Water Resour Manage DOI 10.1007/s11269-014-0749-1

Impacts of Land-use Changes on the Hydrology of the Grande River Basin Headwaters, Southeastern Brazil M. R. Viola & C. R. Mello & S. Beskow & L. D. Norton

Received: 3 August 2013 / Accepted: 2 July 2014 # Springer Science+Business Media Dordrecht 2014

Abstract Land-use changes affect soil water balance. The Upper Grande River Basin (UGRB) headwaters have undergone intense modifications in land use. This study was conducted to simulate, using the LASH model, the impacts on the hydrological regime in the UGRB with five land-use trends: S1 and S2 – reforestation with eucalyptus covering 20 % and 50 %, respectively, from the current grassland area; S3 – reforestation with eucalyptus covering 100 % of the current grassland area only in the sub-basins where this trend is predominant; S4 and S5 deforestation of 30 % and 70 % of the forest remnants in the Mantiqueira Range region for the cultivation of grasslands, respectively. Results demonstrate that runoff would be reduced due to the land-use changes by 51.65 mm yr−1, 110.29 mm yr−1 and 59.48 mm yr−1 for scenarios S1, S2 and S3, respectively. However, scenarios S4 and S5 could increase streamflow by 57.63 mm year−1 and 156.78 mm year−1, respectively. This indicates that land-use changes might make the basin more prone to flooding and other hazards associated with increased runoff. Keywords Land-use changes . Hydrological impacts . Simulation . Sensitivity analysis

1 Introduction Brazil is a continental country with a strong and growing economy, having water resources as the most important source for electric energy supply. Its demographic and socioeconomic development has led to a strong and growing demand for the production of food, fiber, wood, and raw material needed for various uses. Southern Minas Gerais is located in the Upper M. R. Viola : C. R. Mello Soil and Water Engineering Group, Federal University of Lavras, C.P. 3037, CEP 37200-000 Lavras, MG, Brazil S. Beskow (*) Center of Technological Development/Water Resources Engineering, Federal University of Pelotas, Pelotas, RS, Brazil e-mail: [email protected] L. D. Norton Purdue University, Agronomy Department, West Lafayette, IN, USA

M.R. Viola et al.

Grande River Basin (UGRB), which in turn is within the Paraná River Basin. This is the most important Brazilian basin in the context of electrical energy generation. Large reservoirs for hydroelectric power plants have been installed (14 reservoirs) since 1950’s in the UGRB. Among these reservoirs, the Furnas Reservoir has water storage of about 20 km3 which is being directly fed by basins located in UGRB. Approximately 90 % of the total energy consumed in Brazil is generated from hydraulic sources and 65 % of it is produced in the UGRB (Nóbrega et al. 2011). This type of energy generation depends on a linkage between hydrology and land use such that any alteration in the land use in basins can affect electrical energy production. Therefore, the hydrology response of this region is important for water resources management. Watershed plans has been developed in Brazil to assist in water resources management. These plans depend on the relationship between land use and water supply and this has been treated as a key point to enable future water availability and to define limits of land-use occupation by activities that can adversely affect water resources. Therefore, hydrological simulation of trend land-use scenarios can be considered the most viable tool. In studies performed in different parts of the world, hydrological models have been applied to evaluate the impact associated with land-use changes on different spatial scales (Ott and Uhlenbrook 2004; Stackelbert et al. 2007; van Ty et al. 2012; Du et al. 2013). However, in Brazil, these studies have taken into account large basins, such as Grande River Basin (Nóbrega et al. 2011), Uruguay River Basin (Collischonn et al. 2005) and Tocantins River Basin (Costa et al. 2003). An exception in Brazil was a research conducted by Beskow et al. (2013) applying the LASH model for a 32 km2 watershed to simulate land-use change. One of the main characteristics of tropical and subtropical ecosystems is the intra-annual variability of precipitation and this in conjunction with land-use can greatly affect the hydrological regime of a basin (Liu et al. 2014; Shi et al. 2013). Alterations in land-use that affect the UGRB will be reflected not only in local impacts but also throughout the hydroelectric power generation complex installed in the Paraná River Basin (Viola et al. 2013). The objective of this research was to assess the capability of LASH model in simulating possible hydrological impacts resulting from the main land-use change scenarios in the UGRB, in order to provide information for water resources management for a region that is highly strategic for Brazil in the hydro-energy production.

