{"id":757,"date":"2015-09-03T10:07:26","date_gmt":"2015-09-03T10:07:26","guid":{"rendered":"http:\/\/www.vtei.cz\/?p=757"},"modified":"2024-07-16T10:10:23","modified_gmt":"2024-07-16T09:10:23","slug":"recent-developments-of-the-bilan-model","status":"publish","type":"post","link":"https:\/\/www.vtei.cz\/en\/2015\/09\/recent-developments-of-the-bilan-model\/","title":{"rendered":"Recent developments of the BILAN model"},"content":{"rendered":"

<\/i> This article is available in Czech only. For translation or more information on this topic, please contact author.<\/h4>\n

Souhrn<\/h2>\n

Model BILAN je pou\u017e\u00edv\u00e1n k\u00a0simulac\u00edm hydrologick\u00e9 bilance na\u00a0\u010desk\u00fdch i\u00a0evropsk\u00fdch povod\u00edch v\u00a0\u0159ad\u011b projekt\u016f aplikovan\u00e9ho v\u00fdzkumu i\u00a0hydrologick\u00fdch studi\u00ed. P\u0159\u00edsp\u011bvek p\u0159edstavuje v\u00fdvoj tohoto modelu b\u011bhem n\u011bkolika posledn\u00edch let. Popisuje jednak n\u011bkter\u00e9 zm\u011bny v\u00a0modelu samotn\u00e9m (za\u010dlen\u011bn\u00ed u\u017e\u00edv\u00e1n\u00ed vod, mo\u017enost kalibrace soustavy povod\u00ed, ukl\u00e1d\u00e1n\u00ed stavov\u00fdch veli\u010din), jednak se zab\u00fdv\u00e1 p\u0159epracovan\u00fdm u\u017eivatelsk\u00fdm rozhran\u00edm, dopln\u011bn\u00fdm o\u00a0grafick\u00e9 v\u00fdstupy a\u00a0dal\u0161\u00ed prvky pro interaktivn\u00ed pr\u00e1ci s\u00a0modelem. Zm\u00edn\u011bno je tak\u00e9 rozhran\u00ed pro prost\u0159ed\u00ed R a\u00a0jeho webov\u00e1 verze. Z\u00e1v\u011brem jsou uvedeny p\u0159\u00edklady pou\u017eit\u00ed modelu ve\u00a0studi\u00edch vy\u017eaduj\u00edc\u00edch simulaci chov\u00e1n\u00ed povod\u00ed, a\u0165 u\u017e pro sou\u010dasn\u00fd stav, nebo pro podm\u00ednky klimatick\u00e9 zm\u011bny.<\/p>\n

\u00davod<\/h2>\n

Konceptu\u00e1ln\u00ed model BILAN, simuluj\u00edc\u00ed hydrologickou bilanci v\u00a0denn\u00edm \u010di m\u011bs\u00ed\u010dn\u00edm \u010dasov\u00e9m kroku (popis modelu uv\u00e1d\u00ed nap\u0159\u00edklad Tallaksen a\u00a0van Lanen, 2004), je ve\u00a0V\u00fdzkumn\u00e9m \u00fastavu vodohospod\u00e1\u0159sk\u00e9m T. G. Masaryka vyv\u00edjen a\u00a0pou\u017e\u00edv\u00e1n od\u00a090. let 20. stolet\u00ed. V\u00a0roce 2011 byla p\u016fvodn\u00ed softwarov\u00e1 implementace modelu BILAN, napsan\u00e1 v\u00a0jazyce Object Pascal, kompletn\u011b p\u0159eps\u00e1na do\u00a0jazyka C++, \u010d\u00edm\u017e se v\u00fdrazn\u011b zjednodu\u0161il dal\u0161\u00ed v\u00fdvoj modelu. Z\u00e1rove\u0148 byla vytvo\u0159ena dv\u011b rozhran\u00ed k\u00a0modelu (popsan\u00e1 Beranem aj., 2011): grafick\u00e9 u\u017eivatelsk\u00e9 rozhran\u00ed (GUI) zalo\u017een\u00e9 na\u00a0multiplatformn\u00ed knihovn\u011b Qt a\u00a0bal\u00edk pro statistick\u00e9 a\u00a0programovac\u00ed prost\u0159ed\u00ed R (R Core Team, 2015). Ob\u011b rozhran\u00ed se vz\u00e1jemn\u011b dopl\u0148uj\u00ed \u2013 GUI zp\u0159\u00edstup\u0148uje model \u0161irok\u00e9mu spektru u\u017eivatel\u016f a\u00a0d\u00edky vizualizac\u00edm p\u0159edstavuje efektivn\u00ed n\u00e1stroj pro kalibraci povod\u00ed, bal\u00edk pro R umo\u017e\u0148uje pokro\u010dil\u00fdm u\u017eivatel\u016fm vyu\u017e\u00edt v\u00fdhod hromadn\u00e9ho zpracov\u00e1n\u00ed a\u00a0skriptov\u00e1n\u00ed modelu v\u00a0kombinaci s\u00a0rozs\u00e1hl\u00fdmi mo\u017enostmi poskytovan\u00fdmi vlastn\u00edm prost\u0159ed\u00edm R. Model i\u00a0ob\u011b rozhran\u00ed jsou nad\u00e1le udr\u017eov\u00e1ny a\u00a0rozv\u00edjeny podle pot\u0159eb v\u00fdzkumn\u00fdch \u00fakol\u016f a\u00a0po\u017eadavk\u016f u\u017eivatel\u016f. Samostatn\u011b byl pops\u00e1n nov\u011b implementovan\u00fd optimaliza\u010dn\u00ed algoritmus vyu\u017e\u00edvaj\u00edc\u00ed evolu\u010dn\u00ed metody (M\u00e1ca aj., 2013), tento p\u0159\u00edsp\u011bvek poskytuje p\u0159ehled dal\u0161\u00edch nov\u00fdch vlastnost\u00ed a\u00a0mo\u017enost\u00ed modelu. Z\u00e1klady propojen\u00e9ho modelu BILAN jsou uvedeny v\u00a0\u010dl\u00e1nku (Vizina a\u00a0Hanel, 2011). Porovn\u00e1n\u00ed v\u00fdpo\u010detn\u00edho algoritmu pro denn\u00ed a\u00a0m\u011bs\u00ed\u010dn\u00ed verzi je uveden v\u00a0p\u0159\u00edsp\u011bvku (Hor\u00e1\u010dek aj., 2009).<\/p>\n

