Evaluation and hydrological impact of land-use changesin the Longtan basin

GUIYAN MO1, YONGXIANG ZHANG

2,* , YA HUANG3, CHONGXUN MO

3

and QING YANG3

1College of Computer and Information, Hohai University, Nanjing 211 100, China.

2School of Management Science and Engineering, Guangxi University of Finance and Economics,Nanning 530 003, China.3College of Civil and Architectural Engineering, Guangxi University, Nanning 530 004, China.*Corresponding author. e-mail: yunghsiang.chang@hotmail.com

MS received 11 February 2020; revised 23 May 2020; accepted 11 June 2020

Compared to climate change, land-use changes were the main driving factors to short-term hydrologicalvariety. To evaluate land-use types variation and quantify its hydrological impact, this paper identiBedthe temporal-spatial features and simulated the hydrological process of different land-use types over thelast two decades (1990–2010) based on the soil and water assessment tool (SWAT). Meanwhile, thepossible inCuence of land-use changes on streamCow in the next 30 years (2020–2050) was also considered.Results indicated that (1) land-use types mainly constituted of forest land, grassland and cultivated land,which had the largest decreasing rate of 17.34 km2/10a. At the same time, inter-conversion mainlyconducted among the main land-use types and had a similar transfer structure for these two sub-periods(1990–2000, 2000–2010) during 1990–2010, with a more dramatic transformation in 2000–2010.(2) Simulated annual and monthly surface runoA did not vary greatly from 1990 to 2010 and presented arelatively uniform monthly distribution. At the same time, the increased vegetation coverage (forest landand grassland) can not only reduce surface runoA but also prevent peak Cood with increasing anddecreasing steeply. (3) Finally, hydrological variability to the future land-use change will not be intensive,which possibly related to the undeveloped regional economic and insignificant human activities. But italso needed some measures to maintain a balanced nature, such as the soil and water conservationmeasures and returning cultivated land to forest land and grassland.

Keywords. Land-use changes; evaluation; hydrological responses; SWAT model; Longtan basin.

1. Introduction

Vegetation coverage changes as farming, tree/grassplanting, construction, etc., aAect the regionalwater balance/hydrologic cycle as well as the spa-tial distribution of surface latent heat (Gong et al.

2014; Wang et al. 2015). Meanwhile, land-usechanges interfere the catchment’s hydrological andrainfall-runoA process by disordering the equilib-rium relationship between water and energy.Furthermore, land-use variability inCuences thetemporal-spatial distribution of water resources

Supplementary materials pertaining to this article are available on the Journal of Earth Science Website (http://www.ias.ac.in/Journals/Journal˙of˙Earth˙System˙Science).

J. Earth Syst. Sci. (2020) 129:190 � Indian Academy of Scienceshttps://doi.org/10.1007/s12040-020-01458-1 (0123456789().,-volV)(0123456789().,-volV)

and accelerates the complexity and uncertainty ofwater resources variation during hydrologic cyclingprocess (Wang et al. 2006, 2015). Compared to themainly constant catchment characteristics aAect-ing the unique water cycle of a catchment (e.g., soilproperties and topography), land-use change mayunderlie short-term variations (Defries and Eshle-man 2010) and include aAorestation and defor-estation, returning cultivated land to forestland,grassland, etc. For example, urbanization (convertcultivated land/grassland to construction land)increases the impermeable area and further inter-feres the inBltration process of surface runoA,which could bring more peak Cood in the rainyseason (Wang et al. 2015). Obviously, land-usealteration is vital for land evolution, do the regio-nal hydrological cycle and water balance for socialproduction and life. Visibly, simulating and esti-mating runoA response to land-use changes con-tribute to further analyzing the inCuencemechanism of anthropogenic activities on regionalhydrology and water resources (Niu and Sivaku-mar 2013; Karamage et al. 2017). Therefore, rela-ted subjects and studies about land-use change (atregional or global scales) and the changes in waterresources have become issues of increasing concernto experts and scholars in recent years (Zhang et al.2011; Braud et al. 2013; Isik et al. 2013; Yin et al.2017; Gashaw et al. 2018).Land-use variations, aAected by industrial struc-

ture and natural environment during the construc-tion process of the Longtan hydropower dam, arousea series of long-term water resources-related prob-lems (spring/summer drought, Cood disaster, haildamage, storm rainfall, etc.). Meanwhile, simulat-ing and quantifying hydrological response to thepast and future land-use variation can providedecision-makers information to promote vegetationrestoration, water resources management as well assoil and water conservation. However, very limitedresearch has been conducted to evaluate and inves-tigate land-use variability and its contribution tosurface runoA in the Longtan basin (Lei et al. 2007).Not to mention the possible hydrological variationto future land-use patterns. Therefore, the objec-tives of this study were (1) to identify land-usevariation characteristics from 1990 to 2010 (sec-tion 4.1); (2) to calibrate and validate a satisfactorysoil and water assessment tool (SWAT) for surfacerunoA simulation (sections 3.4 and 3.5); (3) toquantify the hydrological response to the past land-use change (section 4.2.1); and (4) to predict howdifferent land-use scenarios aAected streamCow

(section 4.2.2). The results obtained from this studywill help stakeholders with useful information inplanning and making decision for soil and waterresources management and providing a reference toother similar catchments.

