Evaluating the impacts of climate and land-use change on the hydrology and nutrient yield in a transboundary river basin: A case study in the 3S River Basin (Sekong, Sesan, and Srepok)

Science of the Total Environment 576 (2017) 586–598

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Evaluating the impacts of climate and land-use change on the hydrology and nutrient yield in a transboundary river basin: A case study in the 3S River Basin (Sekong, Sesan, and Srepok) Nguyen Thi Thuy Trang a, Sangam Shrestha a,⁎, Manish Shrestha a, Avishek Datta b, Akiyuki Kawasaki c a b c

Water Engineering and Management, School of Engineering and Technology, Asian Institute of Technology, Pathum Thani 12120, Thailand Agricultural Systems and Engineering, School of Environment, Resources and Development, Asian Institute of Technology, Pathum Thani 12120, Thailand Department of Civil Engineering, School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Climate change impact on hydrology and water quality of 3S basin was assessed. • Climate in the study area is projected to be warmer and wetter in future. • Discharge and nutrient loadings is projected to increase in wet season whereas it is projected to decrease in dry season. • Sesan and Srepok rivers of upstream of Vietnam is expected to contribute the highest nutrient loading in future.

a r t i c l e

i n f o

Article history: Received 19 March 2016 Received in revised form 6 October 2016 Accepted 18 October 2016 Available online xxxx Editor: D. Barcelo Keywords: Climate change Land-use change 3S River Basin Nutrient yield Hydrology Transboundary basin

⁎ Corresponding author. E-mail address: [email protected] (S. Shrestha).

http://dx.doi.org/10.1016/j.scitotenv.2016.10.138 0048-9697/© 2016 Elsevier B.V. All rights reserved.

a b s t r a c t Assessment of the climate and land-use change impacts on the hydrology and water quality of a river basin is important for the development and management of water resources in the future. The objective of this study was to examine the impact of climate and land-use change on the hydrological regime and nutrient yield from the 3S River Basin (Sekong, Srepok, and Sesan) into the 3S River system in Southeast Asia. The 3S Rivers are important tributaries of the Lower Mekong River, accounting for 16% of its annual flow. This transboundary basin supports the livelihoods of nearly 3.5 million people in the countries of Laos, Vietnam, and Cambodia. To reach a better understanding of the process and fate of pollution (nutrient yield) as well as the hydrological regime, the Soil and Water Assessment Tool (SWAT) was used to simulate water quality and discharge in the 3S River Basin. Future scenarios were developed for three future periods: 2030s (2015–2039), 2060s (2045–2069), and 2090s (2075–2099), using an ensemble of five GCMs (General Circulation Model) simulations: (HadGEM2-AO, CanESM2, IPSL-CM5A-LR, CNRM-CM5, and MPI-ESM-MR), driven by the climate projection for RCPs (Representative Concentration Pathways): RCP4.5 (medium emission) and RCP8.5 (high emission) scenarios, and two land-use change scenarios. The results indicated that the climate in the study area would generally become warmer and wetter under both emission scenarios. Discharge and nutrient yield is predicted to increase in the wet season and decrease in the dry. Overall, the annual discharge and nutrient yield is projected to increase throughout the twenty-first century, suggesting sensitivity in the 3S River Basin to climate and land-use change.

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

587

The results of this study can assist water resources managers and planners in developing water management strategies for uncertain climate change scenarios in the 3S River Basin. © 2016 Elsevier B.V. All rights reserved.

1. Introduction The 3S Rivers (Sekong, Srepok, and Sesan) are important transboundary tributaries of the Mekong River and flow through three countries, namely Laos, Vietnam, and Cambodia. This basin plays a significant role in the socio-economic development of the region. The 3S River Basin is located in the upstream of the Mekong River, and therefore the water quality and quantity from the 3S Basin outlet will directly impact on water use in the Mekong Delta. The Mekong Delta plays a vital role in food security and the Vietnamese economy, as the country is one of the largest producers and exporters of rice (Oryza sativa L.). Hence, it is vital to assure the quantity and quality of water to the Mekong Delta. Takamatsu et al. (2014) assessed the land-use change trends between 1993 and 1997 in the Mekong River 3S Sub-basins and reported a reduction of 470 km2 of grass/shrub and 303 km2 of forest, with an increase in agricultural area of 741 km2. It is also important to note that the net forest area (high and low-density) is reducing and its quality is degrading (Takamatsu et al., 2014). The Mekong River has been facing environmental degradation over the last 30 years due to multiple pressures such as rapid population growth, accelerated deforestation, industrialisation, and intensive agricultural development. Consequently, there has been a significant degradation in water quality from upstream to downstream in many parts of the basin (Campbell, 2007; Abebe et al., 2010; IUCN, 2014). Moreover, the water quality is also significantly affected by various anthropogenic activities including climate change (Tu, 2009), and the 3S Basin is one of the climate change hotspots in the Lower Mekong Basin (LMB) (Hartman and Carlucci, 2014). The impact of climate change on water has been widely studied in various parts of the world (Middelkoop et al., 2001; Pfister et al., 2004; Xia and Zhang, 2005; Xia and Zhang, 2008; Jun et al., 2010; Sanches Fernandes et al., 2012; Santos et al., 2014; Shrestha et al., 2016a, 2016b). However, most of these studies have focused on water quantity rather than quality. Water quality can be directly affected through several climate related mechanisms, in both the short and long-term (Tu, 2009; Jun et al., 2010). It has been reported that the amount of nutrients transported from land into the water bodies is a result of point-source emissions, atmospheric deposition, subsurface nutrients leaching from soil, and the biochemical deposition of the fresh water system. All of these factors, with the exception of point-source emissions, are strongly influenced by temperature and precipitation (Arheimer et al., 2005). Therefore, water quality is likely to be affected by global climate change. Changes in water quantity also modify its quality through the impact of dilution or the concentration of dissolved nutrients (Jun et al., 2010). The deterioration of water quality in both the main channel and tributaries of the LMB could have a significant negative impact on the sustainable livelihoods of local people and the economic development of the entire region. Wilbers et al. (2014) reported that poor water quality in the Mekong Delta can lead to severe health-related risks. It has been concluded that the lower part of the 3S Basin essentially suffers from the densely populated upstream Sesan and Srepok regions rather than pollution from the local sources (IUCN, 2014). Pollution from non-point sources, especially from agricultural activity, has become the most severe threat to water quality in recent years (Shen et al., 2014). Large amounts of phosphorus (P) and nitrogen (N) resulting from agricultural activity cause various problems such as toxic algal blooms, oxygen depletion, loss of biodiversity as well as degradation of aquatic ecosystems

