The performance of the ensemble Kalman filter (EnKF) in forced, dissipative flow under imperfect model conditions is investigated through simultaneous state and parameter estimation where the source of model error is the uncertainty in the model parameters. Two numerical models with increasing complexity are used with simulated observations. For lower complexity, a two-dimensional, nonlinear, hydrostatic, non-rotating, and incompressible sea breeze model is developed with buoyancy and vorticity as the prognostic variables. Model resolution is 4 km horizontally and 50 m vertically. The ensemble size is set at 40. Forcing is maintained through an explicit heating function with additive stochastic noise. Simulated buoyancy observations on land surface with 40-km spacing are assimilated every 3 hours. Up to six model parameters are successfully subjected to estimation attempts in various experiments. The overall EnKF performance in terms of the error statistics is found to be superior to the worst-case scenario (when there is parameter error but no parameter estimation is performed) with an average error reduction in buoyancy and vorticity of 40% and 46%, respectively, for the simultaneous estimation of six parameters. The model chosen to represent the complexity of operational weather forecasting is the Pennsylvania State University-National Center for Atmospheric Research MM5 model with a 36-km horizontal resolution and 43 vertical layers. The ensemble size for all experiments is chosen as 40 and a 41st member is generated as the truth with the same ensemble statistics. Assimilations are performed with a 12-hour interval with simulated sounding and surface observations of horizontal winds and temperature. Only single-parameter experiments are performed focusing on a constant inserted into the code as the multiplier of the vertical eddy mixing coefficient. Estimation experiments produce very encouraging results and the mean estimated parameter value nicely converges to the true value exhibiting a satisfactory level of variability.

Snow initialization is crucial for weather and seasonal prediction, but the National Centers for Environmental Prediction (NCEP) operational models have been found to produce too little snow water equivalent, partly because they assume a constant and unrealistically low snow density for the snowpack. One possible solution is to use the snow density formulation from the Noah land model used in NCEP operational forecast models. While this solution is better than the constant density assumption, the seasonal evolution of snow density in Noah is still found to be unrealistic, through the evaluation of both the offline Noah model output and the Noah snow density formulation itself. A physically based snow density parameterization is then developed, which performs considerably better than the Noah parameterization based on the measurements from the SNOTEL network over the western United States and Alaska. It also performs better than the snow density schemes used in three other models. This parameterization could be easily implemented in NCEP operational snow initialization. With the consideration of up to 10 snow layers, this parameterization can also be applied to multilayer snowpack initiation or to estimate snow water equivalent from in situ and airborne snow depth measurements.

Computer technology is growing fast in recent years. Modeling technique is used in predicting or in planning engineering works and even in preventing disaster. Modeling is widely used in many domains and unmanned Real-time online operation modeling systems on prediction become popular. Model may become inaccurate due to a number of uncertainties in the approximation and by numerical reasons. Data Assimilation technique is developed to solve this problem. Measured data is used to improve the model results. In this research, the Cressman scheme was chosen as the data assimilation scheme and used for correcting the modeling system. An idealized model was constructed first as Taiwan Strait. In order to test the stability if data assimilation system several geographical variations and data availability cases were designed, eg adding varying bottom topography, an island added in the domain, different measurement data locations. In order to test the model sensibilities an error was inserted to the boundaries. Model results were first corrected with data assimilation system for a period of time, a Harmonic Analysis was, then, used for reanalysis the corrected time series on the boundaries. The new boundary condition is used in the new model run for making predictions. A true topography and island system as Taiwan Strait was tested with the true astronomical tide as the boundary input. The data assimilation system using the Cressman scheme could reduce the RMSE effectively. The factor that affects the efficiency of the data assimilation system is the number and the location of the measurement data.

