Spatial analysis of water quality impact assessment of highway projects in mountainous areas remains largely unexplored. A methodology is presented here for Spatial Water Quality Impact Assessment (SWQIA) due to highway-broadening-induced vehicular traffic change in the East district of Sikkim. Pollution load of the highway runoff was estimated using an Average Annual Daily Traffic-Based Empirical model in combination with mass balance model to predict pollution in the rivers within the study area. Spatial interpolation and overlay analysis were used for impact mapping. Analytic Hierarchy Process- Based Water Quality Status Index was used to prepare a composite impact map. Model validation criteria, cross-validation criteria, and spatial explicit sensitivity analysis show that the SWQIA model is robust. The study shows that vehicular traffic is a significant contributor to water pollution in the study area. The model is catering specifically to impact analysis of the concerned project. It can be an aid for decision support system for the project stakeholders. The applicability of SWQIA model needs to be explored and validated in the context of a larger set of water quality parameters and project scenarios at a greater spatial scale. Keywords Analytic hierarchy process · Environmental impact assessment · Geographic information systems · Sensitivity analysis · Water pollution · Highway Introduction Highways are essential for the development and security of a region. However, understanding the detrimental effects of highway projects is pivotal for environmentally appro- priate decision making. Environmental impact assessment Electronic supplementary material The online version of this (EIA) involves the assessment of impacts of a development article (https ://doi.org/10.1007/s1320 1-018-0699-5) contains project on the environmental attributes, including water supplementary material, which is available to authorized users. resources (Barthwal 2012; Canter 1995). Conventional EIA * Polash Banerjee can be time consuming, expensive, and subjective (Glas- firstname.lastname@example.org son et al. 2005; Takyi 2012). Moreover, conventional EIA Mrinal Kanti Ghose focuses mainly on the temporal aspect of the impacts and email@example.com undermines the importance of their spatial distribution. Ratika Pradhan Geographic information systems (GIS) can overcome these firstname.lastname@example.org limitations and provide an unbiased and easily interpretable 1 EIA (Agrawal 2005). Department of Computer Science and Engineering, SMIT, A highway is essentially a non-point source of water Sikkim Manipal University, Majitar, Sikkim 737136, India 2 pollution. Construction and post-construction conditions Department of Computer Applications, Sikkim University, of a highway generate pollutants, which degrade the water Gangtok, Sikkim 737102, India 3 quality and affect the habitat of the nearby water bodies Department of Computer Applications, SMIT, Sikkim (Wu et al. 1998). Highway runoff contains relatively high Manipal University, Majitar, Sikkim 737136, India Vol.:(0123456789) 1 3 72 Page 2 of 17 Applied Water Science (2018) 8:72 concentration of pollutants as compared to the adjacent studies do not provide a clear methodology to perform Spa- river (USEP 1996; Bingham et al. 2002; Gan et al. 2008). tial Water Quality Impact Assessment (SWQIA). In addi- Statistical models suggest that traffic volume, rainfall char - tion, the reliability of these models is arguable as they lack acteristics, highway pavement type, and properties of pollut- appropriate model validation criteria. Moreover, none of ants and seasons are important determinants of road runoff them have discussed the importance of individual water pol- composition (Aldheimer and Bennerstedt 2003; Forsyth lutants on the overall value of a spatial water quality score et al. 2006; Granato 2013; Kayhanian et al. 2003; Kim et al. or index. Spatial analysis of water quality usually involves 2006; Li and Barrett 2008; Pagotto et al. 2000; Tong and interpolation of individual water pollutants in the study area, Chen 2002; USEP 1996; Yannopoulos et al. 2004, 2013). followed by combining their thematic maps using appropri- These models mostly cater to highway projects of devel- ate water quality index (Gajendra 2011; Yan et al. 2015; oped countries. Applicability of these models in develop- Zhou et al. 2007). Studies show that kriging is an effective ing countries remains largely unexplored. Depending upon interpolation method (Fallahzadeh et al. 2016; Ostovari et al. the nature of the highway runoff study, traffic volume can 2012; Sadat-Noori et al. 2014). be considered in two broad ways, namely, as average daily Spatial Explicit Sensitivity Analysis (SESA) is progres- traffic and vehicles during storm. Average