The Rapid Refresh (RAP), an hourly updated assimilation and model forecast system, replaced the Rapid Update Cycle (RUC) as an operational regional analysis and forecast system among the suite of models at the NOAA/National Centers for Environmental Prediction (NCEP) in 2012. The need for an effective hourly updated assimilation and modeling system for the United States for situational awareness and related decision-making has continued to increase for various applications including aviation (and transportation in general), severe weather, and energy. The RAP is distinct from the previous RUC in three primary aspects: a larger geographical domain (covering North America), use of the community-based Advanced Research version of the Weather Research and Forecasting (WRF) Model (ARW) replacing the RUC forecast model, and use of the Gridpoint Statistical Interpolation analysis system (GSI) instead of the RUC three-dimensional variational data assimilation (3DVar). As part of the RAP development, modifications have been made to the community ARW model (especially in model physics) and GSI assimilation systems, some based on previous model and assimilation design innovations developed initially with the RUC. Upper-air comparison is included for forecast verification against both rawinsondes and aircraft reports, the latter allowing hourly verification. In general, the RAP produces superior forecasts to those from the RUC, and its skill has continued to increase from 2012 up to RAP version 3 as of 2015. In addition, the RAP can improve on persistence forecasts for the 1–3-h forecast range for surface, upper-air, and ceiling forecasts.

Data assimilation (DA) methods for convective‐scale numerical weather prediction at operational centres are surveyed. The operational methods include variational methods (3D‐Var and 4D‐Var), ensemble methods (LETKF) and hybrids between variational and ensemble methods (3DEnVar and 4DEnVar). At several operational centres, other assimilation algorithms, like latent heat nudging, are additionally applied to improve the model initial state, with emphasis on convective scales. It is demonstrated that the quality of forecasts based on initial data from convective‐scale DA is significantly better than the quality of forecasts from simple downscaling of larger‐scale initial data. However, the duration of positive impact depends on the weather situation, the size of the computational domain and the data that are assimilated. Furthermore it is shown that more advanced methods applied at convective scales provide improvements over simpler methods. This motivates continued research and development in convective‐scale DA. Challenges in research and development for improvements of convective‐scale DA are also reviewed and discussed. The difficulty of handling the wide range of spatial and temporal scales makes development of multi‐scale assimilation methods and space–time covariance localization techniques important. Improved utilization of observations is also important. In order to extract more information from existing observing systems of convective‐scale phenomena (e.g. weather radar data and satellite image data), it is necessary to provide improved statistical descriptions of the observation errors associated with these observations.

The effectiveness of the ensemble Kalman filter (EnKF) for assimilating radar observations at convective scales is investigated for cases whose behaviors span supercellular, linear, and multicellular organization. The parallel EnKF algorithm of the Data Assimilation Research Testbed (DART) is used for data assimilation, while the Weather Research and Forecasting (WRF) Model is employed as a simplified cloud model at 2-km horizontal grid spacing. In each case, reflectivity and radial velocity measurements are utilized from a single Weather Surveillance Radar-1988 Doppler (WSR-88D) within the U.S. operational network. Observations are assimilated every 2 min for a duration of 60 min and correction of folded radial velocities occurs within the EnKF. Initial ensemble uncertainty includes random perturbations to the horizontal wind components of the initial environmental sounding. The EnKF performs effectively and with robust results across all the cases. Over the first 18-30 min of assimilation, the rms and domain-averaged prior fits to observations in each case improve significantly from their initial levels, reaching comparable values of 3-6 m s 21 and 7-10 dBZ. Representation of mesoscale uncertainty, albeit in the simplest form of initial sounding perturbations, is a critical part of the assimilation system, as it increases ensemble spread and improves filter performance. In addition, assimilation of ''no precipitation'' observations (i.e., reflectivity observations with values small enough to indicate the absence of precipitation) serves to suppress spurious convection in ensemble members. At the same time, it is clear that the assimilation is far from optimal, as the ensemble spread is consistently smaller than what would be expected from the innovation statistics and the assumed observationerror variance.

