1
Assessing he hyd ological esponse om an ensemble o CMIP5 clima e 1
p ojec ions in he ansi ion zone o he A lan ic egion (Bay o Biscay). 2
Mai e Meau io(a), Ane Zabale a(a), Lau ie Boi hias(b,c), Ane Mi en Epelde(a), Sabine 3
Sau age(b), Jose-Miguel Sanchez-Pe ez(b), Ragha an S ini asan(d), Iñaki 4
An igüedad(a). 5
(a) Hyd ogeology and En i onmen G oup, Science and Technology Facul y, Uni e si y6
o he Basque Coun y UPV/EHU, 48940 Leioa, Basque Coun y, Spain 7
(b) EcoLab, Uni e si é de Toulouse, CNRS, INPT, UPS, 31400 Toulouse, F ance8
(c) Géosciences En i onnemen Toulouse, Uni e si é de Toulouse, CNES, CNRS,9
IRD, UPS, 31400 Toulouse, F ance 10
(d) Spa ial Sciences Labo a o y, Texas A&M Uni e si y (TAMU), 77843 College11
S a ion, Texas, USA. 12
*Co esponding au ho in o ma ion:13
Mai e Meau io 14
e-mail add ess: [email p o ec ed]
15
16
17
18
19
20
21
This is he accep ed manusc ip o he a icle ha appea ed in inal o m in Jou nal o Hyd ology 548 : 46-62 (2017), which
has been published in inal o m a h ps://doi.o g/10.1016/j.jhyd ol.2017.02.029. © 2017 Else ie unde CC BY-NC-ND license
(h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/)
2
Abs ac 22
The clima e changes p ojec ed o he 21s cen u y will ha e consequences on 23
he hyd ological esponse o ca chmen s. These changes, and hei consequences, a e 24
mos unce ain in he ansi ion zones. The s udy a ea, in he Bay o Biscay, is loca ed 25
in he ansi ion zone o he Eu opean A lan ic egion whe e hyd ological impac o 26
clima e change has been sca cely s udied. To assess he hyd ological e ec s o 27
clima e change, 16 clima e scena ios including 5 Gene al Ci cula ion Models (GCM) 28
om he 5º epo o he Coupled Model In e compa ison P ojec (CMIP5), 2 s a is ical 29
downscaling me hods and 2 Rep esen a i e Concen a ion Pa hways we e conside ed 30
in a hyd ological model (SWAT). P ojec ions o u u e discha ge (2011-2100) we e 31
di ided in o h ee 30-yea ho izons (2030s, 2060s and 2090s) and a compa ison was 32
made be ween hese ho izons and he baseline (1961-2000). The esul s show ha he 33
downscaling me hod used esul ed in a highe sou ce o unce ain y han GCM i sel . In 34
addi ion, he unce ain ies inhe en o he me hods used a all he le els do no a ec 35
he esul s equally along he yea . In spi e o hose unce ain ies, gene al ends o he 36
2090s p edic seasonal discha ge dec eases by a ound -17% in au umn, -16% in 37
sp ing, -11% in win e and -7% in summe . These esul s a e in line wi h hose 38
p edic ed o F ance and he Ibe ian Peninsula in he A lan ic egion. T ends o 39
ex eme lows we e also analysed: he mos signi ican end shows an inc ease in he 40
du a ion (days) o low lows. F om an en i onmen al poin o iew, and conside ing he 41
need o mee he objec i es es ablished by he Eu opean Wa e F amewo k Di ec i e 42
(WFD), his would be a d awback o he u u e planning on wa e managemen . 43
Keywo ds: CMIP5, hyd ological end, high low and low low, SWAT model, 44
A lan ic egion, ansi ion zone. 45
Highligh s: 46
3
Hyd ological impac s udies sca ci y in ansi ion zone o A lan ic egion add essed 47
Downscaling me hod used esul ed in a highe sou ce o unce ain y han GCM i sel 48
Resul s in line wi h hose p edic ed o A lan ic egion (F ance, Ibe ian Peninsula) 49
Unce ain ies inhe en o me hods used do no a ec esul s equally along he yea 50
Highes dec ease in low lows is a d awback o u u e planning on wa e managemen 51
1. In oduc ion 52
Clima e change will ha e e ec s on hyd ological sys ems, which in u n will 53
impac ecological, social and economic sys ems (Bende e al., 1984; Dibike and 54
Coulibaly, 2005; B auman e al., 2007; Vö ösma y e al., 2010). These e ec s can be 55
s udied bo h a local and egional le els, p o iding impo an in o ma ion o e i o ial 56
and sec o al planning (Lahme e al., 2001). In some a eas whe e wa e sca ci y is no 57
a key aspec o he e i o ial managemen , as i is he case o he Basque Coun y 58
(Bay o Biscay, Can ab ian Sea), ew s udies ha e been ca ied ou o e alua e he 59
possible e ec s o he clima e change on ca chmen hyd ology. 60
The mos commonly used me hod o e alua ing clima e change impac on 61
hyd ological sys ems is o in oduce Gene al Ci cula ion Models (GCMs) in o 62
hyd ological models (Gosling e al., 2011). This p o ides a good ool o s udying he 63
ela ionship be ween clima e and wa e esou ces, conside ing also he e ec o human 64
ac i i ies (Jo hi yangkoon e al., 2001; Lea esley, 1994). Howe e , GCMs usually ha e 65
li le spa ial esolu ion and i hey a e in oduced di ec ly in o hyd ological models, he 66
pe o mance is poo (Fowle e al., 2007). This is he eason o he need o pe o ming 67
a s a is ical o dynamical downscaling o gene al ci cula ion models. 68
Unce ain ies ela ed o he impac o clima e change appea a he ou le els o 69
he sequence: GCMs, Rep esen a i e Concen a ion Pa hway (RCP) o emission 70
4
scena io, downscaling and hyd ologic p ojec ions. Fo example, Chen e al. (2011) 71
assessed he unce ain y o downscaling me hods and he esul s showed ha impac 72
s udies based on only one downscaling me hod should be in e p e ed wi h cau ion. 73
Wilby (2005) ound ha he unce ain y ela ed o hyd ological model calib a ion is 74
compa able wi h ha in ol ed in g eenhouse and o he pollu an emissions. Wilby and 75
Ha is (2006) de e mined ha he g ea es unce ain ies de i e mo e om he choice o 76
GCM and downscaling me hods and less om he hyd ological models and emission 77
scena ios. Some esea ch suppo s he a gumen ha he choice o hyd ological model 78
has a ela i ely mino impac on he esul s o hyd ological simula ions based on 79
clima e p ojec ions (Boye e al., 2010; Ba es e al., 2008; Kay e al., 2006) and majo 80
unce ain ies come om he GCM s uc u e (A nell, 1999; Be gs om e al., 2011; 81
Nijssen e al., 2001; Kay e al., 2009; Chen e al., 2011; A nell e al., 2011; Teng e al., 82
2012). Fo his eason, he use o an ensemble o clima e models gi es a be e 83
es ima e o unce ain y (e.g. IPCC, 2007, 2013; Johnson and Sha ma, 2009; S ahl e 84
al., 2011). The e o e, he inhe en unce ain ies o he me hods used a each o hese 85
le els p opaga e he unce ain ies o he p e ious le els o he sequence, wi h he 86
esul ha all single unce ain ies a e hen p opaga ed in he hyd ological models 87
(Wilby e al., 2006). 88
Eu ope is a ep esen a i e egion o global changes due o clima e wa ming 89
(Sho house and A nell, 1999). The e is a clea con as be ween he no h and he 90
sou h o he con inen , hence, an inc ease in p ecipi a ion and, he e o e, an inc ease in 91
wa e discha ge, has been poin ed ou in he no h (e.g. A nell, 1998; Kiely, 1999; Xu 92
and Halldin, 1997; IPCC, 2007, 2014). By con as , in he sou h he end is e e sed: a 93
dec ease o p ecipi a ion is p edic ed and, consequen ly, a dec ease in discha ge (e.g. 94
Mimikou e al., 2000; Ayala-Ca cedo and Iglesias, 2000; Ribalaygua e al., 2013; 95
Lespinas e al., 2014; Touhami e al., 2015; Val e de e al., 2015; IPCC, 2014). Due o 96
i s loca ion in he Bay o Biscay (Fig. 1), he a id clima ological condi ions p ojec ed o 97
5
sou he n Eu ope do no seem o be ep esen a i e o he Basque Coun y. Howe e , 98
nei he is i clea ha he hyd ological changes p ojec ed in his a ea will ollow he 99
discha ge inc easing end p ojec ed o no he n Eu ope. In his sense, Coch and 100
Medie o (2015) analyzed low lows o iden i y di e en a eas o hyd ologic ends o he 101
Ibe ian Peninsula and Medie o e al. (2015) in es iga ed he lood-p one egions in 102
Eu ope. Bo h s udies eached he same conclusion: The Basque Coun y a ea would 103
be loca ed in he A lan ic egion. This a ea is cha ac e ized by A lan ic on al sys ems 104
coming om he wes , usually om au umn o sp ing, being summe he d y season. 105
Fu he mo e, he 4 h epo o he In e na ional Panel on Clima e Change (IPCC, 2007) 106