and deterioration of water quality (Hong et al., 2012). Some studies on water quality relate to the 3S Basin; however, most of these focus on individual rivers of a country at local scale. Therefore, it is of utmost importance to study the water quality patterns of the whole basin in a holistic manner. In this context, the objectives of this study were to: (1) simulate the hydrology of the rivers of the 3S Basin; (2) develop the nutrient yield map of nitrogen (N) and phosphorus (P) in the 3S Basin from point sources and non-point sources of pollution; and (3) assess the impact of climate and land-use change on the hydrology and nutrient yield in the 3S River system. The results of this study can support policy makers and planners in proposing suitable management practices to cope with the impact of changing climate on water quantity and quality in the basin. 2. Materials and methods 2.1. Study area and data collection The Sekong, Sesan, and Srepok River Basin, also known as the 3S Basin, is located in the south-eastern part of the Mekong Basin (Fig. 1). The total drainage area of the basin is 78,529 km2, which accounts for 10% of the Mekong River Basin. The elevation of the 3S Basin ranges from 80 to 2040 m above sea level (masl). The 3S Basin covers nine provinces in three countries of the LMB of which two lie in Laos (Attapeu and Sekong), three in Cambodia (Mondulkiri, Ratanakiri, and Stung Treng), and four in Vietnam (Dak Lak, Kontum, Gia Lai, and Lam Dong). The Sesan and Srepok Rivers originate from the central highlands of Vietnam whereas the Sekong River stems from the Annamite Mountains in Laos. The Sesan and Srepok Rivers flow through Cambodia, joining the Sekong River, and finally confluence into the Mekong River at Stung Treng Province in Cambodia. The 3S Basin accounts for 16% of the total annual flow in the Mekong Delta (Räsänen et al. 2012). The physical characteristics of the 3S Basin are presented in Table 1. Fig. 2 shows the DEM, land-use, and soil maps collected from the Mekong River Commission (MRC). Five major types of land-use can be found in the 3S Basin. The largest land-use area is forest, followed by agriculture, grass and shrub, urban, and water resources. The area includes deciduous, evergreen, bamboo, flooded and coniferous forests covering an area of 61,712 km2, accounting for 78.76% of the whole basin. The agriculture land comprises swidden cultivation, rice, field crops, industrial crops, crop mosaic, miscellaneous land, mulberry land, plantains, and perennial land cover of 10,340 km2, which accounts for 13.2% of the entire 3S Basin. Rice is the dominant crop grown in the area, and covers 883 km2, accounting for 1.12% of the total watershed. The urban area occupies only 1075 km2 or 1.37% of the whole basin. Seven major types of soil, namely Acrisols, Cambisols, Ferralsols, Gleysols, Leptosols, Planosols, and Vertisols can be found in the 3S Basin. Such soils can stand alone or be mixed with each other to create 72 soil types. Table 2 shows a summary of the data and corresponding resolution, duration, and source. The basin consists of 16 rainfall stations, six temperature stations, and six stations measuring relative humidity, wind speed, and solar radiation. However, these stations are not homogeneously distributed, with the majority located in the Srepok Basin (Fig. 1). The discharge data was collected from nine stations distributing into the three sub-basins of 3S. The period of data collection from the hydrological stations varies from 1994 to 2008, depending on its availability. Out of the nine hydrological stations, three are located at Sekong,

588

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

r ive gR kon Me Fig. 1. Location map of the 3S Basin showing the hydro-meteorology and water quality monitoring stations.

Table 1 Physiographical characteristics of the 3S Rivers (Räsänen, 2012). Physiography

Sekong

Sesan

Srepok

3S

Total basin area (km2) Percentage of area (%) Average flow (m3/s) Length (km) Average precipitation (mm/year) Average temperature (°C) Average evaporation (mm/year)

28,765.7 36.6 1040 395

18,684.4 23.8 651 396

31,079 39.6 695 457

78,529 100 2386 1248 2270 22.5 1296

two at Sesan, two at Srepok, and two after the confluence of the three rivers. The water quality data for the period from October 2004 to December 2008 was collected from four stations measuring N and P concentrations. The locations of the hydrology and water quality monitoring stations are described in Table 3. 2.2. SWAT model description The SWAT model (Arnold et al., 1998; Bouraoui and Grizzett, 2014) was developed by the United States Department of Agriculture (USDA)

Fig. 2. DEM map, land-use map, and soil map of the 3S Basin.

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

589

Table 2 Summary of data used in the study. Duration

Source

Meteorology

Data

Description

1981–2008

Reservoir data

1994–2008 October 2004–December 2008 1999–2008

Mekong River Commission (MRC), Vietnam Hydro-Meteorological Institute MRC, Hydro-Meteorological Department MRC

Daily data of precipitation, maximum and minimum air temperatures, solar radiation, relative humidity, and wind speed Hydrology Daily discharge from nine stations + Observed water Concentrations of NO− 3 , NH4 and P from four water quality monitoring stations quality

DEM Land-use Soil map Demographic map

Area, conservation level, dead storage level, check water level, total capacity, and active capacity of two reservoirs Topography of the 3S Basin with a resolution of 250 m × 250 m Map of land-use and crop types Soil map and physical properties Population and density map

– 2003 – 2005

MRC MRC MRC MRC MRC

NO3−: nitrate; NH4+: ammonium; P: phosphorus.

