CCE Call for Data 2012-14, Swiss activities Dynamic Modelling Daniel Kurz (EKG Geo-Science, CH), Beat Achermann (FOEN, CH), Beat Rihm (Meteotest, CH) Models from: Luc Bonten & Gerd Jan Reinds (Alterra, NL) Max Posch (CCE, NL) Mattias Alveteg & Harald Sverdrup (LTH, SE) Site data from: Sabine Braun (IAP, CH)
Framework > Studio version of the VSD+ model for geochemical modelling > PROPS (integrated in VSD+) model for ground vegetation modelling, EMNV classification of the sites considered > applied the models to 32 forest sites for which we have > biomass estimates (BHD measurements and allometric functions) > ground vegetation composition data for 28 of the 32 sites (1 year) > soil solution data (10 year period) > calibrated 26 sites of the 32 successfully and finally > 19 sites ran through the whole procedure April 16, 2014 2
Data flow - Model stacking April 16, 2014 3
Data flow - Model stacking Deposition input April 16, 2014 4
Data flow - Model stacking Carbon, nitrogen and base cation fluxes 5
Data flow - Model stacking Climate and hydrology input April 16, 2014 6
Data flow - Model stacking April 16, 2014 7
Running VSD+ Calibration settings > calibrated regarding Gapon exchange coefficients (lgkalbc, lgkhbc) and initial pools of carbon (C) and nitrogen (N) (Cpool_0, CNrat_0) using current base saturation and C and N pools in the topsoil > soil solution data not considered for calibration > settings for the initial distributions Parameter Distribution Mean StdDev lgkalbc normal 0 1 lgkhbc normal 2 1 Cpool_0 normal 10000 5000 Cnrat_0 normal 30 20 > defaults for the rest of calibration relevant parameters > tested different starting years for calibration: 1650, 1750, 1850 8
Running VSD+ Calibration results > Overestimation of C-pool > Substantial scatter of N-pool > Underestimation of C/N ratio > Good fit of base saturation 9
Running VSD+ Calibration issues: Negative N-balance > N balance of the model system (N in -N out ) N n N n n N in,t N out,t t 1 n N t 1 DEP,t N LF,t N UPT,t often is negative in historic times as a result of the combination of allowing growth not limited by N availability (GrowUp) and lowering of historic N deposition > mineralization of the pools is enforced for prolonged periods to balance the missing N, which in turn may lead to conflicts in the calibration procedure > Negative N balance may not only be an issue of the past and the calibration but also of the second half of the modeling period, if background deposition is used as deposition scenario beyond 2020 10
Running VSD+ Soil solution chemistry compared with observations > Site by site comparison > Median and 80% of the value range of the observation period 1998-2008 > Log-log scale > Same order of magnitude shaded > Tracer ions and sodium within same order of magnitude > Ions involved in biological processes and dependent ions show larger scatter > Patterns of different geochemical models mostly pretty comparable 11
Running VSD+ Solid phase chemistry trends 12
Running VSD+ Soil solution chemistry trends 13
Running VSD+ The Nupeff dilemma > Nitrogen uptake efficiency should be set between 0.9 and 1 according to the help file of VSD+ > It alters the nitrate concentration in the soil solution and thereby also affects the concentration of other ions in the soil solution > We have used Nupeff = 0.98 without any scientific underpinning > Nupeff = 1 leads in most of the sites to very low N concentrations and subsequently PROPS refuses to return plant occurrence probabilities other than 0 (when using EMNV classification) 16/04/2014 14
PROPS Regional pattern of default indices > ground vegetation composition changes substantially between 2010 and 2100 > tendency to loose species under GP and rel. low rel. number of species during large parts of the simulation period > hardly any (visible) difference in regional indices patterns between GP and BG deposition scenario 15
PROPS Landolt N index > among the total species modelled and observed on the plots only 10% are in found in both sets > we therefore refrain from comparing modelled and observed plant occurrence > classified the species according to the Landolt N index(1: oligotrophic, 5 eutrophic spezies): group 2 and 3 dominate (75-85%) in both observed and modelled > trends of average normalized Landolt N classified plant occurrence probabilities resulting from applying GP and BG deposition scenarios are comparable 16
Summary > Negative N-balance Does enforced mineralization reflect reality? If not then we have to either reduce the growth (dependent on N-availability?) and/or to lower N contents of the tree tissues and/or to allow an additional N in flux such as N-fixation > Calibration we obviously still have problems to find a common setting for the calibration and alternatives such as uniform prior distribution for the pools should be tested with the aim to improve the calibration result > Nupeff regarding PROPS obviously minimal N concentrations in the soil solution are required how to select appropriate Nupeff value(s) is not really clear but the impact of the (single parameter) Nupeff on the soil solution composition is substantial > Results with the current settings there apparently is hardly any relation between soil solution N and N deposition on a regional scale the plant occurrence probability patterns (indices) are difficult to interpret is a key year (2100) a good choice having the noise in the model input/output in mind April 16, 2014 17