Effect of soil characteristics and agricultural management on Soil Organic Carbon

Get Complete Project Material File(s) Now! »

Site description

In this study, we chose five long-term (up to 10 years) European experimental sites from the CarboEurope Integrated Project (CEIP) ecosystem flux tower network, which have two adjacent contrasting management system plots (a paired-site): Laqueuille (Intensive/extensive grazing and high/no N input grassland, France) sampled in November 2008 and March 2011, Oensingen (high/low N&manure input and cut/no-cut grassland, Switzerland) sampled at the end of August 2012, Reeuwijk (Grazing peat grassland with fertilizer&manure input/ no-input, Netherlands) sampled in January 2011, Hertfordshire (ploughed/non-inverted and high/no N input cropland, United Kingdom) sampled in November 2003 and Carlow (Conventional & N input and reduced tillage & no N cropland, Ireland) sampled in January 2011. At these five sites, paired-plot experiments have been carried out for between 10 and 25 years. For reasonable sample sizes and statistical power, it may take >10 years for significant differences in SOC due to contrasting management to be observed (Smith, 2005).

Grassland

Laqueuille, France
Laqueuille (France) site is situated at 45° 38’N, 2° 44’E. Soil type is an Andosol (Klumpp et al. (2011)). Mean annual temperature is 7.68 °C, mean annual precipitation is 928.5 mm and ele-vation is 1040 m a.s.l. This grassland site is managed by INRA Clermont-Ferrand, UREP since 2002 until the present. Previously, this site was cultivated as a cropland from the beginning of 20th century to the 1950s. Sometime around 1950s, this site was converted to permanent grassland. This site was managed until 1980 by mowing, grazing, cattle manure and slurry ap-plication. Since the paired plot experiment started in 2002, the site was split into two contrasting management plots (intensive management and extensive management systems). These two man-agement systems have been maintained until today. The intensive management system plot is grazed from spring until the summer by livestock (0.9-1.2 livestock unit (LSU)/ha). Ammonium nitrate was applied three times (191.5 kg N ha−1year−1) each year. The extensive plot is also grazed during the same period as the intensive plot by livestock (0.5-0.6 LSU/ha), though it has not been mown. No fertilizers were applied to the extensive plot. The mean soil pH was 5.13 (±0.23) for the intensive plot and 5.25 (±0.1) for the extensive plot in 2008, and 5.46 (±0.24) for intensive plot and 5.40 (±0.2) for extensive plot in 2011. For more detail, see Table 3.1 and Table 3.2 in Section 3.2.1 or Klumpp et al. (2011); Soussana et al. (2007); Allard et al. (2007).
Oensingen, Switzerland
The Oensingen site is situated in central Switzerland at 47°17’ N, 7°44’ E. The soil type is a Eutric-stagnic cambisol and the climate is temperate continental. The mean annual temperature is 9 °C, mean annual precipitation is 1100 mm and elevation is 450 m a.s.l. (Flechard et al. (2007)). This grassland is managed by Agroscope Reckenholz-Tänikon Research Station ART. The previous management of this site was a ley-arable rotation with an 8 year rotation (sum-mer/winter wheat, rape, maize and bi-/tri-annual grass-clover mixture). Annual mean Nitrogen (N) input was 110 N ha−1year−1. The last ploughing took place in November 2000 and the grass was sown after ploughing. The site was split into two contrasting management plots in Novem-ber 2000: intensive and extensive management. Until the present, the site is maintained as a paired plot experiment. The intensive management plot consists of 7 species