{"id":11683,"date":"2025-01-04T22:31:15","date_gmt":"2025-01-04T21:31:15","guid":{"rendered":"http:\/\/www.cmes.cz\/web\/?page_id=11683"},"modified":"2025-01-04T22:31:58","modified_gmt":"2025-01-04T21:31:58","slug":"asimilace-dat","status":"publish","type":"page","link":"http:\/\/www.cmes.cz\/web\/asimilace-dat\/","title":{"rendered":"ASIMILACE DAT"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-page\" data-elementor-id=\"11683\" class=\"elementor elementor-11683\">\n\t\t\t\t\t\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-480fafd elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"480fafd\" data-element_type=\"section\" data-settings=\"{&quot;_ha_eqh_enable&quot;:false}\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-8870eeb\" data-id=\"8870eeb\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-0d4b805 elementor-widget elementor-widget-text-editor\" data-id=\"0d4b805\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.17.0 - 08-11-2023 *\/\n.elementor-widget-text-editor.elementor-drop-cap-view-stacked .elementor-drop-cap{background-color:#69727d;color:#fff}.elementor-widget-text-editor.elementor-drop-cap-view-framed .elementor-drop-cap{color:#69727d;border:3px solid;background-color:transparent}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap{margin-top:8px}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap-letter{width:1em;height:1em}.elementor-widget-text-editor .elementor-drop-cap{float:left;text-align:center;line-height:1;font-size:50px}.elementor-widget-text-editor .elementor-drop-cap-letter{display:inline-block}<\/style>\t\t\t\t<div class=\"content clearfix\"><div class=\"field field-name-body field-type-text-with-summary field-label-hidden\"><div class=\"field-items\"><div class=\"field-item even\"><p><span style=\"color: #003366;\"><strong>Slovo \u00favodem<\/strong><\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Jak zn\u00e1mo, p\u0159edpov\u011b\u010f po\u010das\u00ed je vlastn\u011b probl\u00e9m zad\u00e1n\u00ed po\u010d\u00e1te\u010dn\u00edch podm\u00ednek. Z\u00e1vis\u00ed toti\u017e v prv\u00e9 \u0159ad\u011b na co nejlep\u0161\u00ed znalosti aktu\u00e1ln\u00edho stavu atmosf\u00e9ry. Tak se dost\u00e1v\u00e1me k pojmu anal\u00fdza, kter\u00fd pou\u017e\u00edv\u00e1me v meteorologii ka\u017ed\u00fd den. Anal\u00fdzu m\u016f\u017eeme definovat jako pravidelnou prostorovou reprezentaci z\u00e1visl\u00fdch meteorologick\u00fdch veli\u010din v dan\u00e9m \u010dase. V praxi se anal\u00fdza pou\u017e\u00edv\u00e1 hned trojmo: i) za \u00fa\u010delem diagnostiky sou\u010dasn\u00e9ho po\u010das\u00ed; ii) pro p\u0159edpov\u011b\u010f po\u010das\u00ed; iii) pro n\u00e1slednou verifikaci p\u0159ede\u0161l\u00fdch p\u0159edpov\u011bd\u00ed. Abychom mohli \u201eud\u011blat anal\u00fdzu\u201c, pot\u0159ebujeme vlastn\u011b tak\u00e9 t\u0159i v\u011bci: p\u0159edev\u0161\u00edm s\u00ed\u0165 pozorov\u00e1n\u00ed, d\u00e1le diagnostickou metodu, kter\u00e1 mimo jin\u00e9 zahrnuje vnit\u0159n\u00ed vazby mezi jednotliv\u00fdmi prom\u011bnn\u00fdmi, a v neposledn\u00ed \u0159ad\u011b prognostickou slo\u017eku, kter\u00e1 n\u00e1m poskytne p\u0159edb\u011b\u017en\u00fd odhad na z\u00e1klad\u011b minul\u00fdch pozorov\u00e1n\u00ed.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Z historie v\u00edme, \u017ee prvn\u00ed meteorologick\u00e9 p\u0159\u00edstroje a v\u00edce m\u00e9n\u011b pravideln\u00e1 pozorov\u00e1n\u00ed se objevila v polovin\u011b 18. stolet\u00ed, a \u017ee prvn\u00ed subjektivn\u00ed anal\u00fdze, vlastn\u011b synoptick\u00e9 map\u011b, to trvalo je\u0161t\u011b cel\u00e9 jedno stolet\u00ed, ne\u017e spat\u0159ila sv\u011btlo sv\u011bta. Od dob pana Le Verrier, otce prvn\u00ed organizovan\u00e9 pozorovac\u00ed s\u00edt\u011b a synoptick\u00e9 mapy, se rozvinuly metody objektivn\u00ed anal\u00fdzy a hlavn\u011b se podstatn\u011b zm\u011bnila technologie po\u0159izov\u00e1n\u00ed dat.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">V objektivn\u00ed anal\u00fdze, jej\u00ed\u017e historii zapo\u010dal Richardson\u016fv pokus v roce 1922, jsme se p\u0159es komplikovan\u00e9 grafick\u00e9 metody, polynomi\u00e1ln\u00ed p\u0159ibl\u00ed\u017een\u00ed a metodu postupn\u00fdch korekc\u00ed nakonec dostali k pou\u017eit\u00ed teorie optim\u00e1ln\u00edho odhadu v meteorologii. Prvn\u00ed vla\u0161tovkou byla v tomto sm\u011bru Gandinova metoda optim\u00e1ln\u00ed interpolace z 60. let dvac\u00e1t\u00e9ho stolet\u00ed, kter\u00e1 se v rozvinut\u011bj\u0161\u00edch podob\u00e1ch pou\u017e\u00edv\u00e1 je\u0161t\u011b dnes. Teorie optim\u00e1ln\u00edho odhadu, s podm\u00ednkou platnosti ur\u010dit\u00fdch p\u0159edpoklad\u016f, vede k \u0159e\u0161en\u00ed rovnice pro tzv. BLUE (Best Linear Unbiased Estimate) odhad. Tak\u017ee jednodu\u0161e \u0159e\u010deno, zhruba od 60. let minul\u00e9ho stolet\u00ed je probl\u00e9m anal\u00fdzy pops\u00e1n jednou \u201eBLUE\u201c rovnic\u00ed.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Rovnice BLUE odhadu m\u00e1 dva hlavn\u00ed \u010dleny. Prvn\u00ed popisuje vzd\u00e1lenost optim\u00e1ln\u00edho odhadu od pozorov\u00e1n\u00ed, druh\u00fd zase vzd\u00e1lenost od tzv. p\u0159edb\u011b\u017en\u00e9ho odhadu, kter\u00fdm je zpravidla kr\u00e1tk\u00e1 p\u0159edpov\u011b\u010f modelu z p\u0159edchoz\u00ed anal\u00fdzy, kter\u00fd k \u0159e\u0161en\u00ed nov\u00e9 anal\u00fdzy pou\u017e\u00edv\u00e1me. Jde vlastn\u011b o vyu\u017eit\u00ed minul\u00fdch pozorov\u00e1n\u00ed, ov\u0161em nav\u00edc zat\u00ed\u017een\u00fdch chybou modelov\u00e9 p\u0159edpov\u011bdi, by\u0165 kr\u00e1tk\u00e9. Tato informace je v\u0161ak st\u00e1le cenn\u00e1 pro p\u0159eur\u010den\u00ed \u00falohy, vedouc\u00ed k p\u0159esn\u011bj\u0161\u00edmu koncov\u00e9mu odhadu. D\u00edky pou\u017e\u00edv\u00e1n\u00ed p\u0159edb\u011b\u017en\u00e9ho odhadu s pomoc\u00ed p\u0159edpov\u011bdn\u00edho modelu se dost\u00e1v\u00e1me k pojmu asimilace dat do modelu, kter\u00fd dnes pou\u017e\u00edv\u00e1me v numerick\u00e9 p\u0159edpov\u011bdi po\u010das\u00ed daleko \u010dast\u011bji ne\u017e pojem anal\u00fdza, a tak\u00e9 k pojmu asimila\u010dn\u00edho cyklu. Cel\u00fd dal\u0161\u00ed velk\u00fd d\u00edl v\u00fdzkumu a v\u00fdvoje v discipl\u00edn\u011b asimilace dat je o tom, jak nejl\u00e9pe rovnici BLUE odhadu \u0159e\u0161it, zvl\u00e1\u0161t\u011b s ohledem na to, \u017ee vektor \u0159e\u0161en\u00ed (v\u00edtr, tlak, teplota, vlhkost ve v\u0161ech uzlov\u00fdch bodech modelov\u00e9 m\u0159\u00ed\u017eky) b\u00fdv\u00e1 \u0159\u00e1du asi 106 a\u017e 107.