|Valdes-abellan, Javier - Universidad De Alicante|
|Martinez, Gonzalo - Universidad De Cordoba|
|Pla, Concepcion - Universidad De Alicante|
Submitted to: Vadose Zone Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/17/2019
Publication Date: 4/1/2019
Citation: Valdes-Abellan, J., Pachepsky, Y.A., Martinez, G., Pla, C. 2019. How critical is the frequency of water content measurements for obtaining soil hydraulic properties with data assimilation?. Vadose Zone Journal. 18(1). https://doi.okrg/10.2136/vzj2018.07.0142.
DOI: https://doi.org/10.2136/vzj2018.07.0142 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 values each time as new observations become available. This methodology of updating parameter values became known as data assimilation. It has not been known how the frequency of updates influences the efficiency of the data assimilation. We demonstrated that the optimal frequency is site-specific and depends on soil and climate. Low, larger than week update frequencies were more effective than the high, less than week frequencies in majority of cases we considered. This work is useful for researchers and modelers-practitioners in the field of soil hydrology in that it reveals the role of update frequency as the additional control of data assimilation that has to be optimized to improve the efficiency of the new powerful method of obtaining much needed input for modeling projects.
Technical Abstract: Data assimilation (DA) is a promising alternative to infer soil hydraulic properties from soil water monitoring data. Frequency of measurements and corresponding updates is one of the important controls of the DA efficiency; however, no strict guidance exists on determining the optimal frequency to obtain the most accurate results. In this study, DA is performed with the Ensemble Kalman Filter with a state augmentation approach to update both model states and parameters (i.e., soil water contents and soil hydraulic properties). We analyze results of the data assimilation from synthetic data sets with update frequencies of 1, 2, 3, 5, 7, 9, 11 and 14 days. The synthetic dataset is developed considering one-layer profile, two soil types and four climates. The DA with high frequency (less than 7 days) did not provide better results than those obtained when using low frequencies (greater than 7 day). Moreover, for majority of the studied climate/soil scenarios, assimilation of observed data every 7 or more days yielded better results. Update frequency appears to be a parameter of the data assimilation process that can be optimized. Relatively low frequency may provide satisfactory results. Site specific research with synthetic data may be useful for designing sol water monitoring for data assimilation.