of sown grassland (Ammann et al. (2009)) which is mown 4 times/year. No animal grazing is applied. The mean N input over a 5 year period is 230 kg N ha−1 year−1 divided into 2-3 applications each year. The mean carbon (C) input from 2002 to 2004 was 47 g C m−2year−1 dividing into 1-2 applications each year. The intensive plot was ploughed and the same 7 species were re-sown. The extensive management plot is a mixture of over 30 species of sown grassland and is mown 3 times each year (1st mowing is never before 1st June). There is no grazing by livestock and no fertiliser was applied. The mean soil pH is 6.85 (±0.49) for the intensive plot and 5.81 (±0.65) for the extensive plot. The soil contains carbonate. For more detail, see Table 3.1 and Table 3.2 in Sec-tion 3.2.1 or Ammann et al. (2007, 2009); Flechard et al. (2005); Leifeld et al. (2011); Flechard et al. (2007).
Reeuwijk, Netherlands
Reeuwijk (Netherlands) site is situated at an altitude -1.7 to -1.6 m a.s.l., 52° 2’N and 4° 47’E (intensive plot) and 52° 1’N and 4° 46’E (extensive plot). The soil type is a Fibric Eutric Histosol. The topsoil is peaty and clayey, and the subsoil is Eutrophic peat (Stolk et al. (2011)). The paired plots are located on the polder in the west of the Netherlands. Mean annual temperature is 9.80 °C and mean annual precipitation is 800 m (Jacobs et al. (2007); Veenendaal et al. (2007); Schrier-Uijl et al. (2008, 2010)). This site also has two contrasting management plots, though these plots are further apart than for the other sites (See Figure 2.1).
This site was managed by University of Wageningen, Nature Conservation and the Plant Ecology Group, Wageningen, Netherlands between 2000 and 2004. The intensively managed site is called Oukoop and the extensively managed site is called Stein. Each management plot had an independent eddy-covariance tower. Both plots in Reeuwijk had been under intensive management (deep drainage, application of manure and fertiliser and grazing and harvesting of grass) for 30+ years before the start of the measurement period. About 25 years ago, the extensive plot was taken out of intensive management, and gradually become a meadow bird reserve which was established about 14 years ago (higher water table, no inputs of manure and fertiliser, low grazing, lower harvest rate of grass (see Veenendaal et al. (2007); Schrier-Uijl et al. (2008, 2010)). The paired plot experiment was started in the 1980’s. The intensive management system was dairy farming and mowing, with grazing between the middle of May to the middle of September each year. Mowing occurred 3 times per year, as did 2-3 short periods of grazing by livestock. The mean annual manure application level was 1.67 kgC ha−1year−1 and 309.67 kg N ha−1year−1. The mean annual fertiliser application rate was 88 kg ha−1year−1 divided into 2-3 applications per year. Manure and fertiliser were not applied in winter. Dominant grass species were rye grass and bluegrass and clover occupied less than 1% of the plot. The extensive management plot was a meadow bird reserve. This plot was mown twice after 15th June each year, with a short period of grazing by livestock on most of the plot, with a few areas grazed over the whole summer. This plot was gradually taken out of intensive grassland management after more than 20 years (Schrier-Uijl et al. (2008, 2010)). No manure and no fertilizer were applied to this plot. The mean soil pH is 6.01 (±0.08) for the intensive plot and 5.34 (±0.21) for the extensive plot. For more detail, see Table 3.1 and Table 3.2 in Section 3.2.1 or Jacobs et al. (2007); Veenendaal et al. (2007); Schrier-Uijl et al. (2008, 2010).