Recently there has been a surge in interest in coupling ensemble-based data assimilation methods with variational methods (commonly referred to as 4DVar). Here we discuss a number of important differences between ensemble-based and variational methods that ought to be considered when attempting to fuse these methods. We note that the Best Linear Unbiased Estimate (BLUE) of the posterior mean over a data assimilation window can only be delivered by data assimilation schemes that utilise the 4-dimensional (4D) forecast covariance of a prior distribution of non-linear forecasts across the data assimilation window. An ensemble Kalman smoother (EnKS) may be viewed as a BLUE approximating data assimilation scheme. In contrast, we use the dual form of 4DVar to show that the most likely non-linear trajectory corresponding to the posterior mode across a data assimilation window can only be delivered by data assimilation schemes that create counterparts of the 4D prior forecast covariance using a tangent linear model. Since 4DVar schemes have the required structural framework to identify posterior modes, in contrast to the EnKS, they may be viewed as mode approximating data assimilation schemes. Hence, when aspects of the EnKS and 4DVar data assimilation schemes are blended together in a hybrid, one would like to be able to understand how such changes would affect the mode-or mean-finding abilities of the data assimilation schemes. This article helps build such understanding using a series of simple examples. We argue that this understanding has important implications to both the interpretation of the hybrid state estimates and to their design.

Data assimilation into sea ice models designed for climate studies has started about 15 years ago. In most of the studies conducted so far, it is assumed that the improvement brought by the assimilation is straightforward. However, some studies suggest this might not be true. In order to elucidate this question and to find an appropriate way to further assimilate sea ice concentration and velocity observations into large-scale sea ice models, we analyze here results from a number of twin experiments (i.e. experiments in which the assimilated data are model outputs) carried out with first a simplified model of the Arctic sea ice pack then with NEMO-LIM2, a primitive equation ocean general circulation model coupled to LIM (Louvain-la-Neuve sea ice model). Our objective is to determine to what degree the assimilation of ice velocity and/or concentration data improves the global performance of the model and, more specifically, reduces the error in the computed ice thickness. A simple scheme is used, and outputs from a control run and from perturbed experiments without and with data assimilation are thoroughly compared. Our results indicate that, under certain conditions depending on the assimilation weights and the type of model error, the assimilation of ice velocity data enhances the model performance. The assimilation of ice concentration data also helps in improving the model results, but it has to be handled with care because of the strong link between ice concentration and ice thickness. Therefore, we show that one should conserve the ice thickness (not the ice volume) when ice concentration data are assimilated into the model. We also demonstrate that one should assimilate sea ice concentration and velocity data simultaneously. Finally, we give some concrete keys in order to choose which observational data set to assimilate.

Data assimilation into sea ice models designed for climate studies has started about 15 years ago. In most of the studies conducted so far, it is assumed that the improvement brought by the assimilation is straightforward. However, some studies suggest this might not be true. In order to elucidate this question and to find an appropriate way to further assimilate sea ice concentration and velocity observations into large-scale sea ice models, we analyze here results from a number of twin experiments (i.e. experiments in which the assimilated data are model outputs) carried out with first a simplified model of the Arctic sea ice pack then with NEMO-LIM2, a primitive equation ocean general circulation model coupled to LIM (Louvain-la-Neuve sea ice model). Our objective is to determine to what degree the assimilation of ice velocity and/or concentration data improves the global performance of the model and, more specifically, reduces the error in the computed ice thickness. A simple scheme is used, and outputs from a control run and from perturbed experiments without and with data assimilation are thoroughly compared. Our results indicate that, under certain conditions depending on the assimilation weights and the type of model error, the assimilation of ice velocity data enhances the model performance. The assimilation of ice concentration data also helps in improving the model results, but it has to be handled with care because of the strong link between ice concentration and ice thickness. Therefore, we show that one should conserve the ice thickness (not the ice volume) when ice concentration data are assimilated into the model. We also demonstrate that one should assimilate sea ice concentration and velocity data simultaneously. Finally, we give some concrete keys in order to choose which observational data set to assimilate.