and has been widely used to simulate the quantity and quality of surface water and groundwater. SWAT is a semi-distributed and processoriented hydrological model for use in the modelling of ungauged catchments, and prediction of relative impact scenarios (alternative input data) such as changes in management practices, climate and vegetation on water quality, quantity, land-use, or other variables. SWAT is widely used due to the fact that the tool is freely available and readily applicable through development of the geographic information system (GIS) interfaces, and for its easy linkage to sensitivity, calibration, and uncertainty analysis tools. The sub-basins which possess unique land-use/management/soil attributes are grouped together and known as a hydrological response unit (HRU) (Neitsch et al., 2011). For this study, the USDA-Natural Resource Conservation Service (NRCS) runoff curve number method for moisture condition II (CN2) was used to estimate surface runoff on a daily time step. A detailed description of the model and its components and sub-components can be found in Arnold et al. (1998), Neitsch et al. (2011), and Srinivasan et al. (1998). SWAT models transfer and internally cycle the major forms of N and P. Nutrients are simulated in two phases: (i) N processed in the soil and N transported from the soil to the water bodies. N in the soil has three major forms, namely organic N, the mineral form of N in soil colloids, and the mineral form of N in solutions. The adding and removing process occurs simultaneously. N may be added to soil by fertiliser, manure, fixation, and rain, whereas N is removed as a result of plant uptake, leaching, volatilisation, denitrification, and erosion (Neitsch et al., 2011); (ii) P is not mobile like N, it combines with certain ions to form insoluble compounds. The three main forms of P in mineral soils are organic P, insoluble P, and plant available P. The P is added to soil from

fertiliser, manure, and crop residue application. On the other hand, plant uptake and erosion are the two main processes for removing it from the soil (Neitsch et al., 2011). The process of water quality is influenced by runoff and streamflow. Therefore, the calibration of river discharge must be implemented foremost. River discharge calibration was implemented using SWATCalibration and the Uncertainty Program (CUP) model. The calibration process included sensitivity analysis to define the sensitive parameter with a significant impact on the flow regime. The daily data from five discharge stations for the period from 2000 to 2005 was used for calibration. The principle used in calibration by SWAT-CUP is trial and error. The model was run with various values of parameters and the results were then compared with the observed values by using graphic and statistical parameters such as NashSutcliffe efficiency (NSE), coefficient of determination (R2), and percent bias (PBIAS) as described in Eqs. (1)–(3). The calibrated SWAT model was then validated against the observed discharge for the period from 1994 to 1999 at five stations and for the period from 2006 to 2008 at two stations. Similarly, the observed water quality data from October 2004–2006 and 2007–2008 was used to calibrate and validate the model to simulate nutrient yield. Xn 

2 Q i −Q 0i NSE ¼ 1− X i¼1 2 n Q −Q i i i¼1

ð1Þ

n∑Q i Q 0i −∑Q i ∑Q 0i ffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R2 ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi       2 2 2 n ∑Q i 2 −ð∑Q i Þ  n ∑Q 0i − ∑Q 0i

ð2Þ

Table 3 Hydrology and water quality monitoring stations of the 3S Basin. Hydrological stations No

Code

Station name

Basin

Latitude

Longitude

1 2 3 4 5 6 7 8 9

430,103 430,105 430,106 440,101 440,201 450,101 451,305 14,501 13,901

Chantangoy Attapeu B. Veunkhene Ban Kamphun Kontum Lumphat Ban Don Stungtreng Paske

Sekong Sekong Sekong Sesan Sesan Srepok Srepok Mekong Mekong

13.56 14.81 14.81 13.53 14.34 13.55 12.85 13.55 15.12

106.06 106.84 106.78 106.05 108.01 106.53 107.78 106.02 105.80

Water quality stations No

Code

Station name

Basin

Observed parameter

Latitude

Longitude

1 2 3 4

H430102 H440402 H450101 H451303

Siem Pang Pleicu Lumphat Ban Don

Sekong Sesan Srepok Srepok

NO− 3 P + NO− 3 , NH4 P

14.12 13.95 13.55 12.91

106.39 107.47 106.53 107.74

590

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

Table 4 List of GCMs used in this study. GCMs

Resolution (long by lat)

Scenarios

CanESM2

2.80 × 2.80

RCP4.5RCP8.5 Canadian Centre for Climate Modelling and Analysis, Canada (http://www.cccma.ec.gc.ca/data/cgcm4/CanESM2/index.shtml) RCP4.5 Centre National de Recherches Meteorologiques, France (http://www.cnrm-game-meteo.fr/?lang=fr) RCP8.5 RCP2.6 Hadley Centre, UKMO (http://www.metoffice.gov.uk) RCP4.5 RCP6.0 RCP8.5 RCP2.6 Institute Pierre-Simon Laplace, France (https://www.ipsl.fr/en/) RCP4.5 RCP6.0 RCP8.5 RCP2.6 Max Planck Institute for Meteorology (http://www.mpimet.mpg.de/en/science/models/mpi-esm.html) RCP4.5 RCP8.5

0

0

CNRM-CM5

1.4 × 1.4

HadGEM2-AO

1.90 × 1.250

IPSL-CM5A-LR 3.750 × 1.90

1.90 × 1.90

MPI-ESM-MR

Origin

Xn  PBIAS ¼

 Q i −Q 0i  100 Xn ðQ i Þ i¼1

For precipitation, ð3Þ

i¼1

where, Qi = measured daily discharge, Q′i = simulated daily discharge, 0

Q i = average daily discharge for the observed period, Q i = average daily discharge for the simulated period, n = number of daily discharge values. 2.3. Perturbation method GCM-simulated data is too coarse to be applied directly at basin level. Therefore, the perturbation method of bias correction was applied for correcting the difference between the observed and GCM-simulated data. The method of perturbation has been employed in various hydrological studies (Kim and Kaluarachchi, 2009; Van Roosmalen et al., 2010; Boyer et al., 2010; Khoi and Suetsugi, 2012) because of its simplicity in generating a large range of sensible climate scenarios. There are two fundamental assumptions to this approach: (i) the GCM biases are the same in the baseline period and simulated period; and (ii) during the baseline and future periods the same values are assigned for temporal variability (daily to inter-annual) of the observed climate variables. The principle of creating the future climate data set is to find the difference between the historical and simulated data in the GCMs. Finally, that difference was used to correct the daily rainfall, maximum and minimum temperatures of the GCMs to create the future data set. The following equations were used in the perturbation method: For temperature, −T0GCMref ; T0fut ¼ T0obs C F k ¼ T0GCMfut i;k þ C F k k k k;i

