Analysis of the influence of input data uncertainties on determining the reliability of reservoir storage capacity

J. Hyd ol. Hyd omech., 63, 2015, 4, 287–294 DOI: 10.1515/johh-2015-0036 287 Analysis o he in luence o inpu da a unce ain ies on de e mining he eliabili y o ese oi s o age capaci y Daniel Ma on*, Miloš S a ý, Pa el Menšík B no Uni e si y o Technology, Facul y o Ci il Enginee ing, Ins i u e o Landscape Wa e Managemen , Ve eří 331/95, 616 00, B no, Czech epublic. * Co esponding au ho . Tel.: +420 541 147 773. Fax: +420 541 147 771. E-mails: ma on.[email p o ec ed], s a y.m@ ce. u b .cz, [email p o ec ed] Abs ac : The pape con ains a sensi i i y analysis o he in luence o unce ain ies in inpu hyd ological, mo phological and ope a ing da a equi ed o a p oposal o ac i e ese oi conse a ion s o age capaci y and i s achie ed alues. By in oducing unce ain ies in o he conside ed inpu s o he wa e managemen analysis o a ese oi , he subsequen analysed ese oi s o age capaci y is also a ec ed wi h unce ain ies. The alues o wa e ou lows om he ese oi and he hyd ological eliabili ies a e a ec ed wi h unce ain ies as well. A simula ion model o ese oi beha iou has been compiled wi h his kind o calcula ion as s a ed below. The model allows e alua ion o he solu ion esul s, aking unce ain ies in o conside a ion, in con ibu ing o a educ ion in he occu ence o ailu e o lack o wa e du ing ese oi ope a ion in low-wa e and d y pe iods. Keywo ds: Unce ain ies; Reliabili y; Rese oi s o age capaci y; Mon e Ca lo me hod; Mean mon hly lows; E apo a- ion; Ele a ion– olume cu e; Ele a ion–a ea cu e. INTRODUCTION The cu en knowledge in he ield o clima ology indica es a g adual change in hyd oclima ic condi ions all o e he wo ld. Clima e changes a e e lec ed in he changes in he hyd ological cycle due o he edis ibu ion o p ecipi a ion du ing he yea and hey con ibu e o mo e equen occu ences o ex emes in he o m o loods and d y pe iods. Clea signs o clima ic changes ha e appea ed in he Czech Republic in ecen yea s. I should be no ed ha , om he hy- d ological poin o iew, 2011 and 2012 we e conside ed o be ex emely d y (Zah adníček e al., 2014). The empe a u es in he win e in 2014 we e conside ably abo e a e age. In ha pe iod, he wa e s o age in snow co e was he lowes in he las wen y yea s. The consequences we e ex ao dina y manip- ula ions a some wa e ese oi s. I is appa en ha he subjec o ad anced managemen and con ol o su ace wa e esou ces is becoming mo e and mo e impo an . The manipula ion ules o la ge open wa e ese oi s we e app o ed in he pe iod o cons uc ion o hose wa e wo ks and subsequen ly e iewed o he cu en hyd ological condi ions. I will be necessa y o ca y ou a ho ough e iew in he u u e in ela ion o hei adap abil- i y due o clima e change and hyd ological cyclic e olu ion. The e o e, he asks o wa e managemen analysis o ese oi s o ac i e ese oi conse a ion s o age capaci y will always be necessa y and esea ch in his ield is aluable. In pa icula , he applica ion o new op imiza ion me hods o wa e managemen analysis o ese oi s, new ese oi pe o mance de ini ions and, las bu no leas , he in oduc ion o analysis o unce ain- ies in hese p oblems o combina ions o he abo e men ioned applica ions and knowledge. Unde he gi en condi ions, i is necessa y o in oduce inpu da a unce ain ies in o he p oposal o con ol o ese oi s o age capaci y. The p ocedu es which will be desc ibed below e e o “The in luence o unce ain ies in he calcula ion o mean mon hly discha ges on ese oi s o age” (Ma on e al., 2011). This pape desc ibes, in de ail, he in oduc ion o unce - ain ies o measu emen in de e mining below a e age mon hly lows o e he s age-discha ge cu e in a i e and a numbe o measu emen s o hou ly eco ds o i e s ages in a hyd ome ic p o ile. Fo