Zm\u011bny v\u00a0modelu a\u00a0jeho softwarov\u00e9 implementaci<\/h2>\n

V\u00fdpo\u010detn\u00ed j\u00e1dro<\/h3>\n

Do\u00a0bilan\u010dn\u00edch rovnic modelu byly zahrnuty veli\u010diny u\u017e\u00edv\u00e1n\u00ed vod, co\u017e roz\u0161\u00ed\u0159ilo oblast pou\u017eit\u00ed modelu na\u00a0\u0159e\u0161en\u00ed vodohospod\u00e1\u0159sk\u00e9 bilance. V\u00a0p\u0159\u00edpad\u011b zvolen\u00e9ho u\u017e\u00edv\u00e1n\u00ed do\u00a0modelu vstupuj\u00ed neovlivn\u011bn\u00e9 \u0159ady pozorovan\u00fdch odtok\u016f spolu s\u00a0veli\u010dinami u\u017e\u00edv\u00e1n\u00ed vod, jimi\u017e jsou odb\u011bry z\u00a0povrchov\u00fdch vod, odb\u011bry z\u00a0podzemn\u00edch vod a\u00a0vypou\u0161t\u011bn\u00ed do\u00a0povrchov\u00fdch vod. Experiment\u00e1ln\u011b je dopl\u0148uj\u00ed neevidovan\u00e9 odb\u011bry z\u00a0povrchov\u00fdch vod, kter\u00e9 maj\u00ed b\u00fdt iterativn\u011b odhadov\u00e1ny podle vodohospod\u00e1\u0159sk\u00e9 bilance. Nov\u011b je mo\u017en\u00e9 ukl\u00e1dat aktu\u00e1ln\u00ed hodnoty stavov\u00fdch veli\u010din modelu (z\u00e1soby v\u00a0n\u00e1dr\u017e\u00edch) a\u00a0n\u00e1sledn\u011b ulo\u017een\u00fd stav vyu\u017e\u00edt p\u0159i v\u00fdpo\u010dtu za\u010d\u00ednaj\u00edc\u00edm ve\u00a0zvolen\u00e9m \u010dasov\u00e9m kroku. Toho se vyu\u017e\u00edv\u00e1 nap\u0159\u00edklad p\u0159i simulaci nov\u011b dopln\u011bn\u00e9 \u010d\u00e1sti \u010dasov\u00e9 \u0159ady, je\u017e navazuje na\u00a0st\u00e1vaj\u00edc\u00ed \u0159adu, pro ni\u017e jsou zn\u00e1m\u00e9 parametry modelu. V\u00a0nastaven\u00ed optimalizace je novinkou mo\u017enost stanovit \u0159adu vah, kter\u00e9 budou pou\u017eity p\u0159i v\u00fdpo\u010dtu optimaliza\u010dn\u00edho krit\u00e9ria. Lze tak zv\u00fd\u0161it vliv v\u00fdznamn\u00fdch \u00fasek\u016f \u0159ady, nebo naopak eliminovat obdob\u00ed nespolehliv\u00fdch dat. Optimaliza\u010dn\u00ed krit\u00e9rium se voliteln\u011b po\u010d\u00edt\u00e1 tak\u00e9 z\u00a0\u0159ad z\u00e1kladn\u00edho odtoku (modelovan\u00e9ho a\u00a0ur\u010ditou metodou odvozen\u00e9ho z\u00a0pozorov\u00e1n\u00ed). Model BILAN je koncipov\u00e1n jako celistv\u00fd, s\u00a0parametry reprezentuj\u00edc\u00edmi cel\u00e9 povod\u00ed a\u00a0kalibrovan\u00fdmi pro toto povod\u00ed nez\u00e1visle na\u00a0okoln\u00edch. P\u0159i tomto p\u0159\u00edstupu se tud\u00ed\u017e v\u00a0hodnot\u00e1ch parametr\u016f nemus\u00ed odrazit podobnost sousedn\u00edch povod\u00ed a\u00a0v\u00a0ur\u010dit\u00fdch p\u0159\u00edpadech m\u016f\u017ee model simulovat odtoky z\u00a0mezipovod\u00ed jako z\u00e1porn\u00e9. Tento nedostatek vedl k\u00a0tomu, \u017ee byla p\u0159ipravena optimalizace, kter\u00e1 zohled\u0148uje vztah mezi odtokem z\u00a0v\u00fd\u0161e polo\u017een\u00e9ho a\u00a0n\u00ed\u017ee polo\u017een\u00e9ho povod\u00ed, jeho\u017e je men\u0161\u00ed povod\u00ed sou\u010d\u00e1st\u00ed. Libovoln\u00fd po\u010det povod\u00ed je tak mo\u017eno uspo\u0159\u00e1dat do\u00a0soustavy, v\u00a0n\u00ed\u017e je stejn\u011b jako doposud hydrologick\u00e1 bilance po\u010d\u00edt\u00e1na pro ka\u017ed\u00e9 povod\u00ed zvl\u00e1\u0161\u0165. P\u0159i optimalizaci parametr\u016f t\u011bchto povod\u00ed se vyu\u017e\u00edvaj\u00ed stejn\u00e9 algoritmy jako v\u00a0p\u0159\u00edpad\u011b samostatn\u00fdch povod\u00ed, li\u0161\u00ed se v\u0161ak v\u00a0optimaliza\u010dn\u00edm krit\u00e9riu, kter\u00e9 pro soustavu p\u0159id\u00e1v\u00e1 penalizaci pro z\u00e1porn\u00fd odtok z\u00a0mezipovod\u00ed mezi n\u00ed\u017ee a\u00a0v\u00fd\u0161e polo\u017een\u00fdm povod\u00edm. Krom\u011b t\u011bchto hlavn\u00edch roz\u0161\u00ed\u0159en\u00ed byly dopln\u011bny funkce t\u00fdkaj\u00edc\u00ed se kop\u00edrov\u00e1n\u00ed modelu, na\u010d\u00edt\u00e1n\u00ed a\u00a0ukl\u00e1d\u00e1n\u00ed vstupn\u00edch a\u00a0v\u00fdstupn\u00edch soubor\u016f, transformace veli\u010din, zach\u00e1zen\u00ed s\u00a0chyb\u011bj\u00edc\u00edmi hodnotami apod.<\/p>\n

