RothC-BIOTA. Simulating cropland carbon dynamics. M. Sozanska, P. Smith, R. Milne and T. Brown.

Transcription

1 RothC-BIOTA Simulating cropland carbon dynamics. M. Sozanska, P. Smith, R. Milne and T. Brown.

2 PLAN RothC-BIOTA general overview Monthly estimation of C dynamics in crop and soil pools SOC simulation in two different agri-systems Analysis of the modelling results Immediate solutions and further testing.

3 Modelling developments so far Closely coupled link between BIOTA and RothC, Plant module provides dynamic inputs of monthly C additions to Soil module (based on % debris parameters, SUNDIAL) BIOTA has been parameterised for all main crop types N-cap method to adjust monthly biomass to amount of mineral fertiliser (yield and N uptake relationship, SUNDIAL) Retained RothC function for fitting the model to measurement at equilibrium final adjustments Annual and monthly outputs of all soil and plant C pools, monthly CO 2 soil emissions, GPP and NPP. GIS framework of RothC retained for future spatial modelling

4 Monthly climate data: Temp(av), Rain, Solar radiation, Relative humidity, Pan evaporation. Photosynthesis Respiration NPP Above-ground plant components Below-ground plant components PLANT MODULE Harvest loss (incl. Straw removed at harvest) C flows acc. Runge-Kutta method Respiration = f(leaf conductance) Land management data: Crop type, total amount of mineral N, Organic fertiliser amount, type and month of applic. (debris input). DM *a*ylde/yldmax Ylde=f(Nsoil) RPM f(dm, DPM/RPM) Plant litter = f(biomass,loss factor) Decomposition DPM f(dm, DPM/RPM) C = C*exp (-k*temp*smd*veg/12) SUNDIAL method, C additions to soil as plant debris SOIL MODULE α β Site data: latitude, Atmosperic N deposition Soil data: %Clay, soil depth BIO IOM = exp(som) HUM 1 α β α + β CO 2 =1 α - β = f (clay%) Annual and monthly C pool sizes and C losses GPP, NPP - annual and monthly outputs harvest loss (C loss in plant components removed from field) and monthly C additions to soil from plant debris and organic inputs(annual).

5 Monthly C dynamics in winter wheat, N input > 19 kg/ha. Dynamic C transfers between plant and soil pools simulated by RothC-BIOTA. Winter wheat in intensive management system. 4 6 org C (tc/ha) SOC vegetation pool litter pool simulated months

6 RothC-BIOTA simulation of C dynamics in winter oilseed rape without N input. Winter oilseed rape without N enrichment. Plant and soil C pools - RothCBIOTA org C (tc/ha) SOC plant pool litter pool simulated months

7 Monthly C dynamics in winter beans with atmospheric N uptake. Winter beans without fertiliser added. Dynamic transfers of C between plant and soil pools org C (tc/ha) SOC plant pool litter pool simulated months

8 Broadbalk Wheat Experiment. Location: 4km north of London at IACR-Rothamsted, Latitude: 51ºN49 ; Longitude: ºW21 Climate: cool temperate (mean annual temperature of 9.1 ºC and rainfall 693 mm) Soil: clay with flint over chalk (25 % clay, 57 % silt and 15 % sand) Land use history: five management regimes: continuous wheat from 1844 with occasional fallow Wheat in four different rotations with potatoes, forage maize, winter oats and legumes several treatment types: No additions since 1852, 35t/ha FYM since 1844, Different application levels of NPK (48,96,144,192,244,288 kgn/ha)

9 RothC-BIOTA simulation of SOC under winter wheat on control plot in Broadbalk. Site measurement - Broadbalk, plot 3 (winter wheat, no fertiliser). Different modelling options of RothC-BIOTA SOC (tc /ha) RothC-fitted to measurements measurements RothC-BIOTA without N limitation RothC-BIOTA (N=) RothC-BIOTA (N=48) RothC-BIOTA (N=96) RothC-BIOTA (N=144) RothC-BIOTA (N=192) RothC-BIOTA (N=24) RothC-BIOTA (N=288)

10 RothC-BIOTA results of SOC under winter wheat with mineral fertiliser input. Broadbalk, plot 8 (N= 144kg N/ha). RothC-BIOTA results RothC-BIOTA measurements RothC-fitted to measurements

11 RothC-BIOTA simulation of Broadbalk plot with mineral and organic enrichment. Broadbalk, plot 21 (winter wheat, organic and mineral fertiliser). 9 SOC (tc/ha) RothC fitted to measurements field measurements RothC-BIOTA