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">S takovou obrovskou dimenz\u00ed \u00falohy si v praxi nejl\u00e9pe poradily varia\u010dn\u00ed metody, kter\u00e9 se prosadily v polovin\u011b 80. let minul\u00e9ho stolet\u00ed. Varia\u010dn\u00ed po\u010det dovoluje toti\u017e elegantn\u00ed a kompaktn\u00ed formulaci \u0159e\u0161en\u00ed BLUE odhadu: je to minimum funkcion\u00e1lu, kter\u00fd je ve tvaru kvadratick\u00e9 formy a kter\u00fd m\u011b\u0159\u00ed ji\u017e v\u00fd\u0161e zm\u00edn\u011bnou vzd\u00e1lenost od pozorov\u00e1n\u00ed a p\u0159edb\u011b\u017en\u00e9ho odhadu. Pro praktick\u00e9 \u0159e\u0161en\u00ed varia\u010dn\u00ed \u00falohy je d\u016fle\u017eit\u00e1 metoda sdru\u017een\u00fdch (adjungovan\u00fdch) oper\u00e1tor\u016f, kterou zavedli Le Dimet &amp; Talagrand (1986). P\u0159i hled\u00e1n\u00ed minima funkcion\u00e1lu je toti\u017e pot\u0159eba vypo\u010d\u00edtat jeho gradient podle kontroln\u00edho vektoru (klasicky je kontroln\u00edm vektorem v\u00edtr, teplota, tlak, vlhkost ve v\u0161ech uzlov\u00fdch bodech), kter\u00fd m\u00e1 ji\u017e v\u00fd\u0161e zm\u00edn\u011bnou d\u00e9lku t\u0159eba 107. Zat\u00edmco prvn\u00ed \u010dlen gradientu v\u016f\u010di p\u0159edb\u011b\u017en\u00e9mu odhadu se vlastn\u011b vypo\u010d\u00edt\u00e1v\u00e1 v uzlov\u00fdch bodech m\u0159\u00ed\u017eky a technicky nep\u0159edstavuje velk\u00fd probl\u00e9m, druh\u00fd \u010dlen gradientu v\u016f\u010di pozorov\u00e1n\u00edm u\u017e je o\u0159\u00ed\u0161kem. Klasicky by se musel ka\u017ed\u00fd jednotliv\u00fd p\u0159\u00edsp\u011bvek k celkov\u00e9mu gradientu zvl\u00e1\u0161\u0165 vypo\u010d\u00edt\u00e1vat pro ka\u017ed\u00fd prvek vektoru v\u016f\u010di ka\u017ed\u00e9mu pozorov\u00e1n\u00ed, a to by bylo nesm\u00edrn\u011b n\u00e1ro\u010dn\u00e9. Tento probl\u00e9m \u0159e\u0161\u00ed velmi elegantn\u011b pr\u00e1v\u011b metoda sdru\u017een\u00e9ho zobrazen\u00ed, kter\u00e1 vy\u010d\u00edslen\u00ed gradientu vzd\u00e1lenosti v\u016f\u010di pozorov\u00e1n\u00edm o\u0161et\u0159uje jednoduch\u00fdm a v\u00fdpo\u010detn\u011b levn\u00fdm zp\u016fsobem.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Jak pracuje metoda sdru\u017een\u00fdch zobrazen\u00ed, m\u016f\u017eeme vysv\u011btlit na n\u00e1sleduj\u00edc\u00edm p\u0159\u00edklad\u011b. Datov\u00fd oper\u00e1tor, kter\u00fd po\u010d\u00edt\u00e1 z modelov\u00fdch prom\u011bnn\u00fdch (tlak, teplota, vlhkost) takovou radiometrickou informaci, kterou pozoruje senzor na dru\u017eici, zobrazuje vlastn\u011b informaci z prostoru kontroln\u00ed prom\u011bnn\u00e9 modelu do prostoru pozorov\u00e1n\u00ed. Zde se vypo\u010dte rozd\u00edl modelov\u00e9ho odhadu proti pozorovan\u00e9 radiometrick\u00e9 hodnot\u011b a s n\u00edm spojen\u00fd p\u0159\u00edsp\u011bvek ke gradientu funkcion\u00e1lu, jeho\u017e minimum hled\u00e1me. Sdru\u017een\u00fd datov\u00fd oper\u00e1tor pak zobraz\u00ed tento gradient zp\u011bt do prostoru kontroln\u00ed prom\u011bnn\u00e9, jin\u00fdmi slovy prom\u00edtne jej na gradient v teplot\u011b, vlhkosti, tlaku a tak d\u00e1le. Pou\u017eit\u00ed datov\u00fdch oper\u00e1tor\u016f a k nim sdru\u017een\u00fdch zobrazen\u00ed umo\u017enilo spolu s varia\u010dn\u00ed formulac\u00ed obrovsk\u00fd rozkv\u011bt asimilace nekonven\u010dn\u00edch dat. V p\u0159\u00edpad\u011b konven\u010dn\u00edch dat, jako je t\u0159eba teplota m\u011b\u0159en\u00e1 radiosondou, se datov\u00fd oper\u00e1tor redukuje pouze na prostorovou interpolaci z uzlov\u00fdch bod\u016f modelu do bodu m\u011b\u0159en\u00ed. K t\u00e9to interpolaci tak\u00e9 existuje sdru\u017een\u00e9 zobrazen\u00ed, kter\u00e9 prom\u00edtne gradient funkcion\u00e1lu v teplot\u011b z bodu m\u011b\u0159en\u00ed zp\u011bt do uzlov\u00fdch bod\u016f modelov\u00e9 m\u0159\u00ed\u017eky.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Metoda sdru\u017een\u00e9ho zobrazen\u00ed se d\u00e1 roz\u0161\u00ed\u0159it i na samotn\u00fd model. Tam je ale pot\u0159eba v prvn\u00ed f\u00e1zi odvodit tangentn\u00ed line\u00e1rn\u00ed (TL) model, kter\u00fd je zalo\u017een na principu line\u00e1rn\u00edch perturbac\u00ed. Za z\u00e1klad pro linearizaci slou\u017e\u00ed \u0159e\u0161en\u00ed (trajektorie) \u00fapln\u00e9ho neline\u00e1rn\u00edho modelu. Pomoc\u00ed TL modelu, kter\u00fd je vlastn\u011b line\u00e1rn\u00edm zobrazen\u00edm, m\u016f\u017eeme popsat evoluci perturbac\u00ed prom\u011bnn\u00fdch pod\u00e9l modelov\u00e9 trajektorie. Celkov\u00e9 \u0159e\u0161en\u00ed, kter\u00e9 je sou\u010dtem z\u00e1kladn\u00ed trajektorie a perturbac\u00ed, porovn\u00e1v\u00e1me s pozorov\u00e1n\u00edmi v p\u0159\u00edslu\u0161n\u00fdch \u010dasov\u00fdch okam\u017eic\u00edch. Z\u00e1rove\u0148 vypo\u010d\u00edt\u00e1v\u00e1me p\u0159\u00edsp\u011bvek ke gradientu funkcion\u00e1lu, dan\u00fd vzd\u00e1lenost\u00ed \u0159e\u0161en\u00ed k pozorov\u00e1n\u00edm. N\u00e1sledn\u011b sdru\u017een\u00fd (adjungovan\u00fd; zna\u010den\u00fd tak\u00e9 zkratkou AD) model, kter\u00fd je definov\u00e1n jako sdru\u017een\u00e9 zobrazen\u00ed k modelu TL, slou\u017e\u00ed ke zp\u011btn\u00e9 projekci gradientu funkcion\u00e1lu na po\u010d\u00e1te\u010dn\u00ed stav. Tak se dost\u00e1v\u00e1me k mo\u017enosti roz\u0161\u00ed\u0159it anal\u00fdzu pod\u00e9l \u010dasov\u00e9 osy a k metod\u011b naz\u00fdvan\u00e9 \u010dty\u0159dimenzion\u00e1ln\u00ed varia\u010dn\u00ed asimilace dat, ve zkratce 4DVAR. \u010casov\u00fd interval, pod\u00e9l kter\u00e9ho porovn\u00e1v\u00e1me pozorov\u00e1n\u00ed s modelov\u00fdm \u0159e\u0161en\u00edm a hled\u00e1me optim\u00e1ln\u00ed odhad, naz\u00fdv\u00e1me asimila\u010dn\u00edm oknem.