Prediction and predictability of a catastrophic local extreme precipitation event through cloud-resolving ensemble analysis and forecasting with Doppler radar observations

Local extreme rain usually resulted in disasters such as flash floods and landslides. Upon today, it is still one of the most difficult tasks for operational weather forecast centers to predict those events accurately. In this paper, we simulate an extreme precipitation event with ensemble Kalman filter (EnKF) assimilation of Doppler radial-velocity observations, and analyze the uncertainties of the assimilation. The results demonstrate that, without assimilation radar data, neither a single initialization of deterministic forecast nor an ensemble forecast with adding perturbations or multiple physical parameterizations can predict the location of strong precipitation. However, forecast was significantly improved with assimilation of radar data, especially the location of the precipitation. The direct cause of the improvement is the buildup of a deep mesoscale convection system with EnKF assimilation of radar data. Under a large scale background favorable for mesoscale convection, efficient perturbations of upstream mid-low level meridional wind and moisture are key factors for the assimilation and forecast. Uncertainty still exists for the forecast of this case due to its limited predictability. Both the difference of large scale initial fields and the difference of analysis obtained from EnKF assimilation due to small amplitude of initial perturbations could have critical influences to the event's prediction. Forecast could be improved through more cycles of EnKF assimilation. Sensitivity tests also support that more accurate forecasts are expected through improving numerical models and observations.     

Handling non-linearity in radar data assimilation using the non-linear least squares enhanced POD-4DVar

The Proper Orthogonal Decomposition(POD)-based ensemble four-dimensional variational(4DVar) assimilation method(POD4DEnVar) was proposed to combine the strengths of EnKF(i.e.,the ensemble Kalman filter) and 4DVar assimilation methods.Recently,a POD4DEnVar-based radar data assimilation scheme(PRAS) was built and its effectiveness was demonstrated.POD4 DEnVar is based on the assumption of a linear relationship between the model perturbations(MPs)and the observation perturbations(OPs);however,this assumption is likely to be destroyed by the highly non-linear forecast model or observation operator.To address this issue,using the Gauss-Newton iterative method,the nonlinear least squares enhanced POD4 DEnVar algorithm(referred to as NLS-4DVar) was proposed.Naturally,the PRAS was upgraded to form the NLS-4DVar-based radar data assimilation scheme(NRAS).To evaluate the performance of NRAS against PRAS,observing system simulation experiments(OSSEs) were conducted to assimilate reflectivity and radial velocity individually,with one,two,and three iterations.The results demonstrated that the NRAS outperformed PRAS in improving the initial condition and the forecasting of model variables and rainfall.The NRAS,with a smaller number of iterations,can yield a convergent result.In contrast to the situation when assimilating radial velocity,the advantages of NRAS over PRAS were more obvious when assimilating reflectivity.     

Relative performance of future altimeter systems and tide gauges in constraining a model of North Sea high-frequency barotropic dynamics

We evaluate in this paper the ability of several altimeter systems, considered separately as well as together with tide gauges, to control the time evolution of a barotropic model of the North Sea shelf. This evaluation is performed in the framework of the particular model errors due to uncertainties in bathymetry. An Ensemble Kalman Filter (EnKF) data assimilation approach is adopted, and observing-systems simulation experiments (OSSEs) are carried out using ensemble spread statistics. The skill criterion for the comparison of observing networks is, therefore, not based on the misfit between two simulations, as done in classic twin experiments, but on the reduction of ensemble variance occurring as a consequence of the assimilation. Future altimeter systems, such as the Wide Swath Ocean Altimeter (WSOA) and satellite constellations, are considered in this work. A single WSOA exhibits, for instance, similar performance as two-nadir satellites in terms of sea-level correction, and is better than three satellites in terms of model velocity control. Generally speaking, the temporal resolution of observations is shown to be of major importance for controlling the model error in these experiments. This result is clearly related to the focus adopted in this study on the specific high-frequency response of the ocean to meteorological forcing. Altimeter systems lack adequate temporal sampling for properly correcting the major part of model error in this context, whereas tide gauges, which provide a much finer time resolution, lead to better global statistical performance. When looking into further detail, tide gauges and altimetry are demonstrated to exhibit an interesting complementary character over the whole shelf, as tide gauge networks make it possible to properly control model error in a ∼100-km coastal band, while high-resolution altimeter systems are more efficient farther from the coast.     

