Reflections on the theme of classifying, documenting and exchanging meteorological data
2005; Wiley; Volume: 6; Issue: 3 Linguagem: Inglês
10.1002/asl.112
ISSN1530-261X
Autores Tópico(s)Meteorological Phenomena and Simulations
ResumoThe weather systems on planet Earth all contain strong global elements, and the movements of mass and energy inside the thin envelope of air surrounding this planet do not respect the borders put up by the nations. So, the exchange of meteorological data sets connected to the man-made systems for making quantitative measurements and models for predicting the global weather as well as the regional and local weather is a concern of international character and is connected to many temporal and spatial scales. Three related subjects will be considered in the following text. First, general ideas of classification of meteorological and physical phenomena are presented and connected to a few actual systems for classifying weather and climate. Then, an interpretation of the scientific principle is presented together with a system for documentation of meteorological and physical parameters. In the concluding paragraphs, existing systems for exchanging meteorological data sets is considered and the possible challenges of exchanging such data in the future. The word ‘meteorology’ is of ancient Grecian origin, and etymologically, the word ‘meteoros’ signifies phenomena observed above the ground. The weather phenomena observed has different spatial and temporal dimensions (scale), and they are connected to a great many processes (the development of phenomena in time). The main processes considered in meteorology and in the biological processes involving meteorology are connected to the exchange of heat and water, the movement of weather systems (dynamics) and the water cycle in the atmosphere. The lower boundary of the atmosphere may be covered by vegetation. It may also be covered with bodies of water or ice; or, it may consist of barren land with sand or stone or some artificial man-made constructions. The studies of the dynamics (the movements of the air) of the atmosphere are to a great extent studies of circulations, eddies and waves on very different spatial and temporal scales. By using more ordinary concepts, we characterise the eddies as tropical cyclones, typhoon (western Pacific), hurricanes (in the Atlantic ocean and the Caribbean), extra tropical cyclones, tornados, rainstorms, gusts or turbulence; see (Wallace and Hobbs, 1977). Some circulations are global, and we find the Hadley Circulation, The Ferrel cells and the monsoon circulations; see (Wallace and Hobbs, 1977). Hills and mountains and the great ridges of the mountains of the continents are set up of long wave systems, gravity waves and gravity-inertial waves on different scales; see (Smith, 1979). We also have the wind systems connected to local surface effects and heating differences of local nature; see (Eliassen and Pedersen, 1980). There are a multitude of phenomena connected to the water in the atmosphere, the clouds, the different types of precipitation, evaporation of water from vegetation and the bodies of water on the earth. Classification of the meteorological phenomena mentioned is an important part of the science of meteorology, and the classification of climate is an important part of the studies of climate. The English chemist Luke Howard constructed a classification system of clouds in the beginning of the nineteenth century, and this system is still in use. The concepts of ‘air mass’ and ‘frontal systems’ as well as the basic ideas of synoptic meteorology and construction of synoptic charts were developed in Norway just after the first world war; see Petersen (1958) and Wallace and Hobbs (1977). The synoptic symbols now in use contain both quantitative measurements of meteorological parameters as well as observations of phenomena with rather vague quantitative features. In meteorology, a multitude of weather phenomena have got their names, and classification systems of the phenomena have been developed in dynamic meteorology, in physics of clouds and physics of radiation, in micrometeorology and other parts of physical meteorology, in upper air research, etc through the decades of the last century. But there is no unanimous agreement on the definition of several important meteorological phenomena. Examples are the definition of the temporal and spatial scales connected to the phenomena on the synoptic scale, the meso-scale, the local scale and the micro-scale; see Linacre (1992). One of the most widely used systems for classification of climate was developed in Germany by Wladimir Köppen during the years 1918–1936; see Köppen (1923). This classification system is connected to the temperature and the precipitation/evaporation of a region as well as to the natural vegetative cover of each region. Also, Thorntwait's classification system of climate, connected to the regimes of precipitation and temperature on the globe, is used in certain contexts; see Linacre (1992). According to WMO/OMM/BMO-No.182, 1992, the phrase ‘climatic classification’ is defined in the following way: ‘Division of the Earth's climates into a worldwide system of contiguous regions, each of which is defined by the relative homogeneity of its climatic elements. Examples are Köppen's and Thorntwait's climate classifications.’ I would then like to point at the idea of systematically using the modern tool of object-oriented analysis when constructing classes of meteorological phenomena in numerical models of weather and climate; see Sivertsen (2004, 2005a,b). The basic idea is that in each class or subclass of a phenomenon quantitative parameters/attributes should be attached to the phenomenon to describe it. In order to use a class of a weather phenomenon or a biological phenomenon in a numerical model, one or several numerical attributes have to actually be attached to the class; these attributes usually are called parameters. The attached set of attributes defines each phenomenon and makes it possible to implement it as a submodel of a numerical weather-modelling system. According to WMO/OMM/BMO-No.182, 1992, a ‘climatic element’ is defined in this manner: ‘Any one of the properties or conditions of the atmosphere which together define the climate of a place (e.g. temperature, humidity, precipitation).’ Linacre (1992) makes the following comment on the term ‘element of climate’: ‘It is not possible to measure the climate, but only the individual elements which comprise it. A climate element is any one of the various properties or conditions of the atmosphere which together specify the physical state of the weather or climate at a given place, for any particularly moment or period of time. On the other hand, a climatic factor like latitude or shading is a variable which controls a climatic element.’ I think the examples mentioned support my ideas that it is worthwhile to go on with this conceptual discussion on classification of meteorological phenomena and looking closer at the idea of systematically describing phenomena by numerical attributes/parameters. When doing scientific research in agrometeorology and in meteorology or by using the results of scientific research in practical operational applications, one leans on the scientific principle. In Figure 1, an interpretation of the scientific principle in the field of meteorology is presented. One important idea connected to this discussion is to be able to say something about the temporal and spatial scope of meteorological or agro-meteorological models and other applications used in operational contexts. Sivertsen (2004) discusses aspects of the scope of a specific application of agro-meteorology connected to a simple warning system of late blight in potato, and some general ideas of classification systems of meteorological parameters and meteorological phenomena is presented. Graphic representation of the scientific principle In Figure 1, a scheme containing what is considered the main steps of applied science is presented. By having the whole scheme in mind, it may sometimes be possible, when trying to answer some practical question by using an application in an operational context, to see that this practical question may have implications leading to reconsideration of the parameterisation of the phenomena, the use of ‘basic physical laws’, the mathematical and statistical hypotheses or the way measurements are made. The word ‘parameter’ is derived from two words of ancient Grecian origin; ‘para’ is a prefix which means ‘beside or subsidiary’ and ‘metron’ means ‘measure’ or ‘device for measuring’. The etymological meaning of the word ‘parameter’ then is ‘some element of nature made measurable’. According to the ‘International Meteorological Vocabulary’ (WMO/OMM/BMO-No.182), the concept ‘parameterisation’ is defined in this manner: ‘Approximate representation of subgrid-scale processes in a numerical model in terms of variables which are explicitly calculated’. The use of the term ‘parameterisation’ in this article is more general. When any physical, meteorological or biological phenomenon is described by attaching quantitative attributes to it, it is called parameterisation. The idea of parameterising is thus to construct quantitative models of nature by first classifying the phenomena (giving them names) and then attaching some system of measurable quantities to the phenomena (see Sivertsen, 2004). One of the most elegant ways of constructing models is by using modern object-oriented methods of analysing and programming complex numerical models. In an object-oriented system, the different sub-phenomena of a model are given certain names called classes, and to each class is attached quantitative attributes as well as formulas connecting the quantitative attributes. The classes are often constructed as some nested hierarchy; see (Brown, 1997). Name of the parameter Unit Definition Method(s) for measuring the parameter Representativeness for certain phenomena (models) The attribute above called ‘Representativeness’ will link the actual measured parameter to the different models or phenomena, and this attribute together with the attribute ‘Method(s) for measuring the parameter’ tells us something about the temporal and spatial resolutions of the parameter. Name of the parameter Unit Definition Representativeness of the phenomena of the model considered Representativeness for certain phenomena in other models The attribute ‘unit’ of a meteorological or biological parameter will as a rule contain one of the following units or some combination of the following units: meter (length), second (time), Ampére (strength of electrical current), Celsius (unit of temperature uniquely connected to the Kelvin scale of temperature), mole (amount of matter) and candela (strength of light). In addition to this, the parameter might be a pure integer or a pure real number. This system also could be linked to the CREX, the BUFR or the GRIB system; see Sivertsen (2004) and Section 6. In the frame of the WMO standard classification systems for exchange of meteorological data and related types of about agro-meteorological and biological geophysical data, called CREX and BUFR have been developed. These systems consist of combinations of metadata, classification and documentation of the data as well as standardisation of the format. Systems for the exchange of gridded data sets, called GRIB have also been developed. Conducted by WMO, there are two complementary systems for classification and exchange of meteorological data, the CREX/BUFR and the GRIB systems that have been developed. CREX is an acronym for ‘Character form for the Representation and Exchange of data’, and this is