Cropland

Hertfordshire, United Kingdom
Hertfordshire (UK) arable site is situated at an altitude 140 m a.s.l. and 51° 47’N, 0° 28’ W. The mean annual temperature is 9.46 °C and the mean annual precipitation is 703.5 mm. Soil type is Chromic Luvisols and Orthic Acrisols (silt and clay loam) from FAO soil classification (Berthe-lin (1999)). The clay constitutes about 25% of the top soil and silt about 50%. Stones (flints) occupy about 10.8% of the top soil volume. The site is owned by W. Hill & Sons, Wood Farm in Hertfordshire UK. Soil sampling and eddy-covariance measurements were performed by the De-partment of Sustainable Soils and Grassland Systems, Rothamsted Research, Hertfordshire, UK in a research project funded by the UK Biotechnology and Biological Sciences Research Coun-cil, grant number D16053. This site is also a paired plot site. The experiment was conducted from 2001 to 2008. Dried soil samples were taken in November 2003. One management plot is ploughed and the other is minimum tillage. The site was grassland in the 1940s. Regarding the ploughed plot, the plot was grassland until the late 1960s, then it was converted to cropland (except 1972-1973 and 1976-1978, the plot was ley grassland). From 1980 to 1995, the mean average NPK input was 217.71 kg ha−1year−1 (mostly N:P:K=5:24:24) and mean annual nitro-gen fertiliser input was 193 kg N ha−1year−1. The mean annual yield was 5.77 t ha−1year−1. The crop rotation was around 4 years rotation with winter wheat, winter oilseed rape and spring peas. From 2003 to 2008, in the ploughed plot, the mean annual NPK input was 49.27 kg ha−1year−1 and the NH+4 input was 152.79 kg ha−1year−1. The crop rotation was around 5 years with win-ter wheat, winter oilseed rape and spring peas (as before). The mean annual yield (2003-2008) was 5.9 t ha−1year−1. Concerning to minimum tillage plot, 50% of the plot was converted to arable land. Early to late 1960, all of the plot was converted to arable land (except 1977-1979: ley grassland). From 1980 to 1995, the plot was ploughed and the mean annual NPK input was 211.48 kg ha−1year−1 (mostly N:P:K=5:24:24 or 0:24:24) and the mean annual N fertiliser input was 195.29 kg N ha−1year−1. The Mean annual yield was 5.89 t ha−1year−1. In 1986, the plot was split into two and in 2000 the plot was set-a-side for a year. The crop rotation was around a 5 years rotation with the same crops as the ploughed plot. Minimum tillage management has started from 2001. Annual average (2003-2008) NPK input was 43.50 kg ha−1year−1 and the NH+4 input was 57.63 kg ha−1year−1. The crop rotation was the same as the ploughed plot. The mean annual yield (2003-2008) was 6.28 t ha−1year−1. This soil contains carbonate. Mean soil pH values are 6.94 (±0.23) for the intensive plot and 6.56 (±0.31) for the extensive plot. For more detail, see Table 4.1 and Table 4.2 in Section 4.2.1.
Carlow, Ireland
Carlow (Ireland) site situates at altitude 57 m a.s.l., 52° 52’N and 6° 54’W. The mean annual tem-perature is 9.40 °C and the mean annual precipitation is 824 mm (Van Groenigen et al. (2010a,b); Walmsley et al. (2011)). Soil type is Eutric Cambisol (Osborne et al. (2010)). The site’s land history is described in Van Groenigen et al. (2010a,b). Prior to 1847, the site was pasture, then from 1847 to 1977 the land was used as a school playing field under grassland. From 1977 to 1992, the site was mainly improved grassland with grazing and silage. In 1993, the site was converted to cropland with ploughing and crop rotation. From 2000, the site was cultivated mainly for spring barley for 3 consecutive years. From 2003, the paired plot experiment started. One management is the conventional tillage plot (20-25 cm depth ploughing) and the other is no-tillage plot. Since 2003, the site has been cultivated for spring barley. Regarding the con-ventional tillage plot, the plot is ploughed in March each year. After ploughing, spring barley was sown. The mean annual fertiliser input was 140 kg N ha−1year−1. The annual average grain harvest (2003-2010) was 6.34 t ha−1year−1. Regarding the no-tillage plot, harrowing (10-15 cm depth) was performed once per year in each August or September. Spring barley is sown in March each year. The mean annual fertilizer input is 140 kg ha−1year−1. The mean annual grain harvest (2003-2010) was 5.99 t ha−1year−1. Mean soil pH is 7.40 for the conventional tillage plot and 6.83 for no-tillage plot. For more detail, see Table 3.1 and Table 3.2 in Section 3.2.1 or Walmsley et al. (2011); Osborne et al. (2010); Davis et al. (2010).

