Obtaining soil hydraulic parameters from data assimilation and calibration under different climatic/soil conditions

Authors: VALDES-ABELLAN, JAVIER - Universidad De Alicante; Pachepsky, Yakov; MARINEZ, GONZALO - Universidad De Cordoba

Submitted to: Catena
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/20/2017
Publication Date: 1/16/2018
Citation: Valdes-Abellan, J., Pachepsky, Y.A., Marinez, G. 2018. Obtaining soil hydraulic parameters from data assimilation and calibration under different climatic/soil conditions. Catena. 163:311-320. https://doi.org/10.1016/j.catena.2017.12.022.
DOI: https://doi.org/10.1016/j.catena.2017.12.022

Interpretive Summary: Hydraulic parameters of soils are of paramount importance in projects related to ability of soils to retain and conduct water. A traditional way of finding these parameters is to calibrate water flow model, that is, to find parameter values allowing to this model to reproduce a long term monitoring dataset with the highest possible accuracy. Recently it was shown that instead of waiting until a long term dataset will be accumulated, one can gradually improve the parameter set each time as new observations become available. This methodology became known as data assimilation. The purpose of this work was to compare the calibration and data assimilation methodologies for a the range of soils and climates. We demonstrated that data assimilation provides substantially better results in terms of time needed to reach stable parameter values and closeness of parameter values to true ones. This work useful for researchers and modelers-practitioners in the field of soil hydrology in that it substantiates a new powerful method of obtaining much needed input for modeling projects.

Technical Abstract: Obtaining reliable soil hydraulic properties is essential to correctly simulating soil water content (SWC), which is a key component of countless applications such as agricultural management, soil remediation, aquifer protection, etc. Soil hydraulic properties can be measured in the laboratory; however, the procedures are laborious and costly, and may provide estimates different from those observed in the field. An alternative approach is to obtain soil hydraulic properties using a soil water flow model in conjunction with SWC monitoring data. The goal of the present study was to analyze the efficiency of obtaining hydraulic properties from standard calibration based on inverse modeling (SC), as compared with the joint soil water state utilizing data assimilation (DA) based on the Ensemble Kalman filter method. Two soil textures and four climatic conditions were considered. Our results demonstrate the advantages of data assimilation using the Richards equation soil model, when realistic weather conditions are applied. When observed SWC did not show a broad range of values, both data assimilation and inverse modeling provided sets of properties that led to good Richards model performance, with the RMSE below 0.1 and/or r2 above 0.8 after a period of 100 days. However, DA goodness-of-fit statistics from data assimilation were clearly better than inverse modeling more than 95 percent of the time. Based on the soil and climatic conditions used for simulations, one year was adequate to obtain reliable soil properties with data assimilation but, in general, not with inverse modeling.