his so o compu a ion, he Mon e Ca lo me hod was used. One o he esul s was c ea ing he andom ime se- ies o mean mon hly lows, which was a ec ed by unce ain- ies o measu emen in he hyd ome ic p o ile. The andom se ies o mean mon hly lows se ed as inpu da a o he wa e managemen analysis o he ese oi s o age capaci y, when he spec um o ese oi s o age olumes o he maximum hyd o- logical eliabili y o 100% we e de e mined epea edly using single-pass simula ions o ese oi ope a ion. The inal spec- um o andom ese oi s o age capaci ies was e alua ed s a- is ically and he in e al o possible alues o ese oi s o age olumes was ound. The o he wo k e e ed o in his pape is a pape by Ma on e al. (2014) desc ibing he calcula ion o ese oi s o age capaci y in he condi ions o measu emen unce ain ies and ex ended o using he AR and ARMA models, gene a o s o a i icial s eam low se ies o mean mon hly dis- cha ges. Random discha ge se ies, in his case as da a inpu s o a i icial s eam low se ies gene a o s, we e used, esul ing in andom samples o a i icial s eam low se ies. These andom se ies we e e alua ed using a ese oi simula ion model. The calcula ion esul was a spec um o ese oi s o age capaci ies o maximum eliabili y o 100 %, which was s a is ically e al- ua ed. Bo h pape s ha e indica ed ha he cu en wa e ol- umes in ese oi s can be unde es ima ed and, in d y pe iods, may esul in an unexpec ed ailu e in su ace wa e supply. As ega ds cu en knowledge, unce ain y can be de i ed om se heo y, bu also unce ain y can be de i ed using s a is- ics. F om se heo y, i is necessa y o men ion he applica ion o unce ain y based on he heo y o uzzy se s (Zadeh, 1965) and he heo y o possibili ies (Kli , 2005). Howe e , he i s de ini ion o unce ain y was by Knigh (1921), nowadays known as Knigh ian unce ain y. The unce ain y concep is cu en ly iewed om di e en aspec s, wi h plen y o de ini- ions and poin s o iew, such as he unce ain y associa ed wi h he de ini ion o isk, unce ain ies applied in o ecas ing p ob- lems, and also he unce ain y o measu emen . Unce ain ies Unau hen ica ed Download Da e | 8/10/17 12:51 PM Daniel Ma on, Miloš S a ý, Pa el Menšík 288 applied in hyd ology we e desc ibed, o example, by Be en and Binley (1992). They desc ibed in de ail a me hod called GLUE – gene alized likelihood unce ain y es ima ion. This was ollowed by nume ous publica ions which deal wi h his issue, such as Be en (2007). Unce ain ies in measu emen we e i s o mula ed on he basis o he WECC (1990) ag eemen . A s a is ical app oach using he concep o unce ain y o meas- u emen , which clea ly de ined he in oduc ion and calcula ion o measu emen unce ain ies, was in oduced as he “Guide o Exp ession o Unce ain y in Measu emen ” (GUM 1993). The ISO GUIDE 99998 S anda d (2004) deals wi h he dis ibu ion and p opaga ion o unce ain ies using Mon e Ca lo simula ion. Hyd ological applica ions, including p opaga ion o unce ain- ies in o hyd ological inpu s when measu ing p ecipi a ion, wa e in lows in o ese oi s and e apo a ion in he wa e bal- ances o ese oi s, we e deal wi h by Win e (1981). LaBaugh and Win e (1984) examined he in luence o unce ain ies in he measu emen s o wa e in low in o ese oi s, wa e ou - lows om ese oi s and e apo a ion, and o he hyd ological and ope a ing pa ame e s on he olume and chemical analysis o wa e in ese oi s. The la es publica ions, o example Campos e al. (2014), examined he isks and unce ain ies in, and in luence on ese oi s o age using Mon e Ca lo simula- ions. Ku ia and Vogel (2014) ca ied ou an analysis o unce - ain ies in ese oi s o age using he wa e supply yield model. Coxon e al. (2015) publish pape which apply no el amewo k o discha ge unce ain y o UK gauging s a ions. Ano he e- sea ch a ea in he applica ion o wa e managemen