Soil sampling and soil organic carbon (SOC) analy-sis method

Soil sampling method

In autumn/winter 2011/2012 soil sampling campaigns were carried out for the four agricultural paired-sites (8 plots in total). In each plot, 6 soil cores (60 cm depth for Laqueuille (FR) and Reeuwijk (NL) and 40cm depth for Oensingen (CH) and Carlow (IE)). At the intensive plot in Reeuwijk (NL), 4 soil cores were sampled at equal intervals (10m) along 60cm transect. Each soil core was separated into 0-5 cm, 5-10 cm, 10-20 cm, 20-40 cm and 40-60 cm depth layers that were air dried under laboratory conditions. Regarding Hertfordshire (UK), soil sampling (down to 20 cm soil depth, 8 soil cores each plot) was carried out in November 2003. UK site soil cores were not cut into any strata.
We have to mention that one of the limits of this experimental design using eddy covariance flux tower sites is the absence of replicates, thus a possible pseudo replication of soil samples (i.e. 6 soil cores per field). To overcome this problem, we have chosen a sampling design which covers foot-print area for comparing with flux measurement and model flux output (see Chapter 5). For sampling at Laqueuille (FR), Carlow (IE) and Reeuwijk (NL) sites, we followed to the sampling protocol (See Figure 2.2). For each adjacent field, soil within the footprint of the eddy-flux tower were sampled (i.e. about a soil core every 10m).
Sampled soils were immediately stored in a refrigerator at 4 °C. They processed this on the sampling date. They were thoroughly mixed. After weighing the wet soil weight, they placed soil samples into at 60 °C oven for 24 hours to dry.

Soil pH measurement

We measured soil pH concerning Laqueuille (FR), Oensingen (CH), Reeuwijk (NL) and Carlow (IE) soil samples. Hertfordshire (UK) soil sample’s soil pH have already measured at Depart-ment of Sustainable Soils and Grassland Systems, Rothamsted Research. For soil pH measure-ment, we followed to Soon and Hendershot (1993). We weighted 10g of air-dried mineral soil which was sieved at 2mm. Place them in a small beaker and added 20 mL of ultra pure water. Concerning organic soils (Reeuwijk, NL), we weighted 2g in 20 mL of ultra pure water. We stirred these samples for 1 hour, then we let them stand for 1 hour. We agitated the bottom of samples during this measurement. Hertfordshire (UK) soil sample’s soil pH was measured in water with following to Jones Jr (1999).

Soil fractionation method

We used the method which was improved from Zimmermann et al. (2007) (See Figure2.3) soil fractionation method (Wurster et al. (2010)) concerning the step of separating heavy fraction (S+A) and light fraction (POM) from the >63 µm soil fraction.
For separating S+A and POM, sodium ploytungustate (SPT) at ρ=1.85g/cm3) is used in Zimmermann et al. (2007) method, as SPT salt is non-toxic and can be mixed with deionised water (DI-H2O). We took 30g dry soil for starting soil fractionation. Then, we disrupted the soil in 160mL DI-H2O with using ultrasonic probe (SONICS, vibracell, VCX500) at 34% amplitude and 22J/mL. After the disruption, wet sieving at 63 µm was done.
The fraction <63µm was centrifuged at 1800 rpm for 4 minutes, and the fraction between 63 µm and 0.45 µm was classified as silt and clay, and the <0.45 µm fraction was classified as Dissolved Organic Carbon (DOC).
Silt and clay fraction was oxidised by Sodium Hypochlorite (NaOCl), and the rest fraction after the oxidation is classified as resistant soil organic carbon (rSOC).
The fraction >63 µm was dried in the oven then SPT was added from 20 to 30mL with using a wash bottle for each centrifuge tube in this study. The solution is mixed by turning the centrifuge tube slowly and gently from side to side, until well mixed with enough SPT solution and the >63 µm soil fraction. This process enables the organic material to float on the surface in a centrifuge tube. Separation usually takes place with using light centrifugation. In Zimmermann et al. (2007) method, the centrifugation was taken place at 1800 rpm for 15 minutes then they put the centrifuge tube at stable place for overnight. Then, usually floating light fraction (POM: Particulate Organic Matter) is either poured off or aspirated. However, it is difficult to separate POM and heavy fraction (S+A: Sand and Aggregates) with this method because POM tends to stick to the centrifuge tube walls. Furthermore, this method may remix the separated POM and S+A while pouring POM off or aspirating it (Wurster et al. (2010)). In Wurster et al. (2010), after separating process of >63 µm soil fraction into S+A and POM with SPT, the centrifuge tube was carefully placed in a freezer as upright position. The frozen sample was removed from the freezer and the surface material is immediately thawed off with using DI-H2O from a wash bottle and poured into a beaker. This process should be continued until seeing the frozen surface of S+A in the centrifuge tube. Frozen SPT solution and surface material can be readily recovered with washing off with DI-H2O. S+A was also thawed S+A off with using DI-H2O and washing off for removing SPT by DI-H2O with aspirating. In this study, we used SPT at ρ=1.85 g cm3. At the end of this method, we got 5 fractions: Silt and Clay (s+c), Particulate Organic Matter (POM), Sand and aggregates (S+A), Dissolved Organic Carbon (DOC) and Resistant Soil Organic carbon (rSOC).

