1
Scien i ic RepoR s | 6:33704 | DOI: 10.1038/s ep33704
www.na u e.com/scien i ic epo s
A ibu ion o hyd ological change
in Heihe Ri e Basin o clima e and
land use change in he pas h ee
decades
Kaisheng Luo1,2, Fulu Tao1,3, Juana P. Moiwo4 & Dengpan Xiao5
The con ibu ions o clima e and land use change (LUCC) o hyd ological change in Heihe Ri e Basin
(HRB), No hwes China we e quan i ied using de ailed clima ic, land use and hyd ological da a, along
wi h he p ocess-based SWAT (Soil and Wa e Assessmen Tool) hyd ological model. The esul s showed
ha o he 1980s, he changes in he basin hyd ological change we e due mo e o LUCC (74.5%) han
o clima e change (21.3%). While LUCC accoun ed o 60.7% o he changes in he basin hyd ological
change in he 1990s, clima e change explained 57.3% o ha change. Fo he 2000s, clima e change
con ibu ed 57.7% o hyd ological change in he HRB and LUCC con ibu ed o he emaining 42.0%.
Spa ially, clima e had he la ges e ec on he hyd ology in he ups eam egion o HRB, con ibu ing
55.8%, 61.0% and 92.7% in he 1980s, 1990s and 2000s, espec i ely. LUCC had he la ges e ec on
he hyd ology in he middle-s eam egion o HRB, con ibu ing 92.3%, 79.4% and 92.8% in he 1980s,
1990s and 2000s, espec i ely. In e es ingly, he con ibu ion o LUCC o hyd ological change in he
ups eam, middle-s eam and downs eam egions and he en i e HRB declined con inually o e he
pas 30 yea s. This was he comple e e e se (a sha p inc ease) o he con ibu ion o clima e change o
hyd ological change in HRB.
Wa e is inc easingly a limi ing ac o o socioeconomic de elopmen , especially in a id and semi-a id egions1.
Wa e sca ci y can endange ood secu i y and sus ainable economic de elopmen , as well as he heal h o he
ecosys em2,3. Hence changes in wa e quan i y ha e become he ocus o a en ion in i e basin managemen and
ecological es o a ion.
Add essing he issue o wa e sho age equi es knowledge o he ac o s which d i e hyd ological changes and
he ela ed e ec s on local wa e esou ces. The e ec s o clima e and land use change (LUCC) on wa e esou ces
ha e a ac ed much a en ion o e he yea s4,5. In e ms o global wa ming and ex eme wea he equency and
in ensi y6,7, clima e change has been iden i ied o he educ ions in global wa e esou ces8,9 and exace ba ion o
wa e sho age in a id and semia id egions. Also a ia ions in egional hyd ological cycles a e closely ela ed wi h
LUCC10–12. Tempe a u e inc eases wi h inc easing global wa ming, causing changes in p ecipi a ion pa e n and
in ensi y which in u n signi ican ly a ec egional hyd ological cycle13.
Documen ed li e a u e e eals ha o en, se e al hyd ological s udies only deal wi h speci ic componen s o
wa e balance, e.g., s eam low14–16, g oundwa e echa ge17–20, uno 21–24 and e apo anspi a ion25–28. Land use
a ec s s eam uno 29–31, wa e in il a ion capaci y32,33 and su ace e apo a ion34–36. Howe e , ew s udies ha e
ocused on he e alua ion o basin wa e balance in e ms o he impac o LUCC and clima e change on hyd o-
logical p ocesses.
Re iews o se e al domes ic and in e na ional s udies sugges ha LUCC and clima e change ha e signi ican
e ec s on he hyd ological componen s o a i e basin37,38. Howe e , i emains unclea which o he wo ac o s
(LUCC and clima e change) dominan ly con ibu e o basin hyd ological p ocesses. The e is also a need o know
1Key Labo a o y o Land Su ace Pa e n and Simula ion, Ins i u e o Geog aphic Sciences and Na u al Resou ces
Resea ch, Chinese Academy o Sciences, Beijing 100101, China. 2Uni e si y o Chinese Academy o Sciences, Beijing
100101, China. 3Na u al Resou ces Ins i u e Finland (Luke), 01301 Van aa, Finland. 4Depa men o Ag icul u al
Enginee ing, School o Technology, Njala Uni e si y, Sie a Leone. 5Ins i u e o Geog aphical Sciences, Hebei
Academy o Sciences, Shijiazhuang 050011, China. Co espondence and eques s o ma e ials should be add essed
o F.T. (email: ao l@igsn .ac.cn)
Recei ed: 26 Janua y 2016
Accep ed: 01 Sep embe 2016
Published: 20 Sep embe 2016
OPEN
www.na u e.com/scien i ic epo s/
2
Scien i ic RepoR s | 6:33704 | DOI: 10.1038/s ep33704
he empo al and spa ial changes in he con ibu ions o LUCC and clima e change o basin hyd ological change.