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Varia\u010dn\u00ed formulace m\u00e1 dal\u0161\u00ed v\u00fdhody, nebo\u0165 je aditivn\u00ed. Kvadratick\u00e1 forma m\u016f\u017ee b\u00fdt roz\u0161\u00ed\u0159ena o dal\u0161\u00ed \u010dleny, nap\u0159\u00edklad o m\u011b\u0159en\u00ed vzd\u00e1lenosti k \u0159e\u0161en\u00edm zbaven\u00fdch vysokofrekven\u010dn\u00edho \u0161umu. V jedn\u00e9 varia\u010dn\u00ed \u00faloze je tak o\u0161et\u0159ena i tzv. inicializace dat, odstra\u0148uj\u00edc\u00ed nevyv\u00e1\u017een\u00e9 \u010d\u00e1sti analyzovan\u00fdch p\u0159\u00edr\u016fstk\u016f hmoty a v\u011btru. Na druhou stranu je varia\u010dn\u00ed \u00faloha vysoce po\u010detn\u011b n\u00e1ro\u010dn\u00e1, a proto je aplikov\u00e1na relativn\u011b pozd\u011b vzhledem k znalosti teorie optim\u00e1ln\u00edho odhadu a pot\u0159ebn\u00fdch n\u00e1stroj\u016f line\u00e1rn\u00ed algebry. I kdy\u017e byly vyvinuty velice \u00fa\u010dinn\u00e9 a rychle konverguj\u00edc\u00ed algoritmy pro nalezen\u00ed minima funkcion\u00e1lu, nap\u0159\u00edklad metodou konjugovan\u00e9ho gradientu, je st\u00e1le pot\u0159eba v\u00fdkonn\u00e9ho po\u010d\u00edta\u010de s dostatkem centr\u00e1ln\u00ed pam\u011bti. I tak praktick\u00e1 aplikace, maxim\u00e1ln\u011b vyu\u017e\u00edvaj\u00edc\u00ed tzv. metodu p\u0159\u00edr\u016fstk\u016f, je z\u00e1le\u017eitost\u00ed d\u016fmysln\u00e9ho zvl\u00e1dnut\u00ed v\u0161ech mo\u017en\u00fdch detail\u016f.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Jak jsme se ji\u017e zm\u00ednili, rozvoj zaznamenaly nejen metody anal\u00fdzy, ale i technologie po\u0159izov\u00e1n\u00ed dat. Krom\u011b klasick\u00fdch konven\u010dn\u00edch m\u011b\u0159en\u00ed m\u00e1me k dispozici relativn\u011b obrovsk\u00e9 mno\u017estv\u00ed radiometrick\u00fdch dat z dru\u017eic. P\u0159esto, \u017ee v asimilaci dat do model\u016f je vyu\u017eita relativn\u011b mal\u00e1 \u010d\u00e1st, pomohla tato data pom\u011brn\u011b v\u00fdznamn\u011b k zlep\u0161en\u00ed prediktability atmosf\u00e9ry, zvl\u00e1\u0161t\u011b na ji\u017en\u00ed polokouli. St\u00e1le \u010dast\u011bj\u0161\u00ed jsou pokusy o asimilaci dat z radar\u016f do model\u016f mezo- a konvek\u010dn\u00edho m\u011b\u0159\u00edtka. Objem tzv. nekonven\u010dn\u00edch dat, kter\u00e1 by se dala asimilovat do p\u0159edpov\u011bdn\u00edch model\u016f, zcela ur\u010dit\u011b d\u00e1le poroste.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">O tom, kde se nach\u00e1z\u00edme dnes, jak\u00fd je dne\u0161n\u00ed stav discipl\u00edny asimilace dat, a to nejenom v meteorologii, bylo sympozium, ned\u00e1vno uspo\u0159\u00e1dan\u00e9 v Praze.<\/span><\/p><h2 class=\"rtejustify\"><span style=\"color: #003366;\"><strong>\u010cTVRT\u00c9 WMO MEZIN\u00c1RODN\u00cd SYMPOZIUM O ASIMILACI POZOROV\u00c1N\u00cd V METEOROLOGII A OCE\u00c1NOGRAFII, PRAHA 18.-22. DUBEN 2005<\/strong><\/span><\/h2><p class=\"rtejustify\"><span style=\"color: #003366;\">Ji\u017e v po\u0159ad\u00ed \u010dtvrt\u00e9 sympozium o asimilaci pozorov\u00e1n\u00ed se konalo pod hlavi\u010dkou Sv\u011btov\u00e9 meteorologick\u00e9 organizace v Praze v t\u00fddnu od 18. do 22. dubna. Nav\u00e1zalo tak na p\u0159edchoz\u00ed sympozia, jejich\u017e historie za\u010dala v Clermont Ferrand (1990). D\u00e1 se \u0159\u00edci, \u017ee impulzem pro vznik specializovan\u00e9ho sympozia byl pr\u00e1v\u011b vzestup varia\u010dn\u00edch metod v obdob\u00ed 1985 a\u017e 1990. Dal\u0161\u00ed dv\u011b sympozia se pak po\u0159\u00e1dala v Tokyu (1995) a v Qu\u00e9becu (1999). Pr\u00e1v\u011b b\u011bhem t\u0159et\u00edho sympozia v kanadsk\u00e9m Qu\u00e9becu byla p\u0159edsedou programov\u00e9 komise p\u0159edb\u011b\u017en\u011b navr\u017eena Praha pro po\u0159\u00e1d\u00e1n\u00ed dal\u0161\u00edho setk\u00e1n\u00ed.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\"><img decoding=\"async\" class=\"alignleft\" src=\"http:\/\/www.cmes.cz\/sites\/default\/files\/asimilace%20dat.JPG\" alt=\"\" \/><em>Zah\u00e1jen\u00ed \u010dtvrt\u00e9ho mezin\u00e1rodn\u00edho sympozia v aule Karolina. Zleva: R. Bro\u017ekov\u00e1 (\u010cHM\u00da), E. Manaenkova (WMO), V Pa\u010des (prezident AV \u010cR), I. Obrusn\u00edk (\u0159editel \u010cHM\u00da), Ph. Courtier (WWRP), T. Novotn\u00fd (MZP), A. Frolov (Roshydromet, Rusko). Foto O. \u0160uvarinov\u00e1.<\/em><\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Ofici\u00e1ln\u00ed potvrzen\u00ed ze sekretari\u00e1tu WMO p\u0159i\u0161lo v\u0161ak a\u017e na konci \u0159\u00edjna 2003, kdy byla jmenov\u00e1na mezin\u00e1rodn\u00ed organiza\u010dn\u00ed komise sympozia v \u010dele s Alexandrem Frolovem (Roshydromet, Rusko). V t\u00e9to komisi byli d\u00e1le \u010dty\u0159i spolup\u0159edsedov\u00e9 programov\u00e9 komise: Steve Cohn (NASA, USA), Andrew Lorenc (Met Office, Velk\u00e1 Brit\u00e1nie), Neville Smith (BMRC, Austr\u00e1lie), Chris Snyder (NCAR, USA); d\u00e1le p\u0159edsedkyn\u011b lok\u00e1ln\u00ed organiza\u010dn\u00ed komise Radmila Bro\u017ekov\u00e1 (\u010cHM\u00da) a z\u00e1stupci WMO: Elena Manaenkova (\u0159editelka programu AREP) a Philippe Courtier (p\u0159edseda pilotn\u00ed komise programu WWRP). \u0160lo o pom\u011brn\u011b nezvykl\u00fd form\u00e1t, hlavn\u011b co se t\u00fd\u010de \u010dty\u0159 spolup\u0159edsed\u016f programov\u00e9 komise. Nicm\u00e9n\u011b roz\u0161\u00ed\u0159en\u00fd centr\u00e1ln\u00ed t\u00fdm, funguj\u00edc\u00ed tak v p\u011bti \u010dlenech a vydatn\u011b podpo\u0159en\u00fd Michalem \u017d\u00e1kem (\u010cHM\u00da, \u010dlen lok\u00e1ln\u00ed organiza\u010dn\u00ed komise), se v praxi velmi dob\u0159e osv\u011bd\u010dil.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Proto\u017ee \u0161lo o pom\u011brn\u011b v\u00fdznamn\u00e9 setk\u00e1n\u00ed s o\u010dek\u00e1vanou \u00fa\u010dast\u00ed okolo 200 odborn\u00edk\u016f z cel\u00e9ho sv\u011bta, do organizace se zapojil nejenom \u010cesk\u00fd hydrometeorologick\u00fd \u00fastav, ale tak\u00e9 Matematicko-fyzik\u00e1ln\u00ed fakulta Univerzity Karlovy a \u00dastav fyziky atmosf\u00e9ry Akademie v\u011bd \u010cesk\u00e9 republiky. Nad sympoziem pak p\u0159evzaly z\u00e1\u0161titu v\u00fdznamn\u00e9 v\u011bdeck\u00e9 a politick\u00e9 osobnosti: prof. RNDr. Ivan Wilhelm, CSc., rektor Univerzity Karlovy; prof. RNDr. Helena Illnerov\u00e1, DrSc., p\u0159edchoz\u00ed p\u0159edsedkyn\u011b Akademie v\u011bd \u010cR a n\u00e1sledn\u011b jej\u00ed n\u00e1stupce prof. RNDr. V\u00e1clav Pa\u010des, DrSc.; prof. RNDr. Ivan Netuka, DrSc., d\u011bkan Matematicko-fyzik\u00e1ln\u00ed fakulty UK; RNDr. Libor Ambrozek, ministr vl\u00e1dy \u010cR pro \u017eivotn\u00ed prost\u0159ed\u00ed a ing. Ivan Obrusn\u00edk, DrSc., \u0159editel \u010cHM\u00da.