Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Hydrology and Remote Sensing Laboratory » Research » Publications at this Location » Publication #316678
Title: Impact of model relative accuracy in framework of rescaling observations in hydrological data assimilation studies

Author
item YILMAZ, M.T. - Middle East Technical University
item Crow, Wade
item RYU, D. - University Of Melbourne

Submitted to: Water Resources Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/6/2016
Publication Date: 8/1/2016
Citation: Yilmaz, M., Crow, W.T., Ryu, D. 2016. Impact of model relative accuracy in framework of rescaling observations in hydrological data assimilation studies. Water Resources Research. 17:2245–2257.

Interpretive Summary: Increasingly, soil moisture estimates acquired via modeling and remote sensing are being applied to operationally monitor the extent and severity of agricultural droughts. Better monitoring of existing drought conditions would aid in a number of key agricultural applications including: drought forecasting, irrigation scheduling, and optimizing fertilizer usage. However various water balance modeling and remote sensing approaches can vary significantly with regards to their soil moisture estimates. As a result, new statistical tools are needed to combine soil moisture estimates obtained from a variety of sources into single optimized prediction of soil moisture availability. This paper derives and evaluates mathematical approaches designed to merge multi-source soil moisture datasets into a single unified dataset. These techniques can potentially be used by operational drought monitors to enhance the quality of root-zone soil moisture estimates used in agricultural decision support systems.

Technical Abstract: Soil moisture datasets (e.g. satellite-, model-, station-based) vary greatly with respect to their signal, noise, and/or combined time-series variability. Minimizing differences in signal variances is particularly important in data assimilation techniques to optimize the accuracy of the analysis obtained after merging model and observation datasets. Strategies that reduce these differences are typically based on rescaling the observation time series to match the model, while the impact of the relative accuracy of the reference dataset (i.e. model) component is often neglected. In this study the impacts of the relative accuracies of seasonality and anomaly components of modeled and observation based soil moisture time series are investigated. Experiments are performed both using a well-controlled synthetic and real dataset testbeds. Investigated experiments are based on rescaling observations to a model using strategies with varying aggressiveness: rescaling the entire time-series as one-piece or each month separately, rescaling the seasonality and the anomaly components separately, and inserting the seasonality of the model directly while matching the anomaly component only. Both synthetic and real data assimilation experiments use the simple Antecedent Precipitation Index (API) model and assimilate observations in the Kalman filter framework. For the real data case, the Land Parameter Retrieval Model (LPRM) product based on The Advanced Microwave Scanning Radiometer on the Aqua platform (AMSR-E) observations over four USDA-ARS watersheds are assimilated and the ground-based observations are used to validate the analysis. Results of both synthetic and real assimilation experiments show rescaling observations more aggressively to the model is favorable when the model is more skillful than observations, while rescaling observations strongly to the model degrades the analysis if observations are relatively more accurate.