Thus his s udy quan i ied he con ibu ions o LUCC and clima e change o a ypical basin hyd ological change
in No hwes China empo ally and spa ially, which will deepen he exis ing unde s anding abou he in e ac ion
o he selec ed hyd ological ac o s and he implica ions o wa e esou ces managemen .
Heihe Ri e Basin (HRB), he second la ges inland i e basin in China, was in es iga ed in his s udy (see
Supplemen a y Fig. S1). Because o he agile ecological en i onmen and se e e wa e sca ci y37, e ec i e man-
agemen is c ucial o HRB hyd ology. Wa e sca ci y in he basin has caused signi ican changes in he local
hyd ological en i onmen o e he pas decades; including en i onmen al deg ada ion, saliniza ion and dese -
i ica ion37. The change in hyd ological egimes induced by LUCC and clima e change37 is less unde s ood in he
basin.
The main objec i es o his s udy we e: 1) o quan i y he con ibu ions o LUCC and clima e change o hyd o-
logical p ocesses in HRB in No hwes China; and 2) o de e mine he empo al and spa ial ends in he con ibu-
ions o LUCC and clima e change o he basin hyd ological change. To do his, SWAT (Soil and Wa e Assessmen
Tool) model was adap ed o HRB using clima e, hyd ology, soil, land use and DEM da a, and un o 11 scena io
expe imen al condi ions. To es ima e he con ibu ions o LUCC and clima e change o he basin hyd olog-
ical change, one ac o was changed a a ime wi h he o he s held cons an . Speci ically, 11 model scena ios
(i.e., M1, M2, M3, … , M11) we e used o quan i y he con ibu ions o LUCC and clima e change o he basin
hyd ological change. The Man-Kendall end es was used o analyze he dynamics o empe a u e, p ecipi a ion
and wa e yield in he basin (see Me hods).
Resul s
Model calib a ion and alida ion. Based on he dis ibu ion o he na u al d ainage ne wo k, basin
opog aphy and ain all, he HRB was delinea ed in o 589 sub-basins. The sub-basins we e u he di ided in o
6850 Hyd ologic Response Uni (HURs) based on he basin land use, soil p ope y and slope. The Nash-Su cli e
e iciency coe icien (NS), R2 coe icien , pe cen bias (PBIAS) and RMSE-obse a ion s anda d de ia ion a io
(RSR) we e used o assess he eliabili y and accu acy o he model simula ion. The accu acy measu es a he
yea ly scale we e 0.65 < NS < 0.75, 0.50 < RSR < 0.60, and 10% < PBIAS < 15% o bo h he calib a ion and ali-
da ion pe iods. This sugges ed ha he model pe o mance was ai ly good (Table S1), especially o he ups eam
egion (UHRB), al hough he pe o mance a mon hly scale was no as good as a yea ly scale (Table S2). The
ac ual model pe o mance was bes o he downs eam egion (DHRB). This was because he obse ed i e dis-
cha ge was ei he small o d ied up, making he compa ison impossible wi h he simula ed alues. The measu ed
wa e yield was u he compa ed wi h he simula ed wa e yield a decadal scale. The compa ison showed ha he
SWAT model pe o med ai ly well, wi h a ela i e e o ange o 2.02‒ 3.42% (Table S3).
Land use change (LUCC) and clima e change in 1980‒2000. Compa ed wi h he middle-s eam
(MHRB) and downs eam (DHRB) egions, LUCC was smalles in he ups eam (UHRB) egion in 1980‒ 2009
(see Supplemen a y Fig. S2). Wi h he excep ion o he 1990s, ag icul u al land s eadily expanded while o es
land sh unk.
In he UHRB, he a ea o o es dec eased in he 1980s, 2000s and du ing 1980‒ 2009, bu inc eased in he
1990s (see Supplemen a y Fig. S2). Pas u e land dec eased in he 1980s, bu inc eased in he 1990s and he 2000s
in he UHRB. The a ia ion in land use was la ges in he MHRB du ing 1980‒ 2009, wi h he la ges change in
he 1990s. As shown in Fig. S2 (see Supplemen a y), he la ges luc ua ion was o ag icul u al land. This was
e i ied by published da a in he s a is ics yea book. Ba e land (including dese and he Gobi land) expanded in
he 1980s in he DHRB, bu dec eased since he ea ly 1990s. Ag icul u al land s eadily expanded since he 1980s
in he s udy a ea.