Initial ensemble generation and validation for ocean data assimilation using HYCOM in the Pacific

A method to initialize an ensemble, introduced by Evensen (Physica, D 77:108–129, 1994a; J Geophys Res 99(C5):10143–10162, 1994b; Ocean Dynamics 53:343–367, 2003), was applied to the Ocean General Circulation Model (OGCM) HYbrid Coordinate Ocean Model (HYCOM) for the Pacific Ocean. Taking advantage of the hybrid coordinates, an initial ensemble is created by first perturbing the layer interfaces and then running the model for a spin-up period of 1 month forced by randomly perturbed atmospheric forcing fields. In addition to the perturbations of layer interfaces, we implemented perturbations of the mixed layer temperatures. In this paper, we investigate the quality of the initial ensemble generated by this scheme and the influence of the horizontal decorrelation scale and vertical correlation on the statistics of the resulting ensemble. We performed six ensemble generation experiments with different combinations of horizontal decorrelation scales and with/without perturbations in the mixed layer. The resulting six sets of initial ensembles are then analyzed in terms of sustainability of the ensemble spread and realism of the correlation patterns. The ensemble spreads are validated against the difference between model and observations after 20 years of free run. The correlation patterns of six sets of ensemble are compared to each other. This study shows that the ensemble generation scheme can effectively generate an initial ensemble whose spread is consistent with the observed errors. The correlation pattern of the ensemble also exhibits realistic features. The addition of mixed layer perturbations improves both the spread and correlation. Some limitations of the ensemble generation scheme are also discussed. We found that the vertical shift of isopycnal coordinates provokes unrealistically large deviations in shallow layers near the islands of the West Pacific. A simple correction circumvents the problem.

Sampling strategies and square root analysis schemes for the EnKF

The purpose of this paper is to examine how different sampling strategies and implementations of the analysis scheme influence the quality of the results in the EnKF. It is shown that by selecting the initial ensemble, the model noise and the measurement perturbations wisely, it is possible to achieve a significant improvement in the EnKF results, using the same number of members in the ensemble. The results are also compared with a square root implementation of the EnKF analysis scheme where the analyzed ensemble is computed without the perturbation of measurements. It is shown that the measurement perturbations introduce sampling errors which can be reduced using improved sampling schemes in the standard EnKF or fully eliminated when the square root analysis algorithm is used. Further, a new computationally efficient square root algorithm is proposed which allows for the use of a low-rank representation of the measurement error covariance matrix. It is shown that this algorithm in fact solves the full problem at a low cost without introducing any new approximations.     

Application of data assimilation with the Root Zone Water Quality Model for soil moisture profile estimation in the upper Cedar Creek,Indiana

Data assimilation techniques have been proven as an effective tool to improve model forecasts by combining information about observed variables in many areas. This article examines the potential of assimilating surface soil moisture observations into a field‐scale hydrological model, the Root Zone Water Quality Model, to improve soil moisture estimation. The Ensemble Kalman Filter (EnKF), a popular data assimilation technique for nonlinear systems, was applied and compared with a simple direct insertion method. In situ soil moisture data at four different depths (5, 20, 40, and 60 cm) from two agricultural fields (AS1 and AS2) in northeastern Indiana were used for assimilation and validation purposes. Through daily update, the EnKF improved soil moisture estimation compared with the direct insertion method and model results without assimilation, having more distinct improvement at the 5 and 20 cm depths than for deeper layers (40 and 60 cm). Local vertical soil property heterogeneity in AS1 deteriorated soil moisture estimates with the EnKF. Removal of systematic bias in the forecast model was found to be critical for more successful soil moisture data assimilation studies. This study also demonstrates that a more frequent update generally contributes in enhancing the open loop simulation; however, large forecasting error can prevent more frequent update from providing better results. In addition, results indicate that various ensemble sizes make little difference in the assimilation results. An ensemble of 100 members produced results that were comparable with results obtained from larger ensembles. Copyright © 2011 John Wiley & Sons, Ltd.     

Data assimilation with the EnKF in a 1-D numerical model of a North Sea station

A series of numerical experiments for data assimilation with the Ensemble Kalman Filter (EnKF) in a shallow water model are reported. Temperature profiles measured at a North Sea location, 55°30ˊ North and 0°55ˊ East (referred to as the CS station of the NERC North Sea project), are assimilated in 1-D simulations. Comparison of simulations without assimilation to model results obtained when assimilating data with the EnKF allows us to assess the filter performance in reproducing features of the observations not accounted for by the model. The quality of the model error sampling is tested as well as the validity of the Gaussian hypothesis underlying the analysis scheme of the EnKF. The influence of the model error parameters and the frequency of the data assimilation are investigated and discussed. From these experiments, a set of optimal parameters for the model error sampling are deduced and used to test the behavior of the EnKF when propagating surface information into the water column.     