2. Study site

Longtan watershed, a leading reservoir of thecascade development in the Hongshuihe River basin,stretches from the latitude of 23�110–27�010N andlongitude of 102�140–107�320E and extends over anarea of 98,500 km2 around the Yunnan Province,Guizhou Province and Guangxi Zhuang Autono-mous Region (Bgure 1). Elevation in the study arearanges from 23 to 2,258 m above the sea level, witha mean elevation of 1450 m.Annual mean temperature and precipitation

range from 12.3–21.3�C and 760–1860 mm, respec-tively, which accounts for 89% of annual rainfall inthe Cood season (from April to October, Liu 2015).Land-use types in this basin mainly include culti-vated land, forest landandgrasslandand soil textureare dominated by loam, accounting for 60% soiltypes. Vegetation is mainly composed of evergreenconiferous forest, evergreen and deciduous broad-leaved mixed forest, evergreen broad-leaved forestand subtropical deciduous broadleaf forest, with anaverage forest cover of 34.83%.

3. Methods and materials

To obtain the four intended purposes, this paperwas demonstrated based on the structure shown inBgure 2.

3.1 Land-use change analyses

To illuminate the space-time variation characteris-tics of land-use change experienced betweenthe subsequent periods (1990–2000, 2000–2010,Supplementary Bgure S1), percent of change (Hassenand Assen 2018) and transition matrix (Shiferaw2011)were computed using equation (1) and table 3.(1) Land-use variation amplitude

Pi ¼Xi1 � Xi0

Xi0

� �� 100%; ð1Þ

where Xi0 and Xi1 are the land-use area at thebeginning of the research and the area at the end ofresearch of ith land-use type, respectively.

190 Page 2 of 11 J. Earth Syst. Sci. (2020) 129:190

(2) Land-use transition matrixTransition matrix (tables 3 and 4) between the

two sub-periods (1990–2000 and 2000–2010) wasestablished to distinguish the change of one land-use type to the other ones during a certain period.For example, rows can display the conversion areaof the eight classes in 1990, whereas columns candisplay the conversion area of the eight classes in2000.

3.2 Land-use change prediction

Field investigation showed, when the normal water

level was 375 m, submerged land was about

377 km2, of which paddy Beld, cultivated land,

forest land and grassland took account for 30.55,

40.66, 145.15, and 160.64 km2, respectively.

Meanwhile, reservoir did not have a full waterstorage and there was an overCow power station

Figure 1. Longtan basin (98,500 km2), 25 meteorological stations (black points) and the Tian-e hydrological station (redtriangle).

Meteorological data(Temperature, Precipitation

Wind speed, Related humiditySection 3.4)

Geographic data(DEM, Soil type data

Section 3.4)

SWAT model setup(Section 3.4 + Section 3.5)

Calibration and Validation(Section 3.5)

Streamflow simulation for the past and future land-use change

(section 4.2)

Evaluation and hydrological impact of land-use changes

Land-use data for area andspatial variation during 1990-2010 and future land-use prediction

(Section 4.1 + Section 3.2)

Figure 2. Framework for this paper.

J. Earth Syst. Sci. (2020) 129:190 Page 3 of 11 190

in the upper reaches. To predict the hydrologicalimpact of future land-use change, scenario 1 (RL1)will be set when water level reached the normal one(375 m).Furthermore, changing amplitude of the other

two future land-use scenarios (RL2 and RL3,table 1) was based on the long-term land-useplanning data (Weining Yi, Hui and Miao Auton-omous County 2010; Xingyi Municipal People’sGovernment 2006; Huize Municipal People’sGovernment 2010; Ziyun Municipal People’sGovernment 2010) of the surrounding counties andextreme-value distribution method (R�edei 2008).Specifically, threshold value of built-up area, cul-tivated land, forest land and grassland in the next30 years (2020–2050) could be set to 10%, �1.5%,0.5%, 0.5% for scenario 2 (RL2) and 18%, �3%,0.5%, 1% for scenario 3 (RL3), respectively, com-pared to the baseline land-use type of 2000 (R�edei2008).