An ''additive noise'' method for initializing ensemble forecasts of convective storms and maintaining ensemble spread during data assimilation is developed and tested for a simplified numerical cloud model (no radiation, terrain, or surface fluxes) and radar observations of the 8 May 2003 Oklahoma City supercell. Every 5 min during a 90-min data-assimilation window, local perturbations in the wind, temperature, and water-vapor fields are added to each ensemble member where the reflectivity observations indicate precipitation. These perturbations are random but have been smoothed so that they have correlation length scales of a few kilometers. An ensemble Kalman filter technique is used to assimilate Doppler velocity observations into the cloud model. The supercell and other nearby cells that develop in the model are qualitatively similar to those that were observed. Relative to previous storm-scale ensemble methods, the additive-noise technique reduces the number of spurious cells and their negative consequences during the data assimilation. The additive-noise method is designed to maintain ensemble spread within convective storms during long periods of data assimilation, and it adapts to changing storm configurations. It would be straightforward to use this method in a mesoscale model with explicit convection and inhomogeneous storm environments.

High-resolution Doppler radar observations of tornadoes reveal a distinctive tornado-scale signature with the following properties: a reflectivity minimum aloft inside the tornado core (described previously as an "eye"), a high-reflectivity tube aloft that is slightly wider than the tornado core, and a tapering of this high-reflectivity tube near the ground. The results of simple one-dimensional and two-dimensional models demonstrate how these characteristics develop. Important processes in the models include centrifugal ejection of hydrometeors and/or debris by the rotating flow and recycling of some objects by the nearsurface inflow and updraft.Doppler radars sample the motion of objects within the tornado rather than the actual airflow. Since objects move at different speeds and along different trajectories than the air, error is introduced into kinematic analyses of tornadoes based on radar observations. In a steady, axisymmetric tornado, objects move outward relative to the air and move more slowly than the air in the tangential direction; in addition, the vertical air-relative speed of an object is less than it is in still air. The differences between air motion and object motion are greater for objects with greater characteristic fall speeds (i.e., larger, denser objects) and can have magnitudes of tens of meters per second. Estimates of these differences for specified object and tornado characteristics can be obtained from an approximation of the one-dimensional model.Doppler On Wheels observations of the 30 May 1998 Spencer, South Dakota, tornado demonstrate how the apparent tornado structure can change when the radar-scatterer type changes. When the Spencer tornado entered the town and started lofting debris, changes occurred in the Doppler velocity and reflectivity fields that are consistent with an increase in mean scatterer size.

The feasibility of using an ensemble Kalman filter (EnKF) to retrieve the wind and temperature fields in an isolated convective storm has been tested by applying the technique to observations of the 17 May 1981 Arcadia, Oklahoma, tornadic supercell. Radial-velocity and reflectivity observations from a single radar were assimilated into a nonhydrostatic, anelastic numerical model initialized with an idealized (horizontally homogeneous) base state. The assimilation results were compared to observations from another Doppler radar, the results of dual-Doppler wind syntheses, and in situ measurements from an instrumented tower. Observation errors make it more difficult to assess EnKF performance than in previous storm-scale EnKF experiments that employed synthetic observations and a perfect model; nevertheless, the comparisons in this case indicate that the locations of the main updraft and mesocyclone in the Arcadia storm were determined rather accurately, especially at midlevels. The magnitudes of vertical velocity and vertical vorticity in these features are similar to those in the dual-Doppler analyses, except that the low-level updraft is stronger in the EnKF analyses than in the dual-Doppler analyses. Several assimilation-scheme parameters are adjustable, including the method of initializing the ensemble, the inflation factor applied to perturbations, the magnitude of the assumed observation-error variance, and the degree of localization of the filter. In the Arcadia storm experiments, in which observations of a mature storm were assimilated over a relatively short (47 min) period, the results depended most on the ensemble-initialization method. In the data assimilation experiments, too much northerly storm-relative outflow along the south side of the lowlevel cold pool eventually developed during the assimilation period. Assimilation of Doppler observations did little to correct temperature errors near the surface in the cold pool. Both observational limitations (poor spatial resolution in the radar data near the ground) and model errors (coarse resolution and uncertainties in the parameterizations of moist processes) probably contributed to poor low-level temperature analyses in these experiments.