p o ides a ulne abili y map o Eu ope o he XXI cen u y, whe e he A lan ic egion 107
encompasses he no he n Ibe ian Peninsula, wes e n F ance, he Ne he lands, 108
Belgium, no he n pa o Ge many, wes e n Denma k and he UK. As a gene al end, 109
he epo p edic s an inc ease in he u u e win e s o ms and looding o his egion. 110
Resea ch wo ks conduc ed since 2000 in he A lan ic egion o e alua e he 111
impac s o clima e change in he wa e esou ces a e summa ized in Table 1. This 112
able was done o iden i y possible u u e hyd ological ends in his egion, hus, 113
esea ch wo ks ha we e published in high impac jou nals and some epo s we e 114
collec ed. In addi ion, o ha e a homogeneous iew, he selec ion was made only wi h 115
wo ks ha p esen ed hei esul s in mean discha ge di e ence (%) wi h espec o hei 116
baseline. When necessa y, mean di e ence alues and anges we e calcula ed ( o 117
example o ob ain seasonal alues om mon hly ones). Despi e some seasonal ends 118
in Table 1 a e di icul o in e p e , gene al ends o mean discha ge e olu ion can be 119
de i ed conside ing hose mo e clea ly obse ed. A signi ican dec ease o discha ge is 120
obse ed in all he s udies o summe and sp ing seasons. This dec ease is e en 121
mo e impo an owa ds he end o he cen u y. In win e ends a e no so clea . In he 122
UK (b, in Table 1) and in he Ibe ian Peninsula (d, Table 1), bo h inc ease and 123
dec ease discha ge can be expec ed depending on he s udy, whe eas in F ance (c,
124
6
Table 1) he obse ed end is dec easing. T ends in sp ing a e simila o hose in 125
win e hough dec ease o discha ge p e ails. Annual ends a e de e mined by ends 126
in win e and sp ing. 127
F om he a o emen ioned esea ch wo ks (Table 1), i can be deduced he 128
gene al idea ha a ansi ion zone exis s be ween no he n and sou he n A lan ic 129
egion, whe e expec ed ends in discha ge (in win e and sp ing) change om 130
inc easing in he no h o dec easing in he sou h. I is no an easy ask o loca e his 131
zone, due o he low spa ial esolu ion o clima e models; acco ding o Habe s e al. 132
(2013) his zone is loca ed in no he n F ance and ollowing IPCC (2007) and 133
Goubano a and Li (2007) i would be in he no h o he Ibe ian Peninsula, ha includes 134
he s udy egion. The unce ain ies in ol ed, p ecisely, in he clima e p edic ions o 135
ha ansi ion zone a e la ge and di icul o iden i y. The e o e, i should be an in-dep h 136
s udied a ea. Howe e , compa ed wi h o he Eu opean egions, he numbe o 137
esea ch wo ks in his zone is low (Table 1); his e idences he s eng h o he cu en 138
wo k. 139
In his con ex , he aim o his s udy is o assess he possible u u e e ec s o 140
clima e change on he hyd ology o a ca chmen loca ed in he A lan ic egion o he 141
Ibe ian Peninsula. Fo his pu pose, i e GCMs o he 5º epo o he Coupled Model 142
In e compa ison P ojec (CMIP5), wo downscaling me hods and wo RCPs we e 143
conside ed, using a o al o 16 clima e p ojec ions (Table 2). 144
The p ojec ed clima e a iables we e in oduced in he Soil and Wa e 145
Assessmen Tool o SWAT model (A nold e al., 1998) o e alua e he hyd ological 146
impac o clima e change, ocusing on he ollowing pa ial objec i es: 147
1) To e alua e he baselines o conside ed downscaled p ojec ions o clima e 148
a iables wi h espec o he obse ed da a (1961-2000). 149
7
2) To s udy he a e age hyd ological impac o u u e clima e p ojec ions in h ee 150
ho izons: 2030s o 2011-2040, 2060s o 2041-2070, 2090s o 2071-2100 151
(annual, seasonal and mon hly). 152
3) To assess possible ends in ex eme daily discha ges (2011-2100) and 153
compa e he obse ed igu es o he e e ence pe iod (1961-2000). 154
4) To iden i y he di e ences be ween clima e p ojec ions and iden i y he g ea es 155
unce ain ies in he di e en s eps o he ollowed me hodology. 156
2. Me hodology 157
2.1. Desc ip ion o he s udy a ea 158
The s udy a ea is he ca chmen o he Uppe Ne bioi Ri e (185 km2), which is 159
loca ed in he cen e-wes o he Basque Coun y (Bay o Biscay), a an a e age 160
la i ude o 43º and longi ude o 3º (Fig. 1). I s di ec ion Sou h-No h, is e y common in 161
ca chmen s o he Basque Coun y, as well as i s geology and land use. The ca chmen 162
is loca ed a he in e ace be ween he A lan ic clima e in he no h and he 163
Medi e anean clima e in he sou h. In addi ion, his is one o he ca chmen s wi h 164
longes eco ds o discha ge se ies in his zone. 165
A e age annual ain all is abou 1,000 mm and i is dis ibu ed qui e e enly 166
h oughou he yea : close o 300 mm in win e and au umn; 230 mm in sp ing and 30 167
mm in summe , as an a e age (1961-2014). The mean annual empe a u e is a ound 168
12 °C, being he seasonal a e ages o win e and summe 8 ºC and 20 ºC (1961-169
2014), espec i ely. 170
The mean ele a ion o he ca chmen is a ound 200 m abo e sea le el (m.a.s.l.) 171
(Fig. 1). The li hology is domina ed by sil s ones, clays and sands ones wi h medium-172
low pe meabili y (Geog aphical Da abase o he Basque Go e nmen , 173
www.geoeuskadi.ne ). In he sou heas pa , a an a e age al i ude o 1,100 m.a.s.l., 174
8
he e is a highly pe meable la e C e aceous limes one pla o m. The main soil ypes a e 175
Cambisols, Ranke s and Gleysols (FAO, 1977), which a e cha ac e ized by high clay 176
and sil con en s. Land use in he ca chmen is di ided in o na i e o es s, exo ic 177
plan a ions and pas u elands. The a eas wi h he highes slopes (>35%) a e co e ed 178
by o es and ee plan a ions, while he la e a eas (7-15%) hos pas u elands. The 179
a e age slope in he ca chmen is a ound 17%. I should be no ed ha possible u u e 180
changes in land use ha e no been aken in o conside a ion in his wo k. 181
Mean annual discha ge a he ou le o he ca chmen is 3 m3 s-1. The mean in 182
sp ing and au umn is a ound 3.5 m3 s-1; in win e 5 m3 s-1 and in summe 0.8 m3 s-1 183
(1996-2013). Discha ge da a om a gauging s a ion (Ga dea; h p://www.bizkaia.eus) 184
loca ed a he ou le o he ca chmen (Fig. 1) we e used in his wo k o calib a e and 185
alida e he hyd ological model. Discha ge da a ha e been eco ded a his gauging 186
s a ion since 1995. This s a ion was designed o p ecisely measu e mean and high 187
lows. 188
2.2. Desc ip ion o he hyd ological model: SWAT 189
The SWAT model is a basin-scale con inuous in ime and semi-dis ibu ed 190
model ope a ing on a daily ime s ep. I was de eloped o e alua e he impac o 191
managemen p ac ices on wa e , sedimen and ag icul u al chemical yields in 192
ungauged basins (A nold e al., 1998). I can be used in a b oad ange o condi ions 193
and i has al eady been widely used o s udy he impac s o en i onmen al and clima e 194
change (e.g. Bou aoui e al., 2002; Li e al., 2009; Abbaspou e al., 2009; Bekele and 195
Knapp, 2010; Zhang e al., 2014). 196
SWAT di ides he ca chmen in o sub-basins which a e subdi ided in o 197
Hyd ological Response Uni s (HRUs) wi h homogeneous land use, soil cha ac e is ics 198
and slope g adien . Two me hods a e used o simula e su ace uno in he SWAT 199
9
model: he modi ied SCS cu e numbe (USDA Soil Conse a ion Se ice, 1972) and 200
he G een-Amp in il a ion me hod (G een and Amp , 1911), which equi es a sub-daily 201
p ecipi a ion ime s ep. In his case he obse ed me eo ological da a used o calib a e 202
and alida e he model we e daily and he e o e he SCS me hod was used. The model 203
calcula es he peak uno a e wi h a modi ied a ional me hod (Chow e al., 1988). The 204
la e al subsu ace low in he soil p o ile is de e mined o each soil laye , using he 205
kinema ic s o age ou ing model (Sloan and Moo e, 1984), which is calcula ed 206
simul aneously wi h pe cola ion. The g oundwa e low con ibu ion o he o al 207
s eam low is simula ed by c ea ing shallow aqui e s o age (A nold and Allen, 1996) 208
whe e pe cola ion om he bo om o he oo zone is conside ed as echa ge o he 209
