Sources of variation in simulated ecosystem carbon storage capacity from the 5th Climate Model Intercomparison Project (CMIP5)

Figures & data

Table 1. Full summary of land biogeochemistry models used by ESMs

Model name	Land model	Spatial resolution	Dynamic vegetation	Fire	N cycle	LUC
CanESM 2.2	CTEM/CLASS	2.85°×2.77°	N	Y	N	Y
CCSM4.2	CLM 4.0	1.25°×0.94°	N	Y	Y	Y
IPSL-CM5A-LR	ORCHIDEE	3.75°×1.89°	N	Y	N	Y
IPSL-CM5B-LR
MIROC-ESM	MATSIRO/SEIB-DGVM	2.81°×2.79°	Y	N	N	Y
MIROC-ESM-CHEM
NorESM1-M	CLM 4.0	2.5°×1.89°	N	Y	Y	Y
NorESM1-ME

Fig. 1 The spatial pattern of ecosystem C storage capacity (NPP*MRT_cor, kg C m⁻²) for the five models (modelled time: 1850–1860). Coefficient of variation (CV) was calculated for each grid cell using five models’ results.

Fig. 2 The relationship between mean residence time (MRT, years) and net primary production (NPP, kg C m⁻² yr⁻¹) and the latitude pattern of ecosystem C storage capacity (NPP*MRT_cor) for five models (modelled time: 1850–1860).

Fig. 3 The relationship between ecosystem C storage and soil C across models (a, b, c, d, e) and Taylor diagram for soil C storage at grid scale (f).

Fig. 4 Spatial pattern of net primary production (NPP, kg C m⁻² yr⁻¹) estimated from MODIS and the difference between five models and MODIS (modelled time: 1995–2005).

Fig. 5 Spatial pattern of the NPP/GPP ratio estimated from MODIS and the difference between five models and MODIS (modelled time: 1995–2005).

Fig. 6 Spatial pattern of average mean residence time (MRT, years) of ecosystem for five models and the difference between five models and the average of models (modelled time: 1850–1865).

Table 2. Effects of extreme data on ecosystem mean residence time (MRT) and the publication-based global average

Model	Mean_orig (years)	Percent* eliminated data	Mean_corrected* (years)	Reference
CanESM	36.4	0	36.4
CCSM	23.02	0.8	14.5
IPSL	33.41	0.4	21.3
MIROC	132.63	0.5	44.6
NorESM1	23.02	0.07	18.0
Based on soil respiration measurement			32	Raich and Schlesinger (1992)
Based on ^13C measurement			29	Randerson et al. (1999)
			44	Ciais et al. (1999)
Models			37–60	Meyer et al. (1999)

*A majority of MRTs at all grids for all models at the same spatial resolution are less than 400 yr, so we set the threshold of 1000 yr according as the hist. of data. The percentage of the eliminated data is not more than 1% (the valid data).

Fig. 7 The latitude pattern of NPP, the NPP/GPP ratio and mean residence time (MRT) for five models (each point representing the average over one latitudinal zone). The Data of MODIS was got from Zhang et al. (2009), which is the central tendency produced by average over 5° latitude.

Fig. 8 Spatial pattern of mean residence time (MRT) of ecosystem in the USA from inversed models (Zhou and Luo, 2008; Zhou et al., 2012) and the difference between five models and the estimates of inverse models.

Table 3. Comparison NPP, NPP/GPP and mean residence time (MRT) in the USA and Australia between model-based and publication-based

	NPP (kg C m^−2 yr^−1)
Model	1850–1860	1995–2005	NPP/GPP	MRT cor (year) (USA)	MRT cor (year) (Australia)	Reference
CanESM	0.37	0.42	0.53	34	26
CCSM	0.31	0.34	0.42	24	18
IPSL	0.47	0.54	0.52	29	9
MIROC	0.45	0.49	0.52	49	30
NorESM1	0.17	0.18	0.18	55	40
Inverse model				46	Zhou and Luo (2008)**
Inverse model				56.8±8.8	Zhou et al. (2012)
VAST1.1 with genetic algorithm					78	Barret (2002)**
MODIS-based	0.475 (±0.375) (2000–2005)					Zhao and Running (2010)
MODIS-based	0.55 (2000–2003)		0.5 (2000–2003)			Zhang et al. (2009)

**MRT of Zhou and Luo (2008) and Barrett (2002) was estimated with NPP, so as comparison, MRT is corrected by the NPP/GPP ratio in the USA and in Australia.

Fig. 9 Spatial pattern of coefficient of variation (CV) and standard deviation (SD) of NPP and MRT for all five models (modelled time: 1850–1860).

Table 4. Q₁₀ variance across models (function: )

Model	k (yr)	Q 10	R ^2
CanESM	19.4	1.83	0.61
CCSM	3.5	3.89	0.65
IPSL	13.7	1.43	0.59
MIROC	36.6	1.52	0.58
NorESM1	2.4	5.06	0.63

Raich J. W , Schlesinger W. H . The global carbondioxide flux in soil respiration and its relationship to vegetation and climate. Tellus B. 1992; 44: 81–99.

 Google Scholar

Randerson J. T , Thompson M. V , Field C. B . Linking C-13-based estimates of land and ocean sinks with predictions of carbon storage from CO2 fertilization of plant growth. Tellus. B. 1999; 51: 668–678.

 Google Scholar

Ciais P , Friedlingstein P , Schimel D. S , Tans P. P . A global calculation of the delta C-13 of soil respired carbon: implications for the biospheric uptake of anthropogenic CO2 . Glob. Biogeochem. Cycles. 1999; 13: 519–530.

 Web of Science ®Google Scholar

Meyer R , Joos F , Esser G , Heimann M , Hooss G , co-authors . The substitution of high-resolution terrestrial biosphere models and carbon sequestration in response to changing CO2 and climate. Global Biogeochem. Cycles. 1999; 13: 785–802.

 Web of Science ®Google Scholar

Zhang Y , Xu M , Chen H , Adams J . Global pattern of NPP to GPP ratio derived from MODIS data: effects of ecosystem type, geographical location and climate. Glob. Ecol. Biogeogr. 2009; 18: 280–290.

 Web of Science ®Google Scholar

Zhou T , Luo Y. Q . Spatial patterns of ecosystem carbon residence time and NPP-driven carbon uptake in the conterminous United States. Glob. Biogeochem. Cycles. 2008; 22: GB3032.

 Web of Science ®Google Scholar

Zhou X , Zhou T , Luo Y . Uncertainties in carbon residence time and NPP-driven carbon uptake in terrestrial ecosystems of the conterminous USA: a Bayesian approach. Tellus B. 2012; 64: 17223.

 Google Scholar

Barrett D. J . Steady state turnover time of carbon in the Australian terrestrial biosphere. Glob. Biogeochem. Cycles. 2002; 16(4): 1108.

 Web of Science ®Google Scholar

Zhao M, Running S. W. Drought-induced reduction in global terrestrial net primary production from 2000 through 2009. Science. 2010; 329: 940–943. [PubMed Abstract].

 Google Scholar