Ensemble Kalman filter (EnKF) techniques have been proposed for obtaining atmospheric state estimates on the scale of individual convective storms from radar and other observations, but tests of these methods with observations of real convective storms are still very limited. In the current study, radar observations of the 8 May 2003 Oklahoma City tornadic supercell thunderstorm were assimilated into the National Severe Storms Laboratory (NSSL) Collaborative Model for Multiscale Atmospheric Simulation (NCOMMAS) with an EnKF method. The cloud model employed 1-km horizontal grid spacing, a single-moment bulk precipitationmicrophysics scheme, and a base state initialized with sounding data. A 50-member ensemble was produced by randomly perturbing base-state wind profiles and by regularly adding random local perturbations to the horizontal wind, temperature, and water vapor fields in and near observed precipitation.In a reference experiment, only Doppler-velocity observations were assimilated into the NCOMMAS ensemble. Then, radar-reflectivity observations were assimilated together with Doppler-velocity observations in subsequent experiments. Influences that reflectivity observations have on storm-scale analyses were revealed through parameter-space experiments by varying observation availability, observation errors, ensemble spread, and choices for what model variables were updated when a reflectivity observation was assimilated. All experiments produced realistic storm-scale analyses that compared favorably with independent radar observations. Convective storms in the NCOMMAS ensemble developed more quickly when reflectivity observations and velocity observations were both assimilated rather than only velocity, presumably because the EnKF utilized covariances between reflectivity and unobserved model fields such as cloud water and vertical velocity in efficiently developing realistic storm features.Recurring spatial patterns in the differences between predicted and observed reflectivity were noted particularly at low levels, downshear of the supercell's updraft, in the anvil of moderate-to-light precipitation, where reflectivity in the model was typically lower than observed. Bias errors in the predicted rain mixing ratios and/or the size distributions that the bulk scheme associates with these mixing ratios are likely responsible for this reflectivity underprediction. When a reflectivity observation is assimilated, bias errors in the model fields associated with reflectivity (rain, snow, and hail-graupel) can be projected into other model variables through the ensemble covariances. In the current study, temperature analyses in the downshear anvil at low levels, where reflectivity was underpredicted, were very sensitive both to details of the assimilation algorithm and to ensemble spread in temperature. This strong sensitivity suggests low confidence in analyses of low-level cold pools obtained through reflectivity-data assimilation.

The assimilation of surface observations using an ensemble Kalman filter (EnKF) approach is evaluated for the potential to improve short-range forecasting. Two severe weather cases are examined, in which the assimilation is performed over a 6-h period using hourly surface observations followed by an 18-h simulation period. Ensembles are created in three different ways—by using different initial and boundary conditions, by using different model physical process schemes, and by using both different initial and boundary conditions and different model physical process schemes. The ensembles are compared in order to investigate the role of uncertainties in the initial and boundary conditions and physical process schemes in EnKF data assimilation. In the initial condition ensemble, spread is associated largely with the displacement of atmospheric baroclinic systems. In the physics ensemble, spread comes from the differences in model physics, which results in larger spread in temperature and dewpoint temperature than the initial condition ensemble, and smaller spread in the wind fields. The combined initial condition and physics ensemble has properties from both of the previous two ensembles. It provides the largest spread and produces the best simulation for most of the variables, in terms of the rms difference between the ensemble mean and observations. Perhaps most importantly, this combined ensemble provides very good guidance on the mesoscale features important to the severe weather events of the day.