the former alphanumeric version of ‘Binary Universal Form for Representation of meteorological data’ with the acronym BUFR. There are automated methods for conversion of CREX code to BUFR code and vise versa. GRIB is an acronym for ‘GRIdded Binary’. GRIB is an efficient vehicle for transmitting large volumes of gridded data to automated centers over high-speed telecommunication lines. The gridded data is then the data contained in the numerical weather prediction models. GRIB may also serve as a data storage format. A definition of the BUFR form is given in WMO Manual on Codes, Guide to FM-94 BUFR. –Section number ‘0’ ‘Indicator section’ ‘BUFR’ (coded according to the CCITT International Alphabet, No. 5). –Section number ‘1’ ‘Identification section’ containing length of section, identification of the section message. –Section number ‘2’ ‘Optional section’ containing length of section and any additional items for local use by data-processing centers. –Section number ‘3’ ‘Data description section’ containing length of section, number of data section subsets, data category flag, data compression flag and a collection of data descriptors which define the form and content of individual data elements. –Section number ‘4’ ‘Data section’ containing length of section and binary data. –Section number ‘5’ ‘End section’ ‘7777’ (coded in CCITT International Alphabet, No. 5). The metadata of the BUFR system is contained in Sections 1, 2 and 3. Most of the data are observations, and ‘bit1’ in octet number 7 in Section 3 is set to ‘1’. If ‘bit1’ is set to ‘0’, this refers to other data, usually forecast information from some model. The metadata is contained in several tables, giving information about the ‘category’ of the data and descriptions of the types of quantitative information on the ‘parameter/observation’ considered. –Section number ‘0’ ‘Indicator section’ containing ‘GRIB’ (coded according to the CCITT International Alphabet, No. 5), total length of message in octets and edition number. –Section number ‘1’ ‘Product definition section (PDS)’ containing metadata on the parameters considered, description of the grid and its height, temporal information, forecast time unit, identification of computer center, etc. –Section number ‘2’ ‘Grid description section’ containing information on the grid used (type projection of mapping used), etc. –Section number ‘3’ ‘Bit map section (BMS)–optional’ contains information of parameter fields not defined in certain subsystems of the gridded model by a bit-map system. –Section number ‘4’ ‘Binary data section (BDS)’ contains the numerical data as binary data and the way the numerical data may be represented. –Section number ‘5’ ‘End section’ ‘7777’ (coded in CCITT International Alphabet, No. 5). This is a human-readable indication of the end of a GRIB record. The metadata of the ‘GRIB’ system is mainly contained in section numbers ‘1’ and ‘2’ and the interpretation is given in several tables. The ‘GRIB’ system is tailored for representation and exchange of the content of numerical weather prediction models. The metadata contained in the ‘BUFR’ and ‘GRIB’ systems is called meteorological elements. According to ‘International Meteorological Vocabulary’ (WMO/OMM/BMO-No.182, 1992), a ‘meteorological element’ is defined in the following manner: ‘Atmospheric variable which characterizes the state of the weather at a specific place at a particular time (e.g. air temperature, pressure, wind, humidity, thunderstorm and fog).’ There are several tables giving the interpretation and codes of the meteorological elements attached to ‘BUFR’ and ‘GRIB’. According to my opinion, this classification system consists of a mixture of phenomena and parameters describing the phenomena. This system (which is very flexible and has great scope) is absolutely possible to use; but my message is that this metadata part ought to be reconsidered according to the ideas discussed in Section 2. The conclusion is very short: One probably ought to look at the metadata systems of GRIB and BUFR to see if the classification systems may be constructed in a more logical way using methods from object-oriented analysis of the modern IT world? The methods of meteorology and research on climate are applied and used operationally in many contexts of modern societies, and these methods serves as tools for decision making in human societies. One could then ask, what is the scope of these scientific methods connected to quantitative parameterisation of natural phenomena and the quantitative prognoses of these phenomena? The basic intention of explicitly trying to clarify the different elements of the scientific method and then develop a system for documentation of parameters (measured entities as well as parameters in models) is to determine the scope of the scientific methods. Knowing the scope of the methods in spatial and temporal contexts, one may determine in each case when to use these methods and when not to apply them. The idea is that sometimes the methods are not applied, when they ought to be applied, and sometimes they are used outside their scope. The meteorological phenomena appear on many different temporal and spatial scales; the general scientific methods may be applied on the weather systems anywhere on the globe. The exchange of meteorological sets of data is important to discuss, especially as the possibilities of real-time exchange of data sets are increasing. What is probably possible to develop in the future are systems for exchange of data and information in almost real time between the observation systems on the ground (automated stations, weather radar systems, etc.) and the information from the satellites and the different working numerical weather prediction systems. Tor Håkon Sivertsen*, * The Norwegian Crop Research Institute, Fellesbygget Høgskoleveien 7 N-1432 Ås Norway
Referência(s)