simula ion and modelling is unce ain y associa ed wi h non-s a iona y p ocesses in ime se ies. This issue is oo expensi e o he pu poses o his wo k and needs ex ensi e independen esea ch. The aim o he pape is o p esen ano he possible applica- ion o he Mon e Ca lo me hod o in oducing unce ain ies in o he hyd ological, mo phological and ope a ing inpu da a equi ed o ese oi s o age capaci y design, which is c ucial in low-wa e pe iods. This is also connec ed wi h he calcula ion o eliabili y o wa e ou low om he ese oi in adap i e condi ions. Inpu hyd ological, mo phological and ope a ing da a o he solu ion a e conside ed, especially wa e in low in o he ese oi , wa e losses om he ese oi by e apo a ion om he wa e su ace and by dam seepage, he ese oi ele a- ion– olume cu e and he ese oi ele a ion–a ea cu e. A ese oi s o age model was c ea ed o his pu pose using a single-pass simula ion me hod o de e mine eliabili y o wa e ou lows om he ese oi , bo h conside ing wa e losses om he ese oi and igno ing such losses. By in oducing unce - ain y in o he inpu da a, he Mon e Ca lo me hod is used o de e mine, by epea ed solu ions, he spec um o eliabili y o ese oi s o age capaci y. This is hen e alua ed and app op i- a ely in e p e ed. As he unce ain ies in he inpu da a a e un- known, he wo k ocuses on compiling sensi i i y analysis be ween he unce ain y in he inpu da a o a solu ion o he s o age capaci y o he ese oi and he unce ain y o he achie ed eliabili y o con olled wa e ou lows om he ese - oi . METHOD Mon e Ca lo me hod The gene al p ocedu e o gene a ing unce ain y a ec ed hy- d ological, mo phological and ope a ing inpu da a o he ela - ed wa e managemen analysis o a ese oi o i s s o age capaci y is as ollows. Unce ain ies o inpu quan i ies a e in o- duced in o he calcula ions using he Mon e Ca lo me hod. Using he dis ibu ion cu e F(X), a andom posi ion o alues NXi wi hin he in e al o a gi en unce ain y a e gene a ed as inpu alue Xi. Value Xi is conside ed andom and independen o alues Xi–1 and Xi+1. This p esump ion will allow he in oduc ion o he no mal p obabili y dis ibu ion N( μ (X), σ (X)). Then, each inpu alue Xi is conside ed as a mean alue μ(X) and he amoun o unce ain y is de ined as he s anda d de ia ion σ (X). Subsequen ly, a cumula i e dis ibu ion unc ion Fi(X) o no mal s anda dized p obabili y dis ibu ion is gene a ed o each mean alue μ (Xi). The pseudo andom numbe gene a o gene a es a andom numbe om an in e al in which andom quan i y alue NXi is gene a ed. Fig. 1. P inciple o gene a ing unce ain ies in inpu elemen s using he Mon e Ca lo me hod. Unau hen ica ed Download Da e | 8/10/17 12:51 PM Analysis o he in luence o inpu da a unce ain ies on de e mining he eliabili y o ese oi s o age capaci y 289 Fig. 2. Symbolic in oduc ion o conside ed quan i ies a ec ed wi h unce ain ies. The basic p inciple o gene a ing andom posi ions o poin s in he wo-dimensional coo dina e sys em (NXi,NYi) is iden ical o he heo y desc ibed abo e. The dissimila i y is gi en by plo ing a poin which equi es plo ing wo Mon e Ca lo gene a- o s independen o each o he . Each gene a o p oduces a an- dom posi ion o poin NXi (e.g. wa e le el ele a ion Nhi) and wi h i a andom alue NYi (wa e olume in ese oi NVi). The esul on he ese oi ele a ion– olume cu e is hen a andom poin coo dina e (NVi,Nhi) o he ese oi ele a ion– olume cu e. See Fig. 1. As obse ed, he ese oi in low, wa e su ace e apo a ion, dam seepage and ese oi ele a ion–a ea and ele a ion– olume cu es a e conside ed o be hyd ological and ope a ing inpu s. The p inciple o in oducing unce ain ies in o he calcu- la ion o ese oi s o age capaci y is shown in Fig. 2. The gene a ed andom cu es o wa e in lows in o he es- e oi , wa e e apo a ion om he wa e su ace, seepages and andom a ea and ele a ion– olume cu es se e as inpu alues o a simula ion model which, using single-pass simula ion, simula es he beha iou o he ese oi in he condi ions o da a a ec ed wi h unce ain y. Rese oi simula ion model and ese oi pe o mance calcula ion The basis o a ese oi simula ion model is an adjus ed equa ion in he cumula i e o m con e ed o he ollowing inequali ies (1) S a ý (2005). 