Annual mean empe a u e inc eased signi ican ly in he UHRB, MHRB, DHRB and he en i e HRB du ing
1980‒ 2009 (Figu e S3, Supplemen a y). The la ges change a e (0.62 °C/10a) was in he DHRB, ollowed by he
UHRB (0.56 °C/10a) and hen he MHRB (0.52 °C/10a). Annual p ecipi a ion inc eased sligh ly in he en i e
HRB (7.77 mm/10a) and in he h ee sub- egions, bu none o he inc eases was signi ican (p > 0.05). The max-
imum inc ease was in he UHRB (12.21 mm/10a), ollowed by he MHRB (5.90 mm/10 y ) and hen he DHRB
(5.71 mm/10 y ) (see Supplemen a y Fig. S3).
Change in hyd ological p ocesses. Wa e yield change. Figu e1a depic s he changes in wa e yield and
he ela ed spa ial pa e ns du ing 1980‒ 2009. O e all, he inc ease in wa e yield was mos ob ious in he UHRB
and MHRB du ing 1980‒ 2009. Howe e , he e we e clea di e ences in he spa ial pa e ns, which was mos ob i-
ous o he UHRB (0.3‒ 14.9 mm), ollowed by he MHRB and hen he DHRB. Wa e yield gene ally inc eased in
he en i e HRB in he 1980s, wi h excep ion o he UHRB o which i was − 363‒ 20 mm (Fig.1b). The e was an
o e all decline in wa e yield in he en i e HRB in he 1990s, excep o he UHRB (Fig.1c). The inc ease in wa e
yield in mos o he sub-basins in he UHRB exceeded 80 mm in he 2000s (Fig.1c).
Wa e balance. The changes in wa e balance mainly e lec ed he combined e ec s o LUCC and clima e change
in he basin. Figu e2 depic s he changes in ac ual e apo anspi a ion (ET), su ace uno (SQ), g oundwa e
echa ge (GW), la e al low (LA) and ansmission loss (TL), bo h posi i e and nega i e ends exis ed o 1980‒
2009 in he basin. Unde he combined e ec s o land use and clima e change, he la ges inc ease in ET in he
UHRB was in he 2000s (157.78 mm), ollowed by he 1990s and hen he 1980s. The e was also an inc easing
e ec o LUCC and clima e change on ET in he MHRB. Howe e , because o he small SQ, he change in SQ was
less han 2.5 mm in bo h he MHRB and DHRB.
GW inc eased in he UHRB and MHRB, wi h he la ges echa ge in he 2000s in he UHRB and in he 1980s
in he MHRB, howe e he end weakened wi h ime. GW in he DHRB had dec eased since he 1990s, wi h he
www.na u e.com/scien i ic epo s/
3
Scien i ic RepoR s | 6:33704 | DOI: 10.1038/s ep33704
la ges decline in he i s decade. The end in LA was posi i e o he UHRB in he 1990s and he 2000s, o he
MHRB in he 1980s and o he DHRB in he 1980s and 2000s. Also, i was posi i e o he UHR in he 1980s, o
he MHRB in he 1990s and 2000s, and o he DHRB in he 1980s (Fig.2). O e all, he change in ET due o he
combined e ec o LUCC and clima e change in he h ee egions was la ges du ing 1980‒ 2009, and ha in SQ
was la ges o he MHRB.
Figu e 1. Plo s o β alue o 1980‒ 2009 (a), change in wa e yield in he 1980s (b), 1990s (c) and 2000s
(d) in Heihe Ri e Basin (HRB) in No hwes China. This igu e was gene a ed hough he A cGIS 10.2 so wa e
p o ided En i onmen al Sys ems Resea ch Ins i u e (h p://www.es i.com).
www.na u e.com/scien i ic epo s/
4
Scien i ic RepoR s | 6:33704 | DOI: 10.1038/s ep33704
The con ibu ion o LUCC o he wa e balance is shown in Fig.3. The maximum change in ET (− 85.27 mm)
was in he 1990s o he UHRB. Then i had he la ges e ec in he UHRB, ollowed by he MHRB and DHRB.
The con ibu ion o LUCC o ET change inc eased in 1980‒ 2009 o he MHRB and DHRB. The e ec o LUCC
on SQ was ela i ely small in he MHRB and DHRB. The la ges e ec o land use on GW and TL was in 1980‒
2009 o he MHRB, ollowed by he UHRB and DHRB. Al hough he con ibu ion o LUCC o he ups eam
hyd ological change gene ally dec eased, i had he la ges impac on LA o UHRB, ollowed by he MHRB and
DHRB. This was closely ela ed o he ansi ion end in LUCC, especially wi h dec easing o es and pas u e
lands om he ups eam o he downs eam egions.
Figu e4 shows he con ibu ion o clima e change alone on he en i e wa e balance in HRB. Due o clima e
change, ET inc eased in 1980‒ 2009 o he HRB, in he 1980s o he UHRB and in he 1990s o he MHRB.
