Institute of Geodesy and Geoinformatics
Takizawa, Iwate, Japan
Graduate School of Software and Information Science,
Wroclaw University of Environmental and Life Sciences,
E-mail: [email protected]
The major aim of this project is to present GIS as a useful tool for data verification, analysis and the presentation of spatial distribution, for example of climate parameters which depend on topographic conditions, over medium scale areas and a relatively long period of time. The reliability of the model of phenomena parameters’ distribution, established on the basis of measurements in points, is the one that was obtained for the least reliable parameter put into the model’s creation. It concerns both the location of the measurement’s place and parameter’s value. Subjective experience of an author may influence the modeling on the basis of insufficient number of points. Lack of knowledge on that fact may lead to drawing wrong conclusions when it comes to phenomenon distribution and to elaborating incorrect forecasts. Describing the probability of information transfer, while creating a model on the basis of insufficient number of points, may increase the reliability of that model.
This project analyzes the possibility of applying Geographical Information System (GIS) to spatial distribution of environmental research on the basis of measurements observed in point during the long periods of time. It presents a method based on assumption that if there are similar topographic conditions, it might be suggested that there are also similar conditions in the indicated area. Such assumptions can appear when it comes to the distribution of climate parameters, noise or air pollution. The second principle is a kind of uniformitarianism. If there were similar climatic conditions in the historical period in certain area we can assume, with high probability, that in this area there are still similar climatic conditions, on the condition that the topographic situation has remained unchanged. Due to that fact, in present circumstances, we can transfer information on the basis of fewer number of indicating points. On the basis of these guidelines, the digital maps of climatic conditions can be built. This is a proposition for these meteorological data fields that need to be constructed from point measurements and from spatialisation of obtained values from the whole examined area. Sometimes it is not possible to transfer point information from some places because there are different natural conditions, natural or anthropogenic barriers, or different times of observation than in the places from the nearest neighborhoods. It seems that it would be better to leave some areas terra incognita than to create false model of monitored phenomenon. These places are presently characterized with lack of data. Such a way of interpreting information from phenomenon’s model, indicates areas which need additional researches before making a decision conditioned by examined parameter’s distribution. Such a solution was tested along with developing the Spatial Information System in Department of Geodesy and Cartography of the Office of the Marshal of Lower Silesian Voivodship, Poland. Each region consists of elementary fields of climatic groundwork, which will be assigned to determine the type of climatic conditions with certain degree of climatic risk or described as “lack of data”.
The above mentioned method of spatial interpretation of phenomena measured in points can be also used when monitoring parameters indicated in elaborations for the needs of environment protection, geology, soil science, hydrology etc.
ANALYZING CONDITIONS OF MEASUREMENTS IN POINT
Location of measuring station
Measuring stations (points) should be located while technical and environmental conditions are steady. The information about the type of landscape for each station has been mentioned in the reports. Especially, the following should be given attention to: plain, seaside-plain, plain and lakes, wide river valley, valley slope, hills, steep slope at a lake, hills and lakes, slope of the plateau, foothills, plateau, large clearing. It means that certain measured values are valid only for the same landscape unit and further “spreading out of information” should be conditioned by the analysis of the surroundings. Such characteristics as a shape of the area near the station and convex or concave also have great influence on the data. Conditions of the roughness of the neighborhood, as well as the conditions of moving air masses transformation, the frequency of calms and light winds etc. should be taken into account during the analysis of the representation level of the stations’ location. The change of measuring point’s location can change the conditions of the observation. Stations are moved because of many different reasons. Nevertheless information about the change of the measuring point’s location should be noted in the data base. Changes of the surroundings that occur in time (construction of new buildings, growth of trees, etc.) should also be taken into consideration.
Density of network of measuring stations
Dense network of measuring stations is required especially in mountains and hills, because of very strong influence of the relief. Also location on the northern or southern slope influences measured parameters. The best solution is to place a station in each unit previously separated by different topographic conditions, as well as to make the data representative for the whole area. In figure 1. you can see that existing network does not meet those conditions. Polish network of measuring stations does not represent most of units, which were separated by topoclimatic conditions. Therefore, commonly used interpolation functions do not fulfil any conditions of good climate modelling.
Fig 1. Difficulties with forming spatial distribution of precipitation on the basis of insufficient number of measuring points in each unit separated by different topographic conditions
In the period 1891 – 1930, valuable data from 400 stations in Lower Silesia were accessible for verification but at present period the amount of stations equals only 60. Therefore, the assumption was that decrease in the amount of observation points had no influence on climatic conditions. Localization of those stations is additionally shown in figure 3 which illustrates the usage of historical data. If you would like to find information about historical data concerning stations measuring precipitation in Lower Silesia Region in Poland, see visualization of complex symbols at www.gislab.ar.wroc.pl (click homepage of the author). These information may play a role in navigation in all disseminate precipitation data in Lower Silesia Region. The website is constructed as multidimensional display in a map in Internet. Topographic base as well as measured data are shown in interactive Internet’s maps (Bac-Bronowicz, Cieslinski 2004).