C Fk ¼

P0GCMfut k P0GCMref k

; P0fut ¼ P0obs i;k  C F k k;i

ð5Þ

where CFk = monthly mean change factor at month k, T′GCMfut = k = GCM-simulated temperature for the future period at month k, T′GCMref k GCM-simulated temperature for the reference period at month k, T′fut k,i = future temperature at dayi and month k, T′obs i ,k = observed temperature = GCM-simulated precipitation for a future at dayi and month k, P′GCMfut k

Table 6 The SWAT flow sensitive parameters, their range and fitted value. Parameter

Description

Range

Fitted value

CN2 SOL_K SOL_Z CANMX ALPHA_BF GW_DELAY CH_K2

Initial SCS CN2 Saturated hydraulic conductivity Soil depth (mm) Maximum canopy storage Base flow alpha factor (days) Groundwater delay time (days) Channel effective hydraulic conductivity(mm/h) Manning's value for main channel Moist soil albedo Average slope length

−0.2–0.2 −0.3–0.3 −0.5–0.5 0–10 0.6–1.0 40–50 70–150

0.19 0.04 −0.01 8.93 0.86 49.61 146.61

5–10 0–0.25 30–150

5.98 0.13 143.08

CH_N2 SOL_ALB SLSUBBSN

ð4Þ Table 7 Performance of SWAT model for daily streamflow simulation. Station

Time period

Table 5 Future land-use change scenarios. ID

Description

FG1 Conversion of forest to grassland area at 1% per year FA1 Conversion of forest to agricultural land area at 1% per year

Type

Forest Grassland

Baseline area (km2)

Future area (km2) 2030s

2060s

2090s

51,037.72 20,415.08 7655.66 5103.77 3255.03 33,877.66 46,637.08 49,188.97

Chantangoy Attapeu B. Veunkhene Ban Kamphun Kontum Lumphat

Forest 51,037.72 20,415.09 7655.66 5103.77 Agriculture 12,158.55 42,781.18 55,540.61 58,092.45

Bandon 3S Basin outlet

V: 1994–1999 C: 2000–2005 V: 1994–1999 V: 1994–1999 C: 2000–2005 V: 1994–1999 C: 2000–2005 V: 2006–2008 C: 2000–2005 V: 1994–1999 C: 2000–2005 V: 2006–2008

Note: C: Calibration, V: Validation.

Calibration

Validation

R2

NSE

PBIAS

R2

NSE

PBIAS

– 0.56 – – 0.61

– 0.49 – – 0.45

– −9.36 – – −7.46

0.59 – 0.8 0.80

0.65 – 0.80 0.80

1.61 – 8.01 8.01

0.58

0.60

24.41

0.54

0.54

−6.90 0.58

0.57

−3.82

0.74

0.65

−8.30 0.77

0.80

−0.83

0.72

0.72

8.21 0.68

0.68

−2.86

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

period at month k, P′GCMfut = GCM-simulated precipitation for the referk ence period at month k, P′fut k,i = future precipitation at dayi and month k, Pobs i,k = observed temperature at dayi and month k.

591

The five GCMs used to study the impact of climate change are shown in Table 4, selected on the basis of previous studies conducted in Southeast Asia (Khoi et al., 2015) .

Fig. 3. Observed and simulated daily discharge hydrograph at (a) Ban Veunkhene [Sekong River, Laos], (b) Kontum [Sesan River, Vietnam], (c) Lumphat [Srepok, Cambodia], and (d) 3S Basin outlet.

592

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

2.4. Land-use change scenarios In this study the land-use/land cover change scenarios are constructed on the basis of the historical trend of land-use/land cover change. In the period from 1993 to 1997, 242 km2 and 511 km2 of forest area was converted to grassland and agricultural land, respectively (Takamatsu et al., 2014). This brings the change from forest to grassland to 1.2% and that from forest to agricultural land to 1.47% per year. Therefore, two land-use change scenarios were constructed as follows and described in Table 5: FG1: This scenario assumes that the forest area will be converted to grassland at a rate of 1% per year with other land-use remaining the same. FA1: This scenario is based on the forest area converting to agricultural land at a rate of 1% per year with other land-use remaining the same. 3. Results and discussion 3.1. Calibration and validation for streamflow The most sensitive parameters for flow, range, and fitness value are listed in Table 6. Results of the agreement between the daily simulated and observed flow during the calibration and validation periods, represented by the goodness-of-fit statistics R2, NSE, and PBIAS are presented in Table 7. According to the model performance classified by Moriasi et al. (2007), there was good agreement between the simulated and observed flow at the outlet and Bandon station. At the outlet, the agreement was shown by R2 = 0.72, NSE = 0.72, and PBIAS = 8.21% for the calibration period and R2 = 0.68, NSE = 0.68, and PBIAS = −2.86% for the validation period. In contrast, the model performance in Attapeu and Kontum was satisfactorily based on the R2 and NSE factors. This could be due to the lack of precipitation data for those areas. The values of PBIAS, representing the difference between the average simulated and the observed flow in all stations were in the very good class. The PBIAS values were within the range of ±25%, suggesting that the simulated and observed average flow matched very well. Although agreement between observation and the simulation was achieved, the streamflow peak was not a good fit and mostly underestimated for all stations (Fig. 3). This may have resulted from the lack of precipitation data. A further reason can be attributed to curve number (CN2), which was used to simulate surface runoff. The CN2 method assumes a unique relationship between cumulative rainfall and runoff in the same antecedent moisture conditions (Khoi et al., 2015). However, it was not the objective of this study to predict flood; therefore the mismatch in peak flow is not a significant concern. 3.2. Calibration and validation for water quality and developing nutrient yield map The range and fitted value of the sensitive parameters are presented in Table 8. The most sensitive parameters for the simulation of N were

Table 8 The range and fitted value of the sensitive parameters for water quality simulation in SWAT. Parameter

Description

SHALLST_N Initial concentration of nitrate in shallow aquifer (mg/L) NPERCO N percolation coefficient PHOSKD P soil partitioning coefficient PPERCO P percolation coefficient Initial nitrate in soil (mg/kg) SOL_NO3 SOL_ORGN Initial organic N in soil (mg/kg) SOL_ORGP Initial organic P in soil (mg/kg)