Data assimilation for distributed hydrological catchment modeling via ensemble Kalman filter

Catchment scale hydrological models are critical decision support tools for water resources management and environment remediation. However, the reliability of hydrological models is inevitably affected by limited measurements and imperfect models. Data assimilation techniques combine complementary information from measurements and models to enhance the model reliability and reduce predictive uncertainties. As a sequential data assimilation technique, the ensemble Kalman filter (EnKF) has been extensively studied in the earth sciences for assimilating in-situ measurements and remote sensing data. Although the EnKF has been demonstrated in land surface data assimilations, there are no systematic studies to investigate its performance in distributed modeling with high dimensional states and parameters. In this paper, we present an assessment on the EnKF with state augmentation for combined state-parameter estimation on the basis of a physical-based hydrological model, Soil and Water Assessment Tool (SWAT). Through synthetic simulation experiments, the capability of the EnKF is demonstrated by assimilating the runoff and other measurements, and its sensitivities are analyzed with respect to the error specification, the initial realization and the ensemble size. It is found that the EnKF provides an efficient approach for obtaining a set of acceptable model parameters and satisfactory runoff, soil water content and evapotranspiration estimations. The EnKF performance could be improved after augmenting with other complementary data, such as soil water content and evapotranspiration from remote sensing retrieval. Sensitivity studies demonstrate the importance of consistent error specification and the potential with small ensemble size in the data assimilation system.     

Coupled assimilation for an intermediated coupled ENSO prediction model

The value of coupled assimilation is discussed using an intermediate coupled model in which the wind stress is the only atmospheric state which is slavery to model sea surface temperature (SST). In the coupled assimilation analysis, based on the coupled wind–ocean state covariance calculated from the coupled state ensemble, the ocean state is adjusted by assimilating wind data using the ensemble Kalman filter. As revealed by a series of assimilation experiments using simulated observations, the coupled assimilation of wind observations yields better results than the assimilation of SST observations. Specifically, the coupled assimilation of wind observations can help to improve the accuracy of the surface and subsurface currents because the correlation between the wind and ocean currents is stronger than that between SST and ocean currents in the equatorial Pacific. Thus, the coupled assimilation of wind data can decrease the initial condition errors in the surface/subsurface currents that can significantly contribute to SST forecast errors. The value of the coupled assimilation of wind observations is further demonstrated by comparing the prediction skills of three 12-year (1997–2008) hindcast experiments initialized by the ocean-only assimilation scheme that assimilates SST observations, the coupled assimilation scheme that assimilates wind observations, and a nudging scheme that nudges the observed wind stress data, respectively. The prediction skills of two assimilation schemes are significantly better than those of the nudging scheme. The prediction skills of assimilating wind observations are better than assimilating SST observations. Assimilating wind observations for the 2007/2008 La Niña event triggers better predictions, while assimilating SST observations fails to provide an early warning for that event.     

Impacts of localisation in the EnKF and EnOI: experiments with a small model

The performance of an inexpensive, ensemble-based optimal interpolation (EnOI) scheme that uses a stationary ensemble of model anomalies to approximate forecast error covariances, is compared with that of an ensemble Kalman filter (EnKF). The model to which the methods are applied is a pair of “perfect”, one-dimensional, linear advection equations for two related variables. While EnOI is sub-optimal, it can give results that are comparable to those of the EnKF. The computational cost of EnOI is typically about times less than that of EnKF, where is the ensemble size. We suggest that EnOI may provide a practical and cost-effective alternative to the EnKF for some applications where computational cost is a limiting factor. We demonstrate that when the ensemble size is smaller than the dimension of the model’s sub-space, both the EnKF and EnOI may require localisation around each observation to eliminate effects of sampling error and to increase the effective number of independent ensemble members used to construct an analysis. However, localisation can degrade an analysis if the length-scales of the localising function are too short. We demonstrate that, as the length-scale of the localising function is decreased, localisation can significantly compromise the model’s dynamical balances. We also find that localisation artificially amplifies high frequencies for applications of the EnKF. Based on our experiments, for applications where localisation is necessary, the length-scales of the localisation should be larger than the decorrelation length-scales of the variables being updated.     