Grafick\u00e9 u\u017eivatelsk\u00e9 rozhran\u00ed<\/h3>\n

Jeliko\u017e byla p\u0159edchoz\u00ed verze grafick\u00e9ho u\u017eivatelsk\u00e9ho rozhran\u00ed ur\u010dena pro vyhodnocov\u00e1n\u00ed jedin\u00e9ho povod\u00ed, bylo rozhran\u00ed z\u00e1sadn\u011b upraveno, aby bylo mo\u017eno pracovat s\u00a0v\u00edce povod\u00edmi v\u00a0soustav\u011b. V\u00a0sou\u010dasnosti se skl\u00e1d\u00e1 ze t\u0159\u00ed hlavn\u00edch odd\u00edl\u016f: seznamu na\u010dten\u00fdch povod\u00ed \u010di variant v\u00fdpo\u010dtu t\u00e9ho\u017e povod\u00ed, nastaven\u00ed v\u00fdpo\u010dtu a\u00a0optimaliza\u010dn\u00edho algoritmu (v\u010detn\u011b metody v\u00fdpo\u010dtu potenci\u00e1ln\u00ed evapotranspirace, stanoven\u00ed mez\u00ed parametr\u016f, p\u0159\u00edpadn\u011b uspo\u0159\u00e1d\u00e1n\u00ed povod\u00ed v\u00a0soustav\u011b apod.) a\u00a0p\u0159ehledu v\u00fdsledn\u00fdch hodnot s\u00a0jejich vizualizac\u00ed. Grafick\u00e9 prost\u0159ed\u00ed modelu je zobrazeno na\u00a0obr. 1<\/em>. U\u00a0seznamu povod\u00ed byly dopln\u011bny n\u00e1stroje pro manipulaci se vstupn\u00edmi daty. Je mo\u017en\u00e9 vytv\u00e1\u0159et duplik\u00e1ty vstupn\u00edch soubor\u016f (klonov\u00e1n\u00ed), pop\u0159. vstupn\u00ed data transformovat (a\u00a0to i\u00a0po\u00a0jednotliv\u00fdch m\u011bs\u00edc\u00edch). Tyto mo\u017enosti p\u0159isp\u00edvaj\u00ed k\u00a0efektivn\u00edmu srovn\u00e1v\u00e1n\u00ed variant v\u00fdpo\u010dtu a\u00a0p\u0159\u00edprav\u011b sc\u00e9n\u00e1\u0159ov\u00fdch dat, a\u0165 u\u017e pro meteorologick\u00e9 veli\u010diny, nebo u\u017e\u00edv\u00e1n\u00ed vod. \"bilan-obr1\"<\/a><\/p>\n

