Abstract

Master's Qualification Work

"Weather conditions forecasting by time series of meteorological data"

Introduction, actuality

A well-known expression "Everybody likes to talk about weather..." reflects that great interest, which person shows to his or her surrounding atmospheric environment. This interest touches another people and their communities and nations and human society in general. Daily weather forecast, for example, makes a person to take different decisions; whether to take an umbrella or no or to be in search of shelter of coming tornado.

In developed countries weather and climat a long time ago became economic categories. Every year in the world natural adversity take away 250 000 of human lives, the degree of damage causing to the property, lies within the limits of 50-100 billion dollars of the USA. But the world's statistics shows: if entrust to hydrometeorologyc information and reach upon it sufficiently then it is possible prevent from 30 till 40 per cent of losses and completely avoid victims [1].

Especially noticeable economic effect gives the utilization of meteorological information of aviation, power engineering, building, fishing and navigation, agriculture.

Weather prediction from the scientific point of view - one of the most difficult tasks of physics atmosphere. Exist different methods for prediction meteorological effects and their sizes, for example, synoptical, numerical, statistics methods, but in whole volume no method provides accurate forecasting. Exactly that's why researches in the sphere of forecasting of weather conditions are important and useful and the theme is relevant.

Aims and Tasks

The aim of the master's diploma paper is the study of possibility of forecasting meteooperation factors according to their temporal lines and, ideally, forming and base service their basis of methodic of calculation of short-term weather forecasting. Such weathe lines are given in particular, to automatized meteostations of the type like Vantage Pro 2 [2]. One of such can be found at faculty of computer engineerings and information science DNTU.

The idea of the work consists in analysis and utilization of weather lines meteodegrees for creation of model forecasting adapted to Donetsk region.

Task, which are solving in the master's diploma paper :

• study literary sources according to methods of forecasting of weather lines;
• gathering and analysis of meteodatas for establishment statistics regularities;
• elaboration of the system of forecasting, evaluation of its adequacy;

The valum of the research is the system of meteorological activities of the city.

The object of the research are the models of weather forecasting.

The methods of the research is visual, statistics and spectral analysis of the facts, method of false neighbours determination of arguments of model functions, methods of approximation of functions of many alternatives with the help of generalized multinominals, radial basic functions, artificial neuronic nets.

Science novelty consists in:

• forming method of calculation of short-term weather forecasting on weather lines meteosizes;
• development of programme ensuring, realizing this method.

Research Summary

Despite boudless number of situations, object and aims it's possible to single out main phases of modeling. The work begins from the examination of given information concerning to the object from the point of aim of research, from accepting and preliminary analysis of lines of observing sizes and it finishes utilization of accepted model for solving of concrete problem. But usually this process is iterative - accompanied by repeated repetitions, returns on initialand intermediate phazes, consistent approachings to "good" model.

Let's examine in more detail of each phase [3].

On the first phase is realized receiving of facts and its systematization. Futher on it is used one or several methods of analysis having weather lines of watching meteosizes. This is, for example, a visual analysis in the form of diagrams of dependence variable of time, reconstruction of phase trajectory, spectraland statistic alanalysis and others.

On the crucial scheme of phase 2 is forming the structure of the model. Initially is choosed a type of equation and then is established a connection of dynamic variable with watching sizes.

Forming the structure of model - is the the most difficult and creative phase of the procedure of modeling. On the given phase is chosen a type of equation, a type of coming to it functions and facts.

The task of facts determination consists in determination the least dimension of the models, providing unambiguity of forecasting. For solving this problem are used different methods of evaluation, the method of false neighbours, the method of main components, the Grassberg-Procachia's method, the method of well fit basis.

Futher on goes the phase of determination of the structure modeling equation. For this are used different methods of approximation of functions of many variables: the method of generalized multinominal, using radial basis functions, artificial neuronic nets [4], local models, searching of close neighbours.

After choosing the structure is realized "adjustment of models". For this, as a rule, is realized the search of extreme meaning of some aim function, for example, minimize the sum of squares of solution deviation of model equation from watching units. Necessarily on the given phase are realized preliminary transformations of watching line: noise filtration, numeral differentiation or integration and etc. This, in general, a technical phase of numeral calculations, but even here it has to make choice of principle of calculations operation factors and method for its realization.

The choice has to be made and in final, starting to check up "quality" of model. As usual quality of the model is examined with the utilization of saving for its aim of testing part of the line. It's realized the examination of effectiveness of the model for achievement of necessary accuracy of forecasting [5]. If the model is accepted satisfactory, acceptes construction takes to business, otherwise - comes back on revision on any of phases.

Fig. 1 shows the functional scheme of weather forecast system.

Figure 1 – Functional scheme of weather forecast system
(animation: size - 28,1 КB; shots quantity - 5; infinite number of repetition cycles; delay between shots - 1,3 мs; delay between last and first shot - 1,3 мs)

To the present time the following phases were done:

• watching variables were examined, was established their gathering from local meteostation,which is established in DNTU;
• gathered weather lines of meteosizes were systematize and prepared to the analysis;
• carried out visual, spectral and statistical analysis of the facts and also the analysis of their mutual dependence;
• carried out the attempt of determination of facts functions of model by the method of false neighbours.

In turn - realization the following:

• reconstruction of the model weather forecasting, according to carried researches and its optimization;
• formation the method of calculation of short-term weather forecasting;
• creation of programme guaranteeing, which realizing processing and analysis of weather lines of meteosizes, dynamic determination of facts functions and tuning of parameters of the model, delivery of short-time weather forecasting of the given term.

Conclusions

Thus, the result of diploma work should become methods of short-time weather forecasting according to weather lines of meteoparameters and also its programme realization. Datas of the research and calculations can become useful for meteorologists and can be used for the futher researches and improvements in this sphere.

During the writing of present author's abstract master's diploma paper hasn't finished yet. The final readiness of the work is December 2009. The whole text and materials according to the theme can be received from the author and his or her leader after indicated data.

References

1. Кому и зачем нужен прогноз? Статья. Источник: http://www.primpogoda.ru/articles/prosto_o_pogode/komu_i_zachem_nuzhen_prognoz/
2. Гриценко А.В. Использование локальных метеостанций для прогноза погоды. Метеокомплекс VANTAGE PRO 2. – Охрана окружающей среды и рациональное использование природных ресурсов 2009 / Материалы VIII международной научной конференции аспирантов и студентов. – Донецк, ДонНТУ – 2009.
3. Безручко Б.П., Смирнов Д.А. Математическое моделирование и хаотические временные ряды. Саратов: ГосУНЦ «Колледж», 2005. – 320 с.
4. Горбань А.Н. Функции многих переменных и нейронные сети // Соросовский образовательный журнал. 1998. № 12. С. 105-112.
5. Гриценко А.В. Прогнозирование метеопараметров по временным рядам. – Компьютерный мониторинг и информационные технологии 2008 / Материалы IV международной научной конференции студентов, аспирантов и молодых ученых. – Донецк, ДонНТУ – 2009.