Range

Fitted value

0–3

0.32

0–1 100–200 10–18 0–5 1000–5000 1000–4000

0.93 103.72 13.10 4.93 3872.5 2301

SHALLST_N, NPERCO, SOL_NO3, and SOL_ORGN. Similarly, PHOSKD, PPERCO, and SOL_ORGP were found to be the most sensitive parameters for the simulation of P. The simulated and observed values were compared to confirm the appropriateness of the simulation. The statistical evaluation parameters of R2, NSE, and PBIAS were employed to assess satisfactory levels of simulation (Table 9). The model performed better in the simulations of + − NO− 3 and P than NH4 . For the NO3 simulation, the average values of R2 and NSE were 0.81 and 0.72 for the calibration period, and 0.76 and 0.71 for the validation period, respectively. The model performed very well in the P simulation with the average values of NSE and R2 greater than 0.8 in both calibration and validation phases and the value of PBIAS showed a good rating according to the classification of Moriasi 2 et al. (2007). In the case of NH+ 4 simulation, the values of R and NSE showed a good rating value during the calibration period, whereas the value of the NSE parameter was very low in the validation period. However, the trends of the simulated curve followed those of the measured values. The results of the observed and simulated nutrient yield at various stations are presented in Fig. 4. The relatively poor value of PBIAS between the simulated and observed nutrient values could be due to data limitations. SWAT nutrient calibration requires continuous daily data; however, only the monthly observed water quality data was available. Another limitation could be due to the complex nature of the watershed modelling of nutrient yield (Zhou et al., 2015). The nutrient routines in SWAT might not be adequate to describe the dynamic processes involved (Grunwald and Qi, 2006). However, the performance of the model can be considered satisfactory to simulate water quality with the values of R2 and NSE. The model can therefore be used to simulate the water quality and flow in the 3S Basin.

3.3. Development of nutrient yield map The spatial distribution of the average annual yield of total N and P in the 3S Basin is illustrated in Fig. 5. The average annual yield of N ranged from 1.88 to 345 kg/ha and P from 0.68 to 192.1 kg/ha. The analysis shows that the highest contribution of TN and TP came from the Vietnamese upstream of the Sesan and Srepok Rivers. These higher yields could be attributed to the intensive agricultural and urban areas in Vietnam. The annual total N and P loading was calculated as 732,674 t/year and 360,336 t/year, respectively. Fig. 6 shows the percentage amount of nutrient load from each land-use type in the basin. It can be seen that forest and agriculture are the two major sources of nutrient pollution, accounting for 70.86% N and 70.24% P and 11.74% N and 12.01% P, respectively, of the total nutrient load in the 3S Basin. However, in terms of yield, the urban area exported the highest pollutant load (Fig. 7). The N yield of the urban area was nearly three times, and agricultural land nearly twice the forest area. It can be summarised that human activity might play an important role in the occurrence of point and non-point sources of pollution in the 3S Basin.

Table 9 Performance of SWAT model in simulating monthly nitrate (NO3−), ammonium (NH4+), and phosphorus (P). Station

Parameter

Siem Pang Lumphat Lumphat Pleiku Bandon

NO− 3 NO− 3 NH+ 4 P P

Calibration (2004–2006)

Validation (2007–2008)

R2

NSE

PBIAS

R2

NSE

PBIAS

0.83 0.79 0.70 0.85 0.84

0.79 0.63 0.64 0.80 0.81

27.12 38.54 33.32 1.25 24.57

0.70 0.81 0.55 0.78 0.84

0.69 0.74 −1.51 0.61 0.60

−4.61 −21.1 25.46 −14.43 −28.31

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

Calibration

593

Validation

(a)

(b)

Calibration

Validation

(c)

Calibration

Validation

(d)

Calibration

Validation

(e)

Fig. 4. Comparison of observed and simulated monthly (a) NO3-N at Siem Pang station, (b) NO3-N at Lumphat station, (c) NH4-N at Lumphat station, (d) Total P [TP] at Plieku station, and (e) Total P [TP] at Bandon station.

594

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

Fig. 5. Spatial distribution of annual (a) total N [TN] and (b) total P [TP] yield from 2004 to 2008.

3.4. Climate and land-use change impact on hydrology and nutrient yield 3.4.1. Future climate scenario in the 3S Basin The future climate was depicted by temperature and rainfall parameters. The change in temperature and precipitation showed that the climate in the 3S Basin is projected to change. The perturbation method

Fig. 6. The contribution of N and P loading from each land-use type.

was employed to create future temperatures and rainfall from the five GCMs (HadGEM2-AO, CanESM2, IPSL-CM5A-LR, CNRM-CM5, and MPIESM-MR) under the two emission scenarios RCP4.5 and RCP8.5. The change in average temperature for the 2030s, 2060s, and 2090s compared to the baseline period is presented in Fig. 8. It can be seen that the temperature has an increasing trend in both scenarios for three considered periods in every month. The basin average temperature is projected to increase by 0.89 °C and 1.05 °C in the 2030s, 1.59 °C and 2.31 °C in the 2060s, and 1.97 °C and 3.81 °C in the 2090s under RCP4.5 and RCP8.5, respectively, compared to the baseline period. The highest change in temperature occurred during the dry season, lasting from November to April for both scenarios of climate change. The temperature increased most significantly under the scenario of RCP8.5 in the late twentieth century. The change in seasonal rainfall compared to the baseline period is shown in Fig. 9. Overall, the annual rainfall changes by 0.53% to 2.58% in the 2030s, 5.50% to 6.17% in the 2060s, and 4.35% to 5.89% in the 2090s. It has been observed that the annual rainfall increases more under scenario RCP4.5 than RCP8. The highest increase of rainfall was found in the 2060s under RCP4.5. Rainfall changes during the wet season (May to October) within a range from 2.80% to 1.50% for the 2030s, 6.75% to 7.39% for the 2060s, and 6.54% to 7.22% for the 2090s. In the dry season, rainfall changes from 1.72% to − 3.31% for the 2030s, from 3.86% to − 1.95% for the 2060s, and from 3.36% to

Fig. 7. N and P loading of each land-use type.