Application of data assimilation for improving forecast of water levels and residual currents in Singapore regional waters

Hydrodynamic models are commonly used for predicting water levels and currents in the deep ocean, ocean margins and shelf seas. Their accuracy is typically limited by factors, such as the complexity of the coastal geometry and bathymetry, plus the uncertainty in the flow forcing (deep ocean tide, winds and pressure). In Southeast Asian waters with its strongly hydrodynamic characteristics, the lack of detailed marine observations (bathymetry and tides) for model validation is an additional factor limiting flow representation. This paper deals with the application of ensemble Kalman filter (EnKF)-based data assimilation with the purpose of improving the deterministic model forecast. The efficacy of the EnKF is analysed via a twin experiment conducted with the 2D barotropic Singapore regional model. The results show that the applied data assimilation can improve the forecasts significantly in this complex flow regime.     

An adaptive hybrid EnKF-OI scheme for efficient state-parameter estimation of reactive contaminant transport models

Reactive contaminant transport models are used by hydrologists to simulate and study the migration and fate of industrial waste in subsurface aquifers. Accurate transport modeling of such waste requires clear understanding of the system’s parameters, such as sorption and biodegradation. In this study, we present an efficient sequential data assimilation scheme that computes accurate estimates of aquifer contamination and spatially variable sorption coefficients. This assimilation scheme is based on a hybrid formulation of the ensemble Kalman filter (EnKF) and optimal interpolation (OI) in which solute concentration measurements are assimilated via a recursive dual estimation of sorption coefficients and contaminant state variables. This hybrid EnKF-OI scheme is used to mitigate background covariance limitations due to ensemble under-sampling and neglected model errors. Numerical experiments are conducted with a two-dimensional synthetic aquifer in which cobalt-60, a radioactive contaminant, is leached in a saturated heterogeneous clayey sandstone zone. Assimilation experiments are investigated under different settings and sources of model and observational errors. Simulation results demonstrate that the proposed hybrid EnKF-OI scheme successfully recovers both the contaminant and the sorption rate and reduces their uncertainties. Sensitivity analyses also suggest that the adaptive hybrid scheme remains effective with small ensembles, allowing to reduce the ensemble size by up to 80% with respect to the standard EnKF scheme.     

Assimilation of SLA and SST data into an OGCM for the Indian Ocean

 Remotely sensed observations of sea-level anomaly and sea-surface temperature have been assimilated into an implementation of the Miami Isopycnic Coordinate Ocean Model (MICOM) for the Indian Ocean using the Ensemble Kalman Filter (EnKF). The system has been applied in a hindcast validation experiment to examine the properties of the assimilation scheme when used with a full ocean general circulation model and real observations. This work is considered as a first step towards an operational ocean monitoring and forecasting system for the Indian Ocean. The assimilation of real data has demonstrated that the sequential EnKF can efficiently control the model evolution in time. The use of data assimilation requires a significant amount of additional processing and computational resources. However, we have tried to justify the cost of using a sophisticated assimilation scheme by demonstrating strong regional and temporal dependencies of the covariance statistics, which include highly anisotropic and flow-dependent correlation functions. In particular, we observed a marked difference between error statistics in the equatorial region and at off-equatorial latitudes. We have also demonstrated how the assimilation of SLA and SST improves the model fields with respect to real observations. Independent in situ temperature profiles have been used to examine the impact of assimilating the remotely sensed observations. These intercomparisons have shown that the model temperature and salinity fields better resemble in situ observations in the assimilation experiment than in a model free-run case. On the other hand, it is also expected that assimilation of in situ profiles is needed to properly control the deep ocean circulation. Received: 8 January 2002 / Accepted: 8 April 2002     

Evaluating the utility of the ensemble transform Kalman filter for adaptive sampling when updating a hydrodynamic model

This paper compares two Monte Carlo sequential data assimilation methods based on the Kalman filter, for estimating the effect of measurements on simulations of state error variance made by a one-dimensional hydrodynamic model. The first method used an ensemble Kalman filter (EnKF) to update state estimates, which were then used as initial conditions for further simulations. The second method used an ensemble transform Kalman filter (ETKF) to quickly estimate the effect of measurement error covariance on forecast error covariance without the need to re-run the simulation model. The ETKF gave an unbiased estimate of EnKF analysed error variance, although differences in the treatment of measurement errors meant the results were not identical. Estimates of forecast error variance could also be made, but their accuracy deteriorated as the time from measurements increased due in part to model non-linearity and the decreasing signal variance. The motivation behind the study was to assess the ability of the ETKF to target possible measurements, as part of an adaptive sampling framework, before they are assimilated by an EnKF-based forecasting model on the River Crouch, Essex, UK. The ETKF was found to be a useful tool for quickly estimating the error covariance expected after assimilating measurements into the hydrodynamic model. It, thus, provided a means of quantifying the ‘usefulness’ (in terms of error variance) of possible sampling schemes.     