READ  Portfolio Theory and Capital Asset Pricing Model

Carbon and Nitrogen analysis

SOC and nitrogen analysis have done in INRA Clermont-Ferrand with Thermo Electron Co-operation, NC analyser, Flash EA 1112 series, except total SOC and nitrogen of Oensingen site samples were analysed at Agroscope Reckenholz-Tänikon Research Station ART, Zürich, Switzerland.
Before the analysis, we did decarbonation regarding s+c, S+A, DOC and rSOC fractions because we are interested in analysing soil organic carbon (SOC), and carbonates are inorganic carbon. With regard to soil pH in the Table 3.1 and 4.1, we had several sites’ soil samples which contain carbonate. We also decarbonated to analyse total SOC and soil nitrogen amount. We did not decarbonate POM fraction for avoiding to damage plant materials which are contained a lot in POM fraction. Furthermore, SPT’s density which we use in the soil fractionation is 1.85 g cm−3. Carbonates, principally, consisted by CaCO3 (Calcite) and CaMgCO3 (Dolomite). These two carbonate minerals’ (Calcite and Dolomite) densities are 2.71 g cm−3 and 2.83 g cm−3 respectively. Therefore, S+A, s+c, rSOC and DOC contain carbonates but not POM. They are not in the POM which is lighter than 1.85 g cm−3. Decarbonate process have done by reference to Harris et al. (2001) but with some modification by ourselves. We placed soil fractions in Ag-foil capsules (8 ×5 mm). The capsules were placed in the wells of a microtiter plate. We added 40 µL ultra pure water for each capsule. Then we placed a small beaker with 100 mL of concentrated 12M Hydrochloric acid (HCl) inside the vacuum 5L desiccator. Then we placed the wells of a microtiter plate in the desiccator to expose to HCl vapour. Under the desiccator, we placed hot heater at 40 °C to warm up the desiccator. We left these for 6 hours, then added 25 µL ultra pure water each 2 hours during 4 hours. Then, dry the microtiter at 60 °C for 48 hours1. All of the processes were done under hood. Agroscope Reckenholz-Tänikon Research Station ART treated with HCl in a desiccator to remove carbonate and by elemental analysis (combustion at 1000 °C, released CO2 measured by GC-TCD). The difference between these two methods was between -0.27 and 1.14%. So we can confirm that out HCl vapour method was correct enough to remove decarbonate and to analyse only SOC.

Eddy-covariance flux data

Briefly, Gross Primary Production (GPP) is C fixation by plants and Ecosystem respiration (Reco) is C flow due to respiration by plants, soil and animals. Net ecosystem exchange (NEE) is the difference between GPP and Reco, which includes plant, soil and animal influence on C flow. Each plot was equipped with a meteo station and eddy covariance flux tower, registering half hourly data on meteorological variables (e.g. temperature, radiation, precipitation) and C fluxes. Fluxes data were processed according to EU guidelines (Aubinet et al. (2012)) to obtain NEE, where NEE was further partitioned according to European flux guidelines into GPP and Reco (Reichstein et al. (2005)).

Process-based models

We used four process-based models in this study: Rothamsted Carbon (RothC) model, Pasture Simulation Model (PaSim) and Denitrification / Decomposition Model (DNDC) and DailyCEN-TURY (DayCent).
The RothC model is the earliest and simplest SOC model for agricultural land. In this study, the Zimmermann et al. (2007) soil fractionation method was used, which has been tested as a compatible method for estimating model SOC pool sizes in several studies (Leifeld et al. (2009a); Dondini et al. (2009a); Senapati et al. (2013)). We tested several initialisation methods with the RothC model firstly, using our measured SOC fraction data. We then tested several initialisation methods with PaSim, DNDC and DayCent regarding flux outputs such as GPP, NEE and Reco.
PaSim model is a process-based grassland biogeochemical model which includes mowing and animal grazing modules specific to grassland management regimes. Since the RothC model was constructed as a process-based cropland model, verification with a process-based grassland model was necessary. PaSim was used for validating grassland simulation results. Furthermore, as RothC is a simple SOC model without any other submodels, validation with a more sophisti-cated process-based model containing several submodels was desirable, for its ability to simulate cropland and grasslands. The DayCent model’s carbon and nitrogen submodels are the same as those of PaSim (both based on CENTRY) although DayCent can simulate grassland, cropland, forest and savannah systems. We also tested our initialisation methods using the DayCent model. We also ran the DNDC model due to its ability to simulate varied agricultural land uses. Table 2.1 shows the characteristics of each process-based model.

