|
Testing an Ensemble Kalman Filter for Assimilation of Soil Moisture
into HYDRUS 1D and Coupled Crop Model
Derek
G
Groenendyk, University of Arizona, derek.groenendyk@gmail.com
Ty
Ferre, University of Arizona, tyferre@gmail.com
Kelly
R
Thorp, USDA-ARS, kelly.thorp@ars.usda.gov
(Presenter)
Wade
Crow, USDA ARS HRSL, wade.crow@ars.usda.gov
HYDRUS 1D and other similar hydrologic models can estimate the soil water profile with reasonable accuracy. Uncertainty in soil texture parameters leads to error in water balance simulations, which can affect crop yield predictions when the hydrologic model is coupled with a crop growth model. Assimilation of soil moisture observations into these hydrologic models is a novel method for improving estimates of hydrologic variables and predicting crop yield. The Ensemble Kalman Filter (EnKF) is a computational procedure commonly implemented for these data assimilation problems. Our objective was to test an EnKF for soil moisture data assimilation into HYDRUS 1D coupled with a generic crop growth model. Measurements of in-situ soil moisture were used as the filter observations. These measurements were taken using neutron scattering probes twice weekly during an irrigated wheat cropping experiment conducted at Maricopa, Arizona. The generic crop model coupled to HYDRUS 1D was based on the plant growth module used in the WEPP model. This pairing provided a model with both a rigorous solution for the water balance as well as a simple crop model for yield prediction. Prior to applying the EnKF the crop growth parameters were manually calibrated using biomass, leaf area index, and yield data collected during the field experiment. Implementing the EnKF for the coupled models presented several challenges when modifying model structure and execution. For example, tracking time discretization and time steps was essential for convergence of the numerical solution to the water movement in HYDRUS. Evaluation of the EnKF was based on its ability to accurately estimate soil moisture. The effects of soil moisture assimilation on other simulated hydrological variables and crop yield were also assessed. Further research will explore the value and applicability of remote or in-situ soil moisture measurements for use in EnKF data assimilation algorithms with an overall goal of improving crop yield predictions.
Presentation:
2013_Poster_Groenendyk_80_10.pdf (2843k)
Presentation Type: Poster
Session: Poster Session 2-B
(Wed 4:30 PM)
Associated Project(s):
Poster Location ID: 80
|