2 Materials and methods 2.1 Description of the Grande River Basin (UGRB) headwaters The studied area covers the Aiuruoca River Watershed (ARW), the Grande River Watershed – Madre de Deus (GRW-MD), the Sapucaí River Watershed (SRW), and the Verde River Watershed (VRW), with streamflow gauging stations named as Fazenda Laranjeiras (2,095 km2), Madre de Deus de Minas (2,080 km2), Careaçu (7,325 km2), and Três Corações (4,178 km2), respectively (Fig.1a, b). The climate in this region is Cwb (temperate with rainy and mild summers and dry and cold winters), with an average annual temperature of 17 °C and with minimum and maximum averages of 8 °C and 22 °C, respectively. The rainfall regime presents an average annual rainfall of 1,750 mm and more than 80 % of it is concentrated between November and March (Mello et al. 2012).

Impacts of Land-use Changes on the Hydrology of the Grande River

Fig 1 Location of the Grande River Basin (UGRB) in the Brazilian territory (a), highlighting the watersheds located in the UGRB headwaters and weather stations, rain gauges and water gauging level stations (b), and soil map of each one watershed (c)

The main soils of the ARW and GRW-MD are Cambisols, Latosols, Litolic Neosols, and Fluvic Neosols, with a predominance of the first soil type (Viola et al. 2013). The VRW and SRW are similar in this regard, but they feature a greater percentage of Latosols (Fig. 1c) (Viola et al. 2013). The soil map is useful for the hydrological modeling in terms of the attributes associated with the water balance and water infiltration, especially water storage capacity. Junqueira Junior et al. (2008) determined the soil water availability value (Am) for each soil class through field trials with soils sampled in the study region. For the Latosols, Am ranged from 3.15 to 3.50 mm cm−1; for Cambisols from 4.12 to 4.26 mm cm−1; for Litolic Neosols, 3.90 mm cm−1 and Fluvic Neosols, 3.90 mm cm−1. 2.2 Database Daily rainfall and streamflow data sets were obtained from Agência Nacional de Águas (ANA) for 40 rain gauges and 4 streamflow gauging stations, with historical series for the period from 1990 to 2003 (Fig. 1b). Meteorological data sets were provided by the Instituto Nacional de Meteorologia for stations of São Lourenço, Machado, and Lavras (Fig. 1b), with information on insolation, rainfall, pressure, humidity, wind direction and speed, and maximum, mean, and minimum air temperatures. The DEM was generated with the aid of ArcGIS (Environmental Systems Research Institute ESRI 2004), based on both a contour line map at the scale of 1:50,000 and points of known altitude acquired from Instituto Brasileiro de Geografia e Estatística (IBGE). The resulting DEM was used to numerically generate the drainage network and to delineate all the sub-watersheds for application of the LASH model. The land-use map was generated based on satellite imagery obtained by the Landsat ETM+ sensor, from 2005, using a supervised classification through the maximum likelihood classifier. The land-use map and the percentages of different land uses within the watersheds are presented in Fig. 2a. Land-use maps allow the identification of some vegetation related parameters that are required by the LASH model: leaf area index (LAI), rooting depth, albedo, surface resistance, and height. The corresponding values

M.R. Viola et al.

Fig 2 Map of the current land-use of the watersheds under study and percentage distribution of each land-use in the watersheds (a) and the projections for scenarios S1 (b), S2 (c), S3 (d), S4 (e), and S5 (f)

were extracted from literature and those parameters with considerable temporal variation had values modified over the year (Table 1). 2.3 Description of the LASH model The hydrological simulations were developed using the Lavras Simulation of Hydrology (LASH) model, which was originally designed by Mello et al. (2008b) and Viola (2008) and was improved by Beskow (2009), Beskow et al. (2011a), and Viola (2011). The LASH model is conceptually-based and is distributed either by sub-watersheds or by grid cells, and makes it possible to simulate the behavior of streamflow, evapotranspiration, and soil water storage. This model has been successfully applied at different watershed scales in the southern Minas Gerais (Beskow et al. 2011a, 2013; Viola et al. 2013). More details on the structure of the LASH model can be obtained in Viola et al. (2013), Beskow et al. (2011a) and Beskow (2009). Table 1 Vegetation-related parameters extracted from literature for each land-use category Land-use

LAI (m2.m−2)

Agricultural

0.3-7.0 (1)

Grassland Native Forest Savannah Eucalyptus Bare soils

1.86-3.99 6.25

(3)

1.9

(4)

3.5 0

(5)

(2)

Height (m)

Albedo

0-1.52 (5) 0.5 10

(6) (6)

5

(6)

5 0

(6)

Surface resistance (s.m−1)

Rooting depth (mm)

0.15-0.20 (7)

40 (6)

500 (10)

0.20-0.26

(7)

(6)

600 (11)

0.13-0.18

(5)

0.13-0.18

(5) (5)

0.13-0.18 0.10-0.35 (7)

70

100

(8)

2,000 (11)

150

(9)

2,000 (10)

(7)

1,500 (11) 500 (10)

100 545.3 (7)