Du ing 1980‒ 2009, he inc ease in ET was 261.35 mm o he UHRB, ollowed by he MHRB (217.94 mm) and
DHRB (115.88 mm). O e all, SQ inc eased sligh ly du ing 1980‒ 2009 in he HRB. GW inc eased in he HRB du -
ing 1980‒ 2009 due o clima e change, bu he end was nega i e o he 1980s in he UHRB and o he 1990s and
2000s in he MHRB. The e ec o clima e change was mos ob ious in he UHRB. In he DHRB, clima e change
mainly a ec ed ET, which inc eased by 115.88 mm du ing 1980‒ 2009.
Figu e 2. Plo s o he combined e ec s o land use and clima e change on he hyd ology Heihe Ri e Basin
(HRB) in he 1980s (a–c), 1990s (d– ), 2000s (g–i) and 1980‒ 2009 (j–l). ET is ac ual e apo anspi a ion; SQ is
su ace uno in o each (mm); GW is base- low in o each (mm); TL is ansmission loss; and LA is la e al low
con ibu ion o s eam- low (mm).
www.na u e.com/scien i ic epo s/
5
Scien i ic RepoR s | 6:33704 | DOI: 10.1038/s ep33704
Con ibu ion o LUCC and clima e change o HRB hyd ological change. The con ibu ion o LUCC
o he basin hyd ological change was la ges (92.3%) in he MHRB (Fig.5). Fo clima e change, he con ibu ion
was la ges (55.8%) in he UHRB and smalles (2.0%) in he DHRB. In he UHRB, he con ibu ion o clima e
change o he hyd ological change was 12.9% highe han ha o land use in he 1980s. Thus du ing his pe iod,
he basin hyd ological change was mainly d i en by clima e change. Howe e , LUCC was he d i ing ac o o he
basin hyd ological change in he MHRB and DHRB. The con ibu ion o LUCC o he hyd ological change in he
1980s was 86.4% highe han ha o clima e change in he MHRB and 85.6% highe han ha o clima e change
in he DHRB. Due o unce ain y in he model simula ion, esidual e ec was la ges in he DHRB. This showed
ha in he 1980s he UHRB hyd ological change was d i en mo e by clima e change whe eas ha in he MHRB
and DHRB was d i en mo e by LUCC. Fo he hyd ological change in he whole HRB, he con ibu ion o LUCC
was 53. 2% highe han ha o clima e change.
In he 1990s, con ibu ion o LUCC o hyd ological change was la ges (79.4%) in he MHRB and smalles
(37.0%) in he UHRB. And con ibu ion o clima e change o hyd ological change was la ges (61.0%) in he
UHRB and smalles (7.0%) in he MHRB. The hyd ological change in he UHRB was mainly d i en by clima e
change, accoun ing o 24% mo e han LUCC in he 1990s. Then he hyd ological change in he MHRB and
Figu e 3. Plo s o he e ec s o land use on he hyd ology o he Heihe Ri e Basin (HRB) in he 1980s
(a–c), 1990s (d– ), 2000s (g–i) and 1980‒ 2009 (j–l). ET is ac ual e apo anspi a ion; SQ is su ace uno in o
each (mm); GW is base- low in o each (mm); TL is ansmission loss; and LA is la e al low con ibu ion o
s eam- low (mm).

www.na u e.com/scien i ic epo s/
6
Scien i ic RepoR s | 6:33704 | DOI: 10.1038/s ep33704
Figu e 4. Plo s o he e ec s o clima e change on he hyd ology Heihe Ri e Basin (HRB) in he 1980s
(a–c), 1990s (d– ), 2000s (g–i) and 1980‒ 2009 (j–l). ET is ac ual e apo anspi a ion; SQ is su ace uno in o
each (mm); GW is base- low in o each (mm); TL is ansmission loss; and LA is la e al low con ibu ion o
s eam- low (mm).
Figu e 5. The con ibu ions o land use and clima e change o he en i e hyd ology o he Heihe Ri e Basin
(HRB).
www.na u e.com/scien i ic epo s/
7
Scien i ic RepoR s | 6:33704 | DOI: 10.1038/s ep33704
DHRB was mainly d i en by land use, accoun ing o 72.4% and 34.2% mo e han clima e change, espec i ely. In
he en i e HRB, he con ibu ion o LUCC o hyd ological change was 3.4% g ea e han ha o clima e change.
LUCC was he main d i ing ac o in he 2000s, accoun ing o 92.8% o he hyd ological change in he
MHRB. This ag eed wi h he inc ease in con ibu ion o clima e change in he 2000s, wi h peak le els o 92.7%
and 61.5% in he UHRB and DHRB, espec i ely (Fig.5). The con ibu ions o land use and clima e change o he
hyd ological change in he en i e HRB we e 42.0% and 57.7%, espec i ely. This sugges ed ha he e ec o cli-
ma e change g adually su passed ha o LUCC and became he main ac o in luencing he hyd ological change
in he HRB in No hwes China.