shallow aqui e . The po en ial e apo anspi a ion can be es ima ed using he 210
Ha g ea es (Ha g ea es and Samani, 1985), P ies ley-Taylo (P ies ley and Taylo , 211
1972) and Penman-Mon ei h (Mon ei h, 1965) me hods. This s udy uses Ha g ea es, 212
which only equi es p ecipi a ion and maximum and minimum empe a u e, since hese 213
we e he only me eo ological da a a ailable. The low is ou ed h ough he channel 214
using ei he he a iable s o age coe icien me hod (Williams, 1969) o he Muskingum 215
ou ing me hod (O e on, 1966). In his s udy he o me me hod was used because i 216
be e sui ed he obse ed discha ge. 217
2.3. Hyd ological model inpu and da a sou ce. 218
SWAT equi es opog aphic, land use/co e , soil and me eo ological da a. The 219
sou ce o he Digi al Ele a ion Model (LIDAR 2008, 5x5m), land use classi ica ion 220
(2005, 1: 10,000) and pa o he soil map (1: 25,000) is he Basque Go e nmen ’s 221
Geog aphical Da abase (www.geoeuskadi.ne ). The emainde o he soil map was 222
ob ained om he soil map o A aba p o ince (1:200,000) (Iñiguez e al., 1980). Soil 223
p ope ies we e ob ained om hese wo sou ces and he plan g ow h p ope ies o 224
each land co e we e di ec ly ob ained om he SWAT da abase. 225
16
he es unde es ima es he p obabili y o de ec ing ends. This se ial co ela ion may 382
he e o e in luence he esul s o he es (Douglas e al., 2000). To a oid hese 383
possible e ec s, be o e using he Mann-Kendal es he end ee p e-whi ening 384
app oach de eloped by Yue e al. (2002) was applied o he se ial da a. Wi h he Mann-385
Kendall es i is possible o iden i y inc easing and dec easing ends and he 386
p obabili y o occu ence (P) o hose ends. The alue o P can a y be ween 0 and 1, 387
whe e 0 indica es ha he e is no p obabili y o occu ence in he end and 1 indica es 388
maximum p obabili y. The c i e ia sugges ed by he IPCC (Mas and ea e al., 2010) in 389
i s 5 h epo we e used o e alua e P. In his documen , likelihood e e s o a 390
p obabilis ic assessmen o some well-de ined pas o u u e ou comes. The ca ego ies 391
de ined and used in his esea ch a e: P>0.99, i ually ce ain end; P>0.95, ex emely 392
p obable end; P>0.90, e y p obable end and P>0.66, p obable ends. Values o P 393
below 0.66 a e conside ed o ep esen non-p obable ends in his wo k. 394
3. Resul s and discussion 395
3.1. SWAT calib a ion (1996-2006), alida ion (2007-2013) and simula ion 396
unce ain y 397
The pa ame e s changed in he calib a ion p ocess o SWAT o he Uppe 398
Ne bioi, hei alue- ange used in he au ocalib a ion and he inal alues a e shown in 399
Table 3. These a e some o he mos common pa ame e changes usually pe o med in 400
SWAT o calib a e he model (A nold e al., 2012) and all o hem we e changed aking 401
in o accoun he ca chmen cha ac e is ics. The esul s o he calib a ion (1996-2006) 402
and he alida ion (2007-2013) a e displayed in a daily hyd og aph wi h he obse ed 403
and simula ed discha ge (Fig. 2). The calib a ion can be seen o i he obse ed da a 404
well, al hough he peak magni ude is unde es ima ed in some high lows. In p e ious 405
wo ks ca ied ou wi h he SWAT model (daily ime s ep) in he Basque Coun y, he 406
unde es ima ion o he peak magni ude is usual (Zabale a e al., 2014; Pe aza e al., 407
17
2015; Epelde e al., 2015; Meau io e al., 2015). This inaccu acy may be ela ed o he 408
inabili y o he model o p ope ly conside p ecipi a ion in ensi y and spa ial- empo al 409
dis ibu ion when simula ing apid hyd ological esponses a he daily ime s ep (Qiu e 410
al., 2012). 411
In gene al, simula ed discha ge peaks i obse ed da a be e du ing he 412
alida ion (Fig. 2). The se o s a is ical indices calcula ed o daily discha ge (Table 4) 413
shows ha he model pe o ms sa is ac o ily du ing bo h, calib a ion and alida ion 414
(Mo iasi e al., 2007). 415
Analyzing yea s wi h a low and high a idi y index (AI), i is possible o assess 416
whe he he simula ion is good enough o make long- e m hyd ologic p ojec ions and 417
e alua e when he g ea es unce ain ies a e ound (low o high AI). The se o 418
s a is ical indices (Table 4) shows ha simula ion pe o mance o yea s wi h high and 419
low AI is a leas “good”, al hough some pa ame e s a e sligh ly poo e o yea s wi h 420
high AI (mo e unce ain y). In addi ion, he se o s a is ical indices we e also applied o 421
he en i e pe iod (1996-2013) on a seasonal scale. I is hus possible o e alua e he 422
model pe o mance conside ing low (summe ), in e media e (sp ing and au umn) and 423
high (win e ) lows and de e mining whe e he la ges unce ain ies a e. Win e and 424
sp ing p esen "good" s a is ical esul s, au umn is "a leas sa is ac o y" and he 425
s a is ical indices show ha al hough he hyd og aph seems no o i p ope ly in 426
summe (low 2, NSE and RSR), he wa e yield is simula ed co ec ly (low PBIAS). 427
No e ha summe discha ges, being he lowes , a e mo e ulne able o measu emen 428
e o s. The e o e, al hough he simula ion does pe o m well, summe is he season 429
associa ed o he highes modeling unce ain y. Howe e , acco ding o Mo iasi e al. 430
(2007) he alues o mos o he s a is ical indices shown in Table 4 we e "good" o 431
" e y good" a mon hly ime s ep. Since hese analyses we e made wi h daily alues 432
hey a e conside ed o be a leas "good". Addi ionally, he p- ac o and - ac o 433
18
ob ained wi h he SWAT-CUP p og am ( he ange o he pa ame e s is shown in Table 434
3) o calib a ion and alida ion a e 0.81 and 0.41 espec i ely, which is conside ed 435
"good" (Abbaspou e al., 2015). As a consequence, i can be said ha he 436
pe o mance o he model is good enough o ca ying ou u u e hyd ological 437
p ojec ions wi h a ce ain deg ee o con idence. 438
3.2. Assessmen o he baseline hyd ological p ojec ions 439
The bias-co ec ed p ecipi a ion and maximum and minimum empe a u e o 440
he baseline o each downscaled GCM was in oduced in he calib a ed and alida ed 441
SWAT p ojec . The i s s ep was o assess how he hyd ological simula ions ob ained 442
o baseline (1961-2000) adjus o he ones pe o med using obse ed me eo ological 443
da a (OBS_SIM) o he same pe iod. The mean mon hly discha ges (m3 s-1) ob ained 444
a e shown in Fig. 4. This igu e shows ha as in he case o p ecipi a ion, hyd ological 445
simula ions ob ained using he baseline clima e p ojec ions downscaled wi h he SDSM 446
me hod i much be e o OBS-SIM han hose downscaled wi h he AN me hod. In ac , 447
he adjus men o discha ge se ies ob ained using SDSM downscaling o OBS_SIM is 448
eally good in au umn (-9%) and win e (-2%) whe eas in sp ing and summe he 449
discha ge is unde es ima ed by abou -22% and -71%, espec i ely. Howe e , hose 450
di e ences a e highe in all seasons o he discha ge se ies ob ained using he AN 451
me hod; a ound -22% in au umn, -11% in win e , -46% in sp ing and -83% in summe . 452
The e o e, i is clea ha , in his case, he choice o he downscaling me hod is 453
he cause o a highe unce ain y sou ce in he ob ained discha ge se ies han he 454
choice o he GCM i sel . 455
3.3. Hyd ological impac o u u e clima e scena ios: annual and seasonal scales 456
(2011-2100) 457
19
To e alua e he impac o clima e change on he hyd ology o he Uppe Ne bioi 458
ca chmen a ea, u u e hyd ological p ojec ions di ided in o h ee ime ho izons (2030s, 459
2060s, and 2090s) we e compa ed wi h hei baseline a e age discha ge (1961-2000). 460
The di e ence in a e age discha ge (in %) is shown in Fig. 5. Focusing on he 461
downscaling me hod (AN o SDSM), he hyd ological p ojec ions de i ed om clima e 462
p ojec ions ha use he AN me hod always show a smalle discha ge dec ease han 463
hose downscaled wi h SDSM (wi h he excep ion o CMCC_CESM_AN_R85). 464
Conside ing ha clima e p ojec ions ob ained wi h he SDSM downscaling me hod i 465