Obr. 1. U\u017eivatelsk\u00e9 prost\u0159ed\u00ed modelu BILAN
\nFig. 1. User interface of the BILAN model<\/h6>\n

Novou volbou v\u00a0nastaven\u00ed optimalizace je pou\u017eit\u00ed \u0159ady vah, kter\u00e9 lze automaticky nastavit jako nulov\u00e9 pro chyb\u011bj\u00edc\u00ed \u010di z\u00e1porn\u00e9 hodnoty pozorovan\u00e9ho odtoku, p\u0159\u00edpadn\u011b pro u\u017eivatelem zadan\u00e9 obdob\u00ed. U\u00a0modelu s\u00a0denn\u00edm \u010dasov\u00fdm krokem lze nov\u011b vypo\u010d\u00edtat \u0159adu z\u00e1kladn\u00edho odtoku metodou klouzav\u00fdch minim z\u00a030denn\u00edho \u010dasov\u00e9ho intervalu, kter\u00e1 m\u016f\u017ee b\u00fdt vyu\u017eita p\u0159i takov\u00e9 kalibraci, je\u017e z\u00e1kladn\u00edmu odtoku p\u0159i\u0159azuje ur\u010ditou v\u00e1hu. Proto\u017ee optimalizace soustavy vyu\u017e\u00edv\u00e1 stejn\u00e9 algoritmy jako optimalizace samostatn\u00e9ho povod\u00ed, neli\u0161\u00ed se ani jej\u00ed nastaven\u00ed. Je v\u0161ak samoz\u0159ejm\u011b nutn\u00e9 zadat strukturu soustavy, kter\u00e1 se schematicky zobraz\u00ed jako stromov\u00fd seznam. Nastaven\u00ed optimalizace je mo\u017en\u00e9 ulo\u017eit do\u00a0prost\u00e9ho textov\u00e9ho souboru pro op\u011btovn\u00e9 pou\u017eit\u00ed. V\u00fdsledky modelov\u00e1n\u00ed, ukl\u00e1dan\u00e9 do\u00a0v\u00fdstupn\u00edch soubor\u016f, zahrnuj\u00ed hodnoty kalibrovan\u00fdch parametr\u016f, denn\u00ed \u010di m\u011bs\u00ed\u010dn\u00ed \u010dasov\u00e9 \u0159ady veli\u010din a\u00a0jejich m\u011bs\u00ed\u010dn\u00ed charakteristiky. P\u0159id\u00e1no bylo grafick\u00e9 zn\u00e1zorn\u011bn\u00ed u\u017eivatelem zvolen\u00fdch veli\u010din. K\u00a0dispozici jsou r\u016fzn\u00e9 druhy graf\u016f: denn\u00ed \u010di m\u011bs\u00ed\u010dn\u00ed \u0159ady, m\u011bs\u00ed\u010dn\u00ed charakteristiky (pr\u016fm\u011bry, minima, maxima, krabicov\u00e9 grafy) a\u00a0kvantilov\u00e9 grafy, je\u017e zahrnuj\u00ed \u010d\u00e1ry p\u0159ekro\u010den\u00ed, Gumbelovy extrem\u00e1ln\u00ed grafy (grafy po\u010d\u00e1te\u010dn\u00edch a\u00a0koncov\u00fdch \u00fasek\u016f \u010dar p\u0159ekro\u010den\u00ed) a\u00a0Q-Q graf pozorovan\u00e9ho a\u00a0modelovan\u00e9ho celkov\u00e9ho odtoku; tyto grafy lze vyhodnocovat tak\u00e9 pro zadanou sezonu. V\u0161echny grafy je mo\u017eno exportovat, u\u00a0graf\u016f \u0159ad lze interaktivn\u011b m\u011bnit m\u011b\u0159\u00edtko a\u00a0nastavovat obdob\u00ed pro kalibraci. Grafy umo\u017e\u0148uj\u00ed zobrazit v\u00edce povod\u00ed \u010di variant v\u00fdpo\u010dt\u016f najednou, co\u017e op\u011bt usnad\u0148uje expertn\u00ed kalibraci. Mo\u017enosti (p\u0159\u00edklady) grafick\u00fdch v\u00fdstup\u016f jsou uvedeny na\u00a0obr. 2<\/em>. \"bilan-obr2\"<\/a><\/p>\n