Discussions
LUCC and basin hyd ological change. I was expec ed ha changes in land use will con inue in he
u u e, which will ha e signi ican e ec on egional wa e balance. Human dis u bance o land co e sys ems has
signi ican ly al e ed he Ea h’s land su ace39, subs an i ely eshaping global wa e balance40. The in ensi ica ion
o land use, popula ion g ow h and socio-economic de elopmen con inues o exe a huge p essu e on a ailable
wa e esou ces40.
De o es a ion dec eases soil wa e e en ion capaci y, and inc eases SQ by accele a ing wa e mo emen due
o low esis ance o low and low soil compac ion41. In con as , e o es a ion inc eases wa e e en ion capaci y,
and he eby dec eases SQ by decele a ing wa e mo emen due o high esis ance o low and longe low pa hs42.
Runo was la ges in he UHRB, which is he mos o es ed sub-basin in he HRB. Fo es a ea dec eased in he
1980s, 2000s and in 1980‒ 2009, bu inc eased in he 1990s. These changes, esul ing in dec easing albedo and
inc easing SQ in he 1980s‒ 2000s, led o he decline in ET. Ne e heless, he changes in land su aces, g oundwa-
e and pas u e somehow compensa ed he e ec s o de o es a ion on he basin wa e esou ces.
Su ace oughness, albedo and o he p ope ies ha a ec he exchange o wa e and ene gy be ween he land
su ace and he a mosphe e a e al e ed by LUCC. This esul s in he a iabili y o su ace ene gy and ne adia ion39,
which in u n in luences hyd ological p ocesses. The expansion o ag icul u al lands has a signi ican e ec on
shallow aqui e wa e sys em due o he associa ed inc ease in i iga ion43. O e 80% o i iga ed ag icul u al
lands in he HRB a e loca ed in he MHRB. In he pas 30 yea s, ag icul u al land inc eased by 579.4 km2, and
mos o which (499.7 km2) was in he MHRB. Change in ag icul u al land was he main o m o LUCC in he
middle-s eam egion, wi h a majo impac on wa e balance. Wi h he expansion o ag icul u al land in he 1990s
and 2000s, TL and GW inc eased (due o enhanced in il a ion) and ET inc eased (due o lood i iga ion). In he
1980s, TL and ET dec eased wi h he sh inking o ag icul u al lands. On he a e age, ag icul u al lands expanded
s eadily in he Ejnaqi Oasis by 86.9 km2 in 1980‒ 2009. Also because o his, he oasis apidly deg aded in o dese ,
which in u n a ec ed wa e balance in he DHRB.
I espec i e o he seasons in a yea , deg ada ion o p oduc i e lands in o ba en lands induces he mos
change in o al hea lux. In he HRB, ET was mainly d i en by a ailable ene gy. The con e sion o g assland in o
ba en o spa sely ege a ed lands limi ed a ailable ene gy o ET, which in u n accele a ed uno 34. Dese and
oasis a e he dominan landscape in HRB. Dese is dominan in he MHRB and DHRB, accoun ing o 73.0% in
he DHRB. Dese i ica ion was he e o e he main mode by which LUCC in luenced wa e balance in he s udy
a ea. The expansion o dese s in he pas 30 yea s was 61.9 km2, signi ican ly in luencing wa e a ailabili y in he
DHRB. The e was a clea nega i e co ela ion be ween change in ET and ime du ing 1980‒ 2000. Also o 1980‒
2009, dese i ica ion educed TL and GW, esul ing in low SQ and LA in he DHRB.
The con ibu ion o LUCC o he basin hyd ological change was 87.4% in he 1980s, which d opped o 30.1%
in he 2000s. The a e age con ibu ions o LUCC o ups eam, middle-s eam and downs eam hyd ological
change in 1980‒ 2009 we e 30.3%, 88.2% and 61.1%, espec i ely. Thus unlike he UHRB, LUCC dominan ly
in luenced hyd ological change in he MHRB and DHRB in 1980–2009. Fo es and pas u e accoun ed o o e
90% in he UHRB, which, along wi h spa se popula ion, mi iga ed he e ec o human ac i i y on he basin
hyd ological change. The MHRB is a p edominan ly ag icul u al-based basin wi h a ela i ely dense popula ion,
hence he Gansu side o he Hexi Co ido is a main g ain p oduc ion egion44, and human ac i i y has a majo
impac on he local hyd ological change.
The esul s a e suppo ed by se e al p e ious s udies45–47. Compa ed wi h clima e change, human ac i i y has
la ge e ec on he hyd ological p ocesses in he MHRB45. Due o human ac i i y, he he e ogenei y and di e si y
o he landscape ha e been de e io a ing since he ea ly 1980s37. Human ac i i y was he main d i e o dese i-
ica ion in MHRB in 1860‒ 199937,45, pa icula ly as d i en by he empo al and spa ial pa e n o wa e esou ces
in he egion46. Vege a ion deg ada ion in DHRB was mainly d i en by in ensi e human ac i i y in he egion in
he pas i e decades47.