be e o OBS_SIM se ies da a o he baseline pe iod, i seems mo e easonable o 466
conside hose hyd ological p ojec ions de i ed om SDSM downscaled se ies. 467
I is also impo an o compa e he wo di e en RCPs because, one would 468
expec ha he di e ence be ween he 2060s and 2090s o he p ojec ions wi h RCP 469
4.5 would be minimal, while o 8.5 he di e ence would con inue o inc ease. Indeed, i 470
he p ojec ions o CMCC-CESM a e no conside ed, he di e ence in discha ge a 471
annual scale be ween he baseline and he p ojec ions wi h RCP 4.5 is -6% o he 472
2030s, -8% o he 2060s and -9% o he 2090s, while o RCP 8.5 i is -13% o he 473
2030s; -15% o he 2060s and -20% o he 2090s. In he RCP 4.5 scena io, he 474
seasonal dec ease o discha ge h oughou he cen u y is lowe . In some seasons, as 475
in summe , a s abiliza ion o he discha ge can be obse ed, and in o he s (e.g. 476
au umn) he inc ease in he a e age low in he 2090s almos compensa es o he 477
dec eases obse ed du ing he 2060s. This is no he case o he RCP 8.5 scena io 478
whe e discha ge con inues o dec ease un il he end o he cen u y. 479
Undeniably CMCC_CESM_AN_R85 is mos no ewo hy because i p ojec s 480
highe discha ge han he baseline. CMCC_CESM_SDSM_R85 dec eases espec o 481
he baseline, bu as i happens wi h CMCC_CESM_AN_R85 he discha ge inc eases 482
h oughou he cen u y. The esul s o his GCM we e analysed sepa a ely due o 483
20
hose di e en ends shown. Fig. 6 shows he di e ence (%) be ween CMCC-484
CESM_AN_R85 and CMCC-CESM_SDSM_R85 u u e discha ges wi h ega d o hei 485
baseline on a seasonal and annual scale. In o de o compa e no only he ends bu 486
also he discha ge, in e ms o illus a i e a e age low o each pe iod, he a e age 487
discha ge o each ime ho izon conside ed is also displayed in Fig. 6. These a e he 488
wo p ojec ions ha simula e highes discha ge a annual scale as well as a seasonal 489
scale. 490
As discussed abo e, he downscaling me hod and he selec ion o RCP ha e a s ong 491
in luence on he esul s. The p ojec ions we e he e o e classi ied aking in o 492
conside a ion he downscaling me hod (AN o SDSM) and he scena ios (RCP 4.5 o 493
RCP 8.5) (Fig. 7), analysed in ou di e en g oups (wi h he excep ion o he CMCC-494
CESM p ojec ions). Thus, he esul s a e an ensemble o p ojec ions bu i is possible 495
o analyse di e ences be ween hese ac o s (downscaling me hod and RCP). A 496
annual scale, and o he end o he cen u y (2090s), discha ge dec eases by -9% and -497
20% o RCP 4.5 and RCP 8.5 scena ios, espec i ely, wi h li le di e ences be ween 498
downscaling me hods. These esul s a e consis en wi h mos o he s udies ca ied ou 499
in he A lan ic egion o F ance and in he No h o he Ibe ian Peninsula (Table 1). 500
Howe e , ocusing on he seasonal changes, signi ican di e ences can be ound 501
depending on he use o he downscaling me hod. Summe is he season when he 502
g ea es di e ences can be obse ed; sligh ly inc easing (<5% o 2090s) o clima e 503
p ojec ions de i ed om AN downscaling, and clea ly dec easing (-15 o -25% o 504
2090s) o SDSM-de i ed ones. The AN downscaling me hod simula ed conside ably 505
less discha ge han he SDSM downscaling me hod. Hence, he di e ence dec eases 506
be ween baseline and u u e p ojec ions a e bigge o he SDSM me hod, al hough he 507
p ojec ions downscaled wi h SDSM (independen o RCP) always p ojec ed mo e 508
discha ge han hose downscaled wi h he AN me hod (Fig. 7). In o he seasons 509
(au umn, win e and sp ing), discha ge dec eased ega dless o he me hod chosen, 510
21
wi h highe changes when using he AN me hod in au umn and smalle changes in 511
sp ing. The esul s ob ained using di e en downscaling me hods a e mos simila in 512
win e . This is he season ha has mos weigh in he annual discha ge, hence i s e ec 513
can be obse ed a annual scale. Conside ing all he models, au umn is he season 514
wi h he mos signi ican discha ge dec ease (-17%) ollowed by sp ing (-16%), win e (-515
11%), and summe (-7%) o 2090s. These esul s a e sligh ly di e en om p e ious 516
s udies unde aken in he A lan ic egion o F ance and he no h o he Ibe ian 517
Peninsula (Table 1). In mos o he esea ch wo ks ca ied ou in hese a eas he 518
highes discha ge dec eases occu in summe , and depending on he s udy a e 519
ollowed by au umn o sp ing (Table 1). In he p esen s udy, conside ing he a e age 520
alue, summe is no he mos a ec ed season in pe cen age. This could be explained 521
by he in luence o he p ojec ions downscaled wi h he AN me hod (Fig. 7). 522
Fig. 8 is an ensemble showing a combina ion o he 16 hyd ological p ojec ions 523
analysed (a e age o he mean mon hly discha ge (m3 s-1) ep esen ed in a 524
hyd ological yea ) and hei e olu ion o e ime measu ed a he 3 ime ho izons 525
(2030s, 2060s and 2090s). In o de o conside all he discha ge p edic ions ob ained, 526
he p ojec ions we e no di ided based on he downscaling me hod o he RCP. The 527
highes discha ge alues ep esen he maximum alue o he mean mon hly discha ge 528
o all o he p ojec ions, while he lowes alues ep esen he minimum ones. The 529
possible discha ge ange is he highes in win e and au umn. In hese seasons he 530
possible mean discha ge may a y by 4 m3 s-1. In sp ing he discha ge may ange 531
be ween 1 and 2.1 m3 s-1, while he ange in summe may be he lowes : be ween 0.1 532
and 0.3 m3 s-1. In sp ing, summe and he beginning o au umn, he p ojec ed discha ge 533
is always lowe han he OBS_SIM. Howe e , he esul s o sp ing and summe ha e 534
o be conside ed wi h special ca e because, as discussed p e iously, he baseline o 535
he hyd ological p ojec ions a e unde es ima ed and i is he e o e p obable ha a
536
simila phenomenon happens in he case o u u e p ojec ions. Wi h ega d o he 537
22
e olu ion o he discha ge o e he cen u y, he p ojec ed lowes discha ge dec eases 538
om he 2030s o he 2090s. The p ojec ed highes discha ges show he same end in 539
sp ing and summe , whe eas hey may e en inc ease in au umn and win e . 540
3.4. E alua ion o ends in du a ion and se e i y o ex eme lows 541
The esul s ob ained om end analysis ca ied ou o he du a ion o ex eme 542
lows a e shown in Table 5 and Fig. 9. This analysis has been made o he e e ence 543
pe iod 1961-2000 (Table 5) and he u u e pe iods 2011-2040, 2011-2070 and 2011-544
2100 (Fig. 9). Howe e , he mos signi ican ends appea in he longes pe iod (2011-545
2100) and hence, hese a e he esul s conside ed in his s udy. 546
Analysing he du a ion (in days) o low lows (<Q20) om 1961 o 2000 (Table 547
5), he discha ge simula ed o he e e ence pe iod (OBS_SIM) does no show any 548
signi ican annual end. A seasonal scale, a signi ican end is only de ec ed in sp ing 549
when he low low du a ion shows a "p obable" upwa d end. Howe e , some o he 550
discha ge se ies simula ed using he nine clima e baselines, show signi ican ends; an 551
inc ease a annual scale and, depending on he GCM conside ed, an inc ease o 552
dec ease in sp ing, summe and au umn (Table 5). 553
In he e alua ion o u u e low low du a ion (2011-2100; Fig. 9), a gene al 554
inc easing end can be obse ed, al hough, he e a e a ew dec easing ends. In 555
sp ing and au umn upwa d ends p edomina e. In sp ing he numbe o p ojec ions 556
wi h signi ican inc easing ends is highe unde RCP 8.5 han in RCP 4.5. On he 557
con a y, in au umn his numbe is lowe . Summe p esen andom signi ican ends 558
mos ly unde RCP 8.5, ha in gene al end o be posi i e. The e a e ew p ojec ions 559
wi h signi ican ends in win e , he e o e, is no possible o ob ain clea conclusions o 560
his season. 561
23
OBS_SIM displays dec easing annual and seasonal ends o high low 562
du a ion (abo e Q80). Annually, he dec easing end is "ex emely p obable", on 563
summe is “ e y p obable” and in au umn, win e and sp ing is “p obable” (Table 5). 564
