation, passive diffusion from the symplast, stomatal exchange, substrate loss through grazing and loss by active transport from the apoplast to symplast.
The soil compensation point is modelled as a function of soil pH, and the ammonium concentration in the soil surface solution. This is a fraction of the total ammoniacal N of the soil surface, and is influenced by dissociation constants plus the saturated and actual soil water content of the surface layer.
Decomposition, denitrification and nitrification rates are parameterised by calibrated parameters. Fractions of grazing C converted to methane, fraction of N in urine not volatilised, fraction of grazing C not respired and fraction of excreta N volatilised are all parameters from literature.
h)Ability to deal with abatement processes / See separate table
i)Potential for use in a geo-spatial National Ammonia Inventory (e.g. NARSES) / PaSim can be used to look at the effect of different management routines, different climates and different soil types on the ammonia emission flux. Using the model in this way has the potential to give simple relationships that could be applied to the current estimates to give more information on the spatial and temporal variability across the UK.
j)Potential to improve the current estimates in the National Ammonia Inventory / Providing information on the spatial and temporal variability of ammonia emissions from grassland has the potential to improve the current estimates significantly.
k)Potential for use in national and international ammonia emission and deposition models / The simple relationships of the effect of climate and soil types could be included in the emission process code of emission and dispersion models to give a more accurate spatial and temporal variation of emission fluxes.
l)Level of detail required in input data / Input data for PaSim is on two levels:
1)Field specific data (Soil parameters, Initial C/N status of vegetation, water content/pH of soil)
2)Scenario specific data (meteorology, ammonia/CO2 air concentrations, fertiliser, manure, number animals etc.)
The field specific data is very detailed which makes it difficult to apply the model generally unless this information is known. However once the data for one field has been obtained it may be possible to run it for other similar fields with minimal changes. The scenario-specific data is readily obtained which makes it a straightforward process to run PaSim for several different scenarios. The sensitivity of the modelled emission fluxes to the field specific data is yet to be carried out to find out how accurately these data need to be estimated.
m)Availability of suitable input data in terms of variable type together with spatial and temporal resolution / 1Field specific data
-Soil parameters (bulk density, quartz/clay/silt fraction, water content, air entry potential, hydraulic conductivity, pH). Many of these parameters can be estimated from soil maps.
-Initial C/N status of vegetation. These data are very field/history dependent and therefore would not be easy to estimate from spatial/temporal data. For this reason such initial values are often generated by model ‘spin up’ from e.g. 20 year simulations.
-Soil boundary layer characteristics (water content, temperature). These data could potentially be estimated using soil maps/climate maps.
2Scenario specific data
-Meteorology – these data are available at an hourly temporal resolution and a spatial resolution of approx. 20-30 km.
-The management data could be estimated from fertiliser usage statistics or could be taken from fertiliser usage recommendations (RB209). Stocking density could be estimated from parish census data. Information on grazing periods would be difficult to find as a spatial dataset.
-Air concentration measurements can be taken from national monitoring networks although the spatial resolution would not be sufficient to represent variability at a landscape scale.
n)Ability to use readily available geo-spatial data (e.g. 30 year met. data means, soil textures, vegetation classes on 5km2 grids) / See (m) above
o)Inclusion of supplementary environmental information (e.g. calculation of nitrate leaching) / Since PaSim is a Carbon/Nitrogen/Energy balance simulation, almost any aspect of the system can be studied. These include emissions of methane and N2O and the leaching of nitrates, as well as carbon turnover and CO2 exchange processes.
p)Usefulness in future experimental work / PaSim has and important role in interpreting experimental work when the main focus is on the carbon/nitrogen balances of a grazed field. A disadvantage of the model is the large amount of supplementary information that is required for the model to function effectively. However, this reflects the complexity of the processes treated, and the model is unique in being able to treat the dynamics of ammonia compensation points, cuticular uptake and soil surface emissions in relation to grassland C-N dynamics.
References
Brown, L, Scholefield, D, Jewkes, EC, Lockyer, DR, del Prado, A. (2005) NGAUGE: A decision support system to optimise N fertilisation of British grassland for economic and environmental goals. Agric. Ecosys. and Environ., 109 (1-2): 20-39.
Flechard C., Fowler D., Sutton M.A. and CapeJ.N. (1999) Modelling of ammonia and sulphur dioxide exchange over moorland vegetation. Q. J. R. Met. Soc.125, 2611-2641.
Li, C.S., 2000. Modeling trace gas emissions from agricultural ecosystems. Nutrient Cycling in Agroecosystems, 58(1/3)
Loubet B., Milford C., Hill, P.W. Tang Y.S., Cellier P., and Sutton M.A. (2002) Seasonal variability of apoplastic NH4+ and pH in an intensively managed grassland. Plant and Soil238, 97-110.
Misselbrook TH, Van der Weerden TJ, Pain BF, Jarvis SC, Chambers BJ, Smith KA, Phillips VR, Demmers TGM. (2000) Ammonia emission factors for UK agriculture. Atmos. Env. 34(6): 871-880.Nemitz E., Milford C. and Sutton M.A. (2001) A two-layer canopy compensation point model for describing bi-directional biosphere/atmosphere exchange of ammonia. Q. J. Roy. Meteor. Soc.127, 815-833.
Parton, W.J., D.S. Schimel, C.V. Cole, and D.S. Ojima. 1987. Analysis of factors controlling soil organic matter levels in Great Plains grasslands. Soil Science Society of America Journal 51:1173-1179. 465