Con ibu ion o clima e change o hyd ological change in he en i e basin. Annual mean em-
pe a u e in he HRB inc eased wi h inc easing global wa ming, he a e was much highe (0.054 °C/y ) han
he global mean (0.012 °C/a)48,49, and especially in he DHRB (0.062 °C/y ). Wa ming clima ic condi ions a e
expec ed o inc ease e apo a ion a e50. Al hough o e all ET inc eased no ably in he HRB in 1980‒ 2009, his was
mo e ob ious in he UHRB whe e i had he la ges e ec due o sca ce wa e esou ces. Ob iously, he changes in
ET we e due o clima e change51. Low ET in he 1980s in he UHRB and in he 1990s in he DHRB was due o low
wind speed, which educed he ole o ising empe a u e in he basin hyd ological change.
The e was a weak inc ease in annual p ecipi a ion, esul ing in an o e all inc ease in he basin SQ in 1980‒
2009. The la ges inc ease in SQ was in he MHRB, ollowed by he UHRB and DHRB. In he UHRB, SQ
inc eased due o inc eased summe p ecipi a ion and wa ming win e condi ions52. The widesp ead o o es in
he egion enhanced soil wa e e en ion and he eby dec eased SQ. Cu en ly, in he MHRB, he e a e o e 80%
ag icul u e land and 86% impe ious su ace, which signi ican ly inc ease SQ. Because o high ET (d i en by high
www.na u e.com/scien i ic epo s/
8
Scien i ic RepoR s | 6:33704 | DOI: 10.1038/s ep33704
empe a u e and he as dese and he Gobi), SQ was e y small in he DHRB. Clima e change had he la ges
e ec on GW, LA and TL, and hus on ET and SQ, wi h he la ges e ec in he UHRB.
ET was he main ac o in luencing he con ibu ion o clima e change o he hyd ological change in MHRB
and DHRB. The con ibu ion o clima e change o he hyd ological change in he ups eam basin was 55.8%
(1980s), 61.0% (1990s) and 92.7% (2000s). The con ibu ion o clima e change o he hyd ological change in
middle-s eam basin was 5.9% (1980s), 7.0% (1990s) and 13.0% (2000s). I was 2.0% (1980s), 31.5% (1990s) and
61.5% (2000s) in he downs eam basin. Thus he a e age con ibu ions o clima e change o he hyd ological
change in he uppe , middle and down s eam basins in 1980‒ 2009 we e 21.3%, 57.3% and 57.7%, espec i ely.
Al hough i was only 36.7% o he whole HRB in 1980‒ 2009, i ob iously inc eased wi h clima e wa ming. SQ
and ET inc eased wi h inc easing empe a u e and p ecipi a ion53, u he in luenced uno and hyd ological
p ocesses in he MHRB and HRB54,55.
Tempo al and spa ial changes in analyzed con ibu ions. The e was a la ge spa ial a ia ion in he
con ibu ions o LUCC and clima e change o he hyd ological change in he HRB. In he pas h ee decades, he
e ec o clima e change on he basin hyd ological change was la ges o UHRB, ollowed by DHRB and hen
MHRB. The la ges con ibu ion o LUCC o hyd ological change was in he MHRB, ollowed by he DHRB
and UHRB. The e is li le human ac i i y in he s udy a ea because o he spa se popula ion and ha sh na u al
en i onmen 51. The MHRB has he la ges popula ion and he mos obus economy in he HRB in No hwes
China44,45. In e es ingly, he con ibu ion o LUCC o hyd ological change con inually dec eased in he ups eam,
middle-s eam, downs eam basins and hen in he en i e HRB o e he pas 30 yea s. On he con a y, he con-
ibu ions o clima e change o he hyd ological change in he basin sha ply inc eased. The ends could con inue
in he u u e due o he p ojec ed clima e change56,57.
Adap i e s a egies and op ions. The analysis on he a ibu ion o he basin hyd ological changes o
LUCC and/o clima e change is undamen al o basin wa e esou ces managemen and planning. This s udy
indica ed ha he con ibu ions o LUCC and clima e change o basin hyd ological change a ied wi h egions
due o he he e ogenei y o hyd ological condi ions58. This sugges ed ha he le el o p io i y on LUCC manage-
men and clima e change adap a ion should be made o di e en egions. Fo ins ance, he o es egion o he
UHRB (as he wa e sou ce egion o HRB) should be s ic ly p o ec ed because o i s c i ical ole in p o ec ing
hyd ologic unc ion. In he MHRB, i is u gen o con ol he a e and scale o expansion o ag icul u al lands
because o i s nega i e e ec on hyd ological cycle. Wa e consump ion in indus ial and domes ic sec o s is small
in he HRB, bu a huge 87% o o al wa e use is om he ag icul u al sec o 37. Inc easing ag icul u al wa e use
wi h inc easing popula ion in he MHRB has occupied ecological wa e use in he DHRB37. The cu en lood
i iga ion is he leas e icien way o ag icul u al wa e use, which esul s in a conside able was e o he limi ed
a ailable wa e esou ces in he basin43,44. To add ess he wa e sho age p oblem, some less p oduc i e c oplands
should be ans o med in o g assland and some high wa e consump ion c opping sys ems should be ans o med
in o low wa e consump ion ones. E icien wa e alloca ion should be emphasized o ecological, ag icul u al,
indus ial and human consump ion.