0≤∑󰇛−󰇜∆+󰇛 −󰇜∆ ≤,   (1) whe e Oi is he ese oi ou low, Qi, he ese oi in low o i = 1, …, n, and Δ is he ime s ep o calcula ion (one mon h). Ok+1 is he ou low om he ese oi in he ollowing ime s ep, when in s ep i+1 he alue Oi+1 is i s eplaced wi h he alue o he equi ed ou low Op. The ime cou se o he numbe ed sum simula es he cou se o emp ying he ese oi s o age by ime s eps i = 1, …, k. Fo i = 0 i is necessa y o en e he s a ing solu ion condi ion a e he sum alue. Inequali y (1) is limi ed om bo h he le and he igh . F om he le i is limi ed by alue 0 ( ull s o age capaci y) and om he igh by alue Vz,max (emp y s o age capaci y) cha ac e izing he ese oi s o age capaci y a ailable o he ese oi . By calcula ing he alue o he exp ession, he cu en emp ying o he s o age olume , 󰆒 is ob ained and i is hen es ed as o whe he i lies in a pa icula in e al 〈0, VZ,max〉. I no , i is necessa y o ind alue Oi+1 ( o he sum o he exp ession o be equal o ze o, idle dis- cha ge will occu , o i equal o Vz,max – a ailu e will occu ). The gene al de ini ion o eliabili y was successi ely desc ibed by (Hashimo o e al., 1982; Klemeš, 1967; K i skiy and Menkel, 1952). The classi ica ion o a ailu e in he ese oi s o age capaci y o he ollowing calcula ion o eliabili y is as ollows (2). , = , =1,  ≥  , =0,  <  (2) Z ,i = 1 desc ibes he ese oi s o age capaci y in a no ailu e si ua ion (sa is ac o y s a e). Z ,i = 0 desc ibes he ese oi s o - age capaci y in a ailu e si ua ion (unsa is ac o y s a e). The equi ed eliabili y can be u he calcula ed om alues Z ,i. Gene ally, eliabili y is calcula ed by ime-based eliabili y as empo al eliabili y and occu ence eliabili y, and olume ic eliabili y is calcula ed sepa a ely. The pape uses he o mula o he calcula ion o empo al eliabili y PT (3). = ∑,   (3) whe e k is he numbe o mon hs in he pe iod being sol ed. PRACTICAL APPLICATION The model was applied in p ac ice o he exis ing ese oi , Ví I, which is si ua ed in he Vysočina Region, Czech Republic. This is a mul i-use ese oi se ing mainly as lood p o ec ion and su ace wa e accumula ion o wa e supply and hyd oelec- ic pu poses. The ese oi is buil in he S a ka Ri e basin and has been in ope a ion since 1957. The S a ka is he main in low in o he ese oi . The mean long- e m in low in o he ese oi Qa is 3.34 m3 s–1. Inpu alues o he calcula ion we e made up o a ime se ies o mean mon hly lows o e 60 yea s wi h he measu emen pe iod om 1950 o 2010. The mean annual e apo a ion om he wa e su ace EANNUAL = 613 mm. The mon hly e apo a ion alues om wa e su ace we e de- i ed in a simpli ied manne acco ding o he pe cen age dis i- bu ion o e apo a ion acco ding o he ČSN 75 2405 S anda d (2004) and om he mean annual e apo a ion alues o Ví I ese oi , see Table 1. Unau hen ica ed Download Da e | 8/10/17 12:51 PM Daniel Ma on, Miloš S a ý, Pa el Menšík 290 Table 1. Mon hly dis ibu ion o e apo a ion amoun du ing he calenda yea . Mon h Jan Feb Ma Ap May Jun Jul Aug Sep Oc No Dec Em [%] 6 9 12 14 16 15 11 7 5 2 1 2 Em [mm] 36.78 55.17 73.56 85.82 98.08 91.95 67.43 42.91 30.65 12.26 6.13 12.26 Table 2. Calcula ion wi hou conside ing wa e losses om he ese oi . Measu emen unce ain ies a e applied o wa e in low in o he ese oi . Inpu da a unce ain y Ua = 3 σ ±3% ±6% ±9% ±15% O p [m3 s–1] μ (P T ) U a(P T ) μ (P T ) U a(P T ) μ (P T ) U a(P T ) μ (P T ) U a(P T ) 2.2 100.000 0.000 100.000 0.000 100.000 0.000 99.999 