Production from a hydrocarbon reservoir is typically supported by water or carbon dioxide (CO2) injection. CO2 injection into hydrocarbon reservoirs is also a promising solution for reducing environmental hazards from the release of green house gases into the earth’s atmosphere. Numerical simulators are used for designing and predicting the complex behavior of systems under such scenarios. Two key steps in such studies are forward modeling for performance prediction based on simulation studies using reservoir models and inverse modeling for updating reservoir models using the data collected from field. The viability of time-lapse seismic monitoring using an integrated modeling of fluid flow, including chemical reactions, and seismic response is examined. A comprehensive simulation of the gas injection process accounting for the phase behavior of CO2-reservoir fluids, the associated precipitation/dissolution reactions, and the accompanying changes in porosity and permeability is performed. The simulation results are then used to model the changes in seismic response with time. The general observation is that gas injection decreases bulk density and wave velocity of the host rock system. Another key topic covered in this work is the data assimilation study for hydrocarbon reservoirs using Ensemble Kalman Filter (EnKF). Some critical issues related to EnKF based history matching are explored, primarily for a large field with substantial production history. A novel and efficient approach based on spectral clustering to select ‘optimal’ initial ensemble members is proposed. Also, well-specific black-oil or compositional streamline trajectories are used for covariance localization. Approach is applied to the Weyburn field, a large carbonate reservoir in Canada. The approach for optimal member selection is found to be effective in reducing the ensemble size which was critical for this large-scale field application. Streamline-based covariance localization is shown to play a very important role by removing spurious covariances between any well and far-off cell permeabilities. Finally, time-lapse seismic study is done for the Weyburn field. Sensitivity of various bulk seismic parameters viz velocity and impedance is calculated with respect to different simulation parameters. Results show large correlation between porosity and seismic parameters. Bulk seismic parameters are sensitive to net overburden pressure at its low values. Time-lapse changes in pore-pressure lead to changes in bulk parameters like velocity and impedance.

Tropical cyclone (TC) track and intensity forecasts have improved in recent years due to increased model resolution, improved data assimilation, and the rapid increase in the number of routinely assimilated observations over oceans. The data assimilation approach that has received the most attention in recent years is Ensemble Kalman Filtering (EnKF). The most attractive feature of the EnKF is that it uses a fully flow-dependent estimate of the error statistics, which can have important benefits for the analysis of rapidly developing TCs. We implement the Local Ensemble Transform Kalman Filter algorithm, a variation of the EnKF, on a reduced-resolution version of the National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) model and the NCEP Regional Spectral Model (RSM) to build a coupled global-limited area analysis/forecast system. This is the first time, to our knowledge, that such a system is used for the analysis and forecast of tropical cyclones. We use data from summer 2004 to study eight tropical cyclones in the Northwest Pacific. The benchmark data sets that we use to assess the performance of our system are the NCEP Reanalysis and the NCEP Operational GFS analyses from 2004. These benchmark analyses were both obtained by the Statistical Spectral Interpolation, which was the operational data assimilation system of NCEP in 2004. The GFS Operational analysis assimilated a large number of satellite radiance observations in addition to the observations assimilated in our system. All analyses are verified against the Joint Typhoon Warning Center Best Track data set. The errors are calculated for the position and intensity of the TCs. The global component of the ensemble-based system shows improvement in position analysis over the NCEP Reanalysis, but shows no significant difference from the NCEP operational analysis for most of the storm tracks. The regional component of our system improves position analysis over all the global analyses. The intensity analyses, measured by the minimum sea level pressure, are of similar quality in all of the analyses. Regional deterministic forecasts started from our analyses are generally not significantly different from those started from the GFS operational analysis. On average, the regional experiments performed better for longer than 48 h sea level pressure forecasts, while the global forecast performed better in predicting the position for longer than 48 h.