Me hods
Si e desc ip ion. The HRB lies in No hwes China, wi hin 38‒ 42° N and 98‒ 101° W, and co e s an a ea o
14.31 × 105 km2 (see Supplemen a y Fig. S1). The basin is di ided in o h ee sub-basins — he ups eam basin
(UHRB) o he Qilian Moun ains in Qinghai P o ince; he middle-s eam basin (MHRB) o he oases and i i-
ga ed lands in Gansu P o ince, and he downs eam basin (DHRB) o he Gobi Dese in Inne Mongolia53. The e
a e 17 main ibu a ies (41 pe ennial ibu a ies) in he basin o igina ing om Qilian Moun ains. The es ima ed
mean uno om he ibu a ies is 34. 43 × 108 m3/y . Heihe Ri e ( he main i e in HRB) is he la ges pe en-
nial i e , wi h a o al leng h o 812 km and uno o 15.8 × 108 m3/y , accoun ing o hal o he o al uno in
he basin. As a hin e land i e basin in con inen al Asia, HRB has an a id con inen al monsoon clima e ha is
ex emely ho in summe and se e ely cold in win e , wi h 60–70% o he p ecipi a ion occu ing du ing July o
Sep embe 37.
Hyd ological model se up. The Soil and Wa e Assessmen Tool (SWAT), which is a physically-based,
semi-dis ibu ed hyd ological model was used in his s udy. Because SWAT is a de e minis ic model, each suc-
cessi e model un ha uses he same inpu s gi es he same ou pu s. This ype o model is sui able o isola ing he
e ec s o a single a iable, allowing he impac o change o be isola ed and quan i ied ela i e o he con ibu ions
o o he a iables. Because o his, he SWAT model and i s a ian s a e widely used a ound he wo ld58,59. A com-
p ehensi e index o wa e yield (Ywa e ) was used o cha ac e ize he o al e ec s on hyd ology in he in es iga ed
basin. Ywa e = SQ + L A − TL − Pond: whe e SQ is su ace uno in o each (mm); GW is base low in o each
(mm); TL is ansmission loss [mm]; LA is la e al low con ibu ion o s eam low (mm); and Pond is in e cep ed
wa e om ese oi s/ponds (mm). Rese oi /pond in e cep ion was negligible in his s udy because he limi ed
numbe /a ea o ese oi s/ponds in he basin. The con ibu ions o LUCC and clima e change o hyd ological
changes in he basin we e quan i ied by compa ison wi h di e en clima e scena ios.
The model was p e- un o he pe iod 1975‒ 1981 o s abilize he used pa ame e s. Then da a o 1981‒ 1997
we e used o calib a ion and 1998‒ 2010 o alida ion. The Nash-Su cli e e iciency coe icien (NS) and coe -
icien o de e mina ion (R2) we e used o e alua e he model pe o mance, compa ing simula ed and obse ed
discha ge a bo h mon hly and yea ly scales. The Nash-Su cli e e iciency coe icien (NS), R2 coe icien 59, pe -
cen bias (PBIAS) and RMSE-obse a ion s anda d de ia ion a io (RSR)60,61 we e used o assess he eliabil-
i y and accu acy o he model simula ion. Gene ally, a model simula ion is a ed as good i 0.65 < NS < 0.75,
www.na u e.com/scien i ic epo s/
9
Scien i ic RepoR s | 6:33704 | DOI: 10.1038/s ep33704
0.50 < RSR < 0.60 and ± 10% < PBIAS < ± 15%. A model simula ion is judged as sa is ac o y i 0.50 < NS < 0.65,
0.60 < RSR < 0.70 and ± 15% < PBIAS < ± 25%60.
