『Abstract
　Groundwater contaminant transport models are being used increasingly to simulate the fate and transport of reactive solutes, particularly nitrate, within aquifer systems. These models, however, often are hindered by lack of information regarding parameters, such as kinetic decay rates, that govern the subsistence of the solute within the groundwater. In an overall effort to provide accurate estimates of spatially-variable parameters in numerical reactive transport modeling we employ a data assimilation scheme, the Ensemble Smoother (ES), which yields improved estimates of spatially-variable denitrification rates within an irrigated agricultural river-aquifer system using measurements of (i) nitrate concentration in the groundwater and (ii) mass of nitrate entering the river from the aquifer via groundwater flows. Based on the Kalman Filter methodology, in which distributed uncertain model results are corrected by assimilating measurement data from a reference system, the ES incorporates uncertain parameter values, associated model results, and measurement data into an update algorithm to provide an updated, corrected model state that approaches the reference system state. As an important step in eventually employing the methodology to real-world systems, this study evaluates the parameter estimation scheme for a synthetic aquifer system approaching hydrologic (heterogeneous hydraulic conductivity, cropping patterns, irrigation recharge, canal seepage) and chemical (denitrification, leaching concentrations) complexities expected in aquifers influenced by agricultural practices. Sensitivity analyses are conducted to investigate the influence of (i) the number of measurement data assimilated and (ii) the error assigned to the measurement data. Results indicate that the spatial distribution of denitrification rates can be estimated to a satisfying degree, and when implemented in additional model runs produce (i) simulated values that coincide favorably with measurement values from the reference state, and (ii) spatial distribution of nitrate concentration comparable to that of the reference state.

Keywords: Denitrification; Ensemble Smoother; Agricultural groundwater system,』

1. Introduction
2. Estimation of aquifer parameters using the Ensemble Smoother
　2.1. General forecast step
　2.2. General update step for system-response variables
　2.3. Coupled update of system-response variables and system parameters
　2.4. Forecast and update within the ES framework
　2.5. Evaluating uncertainty in the updated system states
3. Flow and reaction transport simulations and estimation of YλHet
　3.1. Conceptual model of aquifer system
　3.2. Reference state and measurement collection
　3.3. Forecasted ensemble of model states
　3.4. Updated ensembles and sensitivity analysis
　　3.4.1. Conditioning YλHet using CNO3 data
　　3.4.2. Ensemble of YCNO3 using updated YλHet ensemble
　　3.4.3. Conditioning YλHet using RM data
　　3.4.4. Conditioning YλHet in aquifers of uncertain reactivity
4. Discussions
Acknowledgements
References