O uso da técnica de assimilação de dados meteorológicos é extremamente importante para a correção de imprecisões nos dados que compõem as condições iniciais e de fronteira dos modelos de previsão do tempo. Neste trabalho, faz-se uso da técnica de assimilação de dados 3DVAR contida no modelo de mesoescala WRF (Weather Research and Forecasting), o objetivo principal do trabalho é analisar o impacto da assimilação de dados meteorológicos de diversas fontes de dados (GTS – Sistema Global de Telecomunicações, estações automáticas, dados radar) no modelo WRF. Para analisar a consistência da assimilação de dados no WRF verificou-se a diferença entre a análise com e sem assimilação de dados. Confirmando a consistência da mesma, foram realizados os procedimentos necessários para gerar os prognósticos com assimilação de dados para cada caso individualmente. Os experimentos com assimilação de dados foram realizados para cada tipo de dado e em conjunto, possibilitando assim fazer uma análise do impacto que cada dado tem na previsão. Os resultados foram comparados entre si espacialmente utilizando dados do modelo global GFS (Global Forecast System) e satélite da Missão de Medida da Chuva Tropical (TRMM). A variável da precipitação acumulada foi comparada e validada espacialmente com os dados do TRMM, constatou-se para o caso do mês de janeiro uma superestimação dos valores acumulados para algumas regiões e para o caso do mês de abril uma subestimação, isso se deve ao fato da frequência temporal dos dados do satélite TRMM, pois provavelmente elas não foram compatíveis com o horário das precipitações. Quando comparado com o volume de chuva pontual com os dados da estação automática a maioria dos processamentos mostrou-se eficaz. Também no estudo de caso ocorrido no mês de janeiro a inserção de dados assimilados possibilitou uma melhora na intensidade e localização da célula convectiva. As variáveis da temperatura e do vento foram comparadas espacialmente com as análises do modelo GFS. A variável da temperatura ora apresentou valores superiores, ora inferiores ao modelo GFS, mesmo assim os resultados foram satisfatórios, uma vez que, foi possível simular temperaturas superiores antes da passagem do sistema e inferiores após a passagem do mesmo. Para o campo de vento houve uma pequena discrepância em todas as simulações em relação a magnitude, porém a direção do vento foi plotada de forma coerente, simulando até o ciclone presente no caso do mês de abril. Para o perfil vertical da temperatura e temperatura do ponto de orvalho o impacto da assimilação de dados foi pequeno, porém ambas as simulações representaram de forma coesa os perfis quando comparados com o perfil observado. Em suma, o estudo comprova que, embora se tenha algumas incoerências assimilação 3DVAR contribui de modo significativo nas previsões do tempo do modelo WRF. / The use of meteorological data assimilation technique is extremely important for the correction of the imprecisions of observational data for the initial and boundary conditions of weather forecasting models. In the present work it is used the 3DVAR data assimilation technique of the mesoscale model WRF system (Weather Research and Forecasting) aiming the analysis of the impact of the assimilation of meteorological data from several data sources (GTS - Global Telecommunication System, automatic surface stations network and radar) in the WRF model. To analysis the consistency of the data in the WRF assimilation it has been gathered the difference between analysis, with and without data assimilation. Confirming its consistency the procedures required, to generate predictions with data assimilation for each individual case were performed. The data assimilation experiments were performed for each data type as well as including all of them allowing, therefore, the analysis of the impact of each over the forecast. The results were compared and validated using data from the spatially global forecasting model GFS (Global Forecast System), satellite and the mission of the Tropical Rain Measurement (TRMM) data. The cumulative rainfall variable was compared spatially with data from TRMM, where it has been observed, in the case of January, an overestimation of the accumulated values for some regions and an underestimation for the case of April. These have been occurred because of temporal frequency of the TRMM satellite data - which probably because were not compatible with the precipitation time occurrence. Comparison between the accumulated precipitation with data from automatic station presented mostly effective results. Also, in the case study of the January with assimilated data, produced an improvement in the intensity as well as in the location of the convective cell. The wind and temperature variables were compared with the spatially GFS’s analysis. The higher temperature variable values presented alternated, from higher and lower values compared to the GFS results. The results were nevertheless unsatisfactory, because the simulated temperatures presented prior to passing the frontal system and after passing it. For the wind field there was a small discrepancy in all simulations regarding the magnitude, but the wind direction was plotted consistently simulating up to the present in the case of April cyclone. For the vertical profiles of temperature and dew point temperature the impact of data assimilation was small, but both simulations made represented good profiles, compared with the observed values. In summary, the study shows that, although there were some inconsistencies, compared with the observations, the 3DVAR assimilation contributes significantly to WRF model forecasts.