Rothamsted Carbon (RothC) -26.3 model

The RothC-26.3 model simulates agricultural topsoil processes (0-23 cm), and it does not con-tain a plant production submodel (Coleman and Jenkinson (1996, 1999); Coleman et al. (1997)). The model requires weather data (mean monthly air temperature (°C), monthly precipitation (mm) and monthly open pan evaporation (mm)). As site specific data, the model requires clay content of the soil (%), soil cover (vegetated or bare) and depth of soil layer sampled (cm). As site management data, it requires monthly plant residue input (tC/ha) and monthly farmy-fard manure input (tC/ha). The model is constructed with five SOC pools (Decomposable Plant Material (DPM), Resistant Plant Material (RPM), Microbial Biomass (BIO), Humified Organic Matter (HUM) and Inert Organic Matter (IOM). In this model, SOC input is divided into two SOC pools: DPM and RPM. Then, from these two SOC pools, SOC flows to BIO, HUM and CO2. From these BIO and HUM SOC pools, SOC is further divided into BIO, HUM and CO2 (see Figure 2.4). Each SOC pool’s turnover time is DPM (0.165 years), RPM (2.31 years), BIO (1.69 years), HUM (49.5 years) and IOM (50,000 years)(Jenkinson and Rayner (1977); Cole-man and Jenkinson (1999)). RothC model anticipate soil respiration as a flux output. For more detail of this model, see Coleman and Jenkinson (1999); Jenkinson and Coleman (2008).

Denitrification / Decomposition Model (DNDC)

The DNDC model is a general model of C and N biogeochemistry in agricultural ecosystems (Li et al. (1994)). This model predicts crop yield, soil carbon sequestration, nitrogen leaching, and trace gas emissions in agro-ecosystems. DNDC, can run at site or regional scale. The model consists of two components (See Figure 2.5).
User’s Guide for the DNDC Model Version 9.1 (Institute for the Study of Earth, Oceans and Space (2007))
The first component consists of the soil climate, crop growth and decomposition sub-models, predicting soil temperature, moisture, pH, redox potential (Eh) and substrate concentration pro-files, driven by ecological drivers (e.g. climate, soil, vegetation and anthropogenic activity). The second component consists of the nitrification, denitrification and fermentation sub-models, predicting NO, nitrous oxide (N2O), N2, CH4 and NH3 fluxes based on the modelled soil en-vironmental factors (User’s Guide for the DNDC Model Version 9.1 Institute for the Study of Earth, Oceans and Space (2007)). The entire model is driven by four major ecological drivers: climate, soil physical properties, vegetation, and anthropogenic activities. Accurate input data will provide accurate simulation results at either site or regional scale (Li et al. (1994, 1992); Li (2000)). For more detail on this model, see Li et al. (1994, 1992); Li (2000).
In DNDC, SOC is divided into four major pools: plant residue (i.e. litter), microbial biomass, humads (i.e. active humus) and passive humus. Each pool consists of two or three sub-pools with different specific decomposition rates (Figure 2.5). The daily decomposition rate for each sub-pool is regulated by the pool size, the specific decomposition rate, soil clay content, N availability, soil temperature and soil moisture (Li et al. (1994, 1992); Li (2000)). If SOC de-composes, the decomposed C is partially allocated into other SOC pools and partially lost as CO2. Dissolved organic carbon (DOC) is produced as an intermediate during decomposition. However, DOC may be immediately consumed by the soil microbes. During the processes of SOC decomposition, the decomposed organic nitrogen partially transfers to the next organic matter pool.
The organic nitrogen is also partially mineralised to ammonium (NH+4), simulating nitrifi-cation (Li et al. (1994, 1992); Li (2000)). The ammonium concentration is controlled by clay-adsorbed NH+4 and dissolved ammonia (NH3) (Figure 2.5). NH3 emission is controlled by NH3 concentration in the soil water phase and soil environmental factors (e.g. temperature, mois-ture and pH)(User’s Guide for the DNDC Model Version 9.1 Institute for the Study of Earth, Oceans and Space (2007)). The denitrification sub-model of DNDC calculates N2O and NO production, consumption and diffusion, during rainfall, irrigation and flooding events. DNDC simulates relative growth rates of nitrate, nitrite, NO, and N2O denitrificator based on soil het-erotrophic respiration, and concentration of Dissolved Organic Carbon (DOC) and N oxides (Smith et al. (2007c)). DNDC predicts nitrification rates by tracking nitrification activity and NH4+ concentration. Growth and death rates of nitrificating bacteria are calculated as a func-tion of DOC concentration, temperature and moisture, based on Blagodatsky and Richter (1998) and Blagodatsky et al. (1998) (Smith et al. (2007c)). DNDC calculates CH4 production as a function of DOC concentration and temperature, under anaerobic conditions. In DNDC, CH4 oxidation is calculated as a function of soil CH4 and Eh (soil redo potential (mV)). CH4 moves from anaerobic production zones to aerobic oxidation zones via diffusion, which is modelled using concentration gradients between soil layers, temperature and soil porosity (Li et al. (1994, 1992); Li (2000)). DNDC predicts NEE and Reco. Regarding GPP, we calculated using model NEE and Reco results.