The high low du a ions ob ained using clima e baselines, in gene al do no show 565
signi ican ends. Howe e , he mos signi ican ends a e o he model BNU-ESM 566
showing “ e y p obable” o “ i ually ce ain” dec easing ends annually and o 567
au umn. 568
The high low du a ion (Q80) o u u e p ojec ions (2011-2100) show signi ican 569
ends annually and in au umn, howe e , he e a e as many inc easing as dec easing 570
ends (Fig. 9). In sp ing he e a e ew p ojec ions wi h signi ican ends being mos o 571
hem inc easing. In win e unde RCP 8.5 a gene al dec easing end p edomina es. In 572
summe a change in gene al end can be obse ed o m RCP 4.5 o RCP 8.5: unde 573
RCP 4.5 dec easing ends p e ail, while unde RCP 8.5 a e mos ly inc easing. 574
Zabale a e al. (2012) iden i ied hyd ological signs in he ca chmen s o he 575
Basque Coun y. They used obse ed daily discha ge alues in di e en pe iods and 576
ca chmen s ( egional con ex ). The longes analysed pe iod in hei esea ch is 34 577
yea s (1973-2007). Al hough his pe iod does no coincide in ime wi h he e e ence 578
pe iod used in his wo k (1961-2000), simila i y in he inc easing ends o du a ion o 579
low lows (Q20) can be obse ed. The au ho s a ibu ed hese ends o hyd ological 580
signs o clima e change. 581
Unde s anding he lowe pa o he p ojec ed hyd og aphs is essen ial o an 582
assessmen o impac on eshwa e ecosys ems. The e o e, besides knowing he end 583
o he numbe o days wi h high (abo e Q80) and low (below Q20) lows, i is also 584
impo an o know he olume ic de ici (se e i y) end, especially o low lows. The 585
OBS_SIM (1961-2000) low low de ici does no display any signi ican end. Mos o 586
24
he baselines do no show signi ican ends, ne e heless, he ew o hem whe e end 587
is de ec ed, p edic an inc ease in se e i y. 588
Fo 2011-2100 he e a e signi ican ends in mos o he p ojec ed simula ions 589
o se e i y. Howe e , hese ends a e opposi e om each o he and canno be ela ed 590
o he use o gi en GCMs, downscaling me hods o RCP scena ios. As a consequence, 591
se e i y showed e y high unce ain y in u u e hyd ologic p ojec ion in he Uppe 592
Ne bioi ca chmen . 593
4. Conclusions 594
In his s udy, o assess u u e clima e change e ec s (up o yea 2100) on he 595
hyd ological esponse o he Uppe Ne bioi ca chmen ,16 clima e p ojec ions combining 596
i e GCMs (ACCESS1-0, BNU-ESM, MPI-ESM-RL, MPI-ESM-MR, CMCC-CESM), wo 597
downscaling me hods (AEMET analogues -AN- and S a is ical Downscaling Me hod -598
SDSM-) and wo Rep esen a i e Concen a ion Pa hways (RCP 4.5 and RCP 8.5) we e 599
conside ed. Hyd ological simula ion was pe o med using he SWAT model achie ing 600
sa is ac o y esul s o he calib a ion and alida ion pe iods (1996-2013). 601
Di e en sou ces o unce ain ies a e in ol ed in he hyd ologic p ojec ions (GCM, 602
downscaling me hod, RCP, hyd ological model). Some conclusions can be d awn om 603
he ob ained esul s e en i he quan i ica ion o he unce ain ies lies ou side he scope 604
o his s udy. The conside able di e ence be ween he baselines o he clima e models 605
(1961-2000) and he obse ed me eo ological da a ( he models gene ally simula e less 606
ain all, especially in sp ing and summe ) e idences he unce ain y in ol ed o he 607
s udied a ea in he esul s o he GCMs. 608
Howe e , downscaling me hod used esul ed in a highe sou ce o unce ain y han 609
GCM i sel . When simula ed discha ges o he baselines o all clima e p ojec ions we e 610
compa ed wi h he discha ge ob ained om a simula ion made wi h he obse ed 611
25
me eo ological da a (OBS_SIM; 1961-2000), he compa ison shows ha he GCMs 612
downscaled wi h he SDSM me hod achie e a much be e adjus men han hose 613
downscaled wi h he AEMET analogues (AN). Ne e heless, hey all unde es ima e he 614
discha ge amoun . Those unce ain ies inhe en o he me hods used a all he le els 615
do no a ec he esul s equally along he yea . The seasons wi h mos a iable esul s, 616
and so, he ones o which i is he mos di icul o d aw a clea conclusion, a e sp ing 617
and especially summe , o which u u e discha ges could ei he inc ease o dec ease 618
depending on he downscaling me hod. 619
F om esul s ob ained om ou o he analysed GCMs, ACCESS1-0, BNU-ESM, 620
MPI-ESM-RL and MPI-ESM-MR, i can be said ha discha ge would dec ease wi h 621
espec o he baseline a annual scale. This conclusion is consis en wi h he ends 622
ob ained in he A lan ic egion, mos ly in F ance and he Ibe ian Peninsula (c, d; Table 623
1). The spa ially mos homogeneous esul om Table 1 is he dec ease in summe 624
p ojec ions o he en i e A lan ic egion. Howe e , summe is he season ha mos 625
disc epancies show in he s udy a ea; he p ojec ions downscaled wi h he AN me hod 626
p ojec ed a ound 5% mo e discha ge o he 2090s han o he baseline whe eas 627
SDSM p ojec s -15 o -25% less discha ge (Fig 7). In ac , he seasons ha p edic he 628
la ges dec ease in discha ge in he s udy a ea a e au umn and sp ing (a ound -16 % 629
o 2090s). These downwa ds ends a e also de ec ed in he A lan ic egion o F ance 630
and he Ibe ian Peninsula, al hough, hey a e no so s ong. The lowes dec ease as i 631
happens in o he zones o he A lan ic egion, is p ojec ed o win e . 632
Fo he ensemble o he 16 hyd ological p ojec ions analysed in he h ee ho izons, 633
he wides ange be ween he mon hly highes and lowes discha ge alues would 634
occu in win e and au umn (a ound 3-5 m3 s-1) ollowed by sp ing (be ween 0.1-2 m3 s-
635
1) while he na owes one would be in summe (be ween 0.1-0.3 m3 s-1). Fo sp ing, 636
summe and he beginning o au umn, simula ed discha ge is always below he 637
32
Chow, V.T., Maidmen , D.R. and Mays, L.W., 1988.Applied Hyd ology. McG aw-Hill 773
Inc, New Yo k, USA. 774
Cloke, H. L., Je e s, C., We e hall, F., By ne, T., Lowe, J., and Pappenbe ge , F., 775
2010. Clima e impac s on i e low: P ojec ions o he medway ca chmen , UK, wi h 776
UKCP09 and CATCHMOD. Hyd ol. P ocess. 24 (24), 3476-3489. 777
h p://dx.doi.o g.10.1002/hyp.7769. 778
Coch, A. and Medie o, L., 2015. T ends in low lows in Spain in he pe iod 1949–2009. 779
Hyd olog. Sci. J. 1-17. h p://dx.doi.o g//10.1080/02626667.2015.1081202. 780
Da Cunha, L.V., De Oli ei a, R.P., Nascimen o, J., and Ribei o, L., 2007. Impac s o 781
clima e change on wa e esou ces: A case-s udy o Po ugal. Wa e in Cel ic 782
Coun ies: Quan i y, Quali y and Clima e Va iabili y, 310, 37-48. 783
Diaz-Nie o, J. and Wilby, R.L., 2005. A compa ison o s a is ical downscaling and 784
clima e change ac o me hods: Impac s on low lows in he i e Thames, Uni ed 785
Kingdom. Clima ic Change. 69 (2), 245-268. 786
Dibike, Y.B. and Coulibaly, P., 2005. Hyd ologic impac o clima e change in he 787
Saguenay wa e shed: compa ison o downscaling me hods and hyd ologic models. J. 788
Hyd ik, 307 (1–4), 145-163. h p://doi.o g/10.1016/j.jhyd ol.2004.10.012. 789
Douglas, E.M., Voguel, R.M. and K oll, C.N., 2000. T ends in loods and low lows in 790
he Uni ed S a es: Impac o spa ial co ela ion. J. Hyd ol. 240, 90–105. 791
h p://dx.doi.o g/10.1016/S0022-1694(00)00336-X. 792
Ducha ne, A., Habe s, F., Page, C., Sauque , E., Vienno , P., Deque, M., Gascoin, S., 793
Hachou , A., Ma in, E., Oudin, L., Te ay, L. and Thie y, D., 2010. Clima e change 794
33
impac s on wa e esou ces and hyd ological ex emes in no he n F ance. 795
P oceedings o he XVIII In e na ional Con e ence on Compu a ional Me hods in Wa e 796
Resou ces (Cmw 2010), 243-250. 797
Epelde, A.M., Ce o, I., Sanchez-Pe ez, J.M., Sau age, S., S ini asan, R., and 798
An igueedad, I., 2015. Applica ion o he SWAT model o assess he impac o changes 799
in ag icul u al managemen p ac ices on wa e quali y. Hyd ol. Sci. J. 60 (5), 825-843. 800
h p://dx.doi.o g/10.1080/02626667.2014.967692. 801
FAO, 1977. Guidelines o Soil P o ile Desc ip ion. Rome, I aly. 802
