Max Planck Gesellschaft

--- Former member ---

Maarten Braakhekke

PhD student
room: C1.010 (tower)
phone: +49 3641 576266
fax: +49 3641 577200
email: Maarten.Braakhekke(at)bgc-jena.mpg.de

Contents

  1. Research interests
  2. Ph.D. thesis
  3. Peer-reviewed publications
  4. Selected oral presentations
  5. Other activities
  6. Cool graphs

Research interests

  • Biogeochemical and transport processes in soils and ecosystems
  • Dynamic model development
  • Bayesian parameter estimation

Ph.D. thesis

A modelling study on the formation of the vertical soil organic matter profile

The SOMPROF model
The SOMPROF model

In the soil physical, chemical, and biological properties that determine organic matter decomposition change rapidly along the vertical profile. At the same time the concentration and properties of organic matter itself are also highly depth dependent, with average organic carbon turnover rates often varying by several orders of magnitude within the top meter. It has therefore been suggested that the vertical distribution of organic matter constitutes an important factor for soil carbon cycling, which has thus far largely been ignored by modellers who focus usually on the top 30cm, where cycling is most active.

It is known that the vertical distribution of organic matter differs strongly between soil- and ecosystem types, and is mostly determined by biological mixing (bioturbation), downward movement of mobile organic matter fractions with water, and deposition due root turnover and exudation. However, the exact contribution of each of these processes is largely unknown, due to practical difficulties involved in measuring their rates.

During my PhD research I have developed a model structure that dynamically simulates the development of the soil organic matter profile both in the mineral soil and surface organic horizons. This model, called SOMPROF (Braakhekke et al., 2011), is based on simple but mechanistic representations of bioturbation, liquid phase transport and root input of organic matter.

In a subsequent study (Braakhekke et al., 2013), I estimated the parameters of the model for two temperate forest sites, based on in situ measurements. The calibration was performed in a Bayesian framework, allowing the incorporation of prior knowledge and the assessment of posterior parameter uncertainty. The results demonstrate the potential difficulties in such calibration exercises due to convolution of processes. This problem may be (partially) solved by including additional observations and prior knowledge. Furthermore, the results indicated strong differences between the two sites, in terms of the processes involved the vertical organic matter profile formation.

This Ph.D. project is a collaboration between the Max Planck Institute for Biogeochemistry, Jena and the Earth System Science group at Wageningen University, the Netherlands. It is funded by the European Research Council through the QUASOM project.

Supervisors: Christian Beer, Markus Reichstein?, Marcel Hoosbeek*, Bart Kruijt*, Pavel Kabat*
*) Wageningen University, Earth System Science Group, the Netherlands


Peer-reviewed publications

Braakhekke, M. C., C. Beer, M. Schrumpf, A. Ekici?, B. Ahrens, M.R. Hoosbeek, B. Kruijt, P. Kabat, and M. Reichstein? (2013) The use of radiocarbon to constrain current and future soil organic matter turnover and transport in a temperate forest Published online in Journal of Geophysical Research: Biogeosciences

Braakhekke, M. C., T. Wutzler?, C. Beer, J. Kattge, M. Schrumpf, B. Ahrens, I. Schöning, M.R. Hoosbeek, B. Kruijt, P. Kabat, and M. Reichstein? (2013) Modeling the vertical soil organic matter profile using Bayesian parameter estimation. Biogeosciences, 10: 399-420

PDF

Braakhekke, M. C., C. Beer, M. R. Hoosbeek, M. Reichstein?, B. Kruijt, M. Schrumpf, and P. Kabat. (2011) SOMPROF: A Vertically Explicit Soil Organic Matter Model. Ecological Modelling 222,10: 1712-1730

PDF (Preprint)

Mahecha, M. D., M. Reichstein?, M. Jung?, S. I. Seneviratne, S. Zaehle, C. Beer, M. C. Braakhekke, N. Carvalhais?, H. Lange, G. Le Maire, and E. Moors. (2010) Comparing Observations and Process-Based Simulations of Biosphere-Atmosphere Exchanges on Multiple Timescales. Journal of Geophysical Research-Biogeosciences 115