3.3 Land-use change evaluation

Variation percentage for annual runoA underland-use change can be calculated by equation (2)(Silberstein et al. 2012).

b ¼ ðyi � y0Þy0

� 100%; ð2Þ

where yi is the average annual run-oA generated bythe ith land-use change scenario, mm; y0 is theaverage annual runoA generated by the currentland-use change scenario (mm).The average monthly/annual runoA and the

conservation index were evaluated in this paper.The conservation index can reCect the stability ofhydrological process and estimate the demand forwater supply (Pizzolotto and Brandmayr 1996).The index can be measured by equation (3).

b ¼ Qm

Qm; ð3Þ

where Qm and Qv were the streamCow in the driestmonth and the average annual Cow, respectively(m3/s).

3.4 Data prepared for simulation

A SRTM DEM, having a resolution of 90 m anddownloaded from the Geospatial Data Cloud web-site (http://www.gscloud.cn), was used to obtain asatisBed basin-scale DEM map (Bgure 1). Land-usedata for the last two decades (1990–2010) and soil-type map of 2000 in the Longtan basin were bothaccessed from the Resource and Environment DataCloud Platform of China (http://www.resdc.cn)and with a spatial resolution of 1:1,000,000. ThereclassiBed land-use types of forest, grassland,water body, urban area, bare land, paddy andcultivated land were assigned as FRST, PAST,WATR, URBN, BALD, PADY and AGRL,respectively (Wang et al. 2014). Meanwhile, it wasworth noting that the content of clay, loam andsandy soil may be different in the same soil typeand further resulted in different physical propertiesof the same soil type in the same area, which couldderive from the spatial difference of hydrometeo-rological factors. Thus, 35 different soil types in theLongtan basin will be re-classiBed into seven soiltypes in the study watershed according to FAOclassiBcation (Chesworth et al. 2008). These soilswere Haplic Alisols (40.74%), Chromic Cambisols(31.03%), Humic Acrisols (15.44%), Dystric Cam-bisols (5.39%), Rendzic Leptosols (4.82%), Cumu-lic Anthrosols (2.28%) and Ferric Lixisols (0.30%),respectively.Meteorological data (daily average/maximum/

minimum temperature, wind speed, relative humi-dity and precipitation) of the 25 meteorological

Table 1. Land-use area for future three scenarios in the Longtan basin (2020–2050).

Land-use types

RL1 scenario RL2 scenario RL3 scenario

Area (km2) Ratio (%) Area (km2) Ratio (%) Area (km2) Ratio (%)

Forestry 47857.58 48.59 48242.74 48.98 48242.74 48.98

Grassland 26133.55 26.53 26425.66 26.83 26557.13 26.96

Water area 1063.64 1.08 686.98 0.70 686.98 0.70

Built-up area 1105.55 1.12 1046.11 1.06 1122.19 1.14

Bare land 59.10 0.06 2.60 0.003 5.88 0.01

Paddy Beld 8228.10 8.35 8251.22 8.38 8251.22 8.38

Cultivated land 14014.86 14.23 13844.69 14.06 13633.85 13.84

190 Page 4 of 11 J. Earth Syst. Sci. (2020) 129:190

stations (Bgure 1) were collected from the ChinaMeteorological Data Sharing Service System(http://data.cma.cn), covering a period of1959–2013. For the missing climate data of somestations in a short time, this paper used the Krigingmethod to consider the correlation between mete-orological stations for interpolation and supple-mentation, in order to improve the simulationaccuracy of the SWAT model (Liston and Elder2006). StreamCow data, which was collected at theoutlet of the study catchment (Bgure 1) for thesame period of 1959–2013, consisted of the mea-sured monthly inCow after dam construction andmonthly runoA deduced from the Tian-e hydro-logical station before dam construction.

3.5 SWAT model calibration and validation

Based on the data listed above, a SWAT modelwas set up for hydrological simulation under dif-ferent land-use changes in the Longtan basin. Thestudy watershed was delineated into 33 sub-basins,assisted by watershed mask Ble and further sub-divided into 277 HRUs by setting land-use, soil andslope threshold value to 10%, 15% and 10%,respectively. Referring to literature researches onsimilar catchments (Lv et al. 2014; Yang et al.2017; Huang et al. 2018), parameters of SCS runoA

curve number (CN2), base-Cow alpha factor(ALPHA˙BF), groundwater delay (GW˙DELAY)and soil evaporation compensation factor (ESCO)reached top four (Abbaspour et al. 2007; Khalidet al. 2016) in the ranking of sensitivity parameters(Bgure 3).