Fowle , H.J. and Kilsby, C.G., 2007. Using egional clima e model da a o simula e 803
his o ical and u u e i e lows in no hwes England. Clima ic Change. 80 (3-4), 337-804
367. h p://dx.doi.o g/10.1007/s10584-006-9117-3. 805
Fowle , H.J., Blenkinsop, S. and Tebaldi, C., 2007. Linking clima e change modelling o 806
impac s s udies: ecen ad ances in downscaling echniques o hyd ological modelling. 807
In . J. Clima ol. 27 (12), 1547–1578. h p://dx.doi.o g/10.1002/joc.1556. 808
Geog aphical da a base o he Basque Go e nmen , GEOEUSKADI; 809
h p://www.geo.euskadi.ne 810
Gio gi, F. and Mea ns, L.O., 2002. Calcula ion o a e age, unce ain y ange, and 811
eliabili y o egional clima e changes om AOGCM simula ions ia he ‘‘ eliabili y 812
ensemble a e aging’’ (REA) me hod. J. Clima e. 15, 1141–1158. 813
h p://dx.doi.o g/10.1175/1520-0442(2003)016%3C0883:COCOAU%3E2.0.CO;2. 814
Gö gen, K., Bee sma, J., B ahme , G., Bui e eld, H., Ca ambia, M., de Keize , O., 815
K ahe, P., Nilson, E., Lamme sen, R., Pe in, C. and Volken, D., 2010. Assessmen o 816
34
Clima e Change Impac s on Discha ge in he Rhine Ri e Basin: Resul s o he 817
RheinBlick2050 P ojec , CHR Repo , 1–23, 229 pp, Lelys ad, ISBN 978-90-70980-35-818
1. 819
Gosling, S.N., Taylo , R.G., A nell, N.W. and Todd, M.C., 2011. A compa a i e analysis 820
o p ojec ed impac s o clima e change on i e uno om global and ca chmen -scale 821
hyd ological models. Hyd ol. Ea h. Sys . Sc. 15, 279-294. 822
h p://dx.doi.o g/10.5194/hess-15-279-2011. 823
Goubano a, K. and Li, L., 2007. Ex emes in empe a u e and p ecipi a ion a ound he 824
Medi e anean basin in an ensemble o u u e clima e scena io simula ions. Glob. 825
Plane . Chang. 57, 27–42 826
G aham, L.P., And éasson, J. and Ca lsson, B., 2007. Assessing clima e change 827
impac s on hyd ology om an ensemble o egional clima e models, model scales and 828
linking me hods – a case s udy on he Lule Ri e Basin. Clima ic Change. 81 (1), 293-829
307. h p://dx.doi.o g/10.1007/s10584-006-9215-2. 830
G een, W.H. and Amp , G.A., 1911. S udies on soil physics. The Jou nal o Ag icul u al 831
Science. 4 (1), 1-24. h p://dx.doi.o g/10.1017/S0021859600001441. 832
Gup a, H.V., So ooshian, S. and Yapo, P.O., 1999. S a us o au oma ic calib a ion o 833
hyd ologic models: Compa ison wi h mul ile el expe calib a ion. J. Hyd ol.Eng. 4 (2), 834
135-143. 835
Habe s, F., Boe, J., Deque, M., Ducha ne, A., Gascoin, S., Hachou , A., Ma in, E., 836
Page, C., Sauque , E., Te ay, L., Thie y, D., Oudin, L. and Vienno , P., 2013. Impac o 837
clima e change on he hyd ogeology o wo basins in no he n F ance. Clima ic 838
Change. 121 (4), 771-785. h p://dx.doi.o g/10.1007/s10584-013-0934-x. 839
35
Ha g ea es, G. and Samani, Z.A., 1985. Re e ence c op e apo anspi a ion om 840
empe a u e. Appl. Eng. Ag ic. 1 (2), 96-99. h p://dx.doi.o g/10.13031/2013.26773. 841
Hewi son, B.C. and C ane, R.G., 1996. Clima e downscaling: echniques and 842
applica ion. Clim. Res. 7, 85-95. 843
Hiscock, K., Spa kes, R., and Hodgson, A., 2011. E alua ion o u u e clima e change 844
impac s on Eu opean g oundwa e esou ces. Clima e Change E ec s on G oundwa e 845
Resou ces: A Global Syn hesis o Findings and Recommenda ions. 27, 351-365. 846
Hisdal, H., S ahl, K., Tallaksen, L.M. and Demu h, S., 2001. Ha e s eam low d ough s 847
in Eu ope become mo e se e e o equen ? In . J. Clima ol. 21, 317–333. 848
h p://dxdoi.o g/10.1002/joc.619. 849
Iñiguez, J., Sánchez‐Ca pin e o, I., Val, R.M., Romeo, A. and Basconesm, J.C., 1980. 850
Mapa de suelos de Ala a. Vi o ia‐Gas eiz: Dipu ación Fo al de Ala a‐Depa amen o de 851
Eda ología de la Uni e sidad de Na a a. 852
IPCC, 2007. Clima e Change 2007: syn hesis epo . Con ibu ion o wo king g oups I, 853
II and III o he Fou h Assessmen Repo o he In e go e nmen al Panel on Clima e 854
Change. Co e W i ing Team, R.K. Pachau i, and A. Reisinge , eds. Gene a: IPCC 855
Sec e a ia . 856
IPCC, 2013. Clima e Change 2013: The Physical Science Basis. Con ibu ion o 857
Wo king G oup I o he Fi h Assessmen Repo o he In e go e nmen al Panel on 858
Clima e Change. [S ocke , T.F., D. Qin, G.-K. Pla ne , M. Tigno , S.K. Allen, J. 859
Boschung, A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.)]. Camb idge Uni e si y 860
P ess, Camb idge, Uni ed Kingdom and New Yo k, NY, USA, 1535pJohnson, F. and 861
Sha ma, A., 2009. Measu emen o GCM skill in p edic ing a iables ele an o 862
36
hyd oclima ological assessmen s. J. Clima e. 22 (16), 4373-4328. 863
h p://doi.o g/10.1175/2009JCLI2681.1. 864
Jo hi yangkoon, C., Si apalan, M. and Fa me , D.L., 2001. P ocess con ols o wa e 865
balance a iabili y in a la ge semi-a id ca chmen : downwa d app oach o hyd ological 866
model de elopmen . J. Hyd o. 254,174–198. h p://dx.doi.o g/10.1016/S0022-867
1694(01)00496-6. 868
Kay, A.L., Da ies, H.N., Bell, V.A. and Jones, R.G., 2009. Compa ison o unce ain y 869
sou ces o clima e change impac s: lood equency in England. Clima ic Change. 92, 870
41-63. h p://dx.doi.o g/10.1007/s10584-008-9471-4. 871
Kay, A.L., Jones, R.G. and Reyna d, N.S., 2006. RCM ain all o UK lood equency 872
es ima ion II. Clima e change esul s. J. Hyd ol. 318, 163-172. 873
h p://dx.doi.o g/10.1016/j.jhyd ol.2005.06.013. 874
Kendall, M.G., 1975. Rank co ela ion measu es. Cha les G i in, London, UK. 875
Kiely, G., 1999. Clima e change in I eland om p ecipi a ion and s eam low 876
obse a ion. Ad . Wa e . Resou . 23, 141–151. 877
Lahme , W., P u zne , B. and Becke , A., 2001. Assessmen o land use and clima e 878
change impac s on he mesoscale. Physics and Chemis y o he Ea h, Pa B. Hyd ol. 879
Ocean. A mos., 26 (7–8), 565–575. h p://dx.doi.o g/10.1016/S1464-1909(01)00051-X. 880
Lea esley, G.H., 1994.Modeling he e ec s o clima e change on wa e esou ces – a 881
e iew. Clima ic Change. 28, 159–177. 882
37
Lende ink, G., Buishand, A. and an Deu sen, W., 2007. Es ima es o u u e 883
discha ges o he i e Rhine using wo scena io me hodologies: di ec e sus del a 884
app oach. Hyd ol. Ea h Sys . Sc. 11 (3), 1145–1159. h p://dx.doi.o g/10.5194/hess-885
11-1145-2007. 886
Lespinas, F., Ludwig, W. and Heussne , S., 2014. Hyd ological and clima ic 887
unce ain ies associa ed wi h modeling he impac o clima e change on wa e 888
esou ces o small Medi e anean coas al i e s. J. Hyd ol. 511, 403 – 422. 889
h p://dx.doi.o g/10.1016/j.jhyd ol.2014.01.033. 890
Li, Z., Liu, W., Zhang, X. and Zheng, F., 2009. Impac s o land use change and clima e 891
a iabili y on hyd ology in an ag icul u al ca chmen on he Loess Pla eau o China. J. 892
Hyd ol. 377, 35-42. h p://dx.doi.o g/10.1016/j.jhyd ol.2009.08.007. 893
Mann, H.B., 1945. Non-Pa ame ic es s agains end. The Econome ic Socie y. 13, 894
245–259. h p://dx.doi.o g/10.2307/1907187. 895
Ma aun, D., We e hall, F., I eson, A.M., Chandle , R.E., 2010. P ecipi a ion 896
downscaling unde clima e change. Recen de elopmen s o b idge he gap be ween 897
dynamical models and he end use . Re . Geophys. 48 (3), 1944-9208. 898
h p://dx.doi.o g/10.1029/2009RG000314. 899
Mas and ea, M.D., Field, C.B., S ocke , T.F., Edenho e , O., Ebi, K.L., F ame, D.J., 900
Held, H., K iegle , E., Mach, K.J., Ma schoss, P.R., Pla ne , G-K., Yohe, G.W. and 901
Zwie s, F.W., 2010. Guidance No e o Lead Au ho s o he IPCC Fi h Assessmen 902
Repo on Consis en T ea men o Unce ain ies. In e go e nmen al Panel on Clima e 903
Change (IPCC). A ailable in h p://www.ipcc.ch. Accessed 5 Augus 2015. 904
38
Meau io, M., Zabale a, A., U ia e, J.A., S ini asan, R. and An igüedad, I., 2015. 905
E alua ion o SWAT models pe o mance o simula e s eam low spa ial o igin. The 906
case o a small o es ed wa e shed. J. Hyd ol. 525, 326-334. 907
h p://dx.doi.o g/10.1016/j.jhyd ol.2015.03.050. 908
Medie o, L., Kjeldsen, T.R., Macdonald, N., Kohno a, S., Me z, B., Vo ogushyn, S., 909