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

595

Fig. 8. Changes in seasonal average temperature for the 2030s, 2060s, and 2090s compared to the baseline period under (a) RCP4.5 and (b) RCP8.5 scenarios.

Fig. 9. Changes in seasonal rainfall for the 2030s, 2060s, and 2090s compared to the baseline period under (a) RCP4.5 and (b) RCP8.5 scenarios.

−7.00% for the 2090s under RCP4.5 and RCP8.5, respectively. Overall, under the climate change scenario RCP4.5, rainfall is expected to increase for the entire period. Rainfall is predicted to decrease during the dry season and increase during the wet season under the RCP8.5 scenario.

3.4.2. Impact on hydrology The annual discharge of the basin is expected to increase under both RCP 4.5 and RCP8.5 scenarios (Fig. 10). The average annual discharge is projected to increase by 10.7%, 14.8%, and 13.9% under RCP4.5 and 20.9%, 14%, and 10.8% under RCP8.5 scenario for the 2030s, 2060s, and 2090s, respectively. The discharge changes can be explained by increases in precipitation during the wet season. During the wet season, discharge can increase up to 26.9% during the 2060s. However, the basin may face a decrease in discharge during the dry season. The change of discharge in the dry season is − 7.7%, − 3.4%, and − 1.5% under the RCP4.5 scenario, and 1.7%, − 3.9%, and − 6.6% under the RCP8.5 scenario for the 2030s, 2060s, and 2090s, respectively. The combined impact of climate and land-use change shows as increase in discharge through the century (Table 10). Both land-use change scenarios exhibit the same results for the near future. However, a very minor change can be seen between the mid and far future land-

use scenarios. The highest increase in discharge of up to 20% can be seen for FG1 under RCP 8.5 scenario.

3.4.3. Impact on nutrient yield The N and P yield for the 2030s, 2060s, and 2090s under RCP4.5 and RCP8.5 was projected using the calibrated SWAT model (Table 11 and Table 12). In order to perceive the change of nutrient yield by climate change and land-use, other factors were kept constant. The annual nutrient yield is predicted to increase for all periods under the RCP8.5 scenario, with the exception of the 2030s. The highest increase in nutrient yield occurs in the late century under RCP8.5 during the wet season. The increase in yield can be directly associated with the increase in river discharge. Among the three rivers, the Srepork River contributes the highest amount of N and P to the basin, followed by the Sesan River since both rivers originate from intensely cultivated areas in Vietnam. More than 70% of the total nutrients come from these two rivers. The Sekong River contributes on average 40% of the total nutrients in the basin. Similarly, the contribution shared by the Sesan and Sekong Rivers is around 33% and 27%, respectively. Similar trends can be observed in the future under both climate change scenarios (Fig. 11). Since the land-use scenarios are generated to convert forest to agriculture or grassland, compared to other rivers,

Fig. 10. Changes in discharge at the 3S Basin outlet under (a) RCP 4.5 and (b) RCP 8.5 scenarios.

596

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

Table 10 Percentage (%) change in discharge under future climate change and land-use change scenarios.

2030s Annual Wet Season Dry Season 2060s Annual Wet Season Dry Season 2090s Annual Wet Season Dry Season

Baseline (m3/s)

Climate change

RCP4.5 RCP8.5 FG1 FG1 RCP4.5 RCP8.5

FA1 FA1 RCP4.5 RCP8.5

2970 4527

10.7 16.5

20.9 26.9

7.6 9.7

17.4 19.7

7.7 9.7

17.4 19.7

1413

−7.7

1.7

1.0

10.0

1.0

10.0

Climate change and land-use change

2970 4527

14.8 20.5

14.0 19.6

11.6 13.6

10.9 12.7

11.7 13.7

10.9 12.8

1413

−3.4

−3.9

5.3

5.0

5.4

5.1

2970 4527

13.9 18.7

10.8 16.3

10.8 11.9

7.9 9.6

10.9 12.0

8.0 9.6

1413

−1.5

−6.6

7.2

2.4

7.4

2.5

the origin of the Sekong River is covered by thick forest, and produces more nutrients in the future due to land-use change. The contribution of the Sekong River for N yield will increase from an average of 23% to 41% under FA1 scenarios and up to 55% under FG1. Similarly, for the P yield, the contribution of the Sekong River will increase from 31% to 43% under the FA1 scenario and up to 50% under FG1. Under the combined impact of climate change and land-use change, the nutrient yield of the basin will increase in the future. However, it can be seen that converting forest areas to agriculture will yield more nutrients for the rivers compared to land conversion from forest to grassland.

4. Conclusions In this study, the SWAT model was applied to assess the potential impact of climate change and land-use change on water quality and quantity of the 3S River Basin. The model was calibrated against eight hydrological stations and four water quality stations spread within the basin. The calibrated model was then used to estimate the discharge, nitrogen (N) and phosphorus (P) yield under climate change and land-use change scenarios by conversion of forest to agriculture (FA1), and

Table 11 Percentage (%) change in N yield under future climate change and land-use change scenarios. Baseline (tons)

Climate change

Table 12 Percentage (%) change in P yield under future climate change and land − use change scenarios.

Climate change and land-use change

RCP4.5 RCP8.5 FG1 RCP FG1 RCP FA1 RCP FA1 4.5 8.5 4.5 RCP8.5

2030s Annual 1,249,564 7.3 Wet Season 1,119,587 8.2 Dry Season 129,977 −0.2

−6.6 −4.8 −21.6

5.2 4.7 9.6

8.8 1.9 68.8

7.5 7.2 10.9

3.7 −3.6 66.8

2060s Annual 1,249,564 21.9 Wet Season 1,119,587 22.8 Dry Season 129,977 14.3

38.8 41.3 17.2

29.3 30.2 21.4

43.7 46.0 23.6

23.3 23.2 24.3

38.9 40.2 27.7

2090s Annual 1,249,564 28.5 Wet Season 1,119,587 27.8 Dry Season 129,977 34.8

67.3 71.1 34.6

40.6 40.8 38.9

69.0 73.1 33.7

32.1 30.6 45.3

65.3 67.9 43.4

Baseline (tons)