Data from: Zhou et al. (2006)); 2 Manfron et al. (2003)); 3 Shuttleworth (1993)); 4 Collischonn et al. (2005)); 5 Mello et al. (2008b)); Marques Filho et al. (2005)); 7 Miranda et al. (1996); 8 Kuntschik (2004)); 9 Condé (1995); 10 Almeida and Soares (2003)); Beskow et al. (2013))

Impacts of Land-use Changes on the Hydrology of the Grande River

2.4 Calibration and validation With the objective of simulating the hydrological impacts resulting from the land-use scenarios projected for the region, the LASH model was calibrated, in accordance with procedures suggested by Viola et al. (2013), by means of continuous simulation of daily streamflows, using the period from 1990 to 1997, while 1998 to 2003 was the period used for the validation phase. Although longer data sets for simulation involving deterministic and distributed models are preferred, a 14-year period was employed in this study. Considering recent records, this was the longest period with availability of hydrological data with no gaps, thus motivating the use of data sets with less than 20 years of records. An automatic calibration method (Beskow et al. 2011b) was employed in order to seek representative values for the following parameters in the LASH model: hydraulic conductivity of the subsurface reservoir (KSS); hydraulic conductivity of the shallow saturated zone reservoir (KB); response time parameters (CS and CSS); and initial abstraction coefficient (λ). To evaluate the effectiveness of the calibration and validation, the following statistical coefficients were applied: a) Nash-Sutcliffe coefficient (CNS) and its logarithmic version (log (CNS)) (Nash and Sutcliffe 1970); and b) the difference between the observed and simulated streamflows (D) (Liew et al. 2003). 2.5 Scenarios for land-use changes A tool named as Generate Random Points, available in the Hawths Tools extension of ArcGIS (Environmental Systems Research Institute ESRI 2004), was used to delineate land-use scenarios. This tool allowed to randomly choosing a set of points, which had buffer offsets drawn such that the scenarios were exactly the same in area as previously defined for them. The setup of land-use scenarios was based on environmental demands imposed by the agencies responsible for water resources in Minas Gerais state, in an attempt to support both integrated water resources management and alterations in land use in the UGRB. Under this context, results and suggestions published in the Ecologicaleconomic zoning of Minas Gerais state (Mello et al. 2008a), Forest Inventory of Minas Gerais state (Carvalho and Scolforo 2008), and Water Resources Master Plan of the UGRB (Instituto Mineiro de Gestão das Águas/Consório ECOPLAN-LUMESKILL IGAM 2012), were taken into account. Rapid expansion of forests has been occurring in the watersheds, especially for eucalyptus plantation. This expansion aims at meeting the large market demand for raw materials primarily destined for the furniture, construction, charcoal and cellulose. Studies developed by Instituto Mineiro de Gestão das Águas/Consório ECOPLANLUME-SKILL IGAM (2012) in a UGRB watershed, with area of 2,329 km2, indicated that eucalyptus plantation corresponded to 52.37 km2 in 2008 and 79.27 km2 in 2011, representing an expansion of 0.385 % per year. These results originated scenarios S1 and S2, which consider alteration of reforestation by eucalyptus plantation for 20 % and 50 %, respectively, of the current grassland area in all the subwatersheds. Scenarios S1 and S2 represent a planning horizon of 30 and 70 years, respectively. This resulted in a mean alteration in total area of the four watersheds of 11.1 % for S1 and 27.8 % for S2. There have been intensive land-use alterations to eucalyptus plantations in some specific regions at rates greater than those above mentioned. Scenario S3 was established to take such information into account, replacing 100 % of the current grassland area only in the sub-watersheds in which this

M.R. Viola et al.