Argo data assimilation in ocean general circulation model of Northwest Pacific Ocean

The Argo temperature and salinity profiles in 2005–2009 are assimilated into a coastal ocean general circulation model of the Northwest Pacific Ocean using the ensemble adjustment Kalman filter (EAKF). Three numerical tests, including the control run (CTL) (without data assimilation, which serves as the reference experiment), ensemble free run (EnFR) (without data assimilation), and EAKF experiment (with Argo data assimilation using EAKF), are carried out to examine the performance of this system. Using the restarts of different years as the initial conditions of the ensemble integrations, the ensemble spreads from EnFR and EAKF are all kept at a finite value after a sharp decreasing in the first few months because of the sensitive of the model to the initial conditions, and the reducing of the ensemble spread due to Argo data assimilation is not much. The ensemble samples obtained in this way can well represent the probabilities of the real ocean states, and no ensemble inflation is necessary for this EAKF experiment. Different experiment results are compared with satellite sea surface temperature (SST) data and the Global Temperature-Salinity Profile Program (GTSPP) data. The comparison of SST shows that modeled SST errors are reduced after data assimilation; the error reduction percentage after assimilating the Argo profiles is about 10?% on average. The comparison against the GTSPP profiles, which are independent of the Argo profiles, shows improvements in both temperature and salinity. The comparison results indicated a great error reduction in all vertical layers relative to CTL and the ensemble mean of EnFR; the maximum value for temperature and salinity reaches to 85?% and 80?%, respectively. The standard deviations of sea surface height are employed to examine the simulation ability, and it is shown that the mesoscale variability is improved after Argo data assimilation, especially in the Kuroshio extension area and along the section of 10°N. All these results suggest that this system is potentially useful for improving the simulation ability of oceanic numerical models.     

A soil moisture assimilation scheme based on the ensemble Kalman filter using microwave brightness temperature

This study presents a soil moisture assimilation scheme, which could assimilate microwave brightness temperature directly, based on the ensemble Kalman filter and the shuffled complex evolution method (SCE-UA). It uses the soil water model of the land surface model CLM3.0 as the forecast operator, and a radiative transfer model (RTM) as the observation operator in the assimilation system. The assimilation scheme is implemented in two phases: the parameter calibration phase and the pure soil moisture assimilation phase. The vegetation optical thickness and surface roughness parameters in the RTM are calibrated by SCE-UA method and the optimal parameters are used as the final model parameters of the observation operator in the assimilation phase. The ideal experiments with synthetic data indicate that this scheme could significantly improve the simulation of soil moisture at the surface layer. Furthermore, the estimation of soil moisture in the deeper layers could also be improved to a certain extent. The real assimilation experiments with AMSR-E brightness temperature at 10.65 GHz (vertical polarization) show that the root mean square error (RMSE) of soil moisture in the top layer (0–10 cm) by assimilation is 0.03355 m³ · m⁻³, which is reduced by 33.6% compared with that by simulation (0.05052 m³ · m⁻³). The mean RMSE by assimilation for the deeper layers (10–50 cm) is also reduced by 20.9%. All these experiments demonstrate the reasonability of the assimilation scheme developed in this study.     

An effective method based on dynamic sampling for data assimilation in a global wave model

The ensemble Kalman filter (EnKF) performs well because that the covariance of background error is varying along time. It provides a dynamic estimate of background error and represents the reasonable statistic characters of background error. However, high computational cost due to model ensemble in EnKF is employed. In this study, two methods referred as static and dynamic sampling methods are proposed to obtain a good performance and reduce the computation cost. Ensemble adjustment Kalman filter (EAKF) method is used in a global surface wave model to examine the performance of EnKF. The 24-h interval difference of simulated significant wave height (SWH) within 1 year is used to compose the static samples for ensemble errors, and these errors are used to construct the ensemble states at each time the observations are available. And then, the same method of updating the model states in the EAKF is applied for the ensemble states constructed by a static sampling method. The dynamic sampling method employs a similar method to construct the ensemble states, but the period of the simulated SWH is changing with time. Here, 7 days before and after the observation time is used as this period. To examine the performance of three schemes, EAKF, static, or dynamic sampling method, observations from satellite Jason-2 in 2014 are assimilated into a global wave model, and observations from satellite Saral are used for validation. The results indicate that the EAKF performs best, while the static sampling method is relatively worse. The dynamic sampling method improves an assimilation effect dramatically compared to the static sampling method, and its overall performance is closed to the EAKF. In low latitudes, the dynamic sampling method has a slight advantage over the EAKF. In the dynamic or static sampling methods, only one wave model is required to run and their computational cost is reduced sharply. According to the performance of these three methods, the dynamic sampling method can treated as an effective alternative of EnKF, which could reduce the computational cost and provide a good performance of data assimilation.     