DayCent

CENTURY is available for monthly time step simulations, and DayCent is the daily time step version of CENTURY. CENTURY and DayCent were developed to deal with a wide range of cropping system rotations, tillage practices for systems analysis of the effects of management, global change on productivity and sustainability of agroecosystems. The daily time-step version of the CENTURY model, DayCent5, provides enhanced resolution of all the processes simu-lated by CENTURY5: plant production, decomposition, soil hydrology and temperature. Daily weather data is used in the DayCent, although monthly weather or the site parameter’s statis-tical monthly weather can be used. The big difference between CENTURY and DayCent is, DayCent is driven by daily weather data which is better to predict N2O emission. The DayCent model structure is the same as that of CENTURY, although there are additional site parameters in DayCent 5: site parameters for Hydrology and Biophysical Controls and Soil Temperature (see Table 2.2). DayCent predicts only NEE (neither GPP nor Reco) as a C flux output.
DayCent comprises six different submodels: Carbon (C), Nitrogen (N), Phosphorus (P), Sulfur (S), soil water & temperature and plant production (Parton et al. (1987); Parton and Ras-mussen (1994), CENTURY User’s guide and References). The plant production submodel may be a grassland, cropland, forest and savannah systems. The grassland, cropland and forest sys-tem have different plant production submodels, linking to a common SOM submodel (Parton (1996)). Multiple agricultural management systems, including crop rotations, tillage practices, fertilisation, organic matter addition, irrigation, grazing and harvest methods, are simulated by DayCent. Paustian et al. (1992) tested the CENTURY model with long-term SOM experiment data from Sweden. Parton et al. (1987) tested the CENTURY model with SOC and N plant produc-tion data from grassland soils in the United States Great Plains. Del Grosso et al. (2005, 2006) tested DayCent for GHG fluxes (nitrous oxide (N2O), CO2 and CH4) for major crops in the USA. A simple description of the DayCent model (Parton and Rasmussen (1994)) is shown in the Figure 2.6 (Parton (1996)).
In the model, all carbon decomposition is derived from microbial activity and microbial res-piration, and is correlated with each carbon flow (Parton et al. (1987)). Figure 2.7 (CENTURY Tutorial) shows the detail of the carbon submodel in the latest version of DayCent 5. In the DayCent model, plant residue is divided into structural C pools (difficult to decompose) and metabolic C pool (readily decomposable), according to the lignin to N ratio (L/N) of the litter (more structural, higher L/N ratio). Aboveground and belowground C subpools are comprised in the structural and metabolic C pools. The structural C pool comprises cellulose, hemi-cellulose and lignin of plant material, although the metabolic C pool is easily decomposable. The active C pool includes live microbes and microbial products, sharing up to 2% of the toal soil carbon with short turnover time period (1-3 months). The slow C pool, sharing 45-60% of total SOC, represents resistant plant materials originating from strucutural plant materials and physically protected soil microbial products, with turnover times of between 10 and 50 years, depending on the climate. The passive C pool (45-50% of total SOC) contains physically and chemi-cally stabilised C, being resistant to decomposition (Parton et al. (1987); Parton and Rasmussen (1994)) and its turnover time is between 400 and 4000 years. In the CENTURY model, the turnover time of bulk SOC is as a function of turnover time in specific C pools, soil moisture and soil temperature. This factor is calculated multiplying the soil moisture factor (function of precipitation and stored soil water) and soil temperature factor (function of average monthly soil surface temperature) (Parton et al. (1987); Parton and Rasmussen (1994)). Regarding the active C pool’s turnover time, the rate varies according to the soil texture: more rapid for sandy soil texture, while the stabilisation of the active C into slow C is a function of the silt and clay content. This means that higher stabilisation is derived from higher silt and clay content.
The basic N submodel structure is the same as that of the carbon submodel in DayCent (see Figure 2.8, Parton and Rasmussen (1994)). The N flow is calculated from the C/N ratio. The C/N ratio of the different N pools changes according to the soil mineral N. The detailed N flow in the model is explained in section 2.3.4 in the PaSim model. Turnover of active SOC creates most of the soil mineral N (Parton et al. (1987); Parton and Rasmussen (1994), see Figure 2.8).

