Search

A limited‐area model (LAM) is established based on a global model (Global–Regional Integrated Forecast System; GRIST). GRIST–LAM inherits all the technical features of its global counterpart, enabling independent regional weather and climate modeling. The key advancement involves extending the original dynamical core to integrate it under the lateral boundary conditions (LBCs). As an initial development and evaluation study, this paper focuses on the consistency issue between the LAM and the global model. Three perfect‐model tests, using global solutions as LBCs and background truths, were performed to evaluate the LAM behaviors. In the pure dynamical core test, the LBC errors do not compromise the solutions within the interior domain. However, certain configurations can lead to more discontinuous solutions at the domain boundary. The solution error for a specified region decreases as the domain size increases when all other factors are equal. A small error pulse is generated during the initial stage of integration due to the presence of artificial transient waves induced by the LBCs. The model generates fine‐scale details and smaller errors based on coarser‐resolution LBCs. The consistency between LAM and LBC also influences the errors. The climate simulations demonstrate that both hydrostatic and non‐hydrostatic LAMs can reach statistical equilibrium. Regional model climates in the interior domain have higher quality but are sensitive to domain size and LBC configuration. Using a variable LBC coefficient is helpful to alleviate the artificial precipitation at the boundary. In the kilometer‐scale test, the global variable‐resolution model and its LAM counterpart show comparable results. Their performance is competitive with that of a uniform‐resolution global storm‐resolving simulation. Global variable‐resolution and LAM generate higher magnitudes in the tail part of the kinetic energy spectra due to higher local resolution and produce a consistent time evolution of precipitation. The broad implication of this study is also discussed. [ABSTRACT FROM AUTHOR]

It is a common practice to use a buffer zone to damp out spurious wave growth due to computational error along the lateral boundary of limited-area weather and climate models. Although it is an effective technique to maintain model stability, an unintended side effect of using such buffer zones is the distortion of the data passing through the buffer zone. Various techniques are introduced to enhance the communication between the limited-area model's inner domain and the outer domain, which provides lateral boundary values for the inner domain. Among them, scale-selective data assimilation (SSDA) and the spectral nudging (SPNU) techniques share similar philosophy, i.e., directly injecting the large-scale components of the atmospheric circulation from the outer model domain into the interior grids of the inner model domain by-passing the lateral boundary and the buffer zone, but the two methods are taking different implementation approaches. SSDA utilizes a 3-dimensional variational data assimilation procedure to accomplish the data injection objective, whereas SPNU uses a nudging process. In the present study, the two approaches are evaluated comparatively for simulating hurricane track and intensity in a pair of cases: Jeanne (2004) and Irma (2017) using the Weather Research and Forecasting (WRF) model. The results indicate that both techniques are effective in improving tropical cyclone intensity and track simulations by reducing the errors of the large-scale circulation in the inner model domain. The SSDA runs produced better simulations of temperature and humidity fields which are not directly nudged. The SSDA runs also produced more accurate storm intensities in both cases and more realistic structure in Hurricane Jeanne's case than those produced by the SPNU runs. It should be noted, however, that extending these case study results to more general situations requires additional studies covering a large number of additional cases. [ABSTRACT FROM AUTHOR]

Ensembles of data assimilations (EDA) in numerical weather prediction (NWP) are frequently used for both initialization of ensemble prediction systems and provision of background‐error statistics to a deterministic variational data assimilation scheme. The EDA consists of running multiple data assimilation schemes in parallel with perturbed observations and backgrounds. This kind of ensemble is computationally expensive, in particular because it requires the solution of as many linear systems as there are members in the ensemble. Recently, we proposed the use of block Krylov methods to take advantage of the EDA structure to solve all the linear systems simultaneously, allowing us to reduce the number of iterations needed for convergence. This approach is studied further in this article. We develop advanced parallelization strategies for the block Krylov method formulated in observation space using the Object‐Oriented Prediction System (OOPS) jointly developed by the European Centre for Medium‐Range Weather Forecasts, Météo‐France, and their partners. The algorithm is fully tested with the EDA of the limited‐area model AROME of Météo‐France, for different weather situations and configurations representative of the current system, namely assimilating around 104–105 observations together with 25 members. This new block Krylov approach allows us to gain from 13–45% of the computational time of the minimizations. Experiments with extended configurations that assimilate more observations and use more ensemble members confirm the potential of the algorithm in the future, with gains of up to 65%. [ABSTRACT FROM AUTHOR]