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Jako m\u00edsto kon\u00e1n\u00ed bylo vybr\u00e1no konferen\u010dn\u00ed centrum Univerzity Karlovy v Karolinu, vybaven\u00e9 vynikaj\u00edc\u00edm p\u0159edn\u00e1\u0161kov\u00fdm s\u00e1lem a dal\u0161\u00edmi nezbytn\u00fdmi prostory. K hladk\u00e9mu pr\u016fb\u011bhu sympozia p\u0159isp\u011bl profesion\u00e1ln\u011b zdatn\u00fd person\u00e1l konferen\u010dn\u00edho centra a organiza\u010dn\u00edho odd\u011blen\u00ed Univerzity Karlovy. Prost\u0159ed\u00ed velk\u00e9 auly Karolina pak dodalo zah\u00e1jen\u00ed sympozia opravdu slavnostn\u00ed a d\u016fstojnou atmosf\u00e9ru.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Se zaji\u0161t\u011bn\u00edm lok\u00e1ln\u00ed logistiky pom\u00e1hala organiz\u00e1tor\u016fm agentura Carolina. Jeliko\u017e b\u00fdv\u00e1 tradic\u00ed, \u017ee konference WMO jsou prosty registra\u010dn\u00edho poplatku v z\u00e1jmu co nej\u0161ir\u0161\u00ed \u00fa\u010dasti v\u011bdc\u016f z rozvojov\u00fdch zem\u00ed, m\u011bli i v p\u0159\u00edpad\u011b tohoto sympozia organiz\u00e1to\u0159i dosti starost\u00ed s jeho finan\u010dn\u00edm zabezpe\u010den\u00edm. To se nakonec poda\u0159ilo zajistit d\u00edky p\u0159\u00edsp\u011bvk\u016fm od sponzor\u016f, kter\u00fdmi byly:<\/span><\/p><ul><li class=\"rtejustify\"><span style=\"color: #003366;\">NEC, High Performance Computing Europe, GMbH, Dusseldorf, N\u011bmecko<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">Global Ocean Data Assimilation Experiment<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">EUMETSAT, Darmstadt, N\u011bmecko<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">CNRM\/M\u00e9t\u00e9o-France, Francie<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">NASA, USA<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">NCAR, Mesoscale and Microscale Meteorology, USA<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">Cray, Inc., USA<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">Instituto Nazionale di Geofisica e Vulcanologia, It\u00e1lie<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">Centre National d\u2019Etudes Spatiales, Francie<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">Vaisala, Finsko<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">World Climate Research Program<\/span><\/li><li class=\"rtejustify\"><span style=\"color: #003366;\">THORPEX: A World Weather Research Program<\/span><\/li><\/ul><p class=\"rtejustify\"><span style=\"color: #003366;\">O \u00fa\u010dast na sympoziu byl skute\u010dn\u011b obrovsk\u00fd z\u00e1jem, kter\u00fd p\u0159ekonal ve\u0161ker\u00e1 o\u010dek\u00e1v\u00e1n\u00ed; zaregistrovalo se celkem 256 \u00fa\u010dastn\u00edk\u016f z 28 zem\u00ed sv\u011bta s 259 p\u0159\u00edsp\u011bvky (or\u00e1ln\u00edmi a postery). Ve srovn\u00e1n\u00ed s minul\u00fdmi sympozii tato \u010d\u00edsla sv\u011bd\u010d\u00ed o tom, \u017ee obor asimilace dat pro\u017e\u00edv\u00e1 velk\u00fd rozvoj. Krom\u011b velk\u00fdch p\u0159edpov\u011bdn\u00edch center typu ECMWF (European Centre for Medium-Range Weather Forecasts) za\u010d\u00ednaj\u00ed i men\u0161\u00ed t\u00fdmy v r\u016fzn\u00fdch zem\u00edch sv\u011bta pou\u017e\u00edvat modern\u00ed metody asimilace dat. Zde je vhodn\u00e9 je\u0161t\u011b dodat, \u017ee p\u0159edb\u011b\u017en\u011b bylo p\u0159ihl\u00e1\u0161eno dokonce 305 p\u0159\u00edsp\u011bvk\u016f (a 350 \u00fa\u010dastn\u00edk\u016f) a redukce na kone\u010dn\u00fdch 259 p\u0159\u00edsp\u011bvk\u016f p\u0159edstavovala nesnadnou \u00falohu pro programovou komisi. Tato komise m\u011bla celkem 11 \u010dlen\u016f; ka\u017ed\u00fd p\u0159\u00edsp\u011bvek byl vyhodnocen alespo\u0148 t\u0159emi \u010dleny nez\u00e1visle na sob\u011b pro z\u00edsk\u00e1n\u00ed fin\u00e1ln\u00ed \u201ezn\u00e1mky\u201c. Snahou samoz\u0159ejm\u011b bylo akceptovat co nejv\u00edce p\u0159\u00edsp\u011bvk\u016f; organiz\u00e1to\u0159i zavedli i mo\u017enost zapsat se na \u010dekac\u00ed seznam pro p\u0159\u00edpad, kdy se n\u011bkdo jin\u00fd ze sympozia odhl\u00e1s\u00ed. D\u00e1 se \u0159\u00edci, \u017ee nakonec byly vylou\u010deny jenom ty p\u0159\u00edsp\u011bvky, kter\u00e9 sv\u00fdm t\u00e9matem skute\u010dn\u011b nem\u011bly nic spole\u010dn\u00e9ho s asimilac\u00ed dat.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Tradi\u010dn\u00ed form\u00e1t sympozia o asimilaci dat je ten, \u017ee v\u0161echny p\u0159edn\u00e1\u0161ky prob\u00edhaj\u00ed v jednom pl\u00e9nu, nikoliv paraleln\u011b. Je to pr\u00e1v\u011b proto, \u017ee metodika asimilace dat byla nejv\u00edce vypracov\u00e1na v meteorologii, odkud se poznatky p\u0159en\u00e1\u0161ely do jin\u00fdch p\u0159\u00edbuzn\u00fdch obor\u016f, jako je pr\u00e1v\u011b oce\u00e1nografie a te\u010f v posledn\u00ed dob\u011b modelov\u00e1n\u00ed chemick\u00fdch proces\u016f v atmosf\u00e9\u0159e. Asimilace dat tud\u00ed\u017e t\u011bmito obory prol\u00edn\u00e1 a \u00fa\u010delem sympozia je setk\u00e1n\u00ed v\u011bdc\u016f a v\u00fdm\u011bna zku\u0161enost\u00ed a metod z t\u011bchto jednotliv\u00fdch discipl\u00edn; proto tedy nen\u00ed zvykem rozd\u011blovat p\u0159edn\u00e1\u0161ky do paraleln\u011b prob\u00edhaj\u00edc\u00edch sekc\u00ed. Ov\u0161em vzhledem k tomu, jak obrovsk\u00fd byl z\u00e1jem o \u00fa\u010dast na pra\u017esk\u00e9m sympoziu, bude mo\u017en\u00e1 nutn\u00e9 opustit tradici a paraleln\u00ed p\u0159edn\u00e1\u0161kov\u00e9 bloky zav\u00e9st. Fokus sympozia se tak asi do budoucna zm\u011bn\u00ed.