4. Results

Simulated annual runoA (472.35 million m3) was1.03% more than that of the measured annualdischarge (467.53 million m3) in the calibrationperiod (1987–1998). Indices of Ens and R2 for thisperiod were 0.85 and 0.75, respectively. For vali-dation period, it was 3.76% less than the observedone and Ens and R2 were both 0.88. Furthermore,Ens and R2 during the monthly calibrated andveriBed periods were 0.94 and 0.81, 0.94 and 0.91,respectively (Bgure 4). Thus, performance of theSWAT model enabled the applicability to simulatethe hydrological process in the Longtan basin.

4.1 Land-use change analysis

4.1.1 Area variation of land-use

As can be seen from the variation of land-use areain the Longtan basin (table 2), cultivated land and

Figure 3. Sensitivity analysis for parameters in a calibrated SWAT model. (V meant the existing parameter value was to bereplaced by a given value and R meant an existing parameter value was multiple by (1+ a given value). Higher t-Stat in absolutevalues meant higher sensitivity while a p-value of 0 was more significant.)

J. Earth Syst. Sci. (2020) 129:190 Page 5 of 11 190

0

500

1000

1500

2000

2500

Calibration period

1998

1997

1996

1995

1994

1993

1992

1991

1990

1989

1988

1987

Mon

thly

runo

ff/ m

3

Year

ObservedSimulated

0

200

400

600

800

1000

1200

1400

1600

1800

Validation period

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

Mon

thly

runo

ff/ m

3

Year

ObservedSimulated

Figure 4. Simulated and measured monthly runoA during the calibration and validation periods.

Table 2. ReclassiBed land-use area in the Longtan basin from 1990 to 2010.

1990 2000 2010

Area (km2) Ratio (%) Area (km2) Ratio (%) Area (km2) Ratio (%)

Forest land 47955.81 48.69 47812.73 48.54 48002.73 48.73

Grassland 26124.44 26.52 26515.92 26.92 26294.19 26.69

Water area 757.33 0.77 790.26 0.80 868.64 0.88

Built-up area 773.24 0.79 844.57 0.86 951.01 0.97

Bare land 69.27 0.07 68.35 0.07 69.27 0.07

Paddy Beld 8417.64 8.55 8387.64 8.52 8258.65 8.38

Cultivated land 14402.28 14.62 14080.52 14.29 14055.52 14.27

190 Page 6 of 11 J. Earth Syst. Sci. (2020) 129:190

paddy Beld changed more significantly during1990–2000 (decreased by 321.76 and 29.99 km2,respectively) than that in 2000–2010 (decreasedby 25 and 129 km2, respectively). Forest landdecreased 143.07 km2 Brstly for the period of1990–2000 and then increased 190 km2 during2000–2010, with an amplitude of 2.35 km2/year.On the contrary, grassland increased from26124.44 to 26515.92 km2 in 1990–2000 anddecreased to 26294.19 km2 for the period of2000–2010, with a change rate of 8.49 km2/year.At the same time, water area and urban-ruralconstruction land increased continuously, whichranged from 773.24 and 757.53 km2 in 1990 to951.01 and 868.64 km2 in 2010, respectively.However, bare land kept a steady proportion(0.07%) during 1990–2010 and slightly decreasedBrst then increased.

4.1.2 Spatial variation of land-use

Forest land concentrated more in the eastern plain,while less in the western plateau (SupplementaryBgure S1). However, grassland decreased from westto the east during the period of 1990–2010 andconstruction land was mainly concentrated in thewestern region, which was close to water-sourceregion. Meanwhile, water expansion occurredwithin catchment boundary and the other land-usetypes changed not significantly from 1990 to 2010.Possible reasons may be the undramatic humanactivities or poor production.

4.1.3 Inter-conversion between land-use

Conversion amplitude among different land-usetypes was larger in 1990–2000 (table 3) than thatduring 2000–2010 (table 4) and had a similarconversion pattern. From 1990 to 2000, 7021.87,

123.83, 84.22, 12.49, 2228.97 and 5143.07 km2 offorestry was converted from grassland, water area,built-up area (construction land), bare land, paddyBeld and cultivated land, respectively. Meanwhile,a large amount of forestry was also reverted tothe other land-use types. Similarly, 7021.87 and3298.48 km2 grassland were transited to forest landand cultivated land during the period of 1990–2000and 5986.37 and 2676.86 km2 for 2000–2010,respectively. Moreover, paddy Beld and cultivatedland were both mainly transited to forestry andgrassland, in which cultivated land transition wasmore significant. Obviously, land-use conversionmainly occurred among forest land, grass land andcultivated land.