trend is predominant with eucalyptus plantation. Scenario S3 is supposed to be highly relevant to support water resources management in the UGRB, because hydrology practitioners will be able to indicate appropriate limits of land-use occupation with eucalyptus. The southern Minas Gerais is traditionally recognized for its dairy production. This activity, along with the production of beef cattle, stresses the environment due to the predominance of grasslands in the region. Currently, given the inexistence of new agricultural areas in the center-north region of the watersheds, the agricultural frontier has expanded towards the Mantiqueira Range region, mainly by the formation of pastures. Deforestation of Atlantic Forest in Mantiqueira Range region, which contains most of the existing native vegetation in the UGRB is occurring. Analyzing the sub-watersheds ARW and GRW-MD, Carvalho and Scolforo (2008) reported that native vegetation was reduced by 0.0573 % per year between 2005 and 2007. In this regard, scenario S4 was included for planning horizon of 100 years, with 30 % of the forest fragments converted to grasslands. In evaluating the four sub-watersheds, S4 culminated in an average increase of 5.9 % in the area occupied with grassland. A considerable economic pressure in the study region occurs due to the expansion of the agricultural frontier in the direction of the Mantiqueira Range’s springs (Carvalho and Scolforo 2008). This fact motivated the creation of S5 which corresponds to a scenario more drastic than S4 in terms of native vegetation removal, representing a reduction of 0.1432 % per year in the sub-watersheds for a planning horizon of 100 years. It corresponds to a deforestation of 70 % of the forest fragments to grasslands. On average, these watersheds had 13.8 % of the area with alteration in the land use. The projections for scenarios S1, S2, S3, S4, and S5 are presented in Figs. 2b, 2c, 2d, 2e, and 2 f. Table 2 summarizes alterations in the land-use (deforestation and reforestation), relative to the total drainage area, for each watershed. To evaluate the hydrological impacts, the streamflow series simulated by LASH model for the current scenario, with the period from 1997 to 2003 as reference. The impacts were then quantified by comparing the current scenario to the simulated behavior for the projected land-use scenarios. In order to evaluate the sensitivity and to understand the influence of the LASH model parameters that characterize the vegetation, a sensitivity analysis was performed (Collischonn et al. 2005; Beskow et al. 2011a and Mello et al. 2008b). The modified parameters were leaf area index (LAI), soil water storage capacity in the soil (Am), albedo, surface resistance (rs), and average height of the vegetation (hveg).

Table 2 Alterations in land use (reforestation) within each watershed, considering the percentage of area modified in relation to the total drainage area (%), for all the projected scenarios Scenario

Alteration

ARW

GRW-MD

SRW Km2

Km2

%

Km2

%

VRW %

Km2

%

S1

Reforestation

226.7

10.8

195.8

9.4

883.0

12.1

512.1

12.3

S2

Reforestation

566.8

27.1

489.5

23.5

2,207.4

30.1

1,280.1

30.6

S3 S4

Reforestation Deforestation

369.9 117

17.7 5.6

339.8 92.1

16.3 4.4

16.4 7.1

683.5 273.2

16.4 6.5

S5

Deforestation

273.1

13

214.9

10.3

16.7

637.6

15.3

1.203 523.4 1221.4

Impacts of Land-use Changes on the Hydrology of the Grande River

3 Results and discussion 3.1 Calibration and validation of the LASH model The results had CNS values greater than 0.70, thus qualifying it for hydrological simulation. Upon comparing the CNS results with the classification proposed by Moriasi et al. (2007), such results were classified as “very good” (CNS >0.65). The CNS found in the validation phase had high values also, especially in the SRW and VRW, whose CNS values were greater than 0.80. This behavior of the model denotes its capacity to simulate peakflows. Considering the log (CNS) coefficient and the classification proposed by Moriasi et al. (2007), the LASH model also had the capability of LASH to simulate flow declining periods. Additionally, the behavior of mean deviation was classified as very good (<10 %), good (10 %
M.R. Viola et al. Table 3 Observed and simulated annual water yield values and observed and simulated values with respect to hydrological series of minimum and maximum stream flow for the study watersheds Watershed