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Podm\u00ednka jedin\u00e9ho p\u0159edn\u00e1\u0161kov\u00e9ho pl\u00e9na v praxi znamen\u00e1, \u017ee b\u011bhem jednoho t\u00fddne lze vyslechnout pouze relativn\u011b mal\u00e9 mno\u017estv\u00ed p\u0159edn\u00e1\u0161ek, kter\u00e9 ov\u0161em mus\u00ed b\u00fdt pe\u010dliv\u011b vybr\u00e1ny, aby co nejl\u00e9pe reprezentovaly danou sekci. V p\u0159\u00edpad\u011b pra\u017esk\u00e9ho sympozia bylo vytvo\u0159eno celkem deset tematick\u00fdch blok\u016f (aspekty prost\u0159ed\u00ed, tj. hlavn\u011b chemick\u00e9 transportn\u00ed modely apod.; ans\u00e1mblov\u00e9 metody; asimilace dat v modelech oce\u00e1n\u016f a mo\u0159\u00ed; problematika d\u00e9lky tzv. asimila\u010dn\u00edho okna; numerick\u00e1 p\u0159edpov\u011b\u010f po\u010das\u00ed; satelitn\u00ed pozorov\u00e1n\u00ed, asimilace v modelech atmosf\u00e9ry mezo- a konvek\u010dn\u00edho m\u011b\u0159\u00edtka; metodika a statistika; reanal\u00fdza; diagnostika pozorov\u00e1n\u00ed), uveden\u00fdch t\u0159in\u00e1cti pozvan\u00fdmi p\u0159edn\u00e1\u0161kami. Tyto kl\u00ed\u010dov\u00e9 p\u0159ehledov\u00e9 p\u0159edn\u00e1\u0161ky byly dopln\u011bny dal\u0161\u00edmi krat\u0161\u00edmi p\u0159edn\u00e1\u0161kami, kter\u00fdch bylo celkem \u010dty\u0159icet \u010dty\u0159i. Zna\u010dn\u00e1 \u010d\u00e1st p\u0159\u00edsp\u011bvk\u016f, a to v po\u010dtu asi dvou set, byla tak prezentov\u00e1na formou poster\u016f. Vzhledem k jejich zna\u010dn\u00e9mu mno\u017estv\u00ed bylo nutn\u00e9 uzp\u016fsobit program tak, aby \u00fa\u010dastn\u00edci m\u011bli dost \u010dasu si postery prohl\u00e9dnout.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Sympozium tak m\u011blo skute\u010dn\u011b sv\u011btov\u011b \u0161pi\u010dkovou \u00farove\u0148 p\u0159edn\u00e1\u0161ek a splnilo jeden ze sv\u00fdch hlavn\u00edch c\u00edl\u016f: podat souhrnnou, p\u0159ehlednou informaci o posledn\u00edch pokroc\u00edch v asimilaci dat do model\u016f atmosf\u00e9ry a oce\u00e1n\u016f. Co se tedy d\u00e1 \u0159\u00edci o d\u016fle\u017eit\u00fdch trendech za uplynul\u00e9 obdob\u00ed od p\u0159edchoz\u00edho sympozia v kanadsk\u00e9m Qu\u00e9becu? D\u00e1 se \u0159\u00edci, \u017ee ke zna\u010dn\u00e9mu pokroku do\u0161lo ve v\u0161ech aplikac\u00edch asimilace dat. Nap\u0159\u00edklad v roce 1999 to bylo jenom Evropsk\u00e9 centrum pro st\u0159edn\u011bdobou p\u0159edpov\u011b\u010f po\u010das\u00ed (ECMWF), kter\u00e9 v provozn\u00ed praxi m\u011blo ji\u017e od roku 1997 implementovanou \u010dty\u0159dimenzion\u00e1ln\u00ed varia\u010dn\u00ed asimilaci dat (4DVAR), zat\u00edmco letos je 4DVAR operativn\u011b pou\u017e\u00edv\u00e1no v dal\u0161\u00edch glob\u00e1ln\u00edch p\u0159edpov\u011bdn\u00edch modelech M\u00e9t\u00e9o-France (2000), UK Met-Office (2004), v Japonsku (2005) a Kanad\u011b (2005).<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Motivace p\u0159echodu na \u010dty\u0159dimenzion\u00e1ln\u00ed asimilaci dat je z\u0159ejm\u00e1. Jednak dovoluje postihnout v BLUE algoritmu dynamick\u00fd v\u00fdvoj struktury chyb modelov\u00e9ho odhadu stavu atmosf\u00e9ry, jednak umo\u017e\u0148uje l\u00e9pe vyu\u017e\u00edt pozorov\u00e1n\u00ed z tzv. asynoptick\u00fdch term\u00edn\u016f, tedy ta kter\u00e1 jsou k dispozici ve vysok\u00e9m \u010dasov\u00e9m rozli\u0161en\u00ed. Typicky jde hlavn\u011b o nekonven\u010dn\u00ed satelitn\u00ed data, d\u00e1le tak\u00e9 o bulletiny z dopravn\u00edch letadel, v bl\u00edzk\u00e9 budoucnosti o data z meteorologick\u00fdch radar\u016f. Jak jsme se ji\u017e zm\u00ednili v \u00favodu, \u0159e\u0161en\u00ed BLUE odhadu v prostoru i \u010dase v dostate\u010dn\u00e9m rozli\u0161en\u00ed je po\u010detn\u011b nesm\u00edrn\u011b n\u00e1ro\u010dn\u00e9 a bez velmi v\u00fdkonn\u00fdch po\u010d\u00edta\u010d\u016f nemysliteln\u00e9. V sou\u010dasn\u00e9 dob\u011b je d\u00e9lka asimila\u010dn\u00edho okna typicky 6 hodin pro glob\u00e1ln\u00ed modely; pro \u00fa\u010dely st\u0159edn\u011bdob\u00e9 glob\u00e1ln\u00ed p\u0159edpov\u011bdi je asimila\u010dn\u00ed okno modelu ECMWF dlouh\u00e9 a\u017e 12 hodin. Nav\u00edc se varia\u010dn\u00ed \u00faloha \u0159e\u0161\u00ed pro ni\u017e\u0161\u00ed rozli\u0161en\u00ed, ne\u017e ve kter\u00e9m se po\u010d\u00edt\u00e1 produk\u010dn\u00ed p\u0159edpov\u011b\u010f. Jednak je to ur\u010dit\u011b proto, aby se u\u0161et\u0159il v\u00fdpo\u010detn\u00ed \u010das, jednak ze zku\u0161enosti se stejn\u011b v\u00ed, \u017ee spektrum krat\u0161\u00edch vln se v modelu rychle p\u0159izp\u016fsob\u00ed del\u0161\u00edm planet\u00e1rn\u00edm vln\u00e1m a vln\u00e1m synoptick\u00e9ho m\u011b\u0159\u00edtka, a tak by se analyzovan\u00e1 informace obsa\u017een\u00e1 v kr\u00e1tk\u00fdch vln\u00e1ch stejn\u011b ztratila v p\u0159\u00edpad\u011b, kdy by nebyla pln\u011b v konzistenci s vnit\u0159n\u00ed dynamikou a fyzikou modelu.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Jinou alternativou, jak \u0159e\u0161it \u00falohu \u010dty\u0159dimenzion\u00e1ln\u00ed asimilace dat, je metoda Kalm\u00e1nova filtru. Ta toti\u017e umo\u017e\u0148uje vz\u00edt korektn\u011b v \u00favahu v\u00fdvoj chyby modelu, zat\u00edmco v klasick\u00e9m algoritmu 4DVAR se na za\u010d\u00e1tku asimila\u010dn\u00edho okna pou\u017e\u00edv\u00e1 staticky vypo\u010dten\u00fd model kovarianc\u00ed chyb p\u0159edb\u011b\u017en\u00e9ho odhadu. \u0158e\u0161en\u00ed pln\u00e9 \u00falohy Kalm\u00e1nova filtru je velmi drah\u00e9 pro praktickou aplikaci a pokusy o redukci \u0159\u00e1du \u00falohy zat\u00edm nevedly k \u00fasp\u011b\u0161n\u00fdm v\u00fdsledk\u016fm. Na sympoziu byla ale zm\u00edn\u011bna jin\u00e1 mo\u017enost, a sice \u0159e\u0161en\u00ed \u00falohy Kalm\u00e1nova filtru varia\u010dn\u00ed cestou. Je to oblast v\u00fdzkumu, kter\u00e1 se otev\u00edr\u00e1, a kter\u00e1 mo\u017en\u00e1 p\u0159inese dal\u0161\u00ed zdokonalen\u00ed asimila\u010dn\u00edch algoritm\u016f.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Tak\u00e9 jsme se ji\u017e zm\u00ednili, \u017ee zaveden\u00ed varia\u010dn\u00edch metod p\u0159i \u0159e\u0161en\u00ed BLUE odhadu d\u00e1le umo\u017enilo daleko efektivn\u011bj\u0161\u00ed vyu\u017eit\u00ed nekonven\u010dn\u00edch dat. S t\u00edm, jak se rozv\u00edjely algoritmy \u0159e\u0161en\u00ed optim\u00e1ln\u00edho odhadu spolu se st\u00e1le lep\u0161\u00edm vyu\u017e\u00edv\u00e1n\u00edm satelitn\u00edch dat, tak se poda\u0159ilo zlep\u0161ovat prediktabilitu atmosf\u00e9ry v rytmu zhruba jeden den za deset let. Jin\u00fdmi slovy to znamen\u00e1, \u017ee ta sam\u00e1 \u00fasp\u011b\u0161nost p\u0159edpov\u011bdi po\u010das\u00ed na 5. den v roce 1985 je dnes prodlou\u017eena a\u017e na 7. p\u0159edpov\u011bdn\u00ed den! Ohromuj\u00edc\u00ed je tak\u00e9 zlep\u0161en\u00ed prediktability na ji\u017en\u00ed polokouli, a to zvl\u00e1\u0161t\u011b d\u00edky satelit\u016fm, kdy se t\u00e9m\u011b\u0159 set\u0159el rozd\u00edl v\u016f\u010di severn\u00ed polokouli. Asimilace dat tedy hraje v\u00fdznamnou roli ve zlep\u0161ov\u00e1n\u00ed prediktability atmosf\u00e9ry.