Supplementary methods: C dynamics model (S1) Long-term CO2 ... · Supplementary methods: C dynamics model (S1) Long-term CO2 production following permafrost thawing Bo Elberling1,2
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Supplementary methods: C dynamics model (S1)
Long-term CO2 production following permafrost thawing
Bo Elberling1,2 Anders Michelsen3,1, Christina Schädel4, Edward A.G. Schuur4, Hanne H.
Christiansen2,5, Louise Berg1, Mikkel P. Tamstorf6 & Charlotte Sigsgaard1
Data assimilation was used to model decay rates and to optimise parameter (p) estimation. This inverse
modelling approach for incubation data is based on Bayes' theorem (suppl. equation 1), which states that the
posterior probability density function (PPDF) P(p|Z) of model parameters (p) can be obtained from prior
knowledge of parameters, represented by a prior probability density function P(p), and the information that is
contained in the incubation data set, represented by a likelihood function P(Z|p). See S1-S5.
)()|()|( pPpZPZpP (suppl. equation 1)
To perform the Bayesian inversion, we first specified the ranges (lower and upper limits, Supplementary
table 2) of model parameters according to literature values assuming a uniform distribution over the specific
parameter ranges. The likelihood function P(Z|p) (supplementary equation 2) was calculated on the
assumption that errors between observed and modelled values followed a Gaussian distribution where Z(t)
denotes the data obtained from measurements, X(t) is the modelled value and σ is the standard deviation of
the observed decomposition rate.
)(
22 )]()([
21exp)|(
iZobstii tXtZpZP
(suppl. equation 2)
Long-term CO2 production following permafrost thaw
Suppl. Table 2: Prior parameter ranges, maximum likelihood estimates (MLEs) and 95%
probability interval of parameters for each soil type. ki= pool-specific decay rate (day-1); fi = C pool
fractionation coefficient.
* for the poorly constrained parameter kpassive a mean value is calculated instead of a MLE
Parameter Lower Limit Upper Limit MLE 95% Probability Interval
Heath (dry)
ffast (x10-2) 0 5 1.16 (0.16, 4.22)
fslow (x10-1) 0 10 5.55 (4.54, 6.66)
kfast (x10-3) 0 20 2.7 (0.37,11.8)
kslow (x10-4) 0 30 6.97 (5.23, 9.15)
kpassive (x10-7)* 0 11 5.21 (0.76, 9.95)
Grassland (wet)
ffast (x10-2) 0 5 1.06 (0.12, 4.29)
fslow (x10-1) 0 10 1.12 (0.25, 2.63)
kfast(x10-3) 0 20 2.35 (0.35, 10.68)
kslow (x10-4) 0 30 4.29 (0.83, 14.39)
kpassive (x10-7)* 0 11 5.52 (0.8, 10.25)
Grassland (drained)
ffast (x10-2) 0 5 1.11 (0.14, 4.31)
fslow (x10-1) 0 10 7.49 (6.02, 9.02)
kfast (x10-3) 0 20 3.75 (0.53, 15.14)
kslow (x10-4) 0 30 8.55 (6.46, 11.27)
kpassive (x10-7)* 0 11 5.38 (0.75, 10.15)
Zackenberg
Supplementary figure S1: Location of the ZackenbergResearch Station in Northeast Greenland (Elberling et al. Long-term CO2 production following permafrost thawing).
Supplementary Figures:
Long-term CO2 production following permafrost thawing
Supplementary figure S1: Location of the ZackenbergResearch Station in Northeast Greenland (Elberling et al. Long-term CO2 production following permafrost thawing).
Supplementary Figures:
Long-term CO2 production following permafrost thawing
Supplementary figure S2: Zackenberg meteorological station and the ground moraine near the ZEROCALM 1 active layer thickness monitoring grid. The vegetation type is dominated by Cassiope tetragona (Elberling et al. Long-term CO2 production following permafrost thawing).
Supplementary figure S3: ZEROCALM 2 active layer monitoring grid in Zackenberg, illustrating the contrasting vegetation types from well-drained heath (Cassiope tetragona) to poorly drained wet grassland vegetation. The distinct vegetation zonation is controlled by the length of snow cover at this nivation site(Elberling et al. Long-term CO2 production following permafrost thawing).
Supplementary figure S4: Maximum active layer thickness (±SD) between 1997-2012 measured at the end of August at the wet grassland and the well-drained heath sites at Zackenberg. The marked area indicate the depth interval (50-100 cm) used to assessing long changes in soil organic carbon stock associated with increasing maximum active layer thickness (Elberling et al. Long-term CO2 production following permafrost thawing).
Supplementary figure S5: Active layer and permafrost characteristics in Zackenberg (Site 3A,B) in August 2008. Horizontal lines indicate the mean active layer depth (end of August 1997-2008) at wet grassland (solid) and heath sites (dashed). Data from the active layer are based on samples collected from pits, while data from the permafrost are based on intact cores. Modified from Elberling et al. High nitrous oxide production from thawing permafrost. Nature Geosci. 3, 332-335 (2010).
Supplementary figure S6: Soil organic carbon (SOC) stocks based on repeated sampling in 1996, 2008 and 2012 at two contrasting soil types (n=3) in Zackenberg calculated for 0 - 1 m and for 0.5 - 1 m depth intervals. The latter depth interval (see supplementary figure 4S) is used to evaluate changes in SOC stock related to the increasing thickness of the active layer and corresponding permafrost thawing (Elberling et al. Long-term CO2 production following permafrost thawing).
Supplementary figure S7: Longterm incubation of heath site top permafrost. A: basal soil respiration measured as CO2 being produced from intactcores (±SD), B: first 3 years of A and C: decreasing total dissolved C (DOC) and solid organic C (SOC).
Elberling et al. Long-term CO2 production following permafrost thawing.0
Supplementary figure S8: Longterm incubation of wet grassland top permafrost. A: basal soilrespiration measured as CO2 being producedfrom intact cores (±SD) , B: first 3 years of A and C: decreasing total dissolved C (DOC) and solid organic C (SOC).
Elberling et al. Long-term CO2 production following permafrost thawing.0
Supplementary figure S9: Longterm incubation of wet grassland top permafrost (drained). A: basal soil respiration measured as CO2 being produced from intact cores (±SD), B: first 3 years of A and C: decreasing total dissolved C (DOC) and solid organic C (SOC).
Elberling et al. Long-term CO2 production following permafrost thawing.0
Supplementary figure S10: Temperature-dependent CO2 production based on permafrost samples from Zackenberg measured three times during the long term incubation experiment (in 1997 shown as open squares, n=3; in 2003 shown as filled squares, n=1 and in 2008 (filled triangles, n=3). All data are normalized to 5 C and an exponential fit to 2003 data is shown as a dashed line (y = 0.6233e0.0903x, R2 = 0.9825, p< 0.05) indicating a Q10-value of 2.47 (Elberling et al. Long-term CO2 production following permafrost thawing).
Supplementary Figure S11: Potential C pool loss from top permafrost samples over a 50 year incubation period at 5°C for A) dry heath site B) wet grassland site and C) drained grassland site. The black lines depict mean potential C loss and the gray lines 95% confidence interval (Elberling et al. Long-term CO2production following permafrost thawing).