ARW

Year

Obs

GRW-MD

SRW

Sim

Obs

Sim

Obs

VRW Sim

Obs

Sim

Annual water yield (mm) 1998

556.36

459.16

615.07

567.31

487.82

500.9

434.31

408.11

1999 2000

642.71 780.33

634.89 673.92

739.76 762.77

711.51 684.91

560.15 615.31

663.81 506.69

587.87 789.34

587.73 716.87

2001

444.21

422.79

485.25

512.64

364.11

398.66

401.72

378.35

2002

635.69

554.76

591.93

601.51

487.47

521.24

575.4

533.27

2003

656.2

585.85

645.18

692.74

342.92

471.95

478.51

454.59

Minimum stream flow (m3 s−1) 1998

18.10

17.59

16.90

19.88

44.60

54.55

26.48

27.46

1999

9.22

15.83

13.70

16.81

35.65

39.38

26.87

20.16

2000 2001

18.72 14.14

17.32 14.13

19.30 13.32

16.07 15.20

43.51 29.45

37.31 34.77

34.13 26.48

23.33 20.06

2002

16.50

15.49

8.59

16.12

33.18

40.02

26.29

23.35

2003

14.39

15.85

14.04

18.11

31.02

39.65

28.04

23.27

Maximum stream flow (m3 s−1) 1998

125.00

72.13

129.90

155.12

305.80

282.69

186.60

143.23

1999

222.86

150.67

298.90

180.54

470.09

603.73

351.90

289.13

2000

414.52

271.52

232.86

233.75

1029.93

1102.19

553.12

771.00

2001 2002

117.49 171.60

70.63 106.70

111.00 144.47

98.60 139.27

267.02 365.98

236.20 359.43

210.80 319.00

171.53 295.84

2003

268.18

201.09

199.18

231.02

481.66

627.98

396.00

337.28

3.3 Hydrological impacts resulting from land-use change scenarios in UGRB 3.3.1 Water yield The reduction of the annual water yield for S1, S2 and S3 (Table 4) are in accordance with the conclusions obtained by Bosh and Hewlett (1982), Bruijnzeel (1988), and Sahin and Hall (1996) from a summary of experimental results. The average water yield reductions simulated for these scenarios are supported by the modification of the water balance when an alteration in the land-use from grassland to eucalyptus occurs. This reduction is associated with the following features: a) interception is greater for eucalyptus. For grassland, 1.86 m2.m−2 was the minimum LAI while a LAI of 3.5 m2.m−2 was constant for eucalyptus (Table 1). LAI exerts influence on losses by vegetation interception, thus it is expected greater amounts of vegetal interception for S1, S2 and S3 scenarios and consequently a reduction in runoff; b) the Am, in deep soils occupied by eucalyptus, is greater because its root system can reach greater depths (Table 1); c) evapotranspiration: these rates in eucalyptus areas are usually greater than those in other areas because albedo is lesser, resulting in greater energy being made available. Although albedo values for eucalyptus (0.13-0.18) have been lower than those for grassland (0.20-0.26), this parameter presented low influence on runoff simulated by LASH model. Finally, less

Impacts of Land-use Changes on the Hydrology of the Grande River

Fig 3 Sensitivity analysis results of the vegetation-related parameters using LASH model for Aiuruoca River Watershed (a), Grande River Watershed (b), Sapucaí River Watershed (c), and Verde River Watershed (d)

aerodynamic resistance associated with greater LAI results in a more perspiring surface area. Therefore, in areas originally composed of undergrowth vegetation, the effect of reforestation on the water balance is a reduction in streamflow.

M.R. Viola et al. Table 4 Average annual water yield (mm) between 1998 and 2003 for the current land-use (Dave) and the respective deviation (%), in relation to the current land-use for the scenarios S1 (ΔS1), S2 (ΔS2), S3 (ΔS3), S4 (ΔS4), and S5 (ΔS5) Watershed

Dave

ΔS1

ΔS2

ΔS3

ΔS4

ΔS5

ARW

555.2

−10.47

−17.63

−11.36

8.92

26.41

GRW-MD

628.4

−7.57

−16.97

−11.15

7.36

19.11

SRW

510.5

−10.72

−22.98

−10.41

13.90

37.74

VRW

513.2

−9.00

−23.25

−10.06

12.42

32.69

The results obtained by LASH model displayed a similar sign to that obtained by other scale hydrological models used in studies of hydrological impacts associated with reforestation (Santiago 2005; Beskow et al. 2013). The simulated hydrological response for scenarios S4 and S5 was an increase in the streamflow, mainly due to the reduction of evapotranspiration and rainfall interception, favoring the runoff components. Analyses of the historical series observed in experimental watersheds have demonstrated that after deforestation, an increase in the streamflow occurs (Chappell and Tych 2012) due to reduced rainfall interception (Chappell et al. 2001) and evapotranspiration (Chappell et al. 2004). Results obtained by hydrological simulation for watersheds with different drainage areas and climatic characteristics (Liu et al. 2014; Collischonn et al. 2005; Santiago 2005) conform with those presented in this study. 3.3.2 Minimum and maximum stream flows The simulated hydrological impact for the expansion of eucalyptus plantations indicated that the reduction, on average, of the minimum stream flows for the watersheds under study corresponded to 6.64 %, 15.29 %, and 8.35 % for scenarios S1, S2, and S3, respectively (Table 5). The ability of the soil to allow infiltration along with the depth of the root system affects the base flow. Forest land-cover has a deeper root system than grasslands, which allows water to be withdrawn from greater depths, affecting the availability of water to maintain base flow.

Table 5 Average annual minimum stream flows (Qmin) and maximum stream flows (Qmax) for the current scenario (in m3s−1), between 1998 and 2003, and the deviation for the scenarios S1 (ΔS1), S2 (ΔS2), S3 (ΔS3), S4 (ΔS4), and S5 (ΔS5), in %, for ARW, GRW-MD, SRW, and VRW ΔS2

