Process-based modeling of regional water and energy fluxes taking into account multisensor and multiscale observation patterns

Research Area:
Land surface hydrology, data assimilation

Predictions with land surface models are affected by multiple sources of error. Data assimilation techniques can improve these predictions and reduce their uncertainty. Data assimilation is a mathematical-statistical approach to optimally combine measurement data and model predictions. This project focuses especially on soil moisture data to improve predictions with the land surface model CLM. Data assimilation is in this project often done with the Local Ensemble Transform Kalman Filter (LETKF), but also other algorithms were tested.

Soil moisture content is measured very locally (e.g, TDR sensors) and at very large scales of about 40km (e.g., remote sensing satellites like SMOS). The cosmic ray probe measures above-ground neutron counts (which are closely related to soil moisture content at an intermediate scale with a diameter of 600m). In this project a regional network (Rur catchment) of 10 cosmic ray probes was constructed. Soil moisture content could be successfully retrieved from these probes, but it was shown that corrections for a litter layer (in forests) and time variable above-ground biomass (in croplands) have to be made. Different observation operators that link neutron counts and soil moisture content were evaluated and compared with exhaustive data from soil sensor networks. The COSMIC operator was implemented for data assimilation and it was found that the assimilation of neutron counts improves the characterization of states and fluxes of CLM, both in a synthetic case and the real-world case. It was however found that it is important to update soil parameters jointly with the states. In the last year research focuses on the joint assimilation of locally measured soil moisture by TDR-sensors and cosmic ray probes, and soil moisture from satellites. A multi-scale multivariate assimilation algorithm is implemented in combination with the model ParFlow-CLM.

Cooperation partner:

• Prof. Jasper Vrugt (University of Irvine)
• Prof. Hamid Moradkhani (University of Portland)
• Prof. Xin Li (Cold and Arid Regions Environmental and Engineering Research Institute, Lanzhou, China)