Only many years’ averages of precipitation from the same period, obtained from continuous observation for at least thirty and better forty years, can be used for detailed elaborations for all stations located in studied area. Annual precipitations’ sums show the lowest variation in comparison with other periods of observation. In case of shorter period of observations the analyzes of data should be more detailed. In extreme cases, maximal daily precipitations can be used. For a certain number of stations in Lower Silesia the results from 115-year-long precipitation time series concerning years from 1891 to 2006 are available. It happens that even 30-years-long period can be unrepresentative because of uneven distribution of precipitation in longer perspective (for all 115-year period).
DISTINGUISHING PRECIPITATION REGIONS
The next stage of analysis consisted in distinguish the type of precipitation regions. This was achieved using multi-feature analyzes of precipitation conditions in measuring points, and altitude of the station above the sea-level. Next the areas around the station were enclosed in the regions indicated by mentioned above categorization.
The author presents reliable transfer of information to each elementary field around measuring station. Reliability of transfer would be different, depending on the distance and terrain topography. What is really crucial in that case is correct indication of probability distribution (Zhang at a., 2000; Bac-Bronowicz, 2007). As a first step, information is transferred depending on distance. Next the information is corrected according to differences of height between measuring station and basic fields around the point. The last way to transfer information from point to surrounding is meteorological uniformitarianism. If in elaborated area there were similar climatic conditions in the historical period in certain area we can assume (with high probability) that in this area there are still similar climatic conditions. Due to mentioned above facts, the author suggests that transfer of information, on the basis of fewer number of indicating points, is possible. On the basis of these guidelines, the digital map of climatic conditions can be presented.
Determination of typical groups connected with precipitation, based on taxonomic methods
Using multi-feature classification, nine types of regions were selected. Mean magnitude of precipitation in different seasons of the year, and other parameters influencing the precipitation were taken into consideration. Average sums of precipitation in periods: V+VI, VII+VIII, IV-IX and height of points above sea level were chosen to determine precipitation regions. In this elaboration used information is of special importance for agriculture. During elaboration of this article, the author examined various ways to indicate values to divide precipitation into types: field as Voronoi’s mosaic (Thiessen, 1911), cluster analysis (Irvin at a. 1996; Ventura at a. 2000, Zhu at a., 2001) and approximation using spline function. Classifications of the clustering of objects into different groups turned out to be the most accurate. The partitioning of data into clusters was adequately flexible according to users needs. Defining distance between multi-values and other parameters turned out to be effective for different number of points and combination in multi – feature data.
Construction of database
In order to carry out the analyzes whose aim was to separate areas in which the factors have similar influence on spatial information transfer, it was crucial to create spatial information system including chosen attributes necessary for given analyzes and determining the subsequent zones of information transfer to measuring stations.
Selection of reference unit, which is accommodated to the needs, and accuracy of the compilation scales are a very important point in the construction of the spatial information system for climate. After the analyzes of elaborations of parameters’ distribution, the basic fields of 1 km square were chosen. In Poland one of the basic systems is the TEMKART. The initial unit is a trapezium with sides that correspond with one degree in geographic reference system, divided into fields with sides 10′ and 5′. Then the field is divided into 9 rows and 12 columns. In elaborated area units are about 1 km square and fluctuate between 0,981 and 1,022 km square. Further division is possible using quadruple system.
There is one line in tables together with data and metadata for each field. For each field in the system, there was information assigned about its connection with station’s surroundings (number, zone of information transfer etc.), and connection with physico-geographical unit. Connection has been classified into the following division: 30%, 30-50%, 50-70%, 100%. It is significant because it often happens that one basic field appears in the zone of influence of two or more measuring stations. Having tables created in such a way, we can decide from which station information is more reliable because it directly indicates the number of information transfer zone. The smaller it is the more reliable the information is. What is problematic is information transfer if there is the same number of surroundings and in such case, what should also be taken into account is additional information, for example – land cover, climatic parameters distribution in other measurement periods etc. In figure 2. coverage by 9. zones of information transfer of Lower Silesia is presented.
Fig 2. Coverage by 9. zones of Lower Silesia
Ascribing the probability of information transfer to separate zones according to their distance from the station and difference of height
In this proposition, the values of reliability of information transfer from point to surroundings, depend on the distance (between measuring point and elementary field) and terrain topography. What is really crucial in that case is correct indication of probability distribution.
In the first phase, each basic field included in the zones of information transfer of measurement station, has been evaluated depending on the distance from the point which indicated value of its features. In the first zone, that is in the field in which there is meteorological station, information is certain because that is the place in which measurements took place. That is why the probability of information transfer is