ΔS3

ARW

16.03

−7.84

−12.62

−10.6

GRW-MD

17.03

−5.11

−11.17

SRW

40.95

−8.44

VRW Watershed

22.94

−5.17 ΔS1

Watershed

Qmin*

Qmax

ΔS1

ΔS4

ΔS5

9.99

26.03

−7.45

6.54

16.31

−19.01

−8.17

15.27

39.73

−18.35 ΔS2

−7.16 ΔS3

13.93 ΔS4

37.67 ΔS5 79.72

ARW

145.46

−11.54

−19.67

−13.97

17.29

GRW-MD

173.05

−8.12

−18.52

−6.17

4.47

11.07

SRW

535.37

−3.61

−20.76

−1.16

28.77

57.91

VRW

334.67

−24.96

−9.84

9.56

23.82

*

Current scenario

−18.8

Impacts of Land-use Changes on the Hydrology of the Grande River

The results of scenarios S1, S2, and S3 indicate a reduction of the annual maximum stream flows of 10.52 %, 20.98 %, and 7.79 %, on average, respectively. This finding may be explained by the increase in rainfall interception due to the increase in both LAI and Am. This behavior was observed in a study involving the LASH model conducted by Beskow et al. (2013) in a small watershed in Brazil, and also in studies carried out by Ott and Uhlenbrook (2004), and Hundecha and Bárdossy (2004). Results for the minimum and maximum streamflows (Table 5) corresponded to an increase of 11.43 % and 15.02 %, respectively, for scenario S4, and of 29.94 % and 43.13 %, respectively, for scenario S5. These results are supported by the idea of reduction in evapotranspiration, rainfall interception, and soil water storage, although the negative changes in the infiltration pattern that the substitution of the Atlantic Forest by grasslands causes in the hydrological conditions in the UGRB (Alvarenga et al. 2012). Overland and subsurface flows were found to be dependent on the soil moisture and to occur predominantly from December to March. In October, the first rainfall events tend to restore the soil moisture, and during the rainy period, greater overland flow and subsurface flow occur, reaching an apex between December and February. After this period, with decrease in rainfall, a reduction of these water balance components occurs, reaching minimum values during the driest season (June-September). At the end of the rainy season, the aquifer provides greater volumes of base flow, which is the most prevalent streamflow component during the dry season. This hydrological behavior of the region was adequately simulated by the LASH. The overland flow represented a considerable portion of the streamflow in the watersheds. The impacts on this component were reductions for scenarios S 1 (−12.92 %), S 2 (−27.11 %), and S3 (−13.00 %), and increases for scenarios S 4 (10.61 %) and S5 (31.19 %), on average. The subsurface flow represented a small portion of total streamflow simulated for the ARW (7.40 %), GRW-MD (10.33 %), SRW (5.14 %), and VRW (3.70 %). However, it displayed the greatest variations, which corresponded to −12.10 % (S1), −33.08 % (S2), −19.78 % (S3), 20.59 % (S4), and 59.94 % (S5), on average. Li et al. (2007) evaluated modifications of subsurface flow in two basins located in the West of Africa resulting from deforestation and this component represented 18 % of the streamflow under current land-use conditions and increased to 60 % under the deforestation scenario. These authors concluded that the majority of the water that penetrates the soil profile in the tropical forest is used in the transpiration process. The base flow represented a major portion of the streamflow in the watersheds: ARW (58.30 %), GRW-MD (51.62 %), SRW (52.53 %), and VRW (50.40 %). The hydrological impact on it, simulated by the LASH model, indicated that important modifications may exist in the recession periods for the land-use scenarios projected. For scenarios S1, S2, and S3, average reductions of 6.27 %, 13.28 %, and 7.43 % occurred, respectively, mainly due to the increase in evapotranspiration and rainfall interception. However, for scenarios S4 and S5, the results indicated an average increase in the base flow of 9.64 % and 24.40 %, respectively. Studies conducted by Hlavcová et al. (2009), Ott and Uhlenbrook (2004) and Beskow et al. (2013) produced similar findings obtained in this study, in which an reforestation scenario resulted in reduction of both overland flow and base flow.

4 Summary and Conclusions The LASH model simulated adequately different land-use scenarios in the UGRB. The evaluation of the simulated runoff components led to the conclusion that the model was also

M.R. Viola et al.

capable of reproducing the annual variability of the runoff for the four watersheds under study. The sensitivity analysis of the LASH model with respect to its vegetation parameters demonstrated that it was capable of adequately representing hydrological processes when land-use changes were simulated, resulting in an interdependency between simulated streamflow and Am, LAI, rs and hveg. All the statistical coefficients of precision enabled the model to be classified as high performing in all watersheds studied in the UGRB, with adequate representation of the hydrological cycle under present land-use possibilities. Despite having employed data sets of at most 14 years, which were limited by the availability of hydrological data for the study watersheds, readers are encouraged to apply data sets of over 20 years for studies similar to this. In scenarios S1, S2, and S3, which concern reforestation of the current grassland areas with eucalyptus, the simulation indicated, in average, a reduction of the water yield, minimum and maximum streamflows. Analyzing streamflow components, it was found a decrease in overland flow, sub-surface flow and base flow. In scenarios S4 and S5, the simulated hydrological response indicated an increase in the water yield, with an increase in minimum and maximum stream flows. The simulated hydrological response for these scenarios can be explained mainly due to the reduction of evapotranspiration and rainfall interception, favoring the horizontal components of the hydrological cycle. The results for the simulated flow demonstrated the existence of a strong interdependence between land-use and water resources management. The agricultural use of soil has direct consequences for the hydrology of the study watersheds, which may compromise the capacity to generate electric energy in the region or the hydrological regulation in the entire UGRB.