4.2 Hydrological impact of land-use change

4.2.1 Hydrological response to the past land-usechange

(1) Annual scale: Compared to land-use map of1990, rainfall remain unchanged in 2000 and2010, based on the same climate data (table 5).Surface runoA was 162.5 mm under land-usemap of 1990 and gradually increased to162.8 mm in 2000, even to 170.11 mm in 2010(increased 4.68%). Similarly, evapotranspira-tion increased by 0.4 mm in 2000 and 1.0 mmin 2010, respectively. On the contrary, wateryield decreased from 473.9 mm in 1990 to472.9 mm in 2010 as well as the conservationindex, which ranged from 0.193 to 0.216.

(2) Monthly scale: Surface runoA in Cood sea-sons (from May to October, Bgure 5) increasedunder 2010 land-use map, especially in June,July and August. Specially, surface runoA for1990 land-use map was 48.21 mm in June and42.31 mm in July and reached to 50.64 mm(in June) and 44.51 mm (in July) under 2010

Table 3. Transition matrix for land-use conversion from 1990 to 2000 (Unit: km2).

1990

2000

Forestry Grassland Water area Built-up area Bare land Paddy Beld Cultivated land

Forestry 33132.00 7165.91 131.75 107.65 13.54 2261.03 5075.10

Grassland 7021.87 14303.14 106.70 98.98 18.15 1301.25 3298.48

Water area 123.83 93.65 363.96 21.11 0.00 85.64 72.36

Built-up area 84.22 86.53 18.98 133.00 0.41 342.49 110.30

Bare land 12.49 17.03 0.00 0.59 28.38 1.96 8.61

Paddy Beld 2228.97 1339.32 95.29 360.32 4.82 3281.94 1126.36

Cultivated land 5143.07 3523.80 77.50 125.74 3.43 1129.93 4418.42

J. Earth Syst. Sci. (2020) 129:190 Page 7 of 11 190

land-use map, respectively. However, surfacerunoA in the other months changed little forthe past three land-use maps. In short, surfacerunoA under the past land-use maps did notchange considerably on a monthly scale and itrained more in the rainy seasons (from May toOctober), but less in the dry seasons.

4.2.2 Hydrological response to the futureland-use change

(1) Annual scale: Compared to 2010 land-usemap, surface runoA increased to 171.2 and172.3 mm under the future land-use situations,respectively (table 6). Moreover, increasedwater level brought more water yield under theRL1 situation, with a maximum value of474.8 mm. On the other hand, high proportionof forestry caused a higher conservation index(0.217 for RL2 and 0.218 for RL3), but con-sumed more water, which caused water yield0.3 mm less than that in 2010 (472.9 mm).Evapotranspiration increased from 586.9 mmin 2010 to 587.2 mm under the RL3 scenarioand it may be because of the increasedurbanization. Obviously, similar proportion ofcultivated land, forest land and grasslandunder the RL2 and RL3 scenarios made thesimulated surface runoA and conservationindex variated little.

(2) Monthly scale: Similarly, monthly distribu-tion of surface runoA under the future land-use

scenarios (table 7) was similar to the pastland-use types in 4.2.1 section and producedmore in the Cood seasons (from May toOctober). Specifically, RL1 scenario produced162.8 mm surface runoA in the Cood seasonand 163.9 mm for the RL2 and RL3 scenarios.However, it was both 7.4 mm in the dryseasons for these three scenarios.

5. Discussion

5.1 Uncertainties in runoA simulation

Simulation results of the past and the future land-use situations bear great uncertainties as well asmodel calibration and validation. However, theranking of sensitive parameters CN2, ALPHA˙BF,GW˙DELAY and ESCO match good with theresults in similar basins (Lv et al. 2014; Yang et al.2017; Huang et al. 2018). Meanwhile, Ens and R2

coefBcients indicate a satisBed performance ofsimulation, but peak Cow in some dry months isunderestimated before 2002 or overestimated after2002. It may originate from the operation of theLongtan hydro-power dam, which could furtherinCuence the underlying condition. Or it can beattributed to SCS runoA curve number (CN2)method, which ignores the duration and intensityof precipitation. However, rainfall events aremainly composed of high-intensity, short-duration,limited-areal extent summer thunderstorms in theLongtan basin (Wang et al. 2008; Yang et al. 2010).

Table 4. Transition matrix for land-use conversion during 2000–2010 (Unit: km2).