Wi h snow hyd ology module, SWAT is also sui able o simula ing snowmel uno 61,62. In SWAT, he a eal
co e age o snow in a basin is de ined using a eal deple ion cu e, which desc ibes seasonal g ow h and ecession
o snowpack as a unc ion o he amoun o snow in he sub-basin. Snowmel is con olled by ai and snowpack
empe a u e, mel ing a e and a eal snow co e age. Th ee pa ame e s con ol snow all accumula ion — snow all
empe a u e h eshold (SFTMP), a eal snow co e age h eshold a 100% (SNOCOVMX) and a eal snow co e -
age h eshold a 50% (SNO50COV). Ano he ou pa ame e s go e n SWAT snowmel es ima ion — snowpack
empe a u e lag (TIMP), snowmel base empe a u e (SMTMP), maximum mel ac o (SMFMX) and minimum
mel ac o (SMFMN)58. To educe unce ain y, a sui able pa ame e was selec ed h ough sensi i i y analysis
using he La in Hype cube app oach a he 0.001 s ep leng h. Ag icul u al a eas can be i iga ed using di e sions
om wi hin he sub-basin o ou side he sub-basin. All i iga ion pa ame e s adop ed in he model we e he
model de aul alues.
Da a. The soil da a used in he SWAT simula ion we e om Ha monized Wo ld Soil Da abase (HWSD). The
land use da a (including 1985, 1995, 2000, 2005 and 2008) we e om En i onmen al Da a Cen e o Chinese
Academy o Sciences. Then land use da a o 2005 we e used o build he model and he o he s used o cons uc
scena io condi ions. The land use s a us in each decade was ep esen ed using he mid-yea land use maps o he
HRB. The land use maps o 1985, 1995 and 2005 we e conside ed o be su icien ly ep esen a i e o land use in
he 1980s.1990s and 2000s, espec i ely. Wea he da a o he SWAT model simula ion we e ob ained om China
Me eo ological Adminis a ion. Daily clima e da a o 1980‒ 2012 we e om 14 wea he s a ions adminis a ed
by China Bu eau o Me eo ology. Hyd ological da a we e om Heihe Ri e Wa e P o ec ion Bu eau and China
Hyd o-S a is ical Yea book. Mon hly mean discha ge da a used o he model calib a ion and alida ion we e
om 11 gauge s a ions in he basin. Shu le Rade Topog aphy Mission Digi al Ele a ion Model (DEM) wi h a
esolu ion o 30 m was ob ained om Wes Da a Cen e o China (WDCC).
Analysis. To es ima e he con ibu ions o LUCC and clima e change o he hyd ological change in he
HRB, one ac o was changed a a ime and he es held cons an . Speci ically, 11 model scena ios (i.e., M1, M2,
M3, … , M11) we e un o quan i y he con ibu ions o LUCC o clima e change o he basin hyd ological change.
The M01 o M04 scena ios we e used o quan i y he con ibu ions o LUCC and clima e change o HRB basin
hyd ological change in he 1980s. Then scena ios M4 o M7 we e used o quan i y he con ibu ions o LUCC and
clima e change o hyd ological change in he 1990s. Also, scena ios M8 o M11 we e used o he con ibu ions
o LUCC and clima e change o hyd ological change in he HRB in he 2000s (Table1). The con ibu ion o each
decade was e alua ed by compa ing he di e ences o simula ed wa e yield be ween wo scena ios in he HRB
and h ee sub-basins (Table2).
A ep esen a i e land use o he 1980s and clima e da a o 1980‒ 1989 (he ea e called 1980s clima e) we e
deno ed by modeling scena io M1. Then a ep esen a i e land use o he 1990s and clima e da a o 1990‒ 1999
(he ea e called 1990s clima e) we e deno ed by modeling scena io M4. The hyd ological change in 1980s
due o he combined e ec s o LUCC and clima e change was calcula ed as he di e ence be ween M4 and M1
(YM4-YM1). A ep esen a i e land use o he 1980s and clima e da a o 1990‒ 1989 (he ea e called 1990s clima e)
we e deno ed by modeling scena io M2. Also a ep esen a i e land use o 1990s and clima e da a o 1980‒ 1989
(he ea e called 1980s clima e) we e deno ed by modeling scena io M3. The con ibu ion o clima e change o
hyd ological change in he 1980s was hen calcula ed as:
=
−
−×CR Y
YY
YY
() 100%
(1)
clima e 80s M2 M1
M4 M1
Scena io Land use Clima e change
M1 1980s 1980s
M2 1980s 1990s
M3 1990s 1980s
M4 1990s 1990s
M5 1990s 2000s
M6 2000s 1990s
M7 2000s 2000s
M8 2000 2000‒ 2006
M9 2000 2007‒ 2013
M10 2008 2000‒ 2006
M11 2008 2007‒ 2013
Table 1. A lis o he di e en model scena ios used o e alua e he impac s o land use and clima e change
on he hyd ology o Heihe Ri e Basin (HRB) in No hwes China. No e ha land use maps o 1985, 1995
and 2005 we e used o ep esen land use in he 1980s, 1990s, 2000s, espec i ely; clima ic condi ion o he
1980s is deno ed by clima ic da a o 1980–1989; clima ic condi ion o he 1990s is deno ed by clima ic da a o
1990–1999; and hen clima ic condi ion o he 2000s is deno ed by clima ic da a o 2000–2009.