Table of contents :

1 Introduction 
1.1 General Introduction
1.2 Background
1.3 Objective and outline of the thesis
2 Material and Methods 
2.1 Site description
2.1.1 Grassland
Laqueuille, France
Oensingen, Switzerland
Reeuwijk, Netherlands
2.1.2 Cropland
Hertfordshire, United Kingdom
Carlow, Ireland
2.2 Soil sampling and soil organic carbon (SOC) analysis method
2.2.1 Soil sampling method
2.2.2 Soil pH measurement
2.2.3 Soil fractionation method
2.2.4 Carbon and Nitrogen analysis
2.2.5 Eddy-covariance flux data
2.3 Process-based models
2.3.1 Rothamsted Carbon (RothC) -26.3 model
2.3.2 Denitrification / Decomposition Model (DNDC)
2.3.3 DayCent
2.3.4 Pasture Simulation Model (PaSim)
2.3.5 Model initialisation
Applying measured SOC fractions to the model SOC pools
Model initialisation methods
Spin-up run initialisation repeating average recent management
data and 1901-2010 weather data (Control)
Spin-up initialisation reducing or increasing C input (plant litter and organic carbon manure input)
The observed SOC fraction data initialisation (ObSOC)
3 Effect of soil characteristics and agricultural management on Soil Organic Carbon (SOC) pools in European grasslands and cropland – soil profile and soil fractions 
3.1 Introduction
3.2 Materials and Methods
3.2.1 Site description
3.2.2 Soil sampling and analysis method
Soil fractionation
Carbon and Nitrogen analysis
3.2.3 Statistical analysis
3.3 Results
3.4 Discussion
3.5 Conclusion
4 Soil organic carbon (SOC) equilibrium and model initialisation methods for the Rothamsted Carbon (RothC) model -26.3 
4.1 Introduction
4.2 Materials and Methods
4.2.1 Site description
4.2.2 Soil sampling and analysis method
Soil fractionation
Carbon and Nitrogen analysis
4.2.3 Measurement data application to model
Soil Organic Carbon Pools calculation
RothC-26.3 model input data preparation
RothC-26.3 model run
Spin-up run initialisation using the recent management and 1901- 2010 weather data (Control)
Spin-up initialisation involving a reduction or an increase in C input (plant litter and organic carbon manure input)
The observed SOC fraction data initialisation (ObSOC)
4.2.4 Statistical analysis
4.3 Result
4.4 Discussion
4.5 Conclusion
5 Impact of the initialisation method on the simulation of grassland carbon fluxes using detailed soil organic carbon data and alternative models – a case study 
5.1 Introduction
5.2 Materials and Methods
5.3 Results
5.4 Discussion and conclusion
6 Synthesis

GET THE COMPLETE PROJECT

Related Posts