Wilson, D., Albu que que, T., Bloeschl, G., Bogdanowicz, E., Cas ella in, A., Hall, J., 910
Kobold, M., K iauciuniene, J., Lang, M., Madsen, H., Gul, G.O., Pe digao, R.A.P., 911
Roald, L.A., Salinas, J.L., Toumazis, A.D., Veijalainen, N. and Po a insson, O., 2015. 912
Iden i ica ion o cohe en lood egions ac oss Eu ope by using he longes s eam low 913
eco ds. J. Hyd ol.. 528, 341-360. h p://.dx.doi.o g/10.1016/j.jhyd ol.2015.06.016. 914
Meinshausen, M., Smi h, S.J., Cal in, K., Daniel, J.S., Kainuma, M.L.T., Lama que, J-915
F., Ma sumo o, K., Mon zka, S.A., Rape , S.C.B., Riahi, K., Thomson, A., Velde s, 916
G.J.M. and an Vuu en, D.P.P., 2011. The RCP g eenhouse gas concen a ions and 917
hei ex ensions om 1765 o 2300. Clima ic Change. 109 (1-2), 213-241. 918
h p://dx.doi.o g/10.1007/s10584-011-0156-z. 919
Mimikou, M.A., Bal as, E., Va anou, E., Pan azis, K., 2000. Regional impac s o clima e 920
change on wa e esou ces quan i y and quali y indica o s. J. Hyd ol. 234, 95–109. 921
Mon ei h, J.L., 1965. E apo a ion and en i onmen : he s a e and mo emen o wa e in 922
li ing o ganisms. Symposia o he Socie y o Expe imen al Biology.19, 205–234. 923
Mo iasi, D.N., A nold, J.G., Van Liew, M.W., Bingne , R.L., Ha mel, R.D. and Vei h, 924
T.L., 2007. Model e alua ion guidelines o sys ema ic quan i ica ion o accu acy in 925
wa e shed simula ions. Wa e shed Simula ions. T. ASABE. 50 (3), 885–900. 926
39
Moss, R. H., Edmonds, J.A., Hibba d, K.A., Manning, M.R., Rose, S.K., an Vuu en, 927
D.P., Ca e , T.R., Emo i, S., Kainuma, M., K am, T., Meehl, G.A., Mi chell, J.F.B., 928
Nakiceno ic, N., Riahi, K., Smi h, S.J., S ou e , R.J., Thomson, A.M., Weyan , J.P. and 929
Wilbanks, T.J., 2010. The nex gene a ion o scena ios o clima e change esea ch 930
and assessmen . Na u e. 463 (7282), 747-756. h p://dx.doi.o g/10.1038/na u e08823. 931
Mu phy, J.M., Sex on, D.M.H., Ba ne , D.N., Jones, G.S., Webb, M.J. and S ain o h, 932
D.A., 2004. Quan i ica ion o modelling unce ain ies in a la ge ensemble o clima e 933
change simula ions. Na u e, 430, 768–772. h p://dx.doi.o g/10.1038/na u e02771. 934
Nash, J.E. and Su cli e, J.V., 1970. Ri e low o ecas ing h ough concep ual models: 935
Pa 1. A discussion o p inciples. J. Hyd ol.10 (3), 282-290. 936
h p://dx.doi.o g/10.1016/0022-1694(70)90255-6. 937
Nijssen, B., O’donnell, G.M., Hamle , A.F. and Le enmaie , D.P., 2001. Hyd ologic 938
sensi i i y o global i e s o clima e change. Clim. Chang. 50, 143–175. 939
Oua da, T.B.M.J., Cha on, C. and S -Hilai e, A., 2008. S a is ical models and he 940
es ima ion o low lows. Can. Wa e Resou . J. 33 (2), 195-206. 941
h p://dx.doi.o g/10.4296/cw j3302195. 942
O e on, D.E., 1966. Muskingum lood ou ing o upland s eam low. J. Hyd ol. 4, 185-943
200. h p://dx.doi.o g/10.1016/0022-1694(66)90079-5. 944
Pe aza-Cas o, M., Ruiz-Rome a, E., Mon oya-A men a, L.H., Sanchez-Pe ez, J.M. 945
and Sau age, S., 2015. E alua ion o hyd ology, suspended sedimen and nickel loads 946
in a small wa e shed in Basque Coun y (no he n Spain) using eco-hyd ological SWAT 947
model. Ann. Limnol-In . J. Lim. 51 (1), 59-70. h p://dx.doi.o g/10.1051/limn/2015006. 948
40
Pe ez, J., Menendez, M., Mendez, F. and Losada, I., 2014. E alua ing he pe o mance 949
o CMIP3 and CMIP5 global clima e models o e he no h-eas A lan ic egion. Clim. 950
Dynam. 43 (9-10), 2663-2680. h p://dx.doi.o g/10.1007/s00382-014-2078-8. 951
Pe isco, S.E. and Ma ín, J.M., 2006. Escena ios de empe a u a y p ecipi ación pa a la 952
España peninsula y Balea es du an e el pe íodo 2001-2100 basados en “downscaling” 953
es adís ico median e mé odos de análogos. XXIX Jo nadas Cien í icas de la 954
Asociación Me eo ológica Española. Pamplona. 955
P ies ley, C.H.B. and Taylo , R.J., 1972. On he assessmen o su ace hea lux and 956
e apo a ion using la ge-scale pa ame e s. Mon hly Wea he Re iew,100, 81–92. 957
P udhomme, C., Young, A., Wa s, G., Hax on, T., C ooks, S., Williamson, J., Da ies, 958
H., Dadson, S. and Allen, S., 2012. The d ying up o B i ain? A na ional es ima e o 959
changes in seasonal i e lows om 11 egional clima e model simula ions. 960
Hyd ological P ocesses. 26 (7), 1115-1118. h p://dx.doi.o g/10.1002/hyp.8434. 961
Qiu, L.J., Zheng, F.L. and Yin, R.S., 2012. SWAT-based uno and sedimen 962
simula ion in a small wa e shed, he loessial hilly-gullied egion o China: Capabili ies 963
and challenges. In . J. Sedimen Res. 27:226–234. h p://dx.doi.o g/10.1016/S1001-964
6279(12)60030-4. 965
Ribalaygua, J., Pino, M.R., Pó oles, J., Roldán, E., Gai án, E., China o, D. and 966
To es, L., 2013. Clima e change scena ios o empe a u e and p ecipi a ion in A agón 967
(Spain). Sci. To al En i on. 463–464 (0), 1015-1030. 968
h p://dx.doi.o g/10.1016/j.sci o en .2013.06.089. 969
Sho house, C. and A nell, N., 1999. The e ec s o clima e a iabili y on spa ial 970
cha ac e is ics o Eu opean i e lows. Phys. Chem. Ea h 24 (1–2), 7–13. 971
41
Sloan, P.G. and Moo e, I.D., 1984. Modeling su ace and subsu ace s o m low on 972
s eeply sloping o es ed wa e sheds. Wa e Resou . Res. 20 (12), 1815-1822. 973
h p://dx.doi.o g/10.1029/WR020i012p01815. 974
Smakh in, V.Y., 2001. Low low hyd ology: a e iew. J. Hyd ol. 240, 147-186. 975
h p://dx.doi.o g/10.1016/S0022-1694(00)00340-1. 976
S ahl, K., Tallaksen, L.M., Gudmundsson, L. and Ch is ensen, J.H.,2011. S eam low 977
da a om small basins: a challenging es o high esolu ion egional clima e modeling, 978
J. Hyd ome eo ol. 12, 900–912. 979
S eele-Dunne, S., Lynch, P., McG a h, R., Semmle , T., Wang, S., Hana in, J. and 980
Nolan, P., 2008. The impac s o clima e change on hyd ology in I eland. J. Hyd ol. 356 981
(1-2), 28-45. h p://dx.doi.o g/10.1016/j.jhyd ol.2008.03.025. 982
Szêpszô, G., Lingemann, I., Klein, B., and Ko acs, M., 2014. Impac o clima e change 983
on hyd ological condi ions o Rhine and uppe Danube i e s based on he esul s o 984
egional clima e and hyd ological models. Na . Haza ds. 72 (1), 241-262. 985
H p://dx.doi.o g/10.1007/s11069-013-0987-1. 986
Ta akoli, M., De Smed , F., Vans eenkis e, T., and Willems, P., 2014. Impac o clima e 987
change and u ban de elopmen on ex eme lows in he G o e Ne e wa e shed, 988
Belgium. Na . Haza ds. 71(3), 2127-2142. h p://dx.doi.o g//10.1007/s11069-013-1001-989
7. 990
Taylo , K. E., S ou e , R. J. and Meehl, G. A., 2012. An o e iew o CMIP5 and he 991
expe imen design. Ame ican Me eo ological Socie y. 93, 485-498. doi:10.1175/BAMS-992
D-11-00094.1. 993
48
Table 3. Mos sensi i e pa ame e s ( anked om 1 he mos sensi i e and 13 he less 1091
sensi i e) in he Uppe Ne bioi Ri e ca chmen , hei desc ip ion, he ange used o 1092
he au ocalib a ion (p- ac o 0.81 and - ac o 0.41) and he bes alue. 1093
Change
ype
Va iable
name
Desc ip ion
Range
Bes alue
CN2
Cu e numbe
-0.2-+0.2
-0.07
ESCO
Soil e apo a ion compensa ion ac o
0.77-0.86
0.83
GWQMN
Dep h o wa e in he shallow aqui e equi ed o e u n low o occu
614-655
625.36
SOL_AWC
A ailable wa e capaci y
0.1-0.5
0.48
EPCO
Plan up ake compensa ion ac o
0.8-0.95
0.87
REVAPMN
Th eshold wa e in shallow aqui e
768-900
892.21
CH_K2
Main channel conduc i i y
10-44
38.66
ALPHA_BF
Base low alpha ac o
0.6-0.9
0.77
SURLAG
Su ace unno lag coe icien
0.5-2.5
1.32
SMTMP
Snow mel base empe a u e (ºC)
3-9
4.77
GW_DELAY
Delay ime o aqui e echa ge
1-20
1.4
SFTMP
Snow all empe a u e (ºC)
0.39-1.5
0.62
GW_REVAP
G oundwa e “ e ap” coe icien
0.017-0.04
0.026
“ ” means he de aul pa ame e is eplaced by a gi en alue; “ ” means he exis ing pa ame e alue is changed 1094
ela i ely 1095
1096
49
Table 4. Values ob ained o he s a is ical indices used in he e alua ion o he SWAT 1097
model pe o mance a daily ime-s ep. Seasonal s a is ical alues a e calcula ed o he 1098
1996-2013 pe iod. 1099
DISCHARGE
Scale
NSE
2
slope/in .