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Bylo by ale velmi nespravedliv\u00e9 p\u0159isoudit tuto kl\u00ed\u010dovou \u00falohu pouze satelitn\u00edm dat\u016fm. V\u0161echna data jsou toti\u017e d\u016fle\u017eit\u00e1 a pr\u00e1v\u011b na vyu\u017e\u00edv\u00e1n\u00ed dat ze v\u0161ech dostupn\u00fdch kvalitn\u00edch zdroj\u016f stoj\u00ed s\u00edla cel\u00e9ho syst\u00e9mu. Je tak\u00e9 na m\u00edst\u011b zd\u016fraznit, \u017ee satelitn\u00ed data nemohou a je\u0161t\u011b asi hodn\u011b dlouho nebudou moci nahradit data po\u0159\u00edzen\u00e1 radiosondami. Ta maj\u00ed toti\u017e st\u00e1le nejvy\u0161\u0161\u00ed kvalitu a hlavn\u011b slou\u017e\u00ed k odstran\u011bn\u00ed sm\u011brodatn\u00e9 odchylky satelitn\u00edch dat, co\u017e je nutn\u00fd p\u0159edpoklad pro jejich asimilaci. Nap\u0159\u00edklad je tak\u00e9 zaj\u00edmav\u00e9 v\u011bd\u011bt, jak\u00fd maj\u00ed p\u0159\u00ednos individu\u00e1ln\u00ed typy dat pro prediktabilitu. Na severn\u00ed polokouli maj\u00ed radiosondy zhruba stejn\u00fd potenci\u00e1l prediktability jako dru\u017eicov\u00e1 data ATOVS (Advanced TIROS Operational Vertical Sounder ze s\u00e9rie dru\u017eic TIROS), a to 6 a\u017e 24 hodin; je to st\u00e1le v\u00edce ne\u017e data po\u0159\u00edzen\u00e1 komer\u010dn\u00edmi dopravn\u00edmi letadly (6 a\u017e 12 hodin).<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Ke zvy\u0161ov\u00e1n\u00ed prediktability p\u0159isp\u00edvaj\u00ed i dokonalej\u0161\u00ed modely, jejich\u017e kvalita je tak\u00e9 kl\u00ed\u010dov\u00fdm vstupem pro celkov\u00fd \u00fasp\u011bch. Kvalita modelu hraje roli jak v samotn\u00e9 produk\u010dn\u00ed p\u0159edpov\u011bdi, tak i v asimila\u010dn\u00edm cyklu prost\u0159ednictv\u00edm p\u0159edb\u011b\u017en\u00e9ho pole. Kovariance chyb modelu, stejn\u011b jako jeho sm\u011brodatn\u00e1 odchylka od skute\u010dn\u00e9ho stavu atmosf\u00e9ry je ned\u00edlnou sou\u010d\u00e1st\u00ed \u0159e\u0161en\u00ed BLUE odhadu. Pokroku lze dos\u00e1hnout pouze tehdy, zlep\u0161uj\u00ed-li se ob\u011b sou\u010d\u00e1sti p\u0159edpov\u011bdn\u00edho syst\u00e9mu, kter\u00fdmi jsou model v\u010detn\u011b asimilace dat. Zde se samoz\u0159ejm\u011b p\u0159id\u00e1v\u00e1 je\u0161t\u011b dal\u0161\u00ed absolutn\u011b nutn\u00e1 podm\u00ednka: mno\u017estv\u00ed a kvalita pozorov\u00e1n\u00ed atmosf\u00e9ry a oce\u00e1n\u016f.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">M\u016f\u017eeme \u0159\u00edci, a to na sympoziu bylo n\u011bkolikr\u00e1t vysloveno, \u017ee dnes pro\u017e\u00edv\u00e1me ur\u010ditou zlatou \u00e9ru, co se t\u00fd\u010de pr\u00e1v\u011b satelitn\u00edch pozorov\u00e1n\u00ed. Trvalo p\u0159itom t\u00e9m\u011b\u0159 dvacet let, ne\u017e se nap\u0159\u00edklad data z instrumentu TOVS za\u010dala skute\u010dn\u011b efektivn\u011b v modelech vyu\u017e\u00edvat, ne\u017e se vlastn\u011b odstranila v\u0161echna podstatn\u00e1 \u00faskal\u00ed. V dne\u0161n\u00ed dob\u011b jsme tedy schopni kapitalizovat na t\u00e9to dlouhodob\u00e9 investici. Ov\u0161em je ur\u010dit\u00e9 nebezpe\u010d\u00ed, \u017ee zlat\u00e1 \u00e9ra nemus\u00ed trvat i nad\u00e1le. Je to z toho d\u016fvodu, \u017ee se nov\u00e9 sondy vyv\u00edjej\u00ed dnes p\u0159\u00edli\u0161 rychle, ani\u017e by se n\u011bkdo zab\u00fdval ot\u00e1zkou dostate\u010dn\u00e9 \u010dasov\u00e9 stability dostupnosti ur\u010dit\u00e9ho produktu dat. M\u016f\u017ee tak doj\u00edt k tomu, \u017ee b\u011bhem relativn\u011b kr\u00e1tk\u00e9 doby existence partikul\u00e1rn\u00edho typu sondy se asimila\u010dn\u00ed syst\u00e9my nenau\u010d\u00ed dan\u00fd typ dat optim\u00e1ln\u011b vyu\u017e\u00edt, na rozd\u00edl od dne\u0161n\u00edho p\u0159\u00edkladu dat TOVS.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Krom\u011b rozvoje glob\u00e1ln\u00edch syst\u00e9m\u016f asimilace dat pro\u017e\u00edv\u00e1 nev\u00eddan\u00fd v\u00fdvoj asimilace dat do mezom\u011b\u0159\u00edtkov\u00fdch model\u016f atmosf\u00e9ry. Zde ve srovn\u00e1n\u00ed s glob\u00e1ln\u00edmi modely jsou algoritmy \u0159e\u0161en\u00ed pom\u011brn\u011b rozd\u00edln\u00e9. Pokud se v\u016fbec pou\u017e\u00edv\u00e1 \u010dty\u0159- dimenzion\u00e1ln\u00ed metoda, tak potom s velmi kr\u00e1tk\u00fdm asimila\u010dn\u00edm oknem. Je to p\u0159edev\u0161\u00edm proto, \u017ee 4DVAR je velice drah\u00fd n\u00e1stroj a je tak v ostr\u00e9m protikladu k po\u017eadavku na co nejrychlej\u0161\u00ed poskytov\u00e1n\u00ed kr\u00e1tkodob\u00e9 mezom\u011b\u0159\u00edtkov\u00e9 progn\u00f3zy. Jsou tu ale i jin\u00e9 d\u016fvody pro to, aby asimila\u010dn\u00ed okno bylo relativn\u011b kr\u00e1tk\u00e9. V modelech mezom\u011b\u0159\u00edtka je toti\u017e pot\u0159eba co nejv\u00edce vyu\u017e\u00edt i b\u011bhem asimilace bohat\u00fd popis diabatick\u00fdch ireversibiln\u00edch proces\u016f, ke kter\u00fdm se velmi obt\u00ed\u017en\u011b hled\u00e1 sdru\u017een\u00e9 zobrazen\u00ed. Pokud se i povede dan\u00e9 rovnice regularizovat a vhodn\u00e9 sdru\u017een\u00e9 zobrazen\u00ed naj\u00edt, stejn\u011b v\u011bt\u0161inou doch\u00e1z\u00ed k tvorb\u011b nezanedbateln\u00fdch systematick\u00fdch chyb, kter\u00e9 zamezuj\u00ed \u00fasp\u011b\u0161n\u00e9 konvergenci varia\u010dn\u00edho algoritmu. S t\u00edmto probl\u00e9mem bojuj\u00ed zejm\u00e9na glob\u00e1ln\u00ed modely synoptick\u00fdch m\u011b\u0159\u00edtek, kde je zapojen\u00ed diabatick\u00fdch proces\u016f do varia\u010dn\u00edho \u0159e\u0161en\u00ed BLUE odhadu pom\u011brn\u011b minim\u00e1ln\u00ed. Ale pr\u00e1v\u011b v mezom\u011b\u0159\u00edtkov\u00fdch modelech s velmi kr\u00e1tk\u00fdm asimila\u010dn\u00edm oknem systematick\u00e9 chyby nemaj\u00ed \u010das dos\u00e1hnout v\u00fdznamn\u011bj\u0161\u00edch hodnot; t\u00edm p\u00e1dem zapojen\u00ed diabatick\u00fdch proces\u016f do mezom\u011b\u0159\u00edtkov\u00e9ho 4DVAR m\u016f\u017ee b\u00fdt daleko bohat\u0161\u00ed ne\u017e v p\u0159\u00edpad\u011b glob\u00e1ln\u00edch model\u016f s del\u0161\u00edm asimila\u010dn\u00edm oknem.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Tak\u00e9 se st\u00e1le rozv\u00edjej\u00ed postupy asimilace velmi lok\u00e1ln\u00edch dat, nap\u0159\u00edklad z meteorologick\u00fdch radar\u016f, kter\u00e1 se v glob\u00e1ln\u00edch modelech nevyu\u017e\u00edvaj\u00ed. Sympozium d\u00e1le uk\u00e1zalo, \u017ee se asimila\u010dn\u00ed techniky za\u010d\u00ednaj\u00ed pou\u017e\u00edvat i pro chemick\u00e9 transportn\u00ed modely, kde se otev\u00edraj\u00ed mo\u017enosti dal\u0161\u00edch aplikac\u00ed.