0.033 2.3 99.590 0.001 99.591 0.033 99.598 0.095 99.611 0.200 2.5 98.907 0.001 98.906 0.024 98.896 0.109 98.852 0.277 2.7 96.668 0.354 96.688 0.453 96.701 0.540 96.672 0.753 2.9 93.165 0.302 93.230 0.479 93.250 0.621 93.250 0.879 3.0 90.849 0.338 90.923 0.573 90.984 0.749 91.084 0.987 Fig. 3. Rela ion be ween equi ed ou low Op and empo al eliabili y PT wi hou conside ing. The o al ese oi olume is VTOTAL 56.193 x 106 m3, ac i e s o age olume VZ,max is 44.056 x 10 6 m3 and lood ese oi olume VFLOOD is 8.337 x 106 m3. The o al dam heigh is 67.3 m. The ecological low om he ese oi QECO is 0.53 m3 s–1. The alue o seepage h ough he dam was de i ed om empi ical obse a ion and o he g a i y conc e e dam i is 0.15 l s–1 pe 1000 m2. The calcula ion o empo al eliabili y PT o an inc eased ou - low om he ese oi was analysed wi h and wi hou conside - ing wa e losses om he ese oi . When wa e losses we e conside ed in he calcula ions o empo al eliabili y, he de- sc ibed p ocedu es o gene a ing unce ain y-a ec ed hyd olog- ical, mo phological and ope a ing inpu s we e applied. The anal- ysis was ca ied ou o he alues o inc eased equi ed ou low Op lying in he in e al Op∈〈2.1; 3.0〉 m3 s–1. The selec ed num- be o epe i ions using he Mon e Ca lo me hod was 300. Inpu unce ain ies o he analysis anged in in e als ±3, ±6, ±9, and ±15%. The algo i hm simula ing he beha iou o he ese oi hen calcula ed andom discha ges NOi o wa e om he ese - oi and empo al eliabili y o NPT. Then, andom cou ses o mon hly illing and emp ying o he ese oi s o age capaci y we e calcula ed. Fo a be e p esen a ion o he esul s, hese alues we e e alua ed s a is ically. The mean alue μ (X) o each andom se is conside ed o be he esul an alue and he s and- a d de ia ion σ (X) is conside ed o be he s anda d unce ain y ela ed o a pa icula esul . The o al, ex ended unce ain y, ype “Ua” co e ing almos 100% o speci ically 99.97% o occu - ences o he moni o ed quan i y, co esponded o alue μ (X)±3 σ . Sensi i i y analysis was ca ied ou o he calcula ion wi hou conside ing wa e losses om he ese oi , when only Unau hen ica ed Download Da e | 8/10/17 12:51 PM Analysis o he in luence o inpu da a unce ain ies on de e mining he eliabili y o ese oi s o age capaci y 291 in low in o he ese oi was bu dened wi h unce ain y, see Table 2 and Fig. 3. Calcula ions we e also made while conside - ing wa e losses om he ese oi . Fi s , only he e apo a ion alues, ele a ion– olume cu e, ele a ion–a ea cu e and seep- age h ough he dam we e a ec ed wi h unce ain y, see Table 3 and Fig. 4. Then, ese oi in low, e apo a ion, ele a ion– olume cu e, ele a ion-a ea cu e and seepage h ough he dam we e a ec ed wi h unce ain y, see Table 4 and Fig. 5. The shape o he cu es in Figs. 3, 4, and 5 in he ange om Op = 2.3 m3 s–1 o Op = 2.5 m3 s–1 is caused by a la ge ime s ep in he calcula ions (1 mon h), and also by a s ep inc ease in he numbe o ailu e mon hs, which is a small numbe in he gi en Table 3. Calcula ion wi h conside ing wa e losses om ese oi . Unce ain ies conside ed o e apo a ion, ele a ion- olume (a ea) cu es and seepage h ough dam body combina ions. Inpu da a unce ain y Ua = 3 σ ±6% ±9% ±15% O p [m3 s–1] μ (P T ) U a(P T ) μ (P T ) U a(P T ) μ (P T ) U a(P T ) 2.2 100.000 0.000 100.000 0.000 100.000 0.000 2.3 99.590 0.001 99.590 0.001 99.590 0.001 2.5 98.906 0.024 98.899 0.092 98.889 0.140 2.7 96.311 0.001 96.311 0.001 96.311 0.001 2.9 92.896 0.001 92.896 0.001 92.896 0.001 3.0 90.516 0.203 90.514 0.204 90.522 0.224 Fig. 4. Rela ion be ween equi ed ou low Op and empo al eliabili y PT, wi h conside ing wa e losses om ese oi o inpu unce ain- ies ±6, ±9, ±15, ±30% and e apo a ion, ele a ion- olume(a ea) cu es, dam seepage combina ions. Table 4. Calcula ion wi h conside ing wa e losses om ese oi . Unce ain ies conside ed o all in low, e apo a ion, ele a ion- olume (a ea) cu es, seepage h ough dam body combina ions. Inpu da a unce ain y Ua = 3 σ ±3% ±6% ±9% ±15% O p [m3 s–1] μ (P T ) U a(P T ) μ (P T ) U a(P T ) μ (P T ) U a(P T ) μ (P T ) U a(P T ) 2.2 100.000 0.000 100.000 0.000 99.996 0.070 99.978 0.191 2.3 99.587 0.057 99.566 0.155 99.555 0.179 99.539 0.204 2.5 98.859 0.196 98.843 0.215 98.811 0.293 98.721 0.451 2.7 96.316 0.084 96.348 0.338 96.350 0.510 96.300 0.797 2.9 92.939 0.290 92.957 0.454 92.943 0.584 92.896 0.838 3.0 90.556 0.402 90.575 0.591 90.589 0.734 90.608 1.038 Unau hen ica ed Download Da e | 8/10/17 12:51 PM Daniel Ma on, Miloš S a ý, Pa el Menšík 292 Fig. 5. Rela ion be ween equi ed ou low Op and empo al eliabili y PT, wi h conside ing wa e losses om ese oi o inpu unce ain- ies ±3, ±6, ±9 a ±15% and an in low, e apo a ion, ele a ion- olume(a ea) cu es and dam seepage combina ion. Fig. 6. The cou se o illing ese oi s o age capaci y in he condi ions o en e ed inpu da a unce ain ies Ua = ±3% a ±15% o he selec - ed low wa e pe iod. Unau hen ica ed Download Da e | 8/10/17 12:51 PM Analysis o he in luence o inpu da a unce ain ies on de e mining he eliabili y o ese oi s o age capaci y 293 ange o equi ed ou low om he ese oi Op. Fo alues Op = 2.6 m3 s –1 and highe , he e is an appa en inc ease in he ailu e mon hs, due o which he cu es a e smoo h. The mo e signi ican in e al o spacing in he cu es in Fig. 5, unlike Figs. 3 and 4, in he ield o equi ed ou low om he ese oi Op = 2.2 m 3 s –1 is caused by he numbe o ailu e mon hs occu ing in he e alua ed se . The p ocess con ingency applied when unce ain ies a e in oduced in o all inpu da a o he solu ion esul s in a signi ican inc ease in he numbe o ailu e mon hs compa ed o he solu ion in which only unce ain ies o ese oi in low o unce ain ies o e apo a ion, ele a ion– olume (–a ea) cu es and seepage h ough he dam combina ion a e applied. The analysis also included alues o illing he ese oi s o age capaci y. Fig. 6 hen shows he cou se o illing he ese oi o a pa icula numbe o epe i ions and o he selec ed low wa e pe iod. SUMMARY The inal compa ison is om selec ed equi ed ou low om he ese oi Op = 2.5 and Op = 3.0 m3 s–1, whe e he in luence is clea ly g aded. In he a ian wi hou applying wa e losses om he ese oi , he empo al eliabili y o Op = 2.5 m3 s–1 is in in e al PT∈〈98.906%; 98,908%〉, i.e. PT = 98.907% ±0.001% o inpu unce ain y ±3% and in in e al PT∈〈98.575%; 99,129%〉 PT = 98.852% ±0.277% o inpu unce ain y ±15%. The in e al o empo al eliabili y o he Op = 3.0 m3 s–1 ange in in e al PT∈〈90.511%; 91,187%〉 is PT = 90.849% ±0.338% o inpu unce ain y 3% and PT∈〈90.097%; 92,071%〉 PT = 91.084% ±0.987% o inpu unce ain y 15%. In he a ian wi hou applying all combina- ions, i.e. conside ing unce ain ies in bo h in low and wa e losses om he ese oi , he in e al o empo al eliabili y was in PT∈〈98.663%; 99,055%〉 PT = 98.859% ±0.196% o inpu unce ain y ±3% and o unce ain y ±15% in in e al PT∈〈98.27%; 99,172%〉 PT = 98.721% ±0.451%. Fo Op = 3.0 m3 s –1 he in e al o empo al eliabili y acqui ed alues PT∈〈90.154%; 90,958%〉 PT = 90.556% ±0.402% o inpu unce ain y 3% and PT∈〈89.57%; 91,646%〉 PT = 90.608% ±1.038% o inpu unce ain y 15%. The abo e men ioned esul s show a logical conclusion ha wi h inc easing inpu da a unce ain y, he unce ain y in empo al eliabili y also inc eas- es. Con e ed o he numbe o ailu e mon hs, inc eased dis- cha ge Op = 2.5 m3 s–1 in he solu ion wi hou conside ing un- ce ain ies and wi h conside ing ese oi wa e losses, co e- sponds o eigh ailu e mon hs, and o Op=3.0 m3 s –1 o 70 mon hs. When conside ing inpu unce ain ies ±3%, he numbe o ailu e mon hs is om 8 o 9 mon hs o Op = 2.5 m3 s–1 and 66 o 72 mon hs o Op = 3.0 m3 s–1. Fo he inpu da a