In recent years, the Extended Kalman Filter (EKF) has been gaining more attention in the surface data assimilation (DA) community and has already replaced the older Optimal Interpolation (OI) scheme for the vertical component of the land surface DA system in a number of meteorological institutes. An EKF has been developed within the standalone land-surface modelling platform SURFace Externalisée (SURFEX) for the initialisation of soil temperature and soil water content based on screen-level temperature and relative humidity. In this article we present a new combination of the EKF with a basic (using conventional observations only) three-dimensional variational (3D-Var) upper-air assimilation for the limited-area model ALARO coupled to SURFEX. This new combination is compared to an Open Loop experiment where all initial conditions are interpolated from an analysis of the global numerical weather prediction model Action de Recherche Petite Echelle Grande Echelle (ARPEGE) and to an experiment where the surface is initialised using the EKF, while the upper-air initial conditions are interpolated from the ARPEGE analysis. The aim of this article is to examine whether the EKF surface assimilation coupled or not with a basic 3D-Var upper-air assimilation has an added value compared to the Open Loop, in which the more advanced upper-air data assimilation of ARPEGE with more observations used is interpolated onto the limited-area model grid. All set-ups are verified during a 1-year period 2013 against soil measurements, screen-level observations, radiosoundings and merged radar-rain-gauge precipitation observations. Results indicate that the EKF surface assimilation has positive effects on humidity scores and is able to produce similar or improved scores compared to the Open Loop. While the upper-air 3D-Var DA system of ALARO still needs improvements, the potential benefits of the combination of upper-air and surface assimilation are demonstrated through soil moisture and screen-level relative humidity verifications. [ABSTRACT FROM AUTHOR]

To investigate the real-world relevance of idealized-model convective self-aggregation, five 15 day cases of real organized convection in the tropics are simulated. These include multiple simulations of each case to test sensitivities of the convective organization and mean states to interactive radiation, interactive surface fluxes, and evaporation of rain. These simulations are compared to self-aggregation seen in the same model configured to run in idealized radiative-convective equilibrium. Analysis of the budget of the spatial variance of column-integrated frozen moist static energy shows that control runs have significant positive contributions to organization from radiation and negative contributions from surface fluxes and transport, similar to idealized runs once they become aggregated. Despite identical lateral boundary conditions for all experiments in each case, systematic differences in mean column water vapor (CWV), CWV distribution shape, and CWV autocorrelation length scale are found between the different sensitivity runs, particularly for those without interactive radiation, showing that there are at least some similarities in sensitivities to these feedbacks in both idealized and realistic simulations (although the organization of precipitation shows less sensitivity to interactive radiation). The magnitudes and signs of these systematic differences are consistent with a rough equilibrium between (1) equalization due to advection from the lateral boundaries and (2) disaggregation due to the absence of interactive radiation, implying disaggregation rates comparable to those in idealized runs with aggregated initial conditions and noninteractive radiation. This points to a plausible similarity in the way that radiation feedbacks maintain aggregated convection in both idealized simulations and the real world. [ABSTRACT FROM AUTHOR]

In this note the implicit normal mode initialization of a limited-area model is discussed from a different point of view. To that end it is shown that the equations describing the explicit normal mode initialization applied to the shallow water equations in differentiated form on the sphere can readily be derived in normal mode space if the model equations are separable, but only in the case of stationary Rossby modes can be transformed into the implicit equations in physical space. This is a consequence of the simple relations between the components of the different modes in that case. In addition a simple eigenvalue problem is given for the frequencies of the gravity waves.