2000

2010

Forestry Grassland Water area Built-up area Bare land Paddy Beld Cultivated land

Forestry 35742.52 5961.85 133.63 92.75 14.12 1846.01 3954.73

Grassland 5986.37 16541.92 118.36 114.83 12.72 1080.37 2676.86

Water area 106.28 87.14 451.53 14.87 0.10 71.66 62.49

Built-up area 91.16 89.68 10.73 333.33 0.64 230.41 91.32

Bare land 9.19 14.39 0.13 0.54 34.25 0.00 5.42

Paddy Beld 1944.31 1090.19 74.96 281.29 1.30 4143.27 868.10

Cultivated land 4056.75 2528.59 82.56 117.54 6.32 904.17 6413.54

Table 5. Simulated hydrological elements for the past three land-use maps (1990–2010).

Land-use

types

Rainfall

(mm)

Surface runoA

(mm)

Water yield

(mm)

Evapotranspiration

(mm)

Conservation

index

1990 1082.5 162.5 473.9 585.9 0.208

2000 1082.5 162.8 473.5 586.3 0.193

2010 1082.5 170.1 472.9 586.9 0.216

190 Page 8 of 11 J. Earth Syst. Sci. (2020) 129:190

Thus, using average daily rainfall depth as SWATinputs will lead to uncertainty in streamCow sim-ulation. Finally, considering that Cood prediction isnot the objective in this paper, this mismatch inpeak Cow can be ignored.

5.2 Land-use variation analysis

As can be seen from table 3, forest land, grass landand cultivated land are the main land-use types,especially forest land, accounting for 48.73% oftotal land area. According to the research by Xuet al. (2018), social development in southwestChina is low because of geographical and historicalreasons and economic development depends heav-ily on land resources (Dai et al. 2015). On the otherhand, land-use types change not significantly asa whole, but cultivated land decrease 17.34 km2

and construction land increase 8.89 km2 as well asgrassland increase 8.49 km2 for the period of1990–2010. Obviously, conversion among culti-vated land, construction land and grasslandreCects the demand of economic development andthe implementation of the soil and water conser-vation measures, which return cultivated land toforest land or grassland. These results match theresearch by Xu et al. (2018) and Water Conser-vancy Management of GuiZhou (Bureau of WaterConservancy Management of GuiZhou 2005),which demonstrates as ‘Since 1980s, large-scaleecological construction projects were implementedin Guizhou Province and the regional ecological

1 2 3 4 5 6 7 8 9 10 11 120

10

20

30

40

50

60

The

dept

h of

mon

thly

runo

ff /

mm

Month

1990 2000 2010

Figure 5. Average monthly surface runoA under the past three land-use maps (1990–2010).

Table 6. Simulated hydrological elements for future land-use scenarios (2020–2050).

Scenarios

Surface runoA

(mm)

Water yield

(mm)

Evapotranspiration

(mm)

Conservation

index

RunoA change

rate (%)

RL1 171.2 474.8 586.9 0.195 0.23

RL2 172.3 472.6 587.1 0.217 1.29

RL3 172.3 472.6 587.2 0.218 1.29

2010 170.1 472.9 586.9 0.216 –

Table 7. Simulated monthly surface runoA under future land-use scenarios (2020–2050) (Unit: mm).

Month RL1 RL2 RL3

Surface runoA in the dry seasons

11 2.6 2.7 2.7

12 0.7 0.7 0.7

1 0.3 0.3 0.3

2 0.6 0.6 0.6

3 1.0 1.0 1.0

4 2.1 2.1 2.1

Sum 7.4 7.4 7.4

Surface runoA in the rainy seasons

5 15.7 15.7 15.7

6 50.7 51.0 51.0

7 44.5 44.9 44.8

8 27.8 28.0 28.0

9 15.6 15.7 15.7

10 8.6 8.6 8.7

Sum 162.8 163.9 163.9

J. Earth Syst. Sci. (2020) 129:190 Page 9 of 11 190

environment has improved gradually.’ Therefore,future cultivated land can be assumed to decreaseby 13.84–14.23% for three land-use scenarios andconstruction land increase to 1046.11–1122.19 km2.Meanwhile, water expansion within catchmentboundary can be attributed to the operation ofhydro-power dam in September 2006, whichincrease water level.