unce - ain y o ±15%, he numbe o possible ailu e mon hs is 8 o 12 o Op = 2.5 m3 s–1 and 60 o 76 mon hs o Op = 3.0 m3 s–1. The p esen ed esul s show how unce ain ies can in luence he inc ease in ailu e mon hs and which in e als he empo al eliabili y can hen acqui e. CONCLUSIONS In he manipula ion ules o he Ví I ese oi , he s a ed empo al eliabili y o he hyd ological pe iod 1931 o 1991 is PT = 99.59% o a equi ed ese oi ou low Op = 2.5 m3 s–1. This means ha he cu en s a e is unde es ima ed by app oxi- ma ely 1% compa ed o he calcula ions which we e unde aken o ha ese oi . Unde es ima ion can be explained by he leng h o he inpu s eam low se ies in oduced o he calcula- ions, which we e no upda ed un il 2010. In addi ion, he e we e ew low wa e yea s in he i s hal o he 1990s. Fo example, in he p esen ed analysis, he e is he appa en e ec o unde es ima ion o empo al eliabili y o he alue o he equi ed ou low om he ese oi Op = 2.3 m3 s–1, which is, wi h he amoun o empo al eliabili y, nea es o empo al eliabili y acco ding o he manipula ion ules o he ese oi o he solu ion including conside ing inpu da a unce ain ies, and hus he ese oi is classi ied inapp op ia ely in he signi - ican class (A – PT ≥ 99.5%, B – PT ≥ 98.5%, C – PT ≥ 97.5%, D – PT ≥ 95%) acco ding o ČSN 75 2405. Fo he unce ain y o ±6%, he mean alue o empo al eliabili y was PT = 99.566%, which co esponds o he signi icance o he ese oi A – PT ≥ 99.5%. In conside ing unce ain y, he lowe in e al o empo al eliabili y co esponded o alue PT = 99.411% and hus also he ese oi signi icance would all o a lowe signi - ican class co esponding o class B – PT ≥ 98.5%. F om his poin o iew, he e is space o u u e e iews o he manipula- ion ules o ese oi s and possible amendmen o he ČSN 75 2405 s anda d which should ake inpu da a unce ain ies in o conside a ion. In ha case, i will be necessa y o ake he alue o empo al eliabili y PT as he lowe limi o he esul an in e al and hus o incline mo e o he sa e side in he solu ion. Cu en ly, he esul s canno be gene alized, bu he compu a- ional algo i hm is w i en in gene al e ms and i can be applied also o o he ese oi s. When ca ying ou he sensi i i y anal- ysis, he same alue o unce ain ies was always coun ed o all inpu da a. Unde hese condi ions i was shown ha wa e in low was he mos signi ican sou ce o unce ain ies. How- e e , o he inpu measu emen unce ain ies also ha e an in lu- ence on he esul , which mus be aken in o accoun . A he p esen ime, he au ho s a e no awa e o which alues he ele a ion– olume (–a ea) cu es can acqui e, when hei s a ed ac ual cou se is a ec ed by sedimen a ion o he ese oi and o he e ec s. He e i is possible o assume ha highe unce - ain ies o ele a ion– olume (–a ea) cu es can a ec he e- sul s mo e. F om his poin o iew, he esul s may be di e en and hus also he in e als desc ibing he occu ence o he calcula ed empo al eliabili y PT may be di e en . Finally, i mus be s a ed ha he p esen ed sensi i i y analy- sis was only ca ied ou o one ese oi and he esul s canno be gene alized. I can be assumed ha di e en esul s will be ob ained o o he ese oi s and all ese oi sys ems wi h di e en sizes o ese oi s o age capaci y, a ious sou ces o in lows and lood al i ude. In his espec , he e is space o u he esea ch, o example using so wa e based on he pape by Menšík e al. (2015), as well as coope a ion wi h wa e wo ks adminis a o s. Acknowledgemen . This pape was suppo ed by he B no Uni- e si y o Technology’s p ojec “CZ.1.07/2.3.00/30.0039 Ex- cellen young esea che s a B no Uni e si y o Technology” and he speci ic esea ch p ojec FAST-S-15-2694 “Unce ain y p opaga ion in he hyd ological and wa e managemen applica- ions o mi iga ion o d ough on he open wa e ese oi .” REFERENCES Be en, K., 2007. Towa ds in eg a ed en i onmen al models