Submitted to: Soil Science Society of America Journal
Publication Type: Peer reviewed journal
Publication Acceptance Date: 2/1/2011
Publication Date: 5/12/2012
Publication URL: handle.nal.usda.gov/10113/56581
Citation: Pan, F., Pachepsky, Y.A., Jacques, D., Guber, A.K., Hil, R. 2012. Data assimilation with soil water content sensors and pedotransfer functions in soil water flow modeling. Soil Science Society of America Journal. 76:829-844. Interpretive Summary: Soil water modeling is widely used in hydrological, meteorological, agronomical, and many other types of studies and applications. It is well known that soil water models are far from perfect and, therefore, correction of modeling results is needed. One possibility is to make corrections “on the go” using soil moisture sensors. A variety of sensors are available that measure soil moisture at various depths, including on the surface (remote sensing), in the root zone (tensiometers), or in groundwater (diverse methods), raising the issue of which measurements are most reliable. We have demonstrated that it is possible to correct soil water modeling results for the whole soil profile using measurements from only one, or in some instances, two depths. Results of this work are expected to be very useful for applications in environmental modeling in that they show the opportunity of the efficient use of soil moisture monitoring to improve soil water predictions.
Technical Abstract: Soil water flow models are based on a set of simplified assumptions about the mechanisms, processes, and parameters of water retention and flow. That causes errors in soil water flow model predictions. Soil water content monitoring data can be used to reduce the errors in models. Data assimilation (DA) with the ensemble Kalman filter makes it feasible to correct the modeling results based on the information on uncertainty in data and uncertainty in modeling results. The objectives of this study was to evaluate the efficiency of the soil water content sensor data assimilation in soil water flow model using various PTFs to create the ensemble of models for the ensemble Kalman filter application. Sixty two-rod time domain reflectometry (TDR) probes at five depths were installed in the loamy soils of the experimental field at Bekkevoort, Belgium to monitor the soil water content for 384 days. The ensemble of 24 models was developed with 6 water retention PTFs and 4 Ksat PTFs developed from the large database. The data assimilation with measurements at all five depths, one or two depths were simulated to evaluate the efficiency of soil water content sensors to correct the soil water flow model results for weekly and biweekly updates. The assimilating measurements at a single depth provided substantial improvement in the simulations at other observation depths. The soil water content sensors at the top of the profile gave the best assimilation results based on the root-mean-square error (RMSE) between the measured and simulated soil water content at the beginning of each day. The results of this study indicate the data assimilation can be a powerful upscaling method in soil hydrology.