Climate change

Climate change and land−use change

RCP4.5

RCP8.5

FG1 RCP 4.5

FG1 RCP8.5

FA1 RCP 4.5

FA1 RCP8.5

2030s Annual Wet Season Dry Season

459,134 424,874 34,260

5.1 6.0 −5.9

−3.6 −3.5 −4.2

12.6 11.9 20.4

11.7 10.6 25.4

14.9 14.4 21.4

8.8 7.6 23.9

2060s Annual Wet Season Dry Season

459,134 424,874 34,260

17.4 18.0 10.3

30.6 32.2 10.5

32.0 31.8 34.3

43.8 44.6 33.5

29.3 28.6 38.1

42.1 42.5 37.9

2090s Annual Wet Season Dry Season

459,134 424,874 34,260

22.3 21.4 33.5

48.6 50.9 20.1

39.6 38.3 56.1

58.3 60.1 36.1

37.0 34.8 63.7

59.0 60.2 44.9

conversion of forest to grassland (FG1) for the 2030s, 2060s, and 2090s. Data from five GCMs were bias corrected to forecast the future climate condition of the 3S Basin. The analyses suggest that both temperature and rainfall are expected to increase in the future compared to the baseline period. The magnitude of increase varies and depends on the future time period and RCP scenarios. However, rainfall is predicted to decrease during the dry season under RCP8.5. With respect to change in temperature and rainfall, discharge and nutrient yield also alter significantly. The annual water availability of the basin is expected to increase in the future. However, discharge is expected to decrease during the dry season. The nutrient yield map of the basin reveals that the highest contribution of TN and TP comes from Vietnam, upstream of the Sesan and Srepok Rivers. These two rivers contribute more than 70% of the total nutrient yield. This can be attributed to the intensive agricultural and urban areas in Vietnam. Analyses also show that urban areas export the highest rate of nutrients, followed by agriculture. In summary, human activity might play an important role in the occurrence of point sources and non-point sources of pollution in the 3S Basin. The research further suggests that land-use change also has a significant impact on the water quantity and quality of the basin. However, both land-use change scenarios show similar changes in water quantity but their impact on water quality is diverse. Conversion of forest to agriculture (FA1) shows a higher nutrient yield compared to the conversion of forest to grassland (FG1) scenarios. It is also noted that in future the Sekong River will contribute the highest amount of nutrients compared to the Sesan and Srepok Rivers. The results of this study would be useful for understanding the potential impact of global climate change and regional land-use change on both water quantity and quality of the 3S Basin. With increasing discharge and nutrient yield in the future, proper adaptation strategies should be implemented to minimise the negative impact of climate change. Improvement in water quality in the 3S Basin presents a major challenge and strong support and participation from the people and governments of the three nations living in the basin is urgently needed to enhance the effectiveness of the existing monitoring programmes and future management plans for the basin.

Acknowledgements The authors would like to thank the Greater Mekong Subregion (GMS) for providing the fund required for the research. The authors would also like to acknowledge Dr. Dao Nguyen Khoi from the Centre of Water Management and Climate Change (WACC), Vietnam National University for facilitation this research work.

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

Phosphorous

FA1

CC and LUC

FG1

CC

Nitrogen

597

Fig. 11. Percentage contribution of (a) Total N (%) and (b) Total P (%) by Sekong, Sesan, and Srepok (CC: Climate change; LUC: Land-use change; FA1: Conversion of forest land into agriculture at 1% per year; FG1: Conversion of forest land into grassland at 1% per year).

References Abebe, B.A., Haq, S.M.A., Chien, V.Q., Arafat, B., 2010. The Challenges of Integrated Management of Mekong River Basin in Terms of People’s Livelihood. J. Water Resour. Prot. 2, 61–68. Arheimer, B., Andréasson, J., Fogelberg, S., Johnsson, H., Pers, C.B., Persson, K., 2005. Climate change impact on water quality: model results from southern Sweden. Ambio 34 (7):559–566. http://dx.doi.org/10.1579/0044-7447-34.7.559. Arnold, J.G., Srinivasan, R., Muttiah, R.S., Williams, J.R., 1998. Large area hydrologic modeling and assessment part I: model development. J. Am. Water Resour. Assoc. 34 (1): 73–89. http://dx.doi.org/10.1111/j.1752-1688.1998.tb05961.x. Bouraoui, F., Grizzett, B., 2014. Modelling mitigation options to reduce diffuse nitrogen water pollution from agriculture. Sci. Total Environ. 468–469:1267–1277. http://dx. doi.org/10.1016/j.scitotenv.2013.07.066. Boyer, C., Chaumont, D., Chartier, I., Roy, A.G., 2010. Impact of climate change on the hydrology of St. Lawrence tributaries. J. Hydrol. 384:65–83. http://dx.doi.org/10.1016/j. jhydrol.2010. 01.011. Campbell, I.C., 2007. Perceptions, data, and river management: lessons from the Mekong River. Water Resour. Res. 43:1–13. http://dx.doi.org/10.1029/2006WR005130. Grunwald, S., Qi, C., 2006. GIS-based water quality modeling in the Sandusky Watershed. J. Am. Water Resour. Assoc. 42 (4), 957–973. Hartman, P., Carlucci, K., 2014. Climate Change Impact and Adaptation Study for the Lower Mekong Basin. Key Final Results USAID Mekong Adaptation and Resilience to Climate Change (USAID Mekong ARCC). Thailand: Bangkok. http://icem.com.au/ wp-content/uploads/2014/02/Key-Results-10pgs-lowres.pdf. Hong, Q., Sun, Z., Chen, L., Liu, R., Shen, Z., 2012. Small-scale watershed extended method for non-point source pollution estimation in part of the Three Gorges Reservoir Region. Int. J. Environ. Sci. Technol. 9:595–604. http://dx.doi.org/10.1007/s13762-012-0094-y. IUCN, 2014. Water quality summary. http://www.3sbasin.org/knowledge/key-topics/ water-quality.html (Accessed 27 August 2014). Jun, X.I.A., Shubo, C., Xiuping, H.A.O., Rui, X.I.A., Xiaojie, L.I.U., 2010. Potential impacts and challenges of climate change on water quality and ecosystem: case studies in representative rivers in China. J. Resour. Ecol. 1:31–35. http://www.jorae.cn/fileup/PDF/ 2010010104.pdf.