o e e ywhe e: Unce ain y, da a and modelling as a lea ning p ocess. Hyd ol. Ea h Sys . Sci., 11, 460–467. Be en, K.J., Binley, A.M., 1992. The u u e o dis ibu ed models: Model calib a ion and unce ain y p edic ion. Hyd ological P ocesses, 6, 279–298. Unau hen ica ed Download Da e | 8/10/17 12:51 PM Daniel Ma on, Miloš S a ý, Pa el Menšík 294 Campos, J.N.B., Souza Filho, F.A., Lima, H.V.C., 2014. Risks and unce ain ies in ese oi yield in highly a iable in e - mi en i e s: Case o he Cas anhão Rese oi in semi-a id B azil. Hyd ological Sciences Jou nal, 59, 6, 1184–1195. Coxon, G., F ee , J., Wes e be g, I.K., Wagene , T., Woods, R., Smi h, P.J., 2015. A no el amewo k o discha ge unce - ain y quan i ica ion applied o 500 UK gauging s a ions. Wa e Resou . Res., 51, doi:10.1002/2014WR016532. ČSN 75 2405 S anda d, 2004. Wa e Managemen Analysis o Rese oi , ICS 93.160; 13.060.10, The Czech O ice o S anda ds, Me ology and Tes ing, P ague. GUM, 1993. Guide o he exp ession o unce ain y in meas- u emen . 1s ed. In e na ional O ganiza ion o S anda diza- ion, Gene a, 101 pp. ISBN 9267101889. Hashimo o, T., S edinde , J.R., Loucks, D.P., 1982. Reliabili y, esiliency, and ulne abili y c i e ia o wa e esou ce sys- em pe o mance e alua ion. Wa e Resou ces Resea ch, 18, 1, 14–20. ISO GUIDE 99998 S anda d, 2004. Guide o he exp ession o unce ain y in measu emen (GUM) – Supplemen 1: Nu- me ical me hods o he p opaga ion o dis ibu ions. In e - na ional O ganiza ion o S anda diza ion, Gene a. Klemes, V., 1967. Reliabili y es ima es o a s o age ese oi wi h seasonal inpu . Jou nal o Hyd ology, 7, 2, 198–216. Kli , G.J., 2005. Unce ain y and In o ma ion: Founda ions o Gene alized In o ma ion Theo y. John Wiley & Sons, Inc., Hoboken, NJ, 499 pp. Knigh , F.H., 1921. Risk, Unce ain y, and P o i . Bos on, Ha , Scha ne & Ma x; Hough on Mi lin Company, Bos on. K i skiy, S.N., Menkel, M.F., 1952. Wa e Managemen Com- pu a ions. GIMIZ, Lening ad. (In Russian.) Ku ia, F.W., Vogel, R.M., 2014. A global ese oi wa e supply yield model wi h unce ain y. En i onmen al Resea ch Le e s, 9, 095006, doi: 10.1088/1748-9326/9/9/095006. LaBaugh, J.W., Win e , T.C., 1984. The impac o unce ain ies in hyd ologic measu emen on phospho us budge s and em- pi ical models o wo Colo ado ese oi s. Limnology and Oceanog aphy, 2, doi: 10.4319/lo.1984.29.2.0322. Ma on, D., S a ý, M., Menšík, P., 2011. The in luence o unce ain ies in he calcula ion o mean mon hly discha ges on ese oi s o age, Jou nal o Hyd ology and Hyd omechanics, 59, 4, 228–237. Ma on, D., S a ý, M., Menšík, P., 2014. Wa e managemen solu ion o ese oi s o age unc ion unde condi ion o measu emen unce ain ies in hyd ological inpu da a. P o- cedia Enginee ing, 70, 1094–1101, ISSN 1877-7058, h p://dx.doi.o g/10.1016/j.p oeng.2014.02.121. Menšík, P., S a ý, M., Ma on, D., 2015. Wa e managemen so wa e o con olling he wa e supply unc ion o many ese oi s in a wa e shed. Wa e Resou ces, 42, 1, 133–145. WECC, 1990. Guidelines o Exp ession o he Unce ain y in Calib a ions. Doc. 19-1990. In e na ional O ganiza ion o S anda diza ion, Gene a, Swi ze land. Win e , T.C., 1981. Unce ain ies in es ima ing he wa e bal- ance o lakes. JAWRA Jou nal o he Ame ican Wa e Re- sou ces Associa ion, 17, 82–115. doi: 10.1111/j.1752- 1688.1981. b02593.x. Zadeh, L.A., 1965. Fuzzy se s. In o ma ion and Con ol, 8, 3, 338–353. doi: 10.1016/s0019-9958(65)90241-x. Zah adníček, P., T nka, M., B ázdil, R., Možný, M., Š ěpánek, P., Hla inka, P., Žalud, Z., Malý, A., Seme ádo á, D., Do- b o olný, P., Dub o ský, M., Řezníčko á, L., 2014. The ex- eme d ough episode o Augus 2011–May 2012 in he Czech Republic. In . J. Clima ol., doi: 10.1002/joc.4211. Recei ed 20 Ma ch 2015 Accep ed 2 June 2015 Unau hen ica ed Download Da e | 8/10/17 12:51 PM