Obr. 2. P\u0159\u00edklady grafick\u00fdch v\u00fdstup\u016f modelu BILAN: a) \u010dasov\u00e9 \u0159ady (mod\u0159e \u2013 pozorovan\u00fd odtok, \u010derven\u011b \u2013 modelovan\u00fd odtok, hn\u011bd\u011b \u2013 nedostatkov\u00e9 objemy), b) krabicov\u00fd graf, c) Q-Q graf mezi modelovan\u00fdm a\u00a0pozorovan\u00fdm odtokem, d) \u010d\u00e1ry p\u0159ekro\u010den\u00ed pro odtoky a\u00a0nedostatkov\u00fd objem
\nFig. 2. Examples of graphical outputs from the BILAN model: a) time series (blue \u2013 observed runoff, red \u2013 modelled runoff, brown \u2013 deficit volumes), b) boxplot chart type, c) qq plot between modelled and observed runoff, d) flow duration curve for runoffs and deficit volumes<\/h6>\n

Ve\u0161ker\u00e9 v\u00fdstupy mohou b\u00fdt krom\u011b jednotek odtokov\u00e9 v\u00fd\u0161ky uv\u00e1d\u011bny tak\u00e9 v\u00a0jednotk\u00e1ch objemu. Z\u00a0funkc\u00ed j\u00e1dra byly dopln\u011bny tak\u00e9 veli\u010diny u\u017e\u00edv\u00e1n\u00ed vod, k\u00a0nim jsou nav\u00edc dopo\u010d\u00edt\u00e1v\u00e1ny nedostatkov\u00e9 objemy (Bonacci, 1993) za\u00a0p\u0159edpokladu, \u017ee u\u017eivatel ur\u010d\u00ed prahovou hodnotu (nap\u0159\u00edklad minim\u00e1ln\u00ed z\u016fstatkov\u00fd pr\u016ftok), kter\u00e1 m\u016f\u017ee b\u00fdt specifick\u00e1 pro ka\u017ed\u00fd m\u011bs\u00edc. Vliv nastaven\u00ed v\u00fdpo\u010dtu na\u00a0v\u00fdsledky lze n\u00e1zorn\u011b posuzovat v\u00a0re\u017eimu interaktivn\u00edho spou\u0161t\u011bn\u00ed modelu. V\u00a0n\u011bm doch\u00e1z\u00ed k\u00a0automatick\u00e9mu p\u0159epo\u010d\u00edt\u00e1n\u00ed modelu a\u00a0k\u00a0aktualizaci v\u00fdsledk\u016f v\u010detn\u011b graf\u016f p\u0159i jak\u00e9koliv zm\u011bn\u011b vstup\u016f (mezi n\u011b\u017e spadaj\u00ed volby jako nastaven\u00ed optimalizace, vstupn\u00edch parametr\u016f, transformace vstupn\u00edch veli\u010din nebo velikost prahov\u00e9 hodnoty). Grafick\u00e9 u\u017eivatelsk\u00e9 prost\u0159ed\u00ed modelu BILAN je dostupn\u00e9 po\u00a0domluv\u011b s\u00a0autory \u010dl\u00e1nku. K\u00a0dispozici je i\u00a0u\u017eivatelsk\u00e1 p\u0159\u00edru\u010dka dostupn\u00e1 na\u00a0webov\u00fdch str\u00e1nk\u00e1ch bilan.vuv.cz (v\u00a0\u010desk\u00e9m a\u00a0anglick\u00e9m jazyce), je\u017e pokr\u00fdv\u00e1 v\u0161echny vlastnosti programu a\u00a0jej\u00ed\u017e sou\u010d\u00e1st\u00ed je tak\u00e9 popis matematick\u00e9ho konceptu modelu.<\/p>\n

Bal\u00edk pro prost\u0159ed\u00ed R<\/h3>\n

Do\u00a0bal\u00edku pro prost\u0159ed\u00ed R nazvan\u00e9ho \u201ebilan\u201c (Ka\u0161p\u00e1rek aj., 2014) byla p\u0159evzata podpora pro nov\u00e9 vlastnosti j\u00e1dra (u\u017e\u00edv\u00e1n\u00ed vod, aktu\u00e1ln\u00ed stav, soustava povod\u00ed atd.), bal\u00edk na\u00a0rozd\u00edl od\u00a0GUI obsahuje ve\u0161kerou novou funkcionalitu. Bal\u00edk z\u00e1rove\u0148 slou\u017e\u00ed pro experimentov\u00e1n\u00ed s\u00a0nov\u011b p\u0159ipravovan\u00fdmi funkcemi, v\u00a0sou\u010dasnosti je v\u00a0n\u011bm tak mo\u017en\u00e9 pou\u017e\u00edvat libovoln\u00fd \u010dasov\u00fd krok (v\u00a0praxi nap\u0159\u00edklad t\u00fddenn\u00ed), testovat pozm\u011bn\u011bnou strukturu modelu \u010di vyu\u017e\u00edt p\u0159i kalibraci jin\u00e9 veli\u010diny ne\u017e celkov\u00fd a\u00a0z\u00e1kladn\u00ed odtok. Bal\u00edk obsahuje kompletn\u00ed dokumentaci v\u0161ech funkc\u00ed a\u00a0lze jej kompilovat na\u00a0v\u00edce platform\u00e1ch. V\u00a0z\u00e1v\u011bru roku 2013 byl zve\u0159ejn\u011bn na\u00a0ofici\u00e1ln\u00edm \u00falo\u017ei\u0161ti bal\u00edk\u016f pro R zvan\u00e9m CRAN (Comprehensive R Archive Network), kde z\u016fst\u00e1v\u00e1 v\u00a0sou\u010dasnosti archivov\u00e1n, kv\u016fli probl\u00e9m\u016fm nalezen\u00fdm dynamickou anal\u00fdzou k\u00f3du ji\u017e v\u0161ak tato verze nen\u00ed p\u0159\u00edmo p\u0159\u00edstupn\u00e1.<\/p>\n

