Severe weather represents storms, cyclones, fronts, severe wind or thick fog and other phenomena. Limited area models (LAM) can simulate or forecast such phenomena in higher resolution and using dedicated model set-up. This thesis explores the ALADIN (Aire Limitée Adaptation dynamique Développement InterNational) model capabilities to forecast threatening weather conditions for wider area of the Republic of Croatia. The research focuses on the consequences of a fast cyclone entering LAM domain through lateral boundary too quickly to be detected, frequency of such events, mechanism for automatic detection of such events and methods to treat the problem in the operational forecast. The solution will be applied to events with severe weather such as windstorms and/or intensive precipitation. This thesis deals with problems of temporal interpolation of the lateral boundary conditions (LBC) for a limited area model (LAM). The LBCs are taken from a large scale model and usually available with an interval of several hours. However, these data are used at the lateral boundaries every model timestep, which is usually several minutes. Therefore, the LBCs are interpolated in time. In practice, the LBCs are usually interpoated with a 3 h temporal resolution. This can be too infrequent to resolve rapidly moving storms. This problem is expected to be worse with increasing horizontal resolution. In order to detect intensive disturbances in surface pressure moving rapidly through the model domain, a filtered surface pressure field (MCUF - monitoring of the coupling update frequency) is computed operationally in the ARPEGE global model of Météo France. The field is distributed in the coupling files along with conventional meteorological fields used for LBCs for the operational forecast using ALADIN LAM in the Meteorological and Hydrological Service of Croatia (DHMZ). Here an analysis is performed of the MCUF field for the LACE coupling domain for the period since 23rd of January 2006, when it became available, until 15th of November 2014. The MCUF field is a good indicator of rapidly moving pressure disturbances (RMPDs). Its spatial and temporal distribution can be associated to the usual cyclone tracks and areas known to be supporting cyclogenesis. Alternative set of coupling files from IFS operational run in ECMWF is also available operationally in DHMZ with 3 h temporal resolution but the MCUF field is not available. Here, several methods are tested that detect RMPDs in surface pressure a posteriori from the IFS model fields provided in the coupling files. MCUF is computed by running ALADIN on the coupling files from IFS. The coupling error function [There are many functions called error function in the literature, this work focuses on the coupling error function] (that shows when the temporal interpolation misses the storm) is computed using one time step integration of ALADIN on the coupling files without initialization, initialized with digital filter initialization (DFI) or scale selective DFI (SSDFI). Finally, the amplitude of changes in the mean sea level pressure is computed from the fields in the coupling files. The results are compared to the MCUF field of ARPEGE and the results of same methods applied to the coupling files from ARPEGE. Most methods give a signal for the RMPDs, but DFI reduces the storms too much to be detected. The coupling error function without filtering and amplitude have more noise, but the signal of a RMPD is also stronger. The methods are tested for NWP LAM ALADIN, but could be applied to other LAMs and benefit the performance of climate LAMs. Usually, LAMs use higher resolutions and more advanced parameterizations of physical processes than global numerical weather prediction models, but suffer from one additional source of error - the LBCs. The large scale model passes the information on its fields to LAM only over the narrow coupling zone at discrete times separated by a coupling interval of several hours. The LBC temporal resolution can be lower than the time necessary for a particular meteorological feature to cross the boundary. A LAM user who depends on LBC data acquired from an independent prior analysis or parent model run can find that usual schemes for temporal interpolation of large scale data provide LBC data of inadequate quality. The problem of a quickly moving depression that is not recognized by the operationally used gridpoint coupling scheme is examined using a simple one-dimensional model. A spectral method for nesting a LAM in a larger scale model is implemented and tested. Results for a traditional flow-relaxation scheme combined with temporal interpolation in spectral space are also presented. The work presented here shows that more frequent LBCs are important for forecasting small storms even when they develop inside the domain. Missing a storm in a LAM forecast due to infrequent LBCs has lead to a model tuning that enhances storm development. Unfortunately, the same tuning is not very supportive for the fog development.