5.3 Hydrological response to land-use change

Simulated annual surface runoA does not changegreatly as well as evapotranspiration and conser-vation index (table 7), owing to the undevelopedeconomic or unremarkable human activities (Daiet al. 2015). However, surface runoA in the Coodseasons increases slightly as construction landincreases for 2010 land-use map (50.64 mm in Juneand 44.51 mm in July), which possibly due to theincreased impervious region (Herman et al. 2009).Meanwhile, increased vegetation coverage (2010land-use map and the RL2/RL3 scenarios)strengthen soil inBltration and further increase soilwater content, which supplement water demandfor new hydrological process (Loch 2000). Obvi-ously, increased forestland shrinks surface runoAand weakens peak Cood in the rainy seasons andalleviates Cood disaster to a certain extent. Forrain water stored in soil, it can be used for watersupply in the dry seasons and enable a good eco-logical state. However, forestland consumes morewater (2000 and 2010 land-use maps) than grass-land and reduces available water resources.Therefore, it was necessary to Bgure out thedemand of forest land or grassland land and spareno eAort to keep a balanced nature.

6. Conclusion

This paper aimed to evaluate land-use variationcharacteristics and quantify its hydrologicalimpact in the Longtan basin. Some correspondingconclusions reached by this study included thefollowings:

(1) Forest land, grassland and cultivated landwere the main land-use types and cultivatedland decreased most (17.34 km2/10a). On thecontrary, construction land and grasslandincreased 8.89 and 8.49 km2 every decade,respectively. Meanwhile, land-use conversionmainly occurred among forest land, grassland

and cultivated land and transferred moreduring 2000–2010.

(2) Simulated surface runoA under the past andfuture land-use maps did not change signifi-cantly at annual and monthly scale andstreamCow concentrated in the rainy seasons(May–October). Increased forest land couldstrengthen the function of soil and waterconservation which can eAectively store redun-dant rainfall in the Cood seasons and furthersupply water demand in the dry seasons.Meanwhile, high vegetation cover will decreasesurface runoA and weaken Cood discharge toavoid the possible Cood disaster. However,forest land consumed more water, whichresulted in a decrease of available waterresources. Therefore, it was necessary to payattention to soil and water conservation mea-sures and conversion of cropland to forest andgrassland.

Analysis in this paper can Bll the gap in land-usechange variation and its hydrological response forthe study area. Furthermore, it can also make areference for similar watersheds.

Acknowledgements

The authors would like to thank the NationalNatural Science Foundation of China (Grant No.51969004), the Natural Science Foundation ofGuangxi Province (Grant No. 2017GXNSFAA198361) and the Innovation project of GuangxiGraduate Education (Grant No. YCBZ2018023).

Author statement

Guiyan Mo: Formal analysis, Project administration,Writing original draft; Yongxiang Zhang and YaHuang: Methodology; Chongxun Mo and QingYang:Review and editing.

References

Abbaspour K C, Yang J and Maximov I et al. 2007 Modellinghydrology and water quality in the pre-alpine/alpine Thurwatershed using SWAT; J. Hydrol. 333(2) 413–430.

Braud I, Breil P and Thollet F et al. 2013 Evidence of the impactof urbanization on the hydrological regime of a medium-sizedperiurban catchment in France; J. Hydrol. 485(4) 5–23.

Bureau of Water Conservancy Management of GuiZhou 2005The proclamation of soil and water loss in GuiZhou province.

190 Page 10 of 11 J. Earth Syst. Sci. (2020) 129:190

Chesworth W, Arbestain M C and Mac�ıas F et al. 2008ClassiBcation of soils: FAO; pp. 111–113.

Dai M, Zhang J, Wang L and Weici S U 2015 A review onimpacts of land use/land cover change on water resources inkarst areas; Ecol. Sci. 34(3) 189–196.

Defries R and Eshleman K N 2010 Land-use change andhydrologic processes: A major focus for the future; Hydrol.Process. 18(11) 2183–2186.

Gashaw T, Tulu T, Argaw M and Worqlul A W 2018Modeling the hydrological impacts of land use/land coverchanges in the Andassa watershed, Blue Nile Basin,Ethiopia; Sci. Total Environ. 619–620 1394.

Gong T T, Lei H M and Yang D W et al. 2014 EAects ofvegetation change on evapotranspiration in a semiaridshrubland of the Loess Plateau, China; Hydrol. Earth Syst.Sci. Discuss. 11(12) 13,571–13,605.

Hassen E E and Assen M 2018 Land use/cover dynamics andits drivers in Gelda catchment, Lake Tana watershed,Ethiopia; Environ. Syst. Res. 6(1) 4.

Herman C, Liu Z and Kim J 2009 The eAects of climate changeand urbanization on the runoA of the Rock Creek basinin the Portland metropolitan area, Oregon, USA; Hydrol.Process. 23(6) 805–815.

Huang F, Huang B and Qiu J 2018 Simulative study on impactof climate change on runoA within Beijiang River Basin;Water Resour. Hydropower Eng. 49(1) 23–28.

Huize Municipal People’s Government 2010 Overall plan forHuize county.

