Automated distinguishing of hydrographical channels in the georeference database/Automatizuotas hidrografijos kanalu isskyrimas Lietuvos georeferencinio pagrindo duomenu bazeje

Vykdant georeferenciniu duomenu bazes GDB10LT atnaujinimo darbus, linijiniai hidrografiniai objektai buvo skirstomi i tipus: upe, kanalas, griovys. Upes ir upeliai duomenu rinkinyje yra visi objektai, turintys pavadinimus. Kanalo ir griovio tipai isskiriami remiantis siu objektu apibrezimu LR aplinkos ministro isakyme. Duomenu prasme kanalu samparata buvo suformuluota kaip atkarpu, kuriomis galima sujungti du hidrografinius objektus, turincius pavadinimus (upes, upeliai, ezerai), rinkinys. Tokiu atkarpu isskyrima atliekant rankiniu budu, visoje Lietuvos teritorijoje reikalingos dideles darbo sanaudos, be to, galima palikti klaidu del neatidumo, kurias sunku patikrinti. Kanalu isskyrimas atliktas automatizuotai, grindziant grafu analize ir specifiniu tinklo jungu-mo skaiciavimo algoritmu. Buvo isskiriami potencialus kanalu itekejimo i kitus telkinius taskai bei skaiciuojamos visos galimos jungtys tarp siu tasku--tai ir buvo ieskomieji kanalai. Straipsnyje issamiau aptariami su duomenu specifika ir grafu analize susije klausimai, taip pat kanalu isskyrimo klausimai, padedantys identifikuoti duomenu kokybes problemas. Reiksminiai zodziai: hidrografijos tinklas, kanalai, GIS analize, grafu analize, georeferenciniai duomenys. The update of Lithuanian georeference database faced a task to classify linear hydrographical objects into a river, channel and ditch. Regarding the specificity of the obtained data, all hydrographical lines with names were identified as "rivers". The main problem was to separate remaining unnamed hydrographical lines in two categories, including a channel and ditch. The definition of channels and ditches is described in the law on the definition of protection zones for hydrographical objects and is accepted by the Minister of Environment. According to the specificity of the database, the definition of channels was formulated as a set of lines that can interconnect two hydrographic objects already having defined names (rivers, streams, lakes). The manual identification of such structures is very timeconsuming and requires analysis not only for separate objects but also for specific groups; visual analysis shows a high probability of making mistakes. Canal separation is performed in the automated way using geoinformational tools and graph analysis applying specific network connectivity calculation algorithm. The article discusses the issues of the preparation of the data structure for analysis, the specific graph analysis algorithm and the