O uso da técnica de assimilação de dados meteorológicos é extremamente importante para a correção de imprecisões nos dados que compõem as condições iniciais e de fronteira dos modelos de previsão do tempo. Neste trabalho, faz-se uso da técnica de assimilação de dados 3DVAR contida no modelo de mesoescala WRF (Weather Research and Forecasting), o objetivo principal do trabalho é analisar o impacto da assimilação de dados meteorológicos de diversas fontes de dados (GTS – Sistema Global de Telecomunicações, estações automáticas, dados radar) no modelo WRF. Para analisar a consistência da assimilação de dados no WRF verificou-se a diferença entre a análise com e sem assimilação de dados. Confirmando a consistência da mesma, foram realizados os procedimentos necessários para gerar os prognósticos com assimilação de dados para cada caso individualmente. Os experimentos com assimilação de dados foram realizados para cada tipo de dado e em conjunto, possibilitando assim fazer uma análise do impacto que cada dado tem na previsão. Os resultados foram comparados entre si espacialmente utilizando dados do modelo global GFS (Global Forecast System) e satélite da Missão de Medida da Chuva Tropical (TRMM). A variável da precipitação acumulada foi comparada e validada espacialmente com os dados do TRMM, constatou-se para o caso do mês de janeiro uma superestimação dos valores acumulados para algumas regiões e para o caso do mês de abril uma subestimação, isso se deve ao fato da frequência temporal dos dados do satélite TRMM, pois provavelmente elas não foram compatíveis com o horário das precipitações. Quando comparado com o volume de chuva pontual com os dados da estação automática a maioria dos processamentos mostrou-se eficaz. Também no estudo de caso ocorrido no mês de janeiro a inserção de dados assimilados possibilitou uma melhora na intensidade e localização da célula convectiva. As variáveis da temperatura e do vento foram comparadas espacialmente com as análises do modelo GFS. A variável da temperatura ora apresentou valores superiores, ora inferiores ao modelo GFS, mesmo assim os resultados foram satisfatórios, uma vez que, foi possível simular temperaturas superiores antes da passagem do sistema e inferiores após a passagem do mesmo. Para o campo de vento houve uma pequena discrepância em todas as simulações em relação a magnitude, porém a direção do vento foi plotada de forma coerente, simulando até o ciclone presente no caso do mês de abril. Para o perfil vertical da temperatura e temperatura do ponto de orvalho o impacto da assimilação de dados foi pequeno, porém ambas as simulações representaram de forma coesa os perfis quando comparados com o perfil observado. Em suma, o estudo comprova que, embora se tenha algumas incoerências assimilação 3DVAR contribui de modo significativo nas previsões do tempo do modelo WRF. / The use of meteorological data assimilation technique is extremely important for the correction of the imprecisions of observational data for the initial and boundary conditions of weather forecasting models. In the present work it is used the 3DVAR data assimilation technique of the mesoscale model WRF system (Weather Research and Forecasting) aiming the analysis of the impact of the assimilation of meteorological data from several data sources (GTS - Global Telecommunication System, automatic surface stations network and radar) in the WRF model. To analysis the consistency of the data in the WRF assimilation it has been gathered the difference between analysis, with and without data assimilation. Confirming its consistency the procedures required, to generate predictions with data assimilation for each individual case were performed. The data assimilation experiments were performed for each data type as well as including all of them allowing, therefore, the analysis of the impact of each over the forecast. The results were compared and validated using data from the spatially global forecasting model GFS (Global Forecast System), satellite and the mission of the Tropical Rain Measurement (TRMM) data. The cumulative rainfall variable was compared spatially with data from TRMM, where it has been observed, in the case of January, an overestimation of the accumulated values for some regions and an underestimation for the case of April. These have been occurred because of temporal frequency of the TRMM satellite data - which probably because were not compatible with the precipitation time occurrence. Comparison between the accumulated precipitation with data from automatic station presented mostly effective results. Also, in the case study of the January with assimilated data, produced an improvement in the intensity as well as in the location of the convective cell. The wind and temperature variables were compared with the spatially GFS’s analysis. The higher temperature variable values presented alternated, from higher and lower values compared to the GFS results. The results were nevertheless unsatisfactory, because the simulated temperatures presented prior to passing the frontal system and after passing it. For the wind field there was a small discrepancy in all simulations regarding the magnitude, but the wind direction was plotted consistently simulating up to the present in the case of April cyclone. For the vertical profiles of temperature and dew point temperature the impact of data assimilation was small, but both simulations made represented good profiles, compared with the observed values. In summary, the study shows that, although there were some inconsistencies, compared with the observations, the 3DVAR assimilation contributes significantly to WRF model forecasts.