Webov\u00e1 aplikace<\/h3>\n

Pro prost\u0159ed\u00ed R vzniklo n\u011bkolik n\u00e1stroj\u016f ur\u010den\u00fdch pro tvorbu webov\u00fdch aplikac\u00ed, z\u00a0nich\u017e se nejv\u00fdznamn\u011bj\u0161\u00edm stal bal\u00edk Shiny (Chang aj., 2015). Tento bal\u00edk umo\u017e\u0148uje snadno p\u0159ipravit webovou aplikaci sest\u00e1vaj\u00edc\u00ed z\u00a0u\u017eivatelsk\u00e9ho rozhran\u00ed a\u00a0serverov\u00e9 \u010d\u00e1sti, a\u00a0to pouze prost\u0159ednictv\u00edm jazyka R, bez znalosti webov\u00fdch technologi\u00ed. Takovou aplikaci lze provozovat jak lok\u00e1ln\u011b, tak pomoc\u00ed serverov\u00e9ho programu Shiny Server. Pro model BILAN je webov\u00e1 aplikace vyu\u017e\u00edvaj\u00edc\u00ed Shiny dostupn\u00e1 na\u00a0adrese bilan.vuv.cz\/shiny\/bilan. Aplikace obsahuje vzorov\u00e1 data pro povod\u00ed, lze v\u0161ak na\u010d\u00edst i\u00a0data vlastn\u00ed. Nastaven\u00ed kalibrace odpov\u00edd\u00e1 mo\u017enostem bal\u00edku, k\u00a0dispozici je rovn\u011b\u017e vytv\u00e1\u0159en\u00ed v\u00edce model\u016f a\u00a0jejich vz\u00e1jemn\u00e9 srovn\u00e1v\u00e1n\u00ed. To ve\u00a0spojen\u00ed s\u00a0n\u00e1stroji na\u00a0diagnostiku \u00fasp\u011b\u0161nosti modelu (v\u010detn\u011b r\u016fzn\u00fdch typ\u016f graf\u016f) nab\u00edz\u00ed podobn\u011b efektivn\u00ed zp\u016fsob kalibrace povod\u00ed jako desktopov\u00e1 aplikace, nev\u00fdhodou v\u0161ak je pomalej\u0161\u00ed odezva a\u00a0absence interaktivn\u00edho re\u017eimu. P\u0159\u00edklad v\u00fdstupu webov\u00e9 aplikace je uveden na\u00a0obr. 3<\/em>. \"bilan-obr3\"<\/a><\/p>\n

Obr. 3. Webov\u00e1 aplikace \u2013 zobrazen\u00ed empirick\u00fdch vztah\u016f mezi sr\u00e1\u017ekou a\u00a0odtokem, specifick\u00fdm odtokem a\u00a0indexem z\u00e1kladn\u00edho odtoku
\nFig. 3. Web application \u2013 showing the empirical relationships between precipitation and runoff, specific runoff and baseflow index<\/h6>\n