An analysis is presented here of intense convection affecting the Friuli Venezia Giulia region (FVG, northeastern Italy) during the Intensive Observation Period 2b (IOP2b) in the first Special Observation Period (SOP1) of HyMeX (HYdrological cycle in Mediterranean EXperiment). The present study focuses on the first of three severe-convection episodes that affected FVG on the morning of 12 September 2012. In the first episode, a supercell, which produced hail and severe damage to trees and buildings, was observed on the plain of FVG. The available observations are analysed together with a high-resolution mesoscale model, in order to identify the relevant mechanisms for the formation and development of the cell. Six different simulations were performed starting at three different initial times, using respectively two different analysis/forecasts as initial/boundary conditions. A large spread in forecast precipitation is found among the six simulations. Only a few of the simulations were able to reproduce intense rainfall on the plain of FVG during the morning, although with significant differences in the rainfall distribution among them. One of the six simulations well reproduces the observed elongated distribution of the intense rainfall maximum; the characteristics of the cell responsible for this distribution are consistent with those expected for a supercell and its simulated evolution near the Adriatic coast agrees well with the other observations. Some instability parameters over the FVG plain and offshore (over the northern Adriatic Sea) are analysed every 5 min, showing that during this event the potential instability varies significantly over small space and time intervals and among the simulations. The best simulations have the best match to the observed potential instability calculated using the mean characteristics of the lowest 500 m layer. [ABSTRACT FROM AUTHOR]

Three ensemble-based analysis-error representation techniques are applied and compared with the aim of simulating background errors of the mesoscale limited-area model (LAM) Aire Limitée Adaptation Dynamique Développement International (ALADIN) and thus computing appropriate mesoscale static background-error covariances for its three-dimensional variational (3DVAR) data assimilation system. Real data assimilation experiments are performed for the simulation of the background errors, where uncertainties of the initial conditions and lateral boundary conditions are represented, while model errors are not accounted for. The three analysis error-simulation techniques compared are as follows: (i) the downscaling of the analysis error represented by the stochastic ensemble data assimilation (EDA) system of a global model (DSC EDA); (ii) the representation of the analysis error by a LAM stochastic ensemble data assimilation (LAM EDA) system; and (iii) the representation of the analysis error by a LAM ensemble transform Kalman filter (LAM ETKF) system. Spectral diagnostics have been developed in the study, allowing an examination of the simulated background errors at different spatial scales. It is shown that LAM EDA provides more realistic simulated background errors than DSC EDA, particularly at mesoscales. This is also partly the case for LAM ETKF at mesoscales (although to a lesser extent than for LAM EDA). The error variance is underestimated by LAM ETKF, which suggests that the applied inflation technique could be improved. Analysis and forecast impact studies using background-error covariance matrices based on the DSC EDA and LAM EDA error samples confirm the superiority of the latter simulation technique. [ABSTRACT FROM AUTHOR]

Numerical diffusion or filter are used in most numerical models in order to eliminate small-scale (near two–grid intervals in wavelength) waves. However, conventional diffusion or filter schemes introduce the noise, and indeed few people realized, by filters themselves. For instance, most filters are troubled when they are put to use on meteorological fields with sharp gradient or with steep slope and consequently, the recurrence of undesirable numerical high-frequent oscillations (overshooting and undershooting) seems to be inevitable. Particularly when diffusion or filter is implemented in limited-area models, serious side effects on the limited-area boundaries often contaminate the modeling results. The merits and demerits are surveyed for commonly used diffusion or filter operations. A new type of monotonic digit filter is suggested to prevent overshooting and undershooting (due to the computational shock and Gibbs oscillation) nearby the discontinuous or nearly discontinuous locations when the filtering process was carried out, meanwhile the high selective property of damping is retained. Moreover, the new filter is designed on the implicit framework so that it can easily handle the problem of boundary diminishing in limited-area modeling. [ABSTRACT FROM AUTHOR]