Local ensemble assimilation scheme with global constraints and conservation

Ensemble assimilation schemes applied in their original, global formulation respect linear conservation properties if the ensemble perturbations are set up accordingly. For realistic ocean systems, only a relatively small number of ensemble members can be calculated. A localization of the ensemble increment is therefore necessary to filter out spurious long-range correlations. The conservation of the global properties will be lost if the assimilation is performed locally, since the conservation requires a coupling between all model grid points which is removed by the localization. The distribution of ocean observations is often highly inhomogeneous. Systematic errors of the observed parts of the ocean state can lead to spurious adjustment of the non-observed parts via data assimilation and thus to a spurious increase or decrease in long-term simulations of global properties which should be conserved. In this paper, we propose a local assimilation scheme (with different variants and assumptions) which can satisfy global conservation properties. The proposed scheme can also be used for non-local observation operators. Different variants of the proposed scheme are tested in an idealized model and compared to the traditional covariance localization with an ad-hoc step enforcing conservation. It is shown that the inclusion of the conservation property reduces the total RMS error and that the presented stochastic and deterministic schemes avoiding error space rotation provide better results than the traditional covariance localization.     

Calibration of a Land Subsidence Model Using InSAR Data via the Ensemble Kalman Filter

The application of interferometric synthetic aperture radar (InSAR) has been increasingly used to improve capabilities to model land subsidence in hydrogeologic studies. A number of investigations over the last decade show how spatially detailed time‐lapse images of ground displacements could be utilized to advance our understanding for better predictions. In this work, we use simulated land subsidences as observed measurements, mimicking InSAR data to inversely infer inelastic specific storage in a stochastic framework. The inelastic specific storage is assumed as a random variable and modeled using a geostatistical method such that the detailed variations in space could be represented and also that the uncertainties of both characterization of specific storage and prediction of land subsidence can be assessed. The ensemble Kalman filter (EnKF), a real‐time data assimilation algorithm, is used to inversely calibrate a land subsidence model by matching simulated subsidences with InSAR data. The performance of the EnKF is demonstrated in a synthetic example in which simulated surface deformations using a reference field are assumed as InSAR data for inverse modeling. The results indicate: (1) the EnKF can be used successfully to calibrate a land subsidence model with InSAR data; the estimation of inelastic specific storage is improved, and uncertainty of prediction is reduced, when all the data are accounted for; and (2) if the same ensemble is used to estimate Kalman gain, the analysis errors could cause filter divergence; thus, it is essential to include localization in the EnKF for InSAR data assimilation.     

The role of prior model calibration on predictions with ensemble Kalman filter

This paper, based on a real world case study (Limmat aquifer, Switzerland), compares inverse groundwater flow models calibrated with specified numbers of monitoring head locations. These models are updated in real time with the ensemble Kalman filter (EnKF) and the prediction improvement is assessed in relation to the amount of monitoring locations used for calibration and updating. The prediction errors of the models calibrated in transient state are smaller if the amount of monitoring locations used for the calibration is larger. For highly dynamic groundwater flow systems a transient calibration is recommended as a model calibrated in steady state can lead to worse results than a noncalibrated model with a well-chosen uniform conductivity. The model predictions can be improved further with the assimilation of new measurement data from on-line sensors with the EnKF. Within all the studied models the reduction of 1-day hydraulic head prediction error (in terms of mean absolute error [MAE]) with EnKF lies between 31% (assimilation of head data from 5 locations) and 72% (assimilation of head data from 85 locations). The largest prediction improvements are expected for models that were calibrated with only a limited amount of historical information. It is worthwhile to update the model even with few monitoring locations as it seems that the error reduction with EnKF decreases exponentially with the amount of monitoring locations used. These results prove the feasibility of data assimilation with EnKF also for a real world case and show that improved predictions of groundwater levels can be obtained.