Mo\u017enosti pou\u017eit\u00ed modelu BILAN<\/h2>\n

Vyu\u017eit\u00ed modelu BILAN je mo\u017en\u00e9 rozd\u011blit do\u00a0n\u011bkolika \u00farovn\u00ed. Mezi prim\u00e1rn\u00ed pat\u0159\u00ed modelov\u00e1n\u00ed hydrologick\u00e9 bilance na\u00a0povod\u00ed, kter\u00e9 je hydrologicky uzav\u0159en\u00e9. Hydrologick\u00e1 bilance je slo\u017een\u00e1 z\u00a0odtokov\u00fdch (odtok povrchov\u00fd, p\u0159\u00edm\u00fd a\u00a0z\u00e1kladn\u00ed) slo\u017eek a\u00a0slo\u017eek z\u00e1sobn\u00edch (z\u00e1soba vody v\u00a0p\u016fd\u011b, podzemn\u00edch vod\u00e1ch a\u00a0ve\u00a0sn\u011bhu). Vazby mezi jednotliv\u00fdmi slo\u017ekami jsou reprezentov\u00e1ny jednotliv\u00fdmi parametry, kter\u00e9 lze p\u0159en\u00e1\u0161et na\u00a0hydrogeologicky podobn\u00e1 povod\u00ed. Tento p\u0159\u00edstup byl tak\u00e9 zvolen pro modelov\u00e1n\u00ed hydrologick\u00e9 bilance v\u00a0projektu Rebilance z\u00e1sob podzemn\u00edch vod (Ka\u0161p\u00e1rek, Hanel aj., 2014), kde pro cel\u00e9 hydrogeologick\u00e9 rajony, kter\u00e9 nejsou hydrologicky uzav\u0159en\u00e9, byly parametry modelu p\u0159eneseny z\u00a0jednotliv\u00fdch povod\u00ed vyskytuj\u00edc\u00edch se v\u00a0dan\u00e9m rajonu. Pro pilotn\u00ed povod\u00ed v\u00a0\u010cesk\u00e9 republice byly nav\u00edc odvozeny odhady hodnot parametru z\u00e1soby vody v\u00a0p\u016fd\u011b (Spa) a\u00a0parametru ovliv\u0148uj\u00edc\u00edho z\u00e1kladn\u00ed odtok (Grd). V\u00fdpo\u010detn\u00ed algoritmus modelu byl znovu ov\u011b\u0159en t\u00edm, \u017ee v\u00fdsledky byly porovn\u00e1v\u00e1ny s\u00a0v\u00fdstupy model\u016f HBV a\u00a0GR4J (Vizina aj., 2014). Dal\u0161\u00ed mo\u017enost\u00ed vyu\u017eit\u00ed modelu BILAN je modelov\u00e1n\u00ed dopad\u016f zm\u011bn klimatu na\u00a0hydrologickou bilanci, kdy je model nakalibrov\u00e1n pro sou\u010dasn\u00e9 podm\u00ednky a\u00a0n\u00e1sledn\u011b jsou parametry modelu vyu\u017eity pro modelov\u00e1n\u00ed v\u00fdhledov\u00fdch stav\u016f. Vstupem b\u00fdvaj\u00ed v\u00fdstupy z\u00a0region\u00e1ln\u00edch klimatick\u00fdch model\u016f nebo upraven\u00e1 vstupn\u00ed data, nap\u0159. zv\u00fd\u0161en\u00ed teploty vzduchu o\u00a01 \u00b0C. N\u00e1sledn\u011b je provedena anal\u00fdza zm\u011bn hydrologick\u00e9 bilance porovn\u00e1n\u00edm v\u00fdhledov\u00e9ho stavu se sou\u010dasn\u00fdm. Pro sou\u010dasn\u00e9 a\u00a0v\u00fdhledov\u00e9 podm\u00ednky je nav\u00edc mo\u017en\u00e9 modelovat dopady p\u0159i zm\u011bn\u011bn\u00e9m u\u017e\u00edv\u00e1n\u00ed vod, kter\u00e9 je uskute\u010dn\u011bno pomoc\u00ed transformace odb\u011br\u016f z\u00a0povrchov\u00fdch a\u00a0podzemn\u00edch vod. Transformace mohou b\u00fdt provedeny pomoc\u00ed p\u0159i\u010dten\u00ed \u010di ode\u010dten\u00ed v\u00a0jednotliv\u00fdch m\u011bs\u00edc\u00edch nebo pomoc\u00ed n\u00e1sobku jednotliv\u00e9ho odb\u011bru. V\u00a0modelu je t\u00e9\u017e mo\u017en\u00e9 nastavit hodnoty neevidovan\u00fdch odb\u011br\u016f \u010di jejich odhad. N\u00e1sledn\u011b je provedena simulace. Pro posouzen\u00ed, zda je dan\u00fd zdroj pro zvolen\u00fd odb\u011br dostate\u010dn\u00fd, je mo\u017en\u00e9 pou\u017e\u00edt v\u00fdpo\u010det nedostatkov\u00fdch objem\u016f. Nedostatkov\u00e9 objemy lze nastavit na\u00a0libovolnou limitn\u00ed hodnotu, a\u00a0to jako konstantn\u00ed, nebo variabiln\u00ed v\u00a0jednotliv\u00fdch m\u011bs\u00edc\u00edch. Obvykle se nastavuje hodnota minim\u00e1ln\u00edho z\u016fstatkov\u00e9ho pr\u016ftoku nebo pr\u016ftokov\u00e9 kvantily. V\u00a0modelu je mo\u017en\u00e9 simulovat soustavu v\u00edce povod\u00ed. V\u00fdsledky automatick\u00e9 kalibrace pro dv\u011b povod\u00ed jsou uvedeny na\u00a0obr. 4<\/em>. V\u00fdhodou tohoto propojen\u00ed je, \u017ee nedoch\u00e1z\u00ed k\u00a0z\u00e1porn\u00fdm odtok\u016fm na\u00a0mezipovod\u00ed, kter\u00e9 v\u00a0d\u0159\u00edv\u011bj\u0161\u00edch simulac\u00edch jednotliv\u00fdch povod\u00ed zp\u016fsobovalo probl\u00e9my p\u0159i n\u00e1sledn\u00fdch simulac\u00edch vodohospod\u00e1\u0159sk\u00fdch soustav. \"bilan-obr4\"<\/a><\/p>\n