Weber, U.?, M. Jung?, M. Reichstein?, C. Beer, M. C. Braakhekke, V. Lehsten, D. Ghent, J. Kaduk, N. Viovy, P. Ciais, N. Gobron, and C. R. Odenbeck (2009) The Interannual Variability of Africa's Ecosystem Productivity: A Multi-Model Analysis. Biogeosciences 6, 2: 285-295

PDF

Selected oral presentations

Braakhekke, M. C., M. Reichstein, B. Kruijt, M. R. Hoosbeek, C. Beer, C. Reick, M. Heimann, and P. Kabat. 2009.Modelling the vertical profile of soil organic matter in terrestrial ecosystems. International Symposium on Soil Organic Matter Dynamics, Colorado Springs, Colorado, USA, July 6-9, 2009

Braakhekke, M. C., C. Beer, M. R. Hoosbeek, M. Reichstein, J. Kattge, T. Wutzler, B. Kruijt, M. Schrumpf, and P. Kabat. 2010. Mechanistic modeling of the vertical soil organic matter profile in terrestrial ecosystems. EGU General Assembly, Session Soil Organic Matter: structures, functions, management strategies, and C cycle, Vienna, Austria, May 2-7, 2010

Braakhekke, M. C., T. Wutzler, M. Reichstein, J. Kattge, C. Beer, M. Schrumpf, M. R. Hoosbeek, , B. Kruijt, and P. Kabat. 2011. Explaining the vertical SOM profile using Bayesian inversion and lead-210 measurements. EGU General Assembly, Session Soil organic carbon (SOC) dynamics at different spatial scales, Vienna, Austria, April 3-8, 2011

Braakhekke, M. C., T. Wutzler, M. Reichstein, J. Kattge, C. Beer, M. Schrumpf, M. R. Hoosbeek, I. Schöning, B. Kruijt, and P. Kabat. 2011. Modeling the vertical SOM profile using Bayesian inversion and Pb-210 measurements. International Symposium on Soil Organic Matter 2011, Leuven, Belgium, July 11-14, 2011


Other activities

Co-organized workshop: Dynamics of subsoil organic carbon in relation to soil properties, climate and biota at Eurosoil congress, Bari, Italy, 2-6 July, 2012

Co-organized round-table discussion: Current state of knowledge and potential for future experiments to help advance our understanding and model representation of soil organic carbon stabilization processes and dynamics at Eurosoil congress, Bari, Italy, 2-6 July, 2012


Cool graphs

Multi modal MCMC sample

Multimodal MCMC sample
This graph depicts a sample of a multimodal posterior distribution of three parameters of SOMPROF (decomposition rates of 3 organic matter pools). The sample was derived using the DREAM(ZS) algorithm developed by Jasper Vrugt et al. The many colored lines represent 20 Markov chains sampling the distribution and jumping back and forth between the three modes. The research behind this graph is further described in Braakhekke et al., 2013. Click here for a high quality PDF (4.2MB) of this graph.

SOMPROF results for the three modes of the graph above

Simulated carbon stocks and mass fractions by SOMPROF
The plots depict forward results of SOMPROF based on parameter values of the three modes in the graph above. The top graphs depict simulated and measured carbon stocks, and the bottom graphs simulated and measured carbon fractions in the mineral soil. The stacked colors of the model results are the different organic matter fractions in the model. The graph clearly indicate the equifinality of the model: the measurements can be reproduced equally will with three completely different scenarios. The research behind this graph is further described in Braakhekke et al., 2013.

Sample of 3 correlated parameters

Sample of 3 correlated parameters
This graph depicts a sample of three highly correlated parameters of SOMPROF. The sample was derived using the Markov Chain Monte Carlo approach. The colored lines on the axis planes are iso-probability lines of bivariate distributions, derived with kernel density estimation.

Correlation matrix of 13 SOMPROF parameters

Correlation matrix of 13 SOMPROF parameters
This graph depicts the correlation of 13 SOMPROF parameters estimated with Markov Chain Monte Carlo. In the top right triangle the linear correlations are shown, with blue indicating negative correlation and red positive correlation. In the bottom left triangle the bivariate density is depicted, which also reveals possible non-linear correlations. On the diagonal the univariate distributions of the parameters are shown. Compare this graph with the previous to find out for which 3 parameters the distribution is depicted.
Directions | Disclaimer | Data Protection | Contact | Internal | Webmail | Local weather | PRINT | © 2011-2019 Max Planck Institute for Biogeochemistry