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Dal\u0161\u00ed podrobnosti, t\u00fdkaj\u00edc\u00ed se sympozia, lze naj\u00edt na webov sk\u00e9 str\u00e1nce: <a style=\"color: #003366;\" href=\"http:\/\/www.chmi.cz\/meteo\/ok\/dasvmpos\/index.htm\">www.chmi.cz\/meteo\/ok\/dasvmpos\/index.htm<\/a>. na kterou je odkaz i z hlavn\u00ed str\u00e1nky \u010cHM\u00da a kde je mo\u017en\u00e9 nal\u00e9zt or\u00e1ln\u00ed prezentace. \u010cHM\u00da tak\u00e9 vydal sborn\u00edk kr\u00e1tk\u00fdch abstrakt\u016f v\u0161ech p\u0159ijat\u00fdch p\u0159\u00edsp\u011bvk\u016f, kter\u00fd je k dispozici v knihovn\u011b \u00fastavu. Tento sborn\u00edk byl \u00fa\u010dastn\u00edky sympozia velice ocen\u011bn, a to jak po str\u00e1nce edi\u010dn\u00ed, tak po str\u00e1nce estetick\u00e9 a kvality vazby. Vybran\u00e9 p\u0159edn\u00e1\u0161ky ze sympozia budou publikov\u00e1ny jako \u010dl\u00e1nky ve zvl\u00e1\u0161tn\u00edm vyd\u00e1n\u00ed Quarterly Journal of the Royal Meteorological Society.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\"><strong>ZAPOJEN\u00cd \u010cHM\u00da DO ASIMILACE DAT<\/strong><\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Jak jsme se pr\u00e1v\u011b zm\u00ednili, problematika asimilace dat v modelech mezom\u011b\u0159\u00edtka je v n\u011bkter\u00fdch aspektech podstatn\u011b rozd\u00edln\u00e1 od p\u0159\u00edpadu glob\u00e1ln\u00edch model\u016f. Pr\u00e1v\u011b na tomto poli se anga\u017euje odd\u011blen\u00ed numerick\u00e9 p\u0159edpov\u011bdi po\u010das\u00ed \u010cesk\u00e9ho hydrometeorologick\u00e9ho \u00fastavu. P\u0159edpov\u011bdn\u00ed model ALADIN, kter\u00fd se provozn\u011b po\u010d\u00edt\u00e1 v rozli\u0161en\u00ed 9 km pro \u00fazem\u00ed st\u0159edn\u00ed Evropy, kombinuje ve sv\u00e9 po\u010d\u00e1te\u010dn\u00ed podm\u00ednce 4DVAR anal\u00fdzu modelu ARPEGE (M\u00e9t\u00e9o-France) se sv\u00fdm vlastn\u00edm p\u0159edb\u011b\u017en\u00fdm odhadem informace jemn\u00e9ho m\u011b\u0159\u00edtka, kter\u00e9 nen\u00ed glob\u00e1ln\u00edm modelem ji\u017e posti\u017eeno. Specifick\u00e1 verze t\u00e9to metody, kter\u00e9 se v odborn\u00e9m \u017eargonu \u0159\u00edk\u00e1 blending, byla vyvinuta v Praze v r\u00e1mci st\u0159edoevropsk\u00e9 spolupr\u00e1ce LACE (Limited Area modelling in Central Europe).<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Asimila\u010dn\u00ed cyklus blendingu modelu ALADIN je sv\u00e1z\u00e1n s asimila\u010dn\u00edm cyklem ARPEGE, tj. pracuje se 4DVAR anal\u00fdzami spo\u010dten\u00fdmi v re\u017eimu tzv. dlouh\u00e9ho \u201ecut-off\u201c, kdy se \u010dek\u00e1 d\u00e9le na pozorov\u00e1n\u00ed po\u0159izovan\u00e1 v re\u00e1ln\u00e9m \u010dase (zejm\u00e9na dru\u017eice, letadla). Pro implicitn\u00ed blending spekter anal\u00fdzy ARPEGE a p\u0159edb\u011b\u017en\u00e9ho odhadu ALADIN se vyu\u017e\u00edv\u00e1 metoda digit\u00e1ln\u00edho filtru. Ve vlastn\u00ed produk\u010dn\u00ed p\u0159edpov\u011bdi, kter\u00e1 ale nevstupuje do asimila\u010dn\u00edho cyklu, se blending prov\u00e1d\u00ed s produk\u010dn\u00ed anal\u00fdzou ARPEGE (spo\u010dtenou p\u0159i krat\u0161\u00edm cut-off \u010dase pro sb\u011br pozorov\u00e1n\u00ed) a p\u0159edb\u011b\u017en\u00fdm odhadem modelu ALADIN z asimila\u010dn\u00edho cyklu. Aplikace blendingu zlep\u0161uje jak kvalitu po\u010d\u00e1te\u010dn\u00edch podm\u00ednek, tak n\u00e1sledn\u00e9 p\u0159edpov\u011bdi oproti metod\u011b prost\u00e9 dynamick\u00e9 adaptace. V anal\u00fdz\u00e1ch modelu ALADIN se konzistentn\u011b zachov\u00e1vaj\u00ed struktury mezom\u011b\u0159\u00edtka, jak\u00fdmi jsou nap\u0159\u00edklad mo\u0159sk\u00e9 a horsk\u00e9 br\u00edzy, organizovan\u00e9 konvek\u010dn\u00edmi p\u00e1sy a podobn\u011b.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">D\u00e1le se na pra\u017esk\u00e9m pracovi\u0161ti provozuje aplikace naz\u00fdvan\u00e1 DiagPack, kter\u00e1 je vhodn\u00e1 pro okam\u017eitou p\u0159edpov\u011b\u010f po\u010das\u00ed (nowcasting); jde o hodinov\u00e9 objektivn\u00ed anal\u00fdzy, vyu\u017e\u00edvaj\u00edc\u00ed hust\u00e9 s\u00edt\u011b p\u0159\u00edzemn\u00edch pozorov\u00e1n\u00ed. Tyto anal\u00fdzy se po\u010d\u00edtaj\u00ed metodou optim\u00e1ln\u00ed interpolace. Jej\u00ed statistick\u00fd model kovarianc\u00ed chyb modelu je p\u0159itom vylad\u011bn tak, aby anal\u00fdza nebyla optim\u00e1ln\u00ed kombinac\u00ed informace z pozorov\u00e1n\u00ed a p\u0159edb\u011b\u017en\u00e9ho pole, ale aby se co nejv\u00edce p\u0159ibl\u00ed\u017eila pozorov\u00e1n\u00edm. Jde tedy o anal\u00fdzy, kter\u00e9 slou\u017e\u00ed v\u00fdhradn\u011b jako diagnostick\u00fd n\u00e1stroj aktu\u00e1ln\u00edho stavu po\u010das\u00ed; nejsou toti\u017e vhodn\u00e9 jako po\u010d\u00e1te\u010dn\u00ed podm\u00ednka pro model. Je to pr\u00e1v\u011b kv\u016fli v\u00fd\u0161e zm\u00edn\u011bn\u00e9mu p\u0159elad\u011bn\u00ed statistick\u00e9ho modelu. Anal\u00fdza je pak sice bl\u00ed\u017ee pozorov\u00e1n\u00edm, ale d\u00e1le od pomal\u00e9 variety modelu; t\u00edm p\u00e1dem by se nevyv\u00e1\u017een\u00e1 \u010d\u00e1st p\u0159\u00edr\u016fstk\u016f anal\u00fdzy k p\u0159edb\u011b\u017en\u00e9mu odhadu projevila v n\u00e1sledn\u00e9 integraci jako \u0161um a ne jako sign\u00e1l.