Stochastic analysis of terrestrial systems
This research topic focuses on the merging of terrestrial model predictions and terrestrial measurement data, using inverse modelling and data assimilation methods. The aim of these model-data fusion methods is:
- To get better short- and medium term predictions of all possible states of the terrestrial system and fluxes within the terrestrial system. Prominent examples are predictions of soil moisture content (which are for example important as lower boundary condition of atmospheric models and for predictions of agricultural production) and floods.
- To improve parameter estimates in high resolution terrestrial system models, which allows the improvement of long-term predictions with terrestrial system models.
- To confront model predictions and measurement data in a systematic manner, to detect model deficiencies and therefore the potential to improve the representation of model processes.
- To better exploit the value of measurement data. Examples are: (i) Coupling a measurement operator to terrestrial system models; (ii) Exploiting data in cross-compartmental data assimilation, which means that data collected in a certain terrestrial system compartment (e.g., soil moisture) are used to update states of other terrestrial system compartments (e.g., air temperature in lower atmosphere).
- To link short and medium term predictions with near real-time control of water resources management.
In the context of such methods, it is important to characterize model prediction uncertainty and measurement uncertainty correctly. We investigate therefore also how uncertainty of model predictions can be characterized better (for example: modelling of spatial variability of hydraulic conductivity in aquifers and soils) and analyze systematic and random errors in measurement data (for example: eddy covariance data, lysimeter data, soil moisture measurements at different scales).
Ultimately, besides improving the predictive capacity of terrestrial system models on all temporal scales, we want to increase the understanding of complex interactions between different components of the terrestrial system (subsurface, land surface, atmosphere) at intermediate and large spatial scales.
This work is often carried out with the model TerrSysMP (terrestrial systems modeling platform), or other, physics based models for specific terrestrial compartments. We are part of the High Performance Scientific Computing in Terrestrial Systems Centre.
In many studies data from Helmholtz monitoring infrastructures like TERENO, SoilCan or ICOS are used. It is planned to work in the future with data collected in MOSES campaigns.
Currently, the work in this research topic is organized in the following research projects:
- Real-time optimization of irrigation of citrus fields near Valencia
- Data assimilation with TerrSysMP-PDAF
- River-aquifer interaction and the role of streambeds
- Inverse groundwater flow modelling
- Upscaling of biogeochemical fluxes to characterize regional carbon balance
- Long-term analysis of trends in actual evapotranspiration measured by lysimeters
These research projects are linked to Helmholtz projects:
REKLIM
MOSES
EDA (finished project)
DFG-funded projects:
SFB TR32
FOR2131
two other DFG-projects.
EU-funded projects:
eLTER
EoCoE
EXPEER (finished project)
AGADAPT (finished project)
Contact
Prof. Dr. Harrie-Jan Hendricks-Franssen
Head of research group "Stochastic Analysis Terrestrial Systems"
- Institut für Bio- und Geowissenschaften (IBG)
- Agrosphäre (IBG-3)
Raum R 3033