References Almeida AC, Soares JV (2003) Comparação entre uso de água em plantações de eucalipto grandis e floresta ombrófila densa (mata Atlântica) na costa leste do Brasil. Revista Árvore 27(2):159–170 Alvarenga CC, Mello CR, Mello JM, Viola MR (2012) Continuidade espacial da condutividade hidráulica saturada do solo na bacia hidrográfica do Alto Rio Grande, MG. Revista Brasileira de Ciência do Solo 35:914–922 Beskow S (2009) LASH model: a hydrological simulation toll in GIS framework. Federal University of Lavras, Dissertation Beskow S, Mello CR, Norton LD, Silva AM (2011a) Performance of a distributed semi-conceptual hydrological model under tropical watershed conditions. Catena 86:160–171 Beskow S, Mello CR, Norton LD (2011b) Development, sensitivity and uncertainty analysis of LASH model. Scientia Agricola 68:265–393 Beskow S, Norton LD, Mello CR (2013) Hydrological prediction in a tropical watershed dominated by oxisols using a distributed hydrological model. Water Resources Management 27(2):341–363 Bosh JM, Hewlett JD (1982) A review of catchment experiments to determine the effect of vegetation changes on water yield and evapotranspiration. Journal of Hydrology 55:3–23 Bruijnzeel LA (1988) (De)forestation and dry season flow in the tropics: a closer look. Journal of Tropical Forest 1:229–243 Carvalho LMT, Scolforo JRS (2008) Inventário Florestal de Minas Gerais - Monitoramento da Flora Nativa 2005–2007. Lavras: Editora UFLA v 1:357p Chappell NA, Tych W (2012) Identifying step change in single streamflow and evaporation records due to forest cover change. Hydrological Processes 26(1):100–116 Chappell NA, Bidin K, Tych W (2001) Modelling rainfall and canopy controls on net-precipitation beneath selectively-logged tropical forest. Plant Ecology 153:215–229 Chappell NA, Nik AR, Yusop Z, Tych W, Kasran B (2004) Spatially significant effects of selective tropical forestry on water, nutrient and sediment flows: a modelling-supported review. In: Bonell M, Bruijnzeel LA (eds) Forests, water and people in the humid tropics. Cambridge University, Cambridge, pp 513–532