PBIAS
RSR
CALIBRATION
1996-2006
0.63
0.68
0.85/0.35
-1.00
0.61
VALIDATION
2007-2013
0.75
0.77
0.91/0.24
0.17
0.50
HIGH AI
2010-2012
0.67
0.76
1.02/0.16
-10.51
0.58
LOW AI
2003-2005
0.74
0.76
1.01/0.16
-5.16
0.51
ALL 1996-2013
1996-2013
0.69
0.72
0.89/0.3
-0.51
0.56
WINTER
0.66
0.68
0.81/0.62
6.26
0.58
SPRING
0.74
0.76
0.87/0.1
17.74
0.51
SUMMER
0.29
0.40
0.61/0.06
12.23
0.84
AUTUMN
0.61
0.72
0.98/0.77
-26.78
0.63
* Acco ding o Mo iasi e al., (2007) he discha ge simula ion is sa is ac o y a mon hly ime s ep when he 1100
NSE > 0.5, 2 > 0.5, RSR ≤ 0.7, and PBIAS < 25%. The bes alue o slope is 1 and 0 o in e cep (A nold 1101
e al., 2012). 1102
1103
1104
50
Table 5. Sign (+ o -) and p obabili y o occu ence (P) o he annual and seasonal 1105
ends de ec ed o he du a ion (days) o he pe iod below Q20 and abo e Q80 1106
be ween 1961 and 2000. T ends wi h a P highe han 0.66 a e ep esen ed in bold; 1107
posi i e alues a e i alicised. 1108
OBS_SIM
ACCESS1-
0_AN
BNU-
ESM_AN
BNU-
ESM_SDSM
MPI-ESM-
RL_AN
MPI-ESM-
RL_SDSM
MPI-ESM-
MR_AN
MPI-ESM-
MR_SDSM
CMCC-
CESM_AN
CMCC-
CESM_SDSM
Q20
YEAR
0.39
0.58
0.34
0.91
-0.45
0.73
-0.45
0.50
0.93
0.34
AUTUMN
-0.52
0.26
-0.08
0.37
-0.06
0.47
-0.99
0.04
0.65
0.96
WINTER
0.61
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
SPRING
0.86
0.08
0.68
0.16
0.47
0.73
-0.40
-0.81
-0.64
-0.05
SUMMER
0.32
-0.21
0.52
0.85
-0.78
0.45
0.41
0.81
-0.64
-0.41
Q80
YEAR
-0.99
-0.17
-0.91
-0.90
0.02
0.02
0.32
0.47
0.37
0.34
AUTUMN
-0.75
0.50
-0.98
-1.00
-0.75
-0.50
0.25
0.71
-0.62
-0.52
WINTER
-0.68
-0.76
0.49
0.63
0.76
0.62
-0.28
-0.49
-0.62
0.53
SPRING
-0.74
-0.64
0.00
-0.48
-0.42
0.37
0.65
0.37
0.77
0.39
SUMMER
-0.91
0.00
-0.84
-0.85
0.00
0.09
0.00
0.00
-0.35
0.00
1109
1110
51
1111
Fig. 1. A lan ic egion a ea as desc ibed by he IPCC (2007). The loca ion o he 1112
esea ch wo ks summa ized in Table 1 is ep esen ed in by numbe s. The i alicised 1113
numbe s in bold e e o wo ks ca ied ou in mo e han wo ca chmen s. Loca ion o he 1114
Uppe Ne bioi in his con ex and a map o i e ca chmen wi h hyd o-me eo ological 1115
s a ions se ings a e also included. 1116
1117
Fig. 2. Daily obse ed (OBS) and simula ed (SIM) discha ge o bo h he calib a ion 1118
(1996-2006) and he alida ion (2007-2013) pe iods, and he p ecipi a ion (PCP) 1119
obse ed in Amu io s a ion (AEMET 1060). 1120
52
1121
Fig. 3. Di e ence be ween obse ed me eo ological pa ame e s; p ecipi a ion (PCP) 1122
and a e age empe a u e (TMEAN) and clima e baselines (1961-2000) be o e applying 1123
bias co ec ion a annual and seasonal scales. 1124
1125
Fig. 4. Mon hly mean discha ge (m3 s-1) om 1961 o 2000 ob ained om he 1126
hyd ological simula ion wi h obse ed me eo ological da a (OBS_SIM) and om he 1127
hyd ological simula ion wi h he downscaled GCMs baselines. 1128
53
1129
Fig. 5. Annual discha ge di e ence (%) be ween he 16 hyd ological p ojec ions and i s 1130
espec i e baseline simula ions, di ided in o h ee 30-yea ho izons (2030s, 2060s, 1131
2090s). 1132
1133
Fig. 6. Seasonal discha ge di e ence (%) be ween CMCC_CESM_AN_R85 and 1134
CMCC_CESM_SDSM_R85 hyd ological p ojec ions and hei espec i e baselines 1135
di ided in o h ee 30-yea ho izons (2030s, 2060s, 2090s). In addi ion, he mean 1136
annual and seasonal discha ge (m3s-1) is indica ed in each ba . 1137
54
1138
Fig. 7. Annual and seasonal discha ge di e ence (%) be ween hyd ological p ojec ions 1139
and hei espec i e baselines g ouped by downscaling me hod and RCP. The igu e 1140
shows he mean di e ence be ween: 1141
- ACCES1-0_AN_R45, BNU-ESM_AN_R45, MPI-ESM-MR_AN_R45 and MPI-1142
ESM-RL_AN_R45, ep esen ed as AN_R45. 1143
- BNU-ESM_SDSM_R45, MPI-ESM-MR_SDSM_R45 and MPI-ESM-1144
RL_SDSM_R45, ep esen ed as SDSM_R45. 1145
- ACCES1-0_AN_R85, BNU-ESM_AN_R85, MPI-ESM-MR_AN_R85 and MPI-1146
ESM-RL_AN_R85, ep esen ed as AN_R85. 1147
- BNU-ESM_SDSM_R85, MPI-ESM-MR_SDSM_R85 and MPI-ESM-1148
RL_SDSM_R85, ep esen ed as SDSM_R85. 1149
The esul s a e di ided in o 3 ho izons (2030s, 2060s, 2090s). In addi ion, he mean 1150
annual and seasonal discha ge (m3 s-1) is indica ed in each ba . 1151
55
1152
Fig. 8. Mean mon hly discha ge (m3 s-1) simula ed wi h 16 clima e p ojec ions. The 1153
highes discha ge alues ep esen he maximum alue o he mean mon hly discha ge 1154
o all he p ojec ions by mon h, while he lowes alues ep esen he minimum. The 1155
esul s a e di ided in o h ee 30-yea ho izons (2030s, 2060s, 2090s). The g ey colou 1156
ep esen s he ange o possible discha ge alues and he obse ed mean mon hly 1157
discha ge (1961-2000) is shown (OBS_SIM). 1158
1159
1160
Fig. 9. T ends o low low (Q20) du a ion and high low (Q80) du a ion displayed a 1161
annual and seasonal scales o he 2011-2100 pe iod. The p ojec ions unde 1162
Rep esen a i e Concen a ion Pa hway 4.5 (RCP 4.5) and 8.5 (RCP 8.5) a e displayed 1163
sepa a ely. Only alues wi h a p obabili y o occu ence highe han 0.66 a e shown. 1164