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Ve v\u00fdzkumu je t\u00fdm \u010cHM\u00da zapojen do pr\u00e1ce na ur\u010den\u00ed matice kovarianc\u00ed statistick\u00fdch chyb p\u0159edpov\u011bdn\u00edho modelu na omezen\u00e9 oblasti tak, aby se mohla vybudovat mezom\u011b\u0159\u00edtkov\u00e1 asimila\u010dn\u00ed metoda navazuj\u00edc\u00ed na blending, ale aby se p\u0159itom nepo\u0161kodilo dlouhovlnn\u00e9 spektrum vyu\u017e\u00edvaj\u00edc\u00ed v\u00fdsledky 4DVAR anal\u00fdzy ARPEGE. Pro tento \u00fa\u010del byl vypo\u010dten statistick\u00fd model kovarianc\u00ed chyb modelu ALADIN, kter\u00fd eliminuje vliv bo\u010dn\u00edch okrajov\u00fdch podm\u00ednek. Na krok blendingu by tak nav\u00e1zala mezom\u011b\u0159\u00edtkov\u00e1 anal\u00fdza typu 3DVAR vyu\u017e\u00edvaj\u00edc\u00ed t\u011bchto statistik Pro \u00fa\u010dely mezom\u011b\u0159\u00edtkov\u00e9 anal\u00fdzy je ale vhodn\u00e9 vyu\u017e\u00edt pokud mo\u017eno hust\u00e1 lok\u00e1ln\u00ed pozorov\u00e1n\u00ed, kter\u00e1 nebyla ji\u017e zahrnuta v glob\u00e1ln\u00ed 4DVAR anal\u00fdze. Proto se v tomto roce, kter\u00fd byl mo\u017en\u00e1 optimisticky odd\u011blen\u00edm numerick\u00e9 p\u0159edpov\u011bdi po\u010das\u00ed vyhl\u00e1\u0161en rokem asimilace dat, za\u010dalo pracovat na p\u0159\u00edprav\u011b asimilace satelitn\u00edch dat. Konkr\u00e9tn\u011b jde o data ze senzoru SEVIRI (Spinning Enhanced Visible and Infra-Red Imager) dru\u017eice METEOSAT druh\u00e9 generace (MSG), kter\u00e1 maj\u00ed v\u00fdhodu relativn\u011b hust\u00e9ho prostorov\u00e9ho a \u010dasov\u00e9ho rozli\u0161en\u00ed. V\u00fdhledov\u011b se tak\u00e9 po\u010d\u00edt\u00e1 s asimilac\u00ed radarov\u00fdch dat; tady jde ale je\u0161t\u011b o v\u00fdzkumn\u00fd b\u011bh na dlouhou tra\u0165, ne\u017e se bude da\u0159it skute\u010dn\u011b optim\u00e1ln\u011b vyu\u017e\u00edt informaci obsa\u017eenou v t\u011bchto pozorov\u00e1n\u00edch.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Krom\u011b asimilace dat v mezom\u011b\u0159\u00edtku se t\u00fdm \u010cHM\u00da pod\u00edl\u00ed i na vyu\u017eit\u00ed metod asimilace 4DVAR pro \u00fa\u010dely synoptick\u00e9 p\u0159edpov\u011bdi. Jde o techniku, kter\u00e1 umo\u017e\u0148uje asimilovat do modelu pomoc\u00ed 4DVAR algoritmu tzv. opravu, kterou m\u016f\u017ee vygenerovat synoptik na z\u00e1klad\u011b \u010derstv\u00fdch pozorov\u00e1n\u00ed v p\u0159\u00edpad\u011b, kdy se \u0159e\u0161en\u00ed p\u0159edpov\u011bdn\u00edho modelu za\u010d\u00edn\u00e1 od nich odkl\u00e1n\u011bt. Jde hlavn\u011b o porovn\u00e1n\u00ed struktur synoptick\u00e9ho m\u011b\u0159\u00edtka, zejm\u00e9na anom\u00e1li\u00ed potenci\u00e1ln\u00ed vorticity na horn\u00ed hranici mezn\u00ed vrstvy a v hladin\u011b tropopauzy. Metodou inverze potenci\u00e1ln\u00ed vorticity lze potom spo\u010d\u00edtat \u201eopraven\u00e9\u201c pole teploty, v\u011btru a tlaku. Tuto metodu lze tedy pou\u017e\u00edvat pouze v synoptick\u00e9m m\u011b\u0159\u00edtku, nicm\u00e9n\u011b m\u00e1 potenci\u00e1l zlep\u0161it p\u0159edpov\u011b\u010f v\u00fdznamn\u00e9ho po\u010das\u00ed, typicky bou\u0159livou cyklogenezi a podobn\u011b.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">\u010cesk\u00fd hydrometeorologick\u00fd \u00fastav tak\u00e9 navazuje v\u00fdzkumnou spolupr\u00e1ci v oboru asimilace dat s dal\u0161\u00edmi n\u00e1rodn\u00edmi pracovi\u0161ti. Jde zejm\u00e9na o \u00dastav informatiky AV \u010cR, kter\u00fd se zab\u00fdv\u00e1 asimilac\u00ed dat zne\u010di\u0161t\u011bn\u00ed atmosf\u00e9ry a jejich potenci\u00e1lem pro mo\u017enost \u0159edpov\u011bdi kvality ovzdu\u0161\u00ed.<\/span><\/p><p class=\"rtejustify\"><span style=\"color: #003366;\">Asimilace dat je discipl\u00edna, kter\u00e1 ve srovn\u00e1n\u00ed s ostatn\u00edmi discipl\u00ednami numerick\u00e9 p\u0159edpov\u011bdi po\u010das\u00ed m\u00e1 svoje specifick\u00e9 n\u00e1roky jednak na v\u00fdpo\u010detn\u00ed techniku, jednak na p\u0159edb\u011b\u017en\u00e9 zpracov\u00e1n\u00ed zejm\u00e9na nekonve\u010dn\u00edch dat a sestaven\u00ed vhodn\u00e9 datab\u00e1ze pozorov\u00e1n\u00ed. Tyto technick\u00e9 po\u017eadavky, zvl\u00e1\u0161t\u011b t\u00fdkaj\u00edc\u00ed se zpracov\u00e1n\u00ed dat a vhodn\u00e9ho datab\u00e1zov\u00e9ho \u0159e\u0161en\u00ed, p\u0159edstavuj\u00ed nemal\u00fd logistick\u00fd probl\u00e9m pro men\u0161\u00ed meteorologick\u00e9 slu\u017eby. Doufejme proto, \u017ee nov\u00fd ambici\u00f3zn\u00ed datab\u00e1zov\u00fd syst\u00e9m SDNES, budovan\u00fd v \u010cHM\u00da, bude odpov\u00eddat p\u0159\u00ed\u0161t\u00edm provozn\u00edm po\u017eadavk\u016fm asimilace dat, poda\u0159\u00ed-li se v tomto sm\u011bru ud\u011blat pokrok. Jako prvn\u00ed asimila\u010dn\u00ed \u00faloha, kter\u00e1 se p\u0159ipravuje pro provoz modelu ALADIN v \u010cHM\u00da, je anal\u00fdza prom\u011bnn\u00fdch modelu zemsk\u00e9ho povrchu (p\u016fdn\u00ed teploty a vlhkost, sn\u00edh, atd.) metodou optim\u00e1ln\u00ed interpolace. Po\u010d\u00e1te\u010dn\u00ed stav v\u00fd\u0161kov\u00fdch pol\u00ed bude i nad\u00e1le o\u0161et\u0159en pomoc\u00ed spektr\u00e1ln\u00edho blendingu. P\u0159ed pl\u00e1novan\u00fdm spu\u0161t\u011bn\u00edm paraleln\u00edho testu zb\u00fdv\u00e1 je\u0161t\u011b do\u0159e\u0161it n\u011bkter\u00e9 algoritmick\u00e9 detaily napojen\u00ed t\u00e9to asimilace na ostatn\u00ed kroky p\u0159\u00edpravy po\u010d\u00e1te\u010dn\u00ed podm\u00ednky. Nebo\u0165, jak s oblibou \u0159\u00edkaj\u00ed specialist\u00e9 na asimilaci dat, \u010f\u00e1bel je schov\u00e1n v mali\u010dkostech.<\/span><\/p><p class=\"rteright\"><span style=\"color: #003366;\"><em>Radmila Bro\u017ekov\u00e1, MZ 2005\/3, ro\u010dn\u00edk 58, str. 85-89<\/em><\/span><\/p><\/div><\/div><\/div><\/div>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Slovo \u00favodem Jak zn\u00e1mo, p\u0159edpov\u011b\u010f po\u010das\u00ed je vlastn\u011b probl\u00e9m zad\u00e1n\u00ed po\u010d\u00e1te\u010dn\u00edch podm\u00ednek. Z\u00e1vis\u00ed toti\u017e v prv\u00e9 \u0159ad\u011b na co nejlep\u0161\u00ed znalosti aktu\u00e1ln\u00edho stavu atmosf\u00e9ry. Tak se dost\u00e1v\u00e1me k pojmu anal\u00fdza, [&hellip;]<\/p>\n","protected":false},"author":3,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-11683","page","type-page","status-publish","hentry","entry"],"_links":{"self":[{"href":"http:\/\/www.cmes.cz\/web\/wp-json\/wp\/v2\/pages\/11683","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.cmes.cz\/web\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"http:\/\/www.cmes.cz\/web\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"http:\/\/www.cmes.cz\/web\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"http:\/\/www.cmes.cz\/web\/wp-json\/wp\/v2\/comments?post=11683"}],"version-history":[{"count":4,"href":"http:\/\/www.cmes.cz\/web\/wp-json\/wp\/v2\/pages\/11683\/revisions"}],"predecessor-version":[{"id":11687,"href":"http:\/\/www.cmes.cz\/web\/wp-json\/wp\/v2\/pages\/11683\/revisions\/11687"}],"wp:attachment":[{"href":"http:\/\/www.cmes.cz\/web\/wp-json\/wp\/v2\/media?parent=11683"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}