Modification of meteorological forecast preparation technology feasible following implementation of the numerical forecasing models

Weather forecasting is a complex learning process, involving both objective methods and subjective interpretation of numerical modelling results. The increasing demand of the users, who require more exact forecast timing and more precise territory indication as well as the growing dependency of the economy and population upon the weather and natural environment conditions necessitate continuous improvement of the meteorological forecasting system. These requirements can be met exclusively through the launch and routine operation by IMGW of the numerical mezoscale forecasting model with the use of the IMGW own computer hardware.

The aim is to put in place the hydrodynamic, non-hydrostatic forecasting model with the calculation grid in the order of 2.5 x 2.5 km and from 35 to 50 atmosphere layers, adjusted to the orography conditions in Poland. The com-puter hardware should enable 4 recalculation cycles of one or a few models every 24 hours, including hydrological models and the models of pollutant distribution in the atmosphere.

Given the current development in the syn-optic meteorology any further improvement of weather forecasting, in particular the quantitative precipitation forecast is not feasible without complex systems, supported with modern technical measures for data collection and processing and hydrodynamic simulation methods, i.e. the objective forecasting methods. Today the IMGW does not operate any proprietary numerical mezoscale model, although it participates in the development of two such models.

In 1997, thanks to the use of data generated by the German (EM), British (UMPL) operated in Poland and French (ALADIN) numerical model calculated in the Czech Republic there were forecasted with the two days notice two precipitation waves, i.e. those starting on the 6 and 18 July. In respect of the second case, the models enable accurate forecast of the precipitation wave time, territory and intensity. The precipitation could not be forecast with such notice with the use of traditional synoptic methodologies. The decision making process, which preceded issue of the warning was strongly im-pacted by the fact that two (in some cases three) independent models confirmed the forecast of the precipitation time, territory and intensity.

Similarly in 1998 the IMGW hydrological and meteorological sevice using results of numerical models accurately forecasted a number of extreme meteorological and hydrological phenomena.

The system of numerical mezoscale forecasting compared to the other existing models, features high time and spatial resolution. Currently forecasts are calculated based of the 17 x 17 km grid (150 x 150 km in the hitherto models) and the time raster of over 10 minutes (between 6 and 12 hours in the hitherto models). Hence, the model enables objective forecasting of atmospheric phenomena, well identified in terms of time and territory. The capacity of quantitative forecasting of the phenomena like precipitation, wind velocity or vertical air streams is an important feature of such models. The development of the research concerning models and the increase in the processing capacity of computers leads to the development of the new class of models that is the non-hydrostatic models with higher time and spatial resolution and greater forecasting accuracy.

Therefore, the production start-up of the mezoscale model and its development into the non-hydrostatic model, fully adjusted to the Polish conditions is one of the prerequisites for the increase of the meteorological and hydrological function credibility. The task includes practical implementation of mezoscale, ultimately non-hydrostatic models with 2.5 x 2.5 spatial resolution and the vertical resolution of up to 50 atmosphere layers. The model should cover the entire area of Poland, partially reaching the territories of the neighbouring countries.

Cooperation will be continued with the meteorological organisations involved in the development of the ALADIN mezoscale meteorological model and with the Interdisciplinary Mathematical Modelling Centre of the Warsaw University and the United Kingdom Meteorological Office (in respect of the UKMO model). Moreover, cooperation will enhance with other centres performing similar exercise to ensure the model development in line with the latest scientific achievements. Under its research programme IMGW will make efforts to use in an optimum manner the results of the meteorological model in hydrological and other related models.

Training on the modelling methodology, numerical methods and forecast models should be delivered in the leading international research centres of WMO and other institutions. Due to lasting duration of the Project, the training should not be a one-time exercise, but rather a process of adjustments to the modified organisa-tion, methodology and technology requirements.

Live calculation of the model, particularly in case of any atmospheric hazard requires installation and purchase of a super-computer with the processing capacity adjusted to the requirements resulting from the parameters of non-hydrostatic model, other models to be calculated in extreme situations and in the case of the research and development work required. The hardware for the numerical model must have the capacity corresponding to the model type and number of the processing cycles per 24 hours. It should also feature the development potential and cost corresponding to the need of the processing capacity increase at the subsequent stages of the numerical model upgrade. It is important to note that the processing load is unevenly spread over time depending on cycles of data collection and access to boundary conditions of other numerical models.

In the 3-year period scheduled for the project delivery substantial progress in terms of computer processing capacity is envisaged. Hence the potential of significant development of the mezoscale, ultimately non-hydrostatic models. Thus, the computer equipment procured should have a clear development track over the coming 3 to 4 years.

During the July 1997 flood weather forecasting was based on up to 48 hours numerical weather forecasting models: EM-DWD (1), UMPL (2), ALADIN (3)

TCP/IP will be the standard protocol for data transmission from the model to the national data transmission system. External users will access the desired model results through the Client Service System (SOK). The format for the data and data exchange through the Global Data Processing System (the GDPS) will comply with the GTS standard. In view of Poland's participation in the GTS upgrade programme the results should be transmitted by means of the TCP/IP and X.400.

The model results will be accessible for other, e.g. hydrological models through the national data transmission system. Following completion of processing, the selected group of the model results will; be transferred through the LAN (an element of the data transmission system) and made available to the external users, particularly those logged into the Internet through the Client Service System. The new telecommunications infrastructure should be put in place to enable parallel processing of a few numerical models and the process's switch over to other centres, outside of the IMGW.