A physical-statistical approach to the severe squall forecasting is considered. It is based on the output data of the regional hydrodynamic model of Hydrometeorological Center of Russia. By using the multi-dimensional discriminant analysis of a chosen set of predictors, a number of relationships are developed for the severe squall prediction; estimates of forecasting efficiency are presented for these relationships on the developmental sample. Physical interpretation of the obtained relationships is given.Squalls represent a phenomenon that causes natural extraordinary situations and fatalities. Up to date, the squall forecasting efficiency is not high enough. This can be partially explained by the fact that the squalls are mainly missed by the sparse observational network, low accuracy of measurements and indirect estimation of wind speed, and a higher complexity of the process, its parametrization, and a whole complex of factors which cause formation of squalls. Development of synoptic-physical-statistical approaches based on the modern hydrodynamic models with high spatial and temporal resolution helps to solve the problem of the squall automated forecasting and to improve its forecasting techniques. In the present paper, such an approach is presented.
INITIAL DATAConditions of severe squalls origination and their short-range forecasting are studied within the framework of the Model Output Statistics (MOS) concept, based on the output data of the regional model operatively used in Hydrometeorological Center of Russia [5] with 75-km horizontal resolution. This model has been previously used with a success for development of three methods to forecast adverse convective weather [2]; they have been recommended by Central Methodical Commission for practical use. In the presented paper, the 12-hour model forecasts from 00:00 UTC for the summer period (May 15 to September 15) of 2004-2006 are used for Western Europe and European Russia (30°-65°N, 0°-55°E). Information on squalls (i.e., actual wind speed, time, and place of its formation) was taken from the journal of reports on dangerous weather issued by the Hydrometeorological Center of Russia, from tables of dangerous weather for the Federation members, and from the web-sites dealing with natural disasters. The total sample consists of 127508 cases (one case corresponds to a single gridpoint), from which, including 211 cases of the squall occurrence (93 cases with wind speed from 15 to 24 m/s, 112 cases from 25 to 32 m/s, 6 cases of 33 m/s or more). In the points of the model grid, for the cases of squall occurrence, the real surface wind speeds are given; for the cases of non-occurrence, wind speeds from the database of operative objective analysis (OA) of the Hydrometeorological Center of Russia for 12:00 UTC were used. The OA data are interpolated in the points of the model grid. To include the real data on squall wind (which are collected manually) in the corresponding model gridpoints, the observation site coordinates being given, a special program is used.
S...