Impacts of Land-use Changes on the Hydrology of the Grande River Collischonn W, Tucci CEM, Haas R, Andreolli I (2005) Forecasting River Uruguay flow using rainfall forecasts from a regional weather-prediction model. Journal of Hydrology 305:87–98 Condé RCC (1995) Fluxos de vapor d’água em um cerrado sensu scrito do Distrito Federal. University of Brasília, Dissertation Costa MH, Botta A, Cardille JA (2003) Effects of large-scale changes in land cover on the discharge of the Tocantins River, Southeastern Amazonia. Journal of Hydrology 283:206–217 Du J, Rui H, Zuo T, Li Q, Zheng D, Chen A, Xu Y, Xu CY (2013) Hydrological simulation by SWAT model with fixed and varied parametrization approaches under land use change. Water Resources Management 27: 2823–2838 Hlavcová K, Szolgay J, Kohnová S, Horvát O (2009) The limitations of assessing impacts of land use changes on runoff with a distributed hydrological model: case study of the Hron River. Biologia 64(3):589–593 Hundecha Y, Bárdossy A (2004) Modeling of the effect of land use changes on the runoff generation of a river basin through parameter regionalization of a watershed model. Journal of Hydrology 292:281–295 Environmental Systems Research Institute (ESRI) (2004) Redands, CA, 265p. Junqueira Junior JA, Silva AM, Mello CR, Pinto DBF (2008) Continuidade espacial de atributos físico-hídricos do solo em sub-bacia hidrográfica de cabeceira. Ciência e Agrotecnologia 32:914–922 Kuntschik G (2004) Estimativa de biomassa vegetal lenhosa em cerrado por meio de sensoriamento remoto óptico e de radar. University of São Paulo, Thesis Li KY, Coe MT, Ramankutty N, de Jong R (2007) Modeling the hydrological impact of land-use change in West Africa. Journal of Hydrology 337(3/4):258–268 Liew MW, Arnold JG, Garbrecht JD (2003) Hydrologic simulation on agricultural watersheds: Choosing between two models. Transactions of the ASAE 46:1539–1551 Liu J, Wang S, Li D (2014) The analysis of the impact of land-use changes on flood exposure of Wuhan in Yangtze river basin, China. Water Resources Management 28:2507–2522 Manfron PA, Dourado Neto D, Pereira AR, Bonnecarrére RAG, Medeiros SLP, Pilau FG (2003) Modelo do índice de área foliar da cultura do milho. Revista Brasileira de Agrometeorologia 11(2):333–342 Marques Filho AO, Dallarosa RG, Pachêco VB (2005) Radiação solar e distribuição vertical de área foliar em fl oresta : reserva biológica do Cuieiras, ZF2, Manaus. Acta Amazônica 35(4):427–436 Mello CR, Silva AM, Coelho G, Marques JJGSM, Campos CMM (2008a) Recursos Hídricos. In: Scolforo JRS, Carvalho LMT, Oliveira AD (Org.). Zoneamento ecológico-econômico do Estado de Minas Gerais: componentes geofísicos e biótico. Lavras, MG: Editora UFLA, 1:103–135. Mello CR, Viola MR, Norton LD, Silva AM, Weimar FA (2008b) Development and application of a simple hydrologic model simulation for a Brazilian headwater basin. Catena 75:235–247 Mello CR, Norton LD, Curi N, Yanagi SNM (2012) Sea surface temperature (SST) and rainfall erosivity in the Upper Grande River Basin, Southeast Brazil. Ciência e Agrotecnologia 36:53–59 Instituto Mineiro de Gestão das Águas/Consórcio ECOPLAN-LUME-SKILL (IGAM) (2012) Plano Diretor de Recursos Hídricos e Enquadramento de Corpos de Água da Bacia Hidrografia do Alto Rio Grande. Belo Horizonte, 465p. Miranda AC, Miranda HS, Lloyd J, Grace J, Mcyntire JA, Meir P, Riggan P, Lockwood R, Brass J (1996) Carbon dioxide fluxes over a cerrado sensu stricto in central Brazil. In: Gash JHC, Nobre CA, Roberts JM, Victoria RL (eds) Amazonian deforestation and climate. Wiley, New York, pp 353–364 Moriasi DN, Arnold JG, Liew MW, Binger RL, Harmel RD, Veith T (2007) Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the ASABE 50:885–900 Nash JE, Sutcliffe JV (1970) River flow forecasting through conceptual models Part I - A discussion of principles. Journal of Hydrology 10:282–290 Nóbrega MT, Collischonn W, Tucci CEM, Paz AR (2011) Uncertainty in climate change impacts on water resources in the Rio Grande Basin, Brazil. Hydrology and Earth System Sciences 15:585–595 Ott B, Uhlenbrook S (2004) Quantifying the impact of land-use changes at the event and seasonal scale using a process-oriented catchment model. Hydrology and Earth System Sciences 8:62–78 Sahin V, Hall MJ (1996) The effects of afforestation and deforestation on water yield. Journal of Hydrology 178: 293–309 Santiago AV (2005) Simulações dos efeitos da cobertura vegetal no balanço hídrico da bacia do Rio Ji-Paraná, RO. Dissertation, Escola Superior de Agricultura “Luiz de Queiroz”. Shi P, Ma X, Hou Y, Li Q, Zhang Z, Qu S, Chen C, Cai T, Fang X (2013) Effects of land-use and climate change on hydrological processes in the Upstream of Huai river, China. Water Resources Management 27:1263–1278 Shuttleworth WJ (1993) Evaporation. In: Maidment DR (ed) Handbook of hydrology. McGraw-Hill, New York, pp 4.1–4.53 Van Ty T, Sunada K, Ichikawa Y, Oishi S (2012) Scenario-based impact assessment of land use/cover and climate changes on water resources and demand: a case study in the Srepok river basin, Vietnam-Cambodia. Water Resources Management 26:1387–1407

M.R. Viola et al. Viola MR (2008) Simulação hidrológica na região Alto Rio Grande a montante do Reservatório de Camargos/ CEMIG. Federal University of Lavras, Thesis Viola MR (2011) Simulação hidrológica na cabeceira da bacia hidrográfica do Rio Grande de impactos de uso do solo e cenários de mudanças climáticas A1B. Federal University of Lavras, Dissertation Viola MR, Mello CR, Beskow S, Norton LD (2013) Applicability of the Lash Model for Hydrological Simulation of the Grande River Basin, Brazil. Journal of Hydrologic Engineering 18(12):1639–1652 Zhou MC, Ishidaira H, Hapuarachchi JM, Kiem AS, Takeuchi K (2006) Estimating potential evapotranspiration using Shuttleworth-Wallace model and NOAA-AVHRR NDVI data to feed a distributed hydrological model over the Mekong River basin. Journal of Hydrology 327(1/2):151–173