Khoi, D.N., Suetsugi, T., 2012. Hydrologic response to climate change: a case study for the Be river catchment, Vietnam. J. Water Clim. Change 3 (3):107–124. http://dx.doi.org/ 10.2166/wcc.2012.035. Khoi, D.N., Thom, V.T., Linh, D.Q., Quang, C.N.X., Phi, H.L., 2015. Impact of climate change on water quality in the upper Dongnai river basin, Vietnam. Proceeding of the 36th, IAHR World Congress, 28 June-3 July 2015. The Netherland, The Hague. Kim, U., Kaluarachchi, J.J., 2009. Climate change impacts on water resources in the upper Blue Nile river basin, Ethiopia. J. Am. Water Resour. Assoc. 45 (6):1361–1378. http:// dx.doi.org/10.1111/j.1752-1688.2009.00369.x. Middelkoop, H., Daamen, K., Gellens, D., et al., 2001. Impact of climatic change on hydrological regimes and water resources management in the Rhine basin. Clim. Chang. 49, 105–128 (10.1023%2FA%3A1010784727448). Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Binger, R.L., Harmel, R.D., Veith, T.L., 2007. Model evaluation guidelines for systematic quantification of accuracy in watershed simulation. Trans. ASABE 50 (3), 885–900. Neitsch, S.L., Arnold, J.G., Kiniry, J.R., Williams, J.R., King, K.W., 2011. Soil and Water Assessment Tool Theoretical Documentation Version 2009. Texas Water Resources Institute Technical Report No. 406. Texas A&M University System, College Station, TX. http://swatmodel.tamu.edu/. Pfister, L., Kwadijk, J., Musy, A., et al., 2004. Climate change, land use change and runoff prediction in the Rhine-Meuse basins. River Res. Appl. 20:229–241. http://dx.doi. org/10.1002/rra.775. Räsänen, T.A., 2012. Baseline Hydrology of the 3S Basin: Sesan-Sekong-Srepok VMod Hydrological Modelling Report: Challenge Program on Water & Food Mekong Project MK3 “Optimizing the Management of a Cascade of Reservoirs at the Catchment Level”. ICEM-International Centre for Environmental Management, Hanoi, Vietnam. Räsänen, T.A., Koponen, J., Lauri, H., Kummu, M., 2012. Downstream hydrological impacts of hydropower development in the Upper Mekong Basin. Water Resour. Manag. 26 (11), 3495–3513. Sanches Fernandes, L.F., Seixas, F., Oliveira, P., Leitao, S., Moura, J., 2012. Climate Change impacts on nitrogen in a hydrographical basin in the northeast of Portugal. Fresenius Environ. Bull. 21 (12), 3643–3650. Santos, R.M.B., Sanches Fernandes, L.F., Moura, J.P., Pereira, M.G., Pacheco, F.A.L., 2014. The impact of climate change, human interference, scale and modeling uncertainties on

598

N.T.T. Trang et al. / Science of the Total Environment 576 (2017) 586–598

the estimation of aquifer properties and river flow components. J. Hydrol. 519, 1297–1314. Shen, Z., Qiu, J., Hong, Q., Chen, L., 2014. Simulation of spatial and temporal distributions of non-point source pollution load in the Three Gorges Reservoir Region. Sci. Total Environ. 493:138–146. http://dx.doi.org/10.1016/j.scitotenv.2014.05.109. Shrestha, S., Shrestha, M., Babel, M.S., 2016a. Assessment of climate change impact on water diversion strategies of Melamchi Water Supply Project in Nepal. Theor. Appl. Climatol. http://dx.doi.org/10.1007/s00704-015-1713-6. Shrestha, S., Shrestha, M., Babel, M.S., 2016b. Modelling the potential impacts of climate change on hydrology and water resources in the Indrawati River Basin, Nepal. Environ. Earth. Sci. 75:280. http://dx.doi.org/10.1007/s12665-015-5150-8. Srinivasan, R., Ramanarayanan, T.S., Arnold, J.G., Bednarz, S.T., 1998. Large area hydrologic modeling and assessment part II: model application. J. Am. Water Resour. Assoc. 34: 91–101. http://dx.doi.org/10.1111/j.1752-1688.1998.tb05962.x. Takamatsu, M., Kawasaki, A., Rogers, P.P., Malakie, J.L., 2014. Development of a land-use forecast tool for future water resources assessment: case study for the Mekong River 3S sub-basins. Sustain. Sci. 9, 157–172. Tu, J., 2009. Combined impact of climate and land use changes on streamflow and water quality in eastern Massachusetts, USA. J. Hydrol. 379:268–283. http://dx.doi.org/10. 1016/j.jhydrol.2009.10.009.

Van Roosmalen, L., Christensen, J.H., Butts, M.B., Jensen, K.H., Refsgaard, J.C., 2010. An inter-comparison of regional climate model data for hydrological impact studies in Denmark. J. Hydrol. 380 (3–4):406–419. http://dx.doi.org/10.1016/j.jhydrol.2009.11. 014. Wilbers, G., Becker, M., Sebesvari, Z., Renaud, F.G., 2014. Spatial and temporal variability of surface water pollution in the Mekong Delta, Vietnam. Sci. Total Environ. 485–486: 653–665. http://dx.doi.org/10.1016/j.scitotenv.2014.03.049. Xia, J., Zhang, Y.Y., 2008. Water security in North China and countermeasure to climate change and human activities. Phys. Chem. Earth 33:359–363. http://dx.doi.org/10. 1016/j.pce.2008. 02.009. Xia, J., Zhang, L., 2005. Climate change and water resources security in North China. In: Wagener, T. (Ed.), Regional Hydrological Impacts of Climatic Change: Impact Assessment and Decision Making. IAHS Publication, Wallingford, pp. 167–173. Zhou, X.V., Clark, C.D., Nair, S.S., Hawkins, S.A., Lambert, D.M., 2015. Environmental and economic analysis of using SWAT to simulate the effects of switchgrass production on water quality in an impaired watershed. Agric. Water Manag. 160:1–13. http:// dx.doi.org/10.1016/j. agwat.2015.06.018.