Using atmospheric carbon dioxide in evaluating global soil carbon model

The soil carbon is an important part of the global carbon cycle. The carbon storage in soil is larger than that of the atmosphere and as the soil carbon is potentially released to the atmosphere during warming, the increased carbon dioxide in atmosphere will further accelerate the climate change. Therefore it is important to improve the description of the soil carbon in global biosphere models, to ensure that the future predictions of the soil carbon’s fate are as good as possible.

Comparison of soil carbon models at global scale is complicated. The amount of carbon in soil can be compared to global products based on measurements. The estimation of the carbon dioxide released to atmosphere from the soil carbon is more challenging. The atmospheric carbon dioxide concentration consists of signals from photosynthesis and breathing by the vegetation, anthropogenic emissions and ocean fluxes. Therefore it is challenging to estimate the emissions from the soil carbon, since their contribution is only one part of the whole signal.

Comparison revealed differences between models

In this study we used a global biosphere model JSBACH, which included two different soil carbon models. Another one was the original soil carbon model of JSBACH and the other one was the Yasso model developed in Finland. The soil carbon models used different input information (soil temperature vs. air temperature and soil moisture vs. precipitation) and they also had different response functions to these environmental variables. Since the soil carbon models did not have feedback with the other parts of the biosphere model, the differences in the JSBACH model versions were solely caused by the soil carbon models.

From the JSBACH model simulations we obtained information on the terrestrial fluxes and sources of carbon. We passed on this information to an atmospheric transfer model, as well as information on ocean surface fluxes and human-induced emissions. The three-dimensional atmospheric model transmitted and mixed these different carbon dioxide concentrations in the atmosphere and we were able to compare the model’s results with the surface and satellite concentrations of carbon dioxide.

The two different soil carbon models resulted in different seasonal cycles of the carbon dioxide. The Yasso soil model replicated better the seasonal cycle observations in the northern regions, while the original soil model of JSBACH performed better in the tropics and temperature regions. A more detailed analysis revealed, that the response functions of the Yasso model are not optimal for estimating the seasonal cycle in the tropics, even though it estimated more realistic soil carbon distribution globally than the other model.

This method showed that the developed evaluation method is successful and that it can be used also in future. In addition we used both surface and satellite observations. The satellite observations provide information on larger regions, covering areas that do not necessarily have surface observation network. The observation time series from satellites are now long enough that they can be used in carbon cycle research.

Several people from the Finnish Meteorological Institute participated in the study and data from the ICOS station at Pallas was used. In addition, researchers from Germany, the Netherlands, Japan and the United States were involved. The study was part of the postdoctoral fellowship of Tea Thum funded by the Academy of Finland.

Further information:

Researcher Tea Thum, phone 050 592 7359, tea.thum@fmi.fi.

Thum, T., Nabel, J. E. M. S., Tsuruta, A., Aalto, T., Dlugokencky, E. J., Liski, J., Luijkx, I. T., Markkanen, T., Pongratz, J., Yoshida, Y., and Zaehle, S.: Evaluating two soil carbon models within the global land surface model JSBACH using surface and spaceborne observations of atmospheric CO2, Biogeosciences, 17, 5721–5743, https://doi.org/10.5194/bg-17-5721-2020, 2020.