12 Bad Lauchstaedt Experiment. Location: near Leipzig, Latitude: 51ºN24 ; Longitude: 11ºE53 Climate: cool temperate (annual average temperature 8.7 ºC, mean annual rainfall 483mm) Soil: Haplic Phaeozem, silt loam (21 % clay, 68 % silt and 11% sand) Land use: crop rotation - sugar beet,spring barley,potatoes,winter wheat fertilising: no FYM, 2t FYM/2yrs, 3tFYM/2yrs Increasing NPK input (higher in non-organic years)

13 Bad Lauchstaedt - modelling results show greatly underestimated SOC at equilibrium. BadLauchstaedt - fitted RothC compared with RothC-BIOTA 12 1 SOC (tc/ha) RothC-BIOTA measured RothC-fitted (spr barley) simulation year

14 Reasons for RothC-BIOTA failure: Incorrect proportion of DM input to soil Wrong DM estimated by the vegetation model component Equilibrium levels are not predicted correctly with the BIOTA method as a result of unknown management (eg. Straw incorporation, fertiliser input, previous crop history)

15 A. Incorrect proportion of DM input to soil Treatment control pl. 144kgN/ha Measured ratio 61.9% 2.2% Modelled ratio 19.6% 19.4% Case: Broadbalk plot 3 and 8 Too low ratio for control plot suggests that we do not input enough of debris when there is restricted N. This might be because the N cap is too sharp (as shown later). B. Wrong DM estimation by the vegetation model component Organic C content of above ground plant matter on selected plots in Broadbalk. DM = grain + straw [tc/ha] RothC-BIOTA, plot 3 (N=) RothC-BIOTA, plot 8 (N=144 kgn/ha) measurements, plot 3 measurements, plot 8

16 C Equibrium levels are not predicted correctly or are they? Broadbalk, plot 3 (winter wheat, no fertiliser). Tests of Equilibrium level, N cap method and debris inputs. SOC (tc /ha) measurements RothC-BIOTA with full N limitation RothC fitted to measurements RothC-BIOTA, fitted to equlibrium and reduced debris inputs under fallow fitted to 32. tc (higher eq.) and reduced N limit by 1% Increase of RothC-BIOTA after 192 was due to very high debris input in fallow years (3.2 tc/annum) Which was much higher than for winter wheat (N=).83tC/annum. Input from fallow (harvest loss in the model) was reduced to 1% - biomass growth in future needs to be adjusted.

17 . Fitting the model to the equilibrium level is not always the best. Broadbalk, plot 21 (organic and mineral fertiliser input). Effect of fitted equilibrium and reduced N limit. SOC (tc/ha) RothC fitted to measurements field measurements RothC-BIOTA with full N limitation RothC-BIOTA, N limit effect reduced by 5% RothC-BIOTA, fitted equil RothC-BIOTA unfitted to equilibrium, N limit effect reduced by 5%. Or maybe the equilibrium levels are too low in RothC (as suggested in plot 3, and plot 21

18 . But then again, fitting the model to equilibrium gives the best fit for very intensive agricultural systems. Case: Bad Lauchstaedt. BadLauchstaedt - effect of equilibrium level, biomass production and N limitation. (RothC-BIOTA fitted to equilibrium). 12 RothC-BIOTA SOC (tc/ha) simulation year measured RothC-fitted (for spring barley) RothC-BIOTA fitted to equilibrium with full N limit

19 In Bad Lauchstaedt DM production needs to be magnified by a factor of BadLauchstaedt - effect of increased biomass production (BIOTA) 12 1 RothC-BIOTA SOC (tc/ha) measured RothC-fitted (spr barley) simulation year RothC-BIOTA, assimil*4.5

20 the best fit obtained with a combination of factors fitting and increased assimilation rate BadLauchstaedt - effect of equilibrium level, biomass production and N limitation. (RothC-BIOTA fitted to equilibrium). 12 RothC-BIOTA SOC (tc/ha) simulation year measured RothC-fitted (for spring barley) RothC-BIOTA fitted to equilibrium with full N limit

21 Immediate actions from this study Total DM production needs further parameterisation N cap method has to be revisited In majority of cases the model will benefit from fitting to equilibrium (RothC alone), but the equilibrium levels need to be carefully chosen

22 Further work Evaluation of the model with other systems including grasslands and semi-natural ecosystems Preparing the model for spatial applications

23 Acknowledgements. We would like to thank Gill Tuck for providing DM data (Broadbalk), Kevin Coleman for SOC data (Broadbalk), SOMNET data providers (Martin Koerschens Bad Lauchstaedt and Paul Poulton Broadbalk). Great thanks to DEFRA for money.