Obr. 4. P\u0159\u00edklad optimalizace dvou povod\u00ed DBC 0170 a\u00a0DBC 0180 na\u00a0toku Metuje (\u010derven\u011b jsou zobrazeny odtoky pro ob\u011b povod\u00ed, mod\u0159e pozorovan\u00e9 odtoky)
\nFig. 4. An example of optimization of two river basins DBC 0170 and DBC 0180 of the Metuje River (red \u2013 modelled runoff for the both basins, blue \u2013 observed runoff)<\/h6>\n

Z\u00e1v\u011br<\/h2>\n

Model hydrologick\u00e9 bilance BILAN je vyv\u00edjen ve\u00a0V\u00fdzkumn\u00e9m \u00fastavu vodohospod\u00e1\u0159sk\u00e9m v\u00a0Praze od\u00a090. let 20. stolet\u00ed, av\u0161ak za\u00a0posledn\u00ed roky doznal v\u00fdrazn\u00fdch zm\u011bn, a\u00a0to p\u0159edev\u0161\u00edm v\u00a0pou\u017eitelnosti pro koncov\u00e9 u\u017eivatele. Hlavn\u00ed v\u00fdhodou modelu je nen\u00e1ro\u010dnost na\u00a0vstupn\u00ed data a\u00a0robustnost. Pro modelov\u00e1n\u00ed sta\u010d\u00ed denn\u00ed \u010di m\u011bs\u00ed\u010dn\u00ed \u0159ady teplot vzduchu a\u00a0sr\u00e1\u017ekov\u00fdch \u00fahrn\u016f. Sou\u010dasn\u00e1 verze modelu BILAN poskytuje plnohodnotn\u00e9 grafick\u00e9 u\u017eivatelsk\u00e9 prost\u0159ed\u00ed, kter\u00e9 prost\u0159ednictv\u00edm interaktivn\u00edch prvk\u016f a\u00a0grafick\u00fdch v\u00fdstup\u016f umo\u017e\u0148uje pohodlnou expertn\u00ed kalibraci modelu. V\u00fdhodou je tak\u00e9 mo\u017enost p\u0159izp\u016fsobit optimaliza\u010dn\u00ed algoritmus, nap\u0159\u00edklad pro optimalizaci n\u00edzk\u00fdch \u010di vysok\u00fdch pr\u016ftok\u016f. Hromadn\u00e9 zpracov\u00e1n\u00ed dat (nap\u0159. modelov\u00e1n\u00ed dopad\u016f klimatick\u00e9 zm\u011bny na\u00a0hydrologickou bilanci) usnad\u0148uje prost\u0159ed\u00ed R, v\u00a0n\u011bm\u017e lze model pou\u017e\u00edt ve\u00a0form\u011b bal\u00edku. Model je k\u00a0dispozci tak\u00e9 jako webov\u00e1 aplikace. Jednotliv\u00e9 verze modelu jsou dostupn\u00e9 po\u00a0domluv\u011b u\u00a0autor\u016f p\u0159\u00edsp\u011bvku.<\/p>\n

Pod\u011bkov\u00e1n\u00ed<\/h3>\n

\u010cl\u00e1nek vznikl na\u00a0z\u00e1klad\u011b v\u00fdsledk\u016f projektu \u201eUdr\u017eiteln\u00e9 vyu\u017e\u00edv\u00e1n\u00ed vodn\u00edch zdroj\u016f v\u00a0podm\u00ednk\u00e1ch klimatick\u00fdch zm\u011bn\u201c (TA01020508), kter\u00fd byl financov\u00e1n Techno-logickou agenturou \u010cesk\u00e9 republiky.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"

The BILAN model has been used in a number of research projects and hydrological studies dealing with estimation of water balance for European catchments. This paper is focused on the recent development of the model during last years. The model changes include both core development (new variables representing water use, optional calibration for a system of catchments, saving of state variables) and enhanced user interface that was extended by new graphical outputs and controls allowing interactive use of the model.<\/p>\n","protected":false},"author":8,"featured_media":742,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[86],"tags":[83,84,82,85],"coauthors":[27,33,34],"class_list":["post-757","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-hydraulics-hydrology-and-hydrogeology","tag-bilan-model","tag-graphical-user-interface","tag-hydrological-balance-modelling","tag-web-application"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/posts\/757","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/comments?post=757"}],"version-history":[{"count":2,"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/posts\/757\/revisions"}],"predecessor-version":[{"id":30293,"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/posts\/757\/revisions\/30293"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/media\/742"}],"wp:attachment":[{"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/media?parent=757"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/categories?post=757"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/tags?post=757"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/www.vtei.cz\/en\/wp-json\/wp\/v2\/coauthors?post=757"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}