Isik S, Kalin L and Schoonover J E et al. 2013 Modeling eAectsof changing land use/cover on daily streamCow: AnArtiBcial Neural Network and curve number-based hybridapproach; J. Hydrol. 485(485) 103–112.

Karamage F, Zhang C and Fang X et al. 2017 Modelingrainfall-runoA response to land use and land cover change inRwanda (1990–2016); Water-Sui 9(3) 147.

Khalid K, Ali M F and Rahman N F A et al. 2016 Sensitivityanalysis in watershed model using SUFI-2 algorithm; Proc.Eng. 162 441–447.

Lei P, Shi J and Cai L 2007 Application of 3S techniques in thetype monitoring of Longtan reservoir’s land utilization;J. Mianyang Normal Univ. 26(5) 86–90.

Liston G E and Elder K 2006 A meteorological distributionsystem for high-resolution terrestrial modeling (MicroMet);J. Hydrometeorol. 7(2) 217–234.

Liu L 2015 Flood season staging operation research of Longtanreservoir based on change-point and fuzzy theory; GuangxiUniversity.

Loch R J 2000 EAects of vegetation cover on runoA and erosionunder simulated rain and overland Cow on a rehabilitatedsite on the Meandu Mine, Tarong, Queensland; Austr.J. Soil Res. 38(38) 299–312.

Lv L et al. 2014 RunoA simulation of Dongjiang River basinbased on the soil and water assessment tool; J. Nat. Res. 101746–1757.

Niu J and Sivakumar B 2013 Study of runoA response to landuse change in the East River basin in South China; Stoch.Environ. Res. Risk Assess. 28(4) 857–865.

Pizzolotto R and Brandmayr P 1996 An index to evaluatelandscape conservation state based on land-use patternanalysis and geographic; Infor. Syst. Tech. 11 37–44.

R�edei G P 2008 Extreme-value distribution theory; Springer,Dordrecht.

Shiferaw A 2011 Evaluating the land use and land coverdynamics in Borena Woreda of South Wollo Highlands,Ethiopia; J. Sustain. Dev. Africa 13(1) 69–109.

Silberstein R P, Aryal S K and Durrant J et al. 2012 Climatechange and runoA in south-western Australia; J. Hydrol.475(12) 441–455.

Wang G, Yu Z, Liu G and Lin C 2006 Impact of land-usechange on hydrological processes in the Maying Riverbasin, China; Sci. China 49(10) 1098–1110.

Wang W, Chen X, Shi P and Phajmvan G 2008 Detectingchanges in extreme precipitation and extreme streamCow inthe Dongjiang River Basin in southern China; Hydrol.Earth Syst. Sci. 12(1) 207–221.

Wang G, Yang H and Wang L et al. 2014 Using the SWATmodel to assess impacts of land use changes on runoAgeneration in headwaters; Hydrol. Process. 28(3) 1032–1042.

WangK, ZhangQ,ChenYD and SinghVP 2015 EAects of land-use/cover change on hydrological processes using a GIS/RS-based integrated hydrological model: Case study of the EastRiver,China; Int.Assoc. Sci.Hydrol.Bull.60(10) 1724–1738.

Weining Yi, Hui and Miao Autonomous County 2010 Generalplanning for Weining Yi, Hui and Miao AutonomousCounty (2018–2035).

Xingyi Municipal People’s Government 2006 Overall plan forland utilization (2006–2020).

Xu Y, Wang S and Bai X et al. 2018 RunoA response to climatechange and human activities in a typical karst watershed,SW China; Plos One 13(3) e193073.

Yang T, Shao Q and Hao Z C et al. 2010 Regional frequencyanalysis and spatio-temporal pattern characterization ofrainfall extremes in the Pearl River Basin, China; J. Hydrol.380(3) 386–405.

Yang M et al. 2017 Impact of climate change on water resourcein coastal areas of South China – A case study ofNanliujiang River catchment; J. Guangxi Univ. (NaturalScience Edition) 42(5) 1951–1959.

Yin J, He F, Jiu Xiong Y and Qiu G Y 2017 EAects of landuse/land cover and climate changes on surface runoA in asemi-humid and semi-arid transition zone in northwestChina; Hydrol. Earth Syst. Sci. 21 1–23.

Zhang L, Zhao F, Chen Y and Dixon R N M 2011 EstimatingeAects of plantation expansion and climate variability onstreamCow for catchments in Australia;Water Resour. Res.47(12) 12539.

Ziyun Municipal People’s Government 2010 Overall plan forland utilization (2013–2020).

Corresponding editor: SUBIMAL GHOSH

J. Earth Syst. Sci. (2020) 129:190 Page 11 of 11 190