daily traffic is a sively becoming an essential component of spatial model good predictor of water pollutants like Chemical Oxygen validation. It is the measurement of variation in the model Demand (COD), Total Suspended Solids (TSS), and Zinc outputs explained by the variation in the model inputs (Chen (Zn), while it poorly predicts the levels of lead, copper, and et al. 2010, 2011; Crosetto et al. 2000; Feizizadeh et al. oil and grease (Thomson et al. 1997; Venner 2004). In con- 2014; Lilburne and Tarantola 2009; Qi et al. 2013; Xu and trast, oil and grease has a significant relationship with the Zhang 2013). The outputs of a robust spatial model show vehicles during storm rather than average daily traffic (Sten- marginal perturbations to changes in the model inputs. How- strom et al. 1982; Venner 2004). Correlation studies of river ever, the computational cost associated with SESA is a major water have shown COD is strongly correlated topercent Dis- constraint for its inclusion in spatial analysis. solved Organic Carbon, Dissolved Oxygen, and Total Dis- The aim of this study is to address the lack of appropri- solved Solids (TDS). Moreover, TSS is strongly associated ate understanding and methods to assess the spatiotempo- with pH and TDS (Bhandari and Nayal 2008; Waziri and ral impact of highway construction on water quality in a Ogugbuaja 2010). developing country. The SWQIA model performs a project- A wide variety of water quality indices are used in the specific spatiotemporal assessment of traffic-induced water impact assessment of a highway project. However, except for pollution in East Sikkim. With further validation, in terms Water Quality Status Index (WQSI), all other water indices of a wider study area and comprehensive water quality involve predetermined weight or the importance of water parameters, it can be used in developing countries to assess pollutants, which cannot be manipulated to see the impact impacts of highway construction on water quality. It acts of the change in water pollutant weight on the overall impact as a geovisualization and temporal extrapolation tool for on water quality. (Karbassi et al. 2011; Mushtaq et al. 2015; traffic-induced water pollution. Moreover, SWQIA model Li et al. 2009; Yan et al. 2015). The weight of a pollutant can capture people’s perception of the project impacts. The can be determined using Multi-criteria Decision-Making results of the model are encouraging and show that it is a (MCDM) methods like Delphi Method and Analytic Hier- robust model with good prediction capability. archy Process (AHP) that use expert opinion-based criteria weighing (Mushtaq et al. 2015; Kumar and Alappat 2009). AHP decomposes the decision process into several levels Materials and methods of hierarchy. Based on a pairwise comparison of criteria for alternatives, a comparison matrix is made for the evalua- Study area tion of criteria weights (Saaty 1980, 1990; Saaty and Vargas 1994). The data requirement for AHP is less data-intensive The study area stretches from Rangpo (27 10′31.26″ N, ° ° than classic statistical methods, which are based on histori- 88 31′44.43″E, Elevation 300 m) to Ranipool (27 17′28.74″ cal data (Arriaza and Nekhay 2008). Use of AHP-based N, 88 35′31.11″E, Elevation 847 m) in the East district of weighing of pollutants helps in giving due importance to Sikkim, a stretch of 27 km along the route of NH 10 high- the characteristics and conditions of the study area, which way. It is the main route which connects Sikkim with the rest may not be reflected in non-expert opinion-based criteria of India. In 2008–09, the broadening of NH 10 had com- weighing (Karbassi et al. 2011). menced to promote defence and economic growth in Sik- Limited literature is available on GIS-based water qual- kim. The highway has been broadened from its present width ity impact assessment of highway projects (Agrawal 2005; of 7–12 m. This broadening of the highway will cause an Banerjee and Ghose 2016; Brown and Affum 2002). These increase in traffic volume. The project stretches from Sevok 1 3 Applied Water Science (2018) 8:72 Page 3 of 17 72 in West Bengal to North Sikkim. However, the road cor- scenario, and 2039 as post-project scenario (Fig. 3). The ridor chosen for the study is relatively much smaller than changes considered from ‘pre-project’ to ‘project imple- the actual stretch of the highway because of its relatively mentation’ scenarios included changes in AADT and LULC. homogenous geography. A drainage area of 147 km was While only change in AADT was considered for ‘post-pro- delineated to include all the micro-catchments providing ject year’ scenario. AADT for ‘post-project year’ scenario runoff to the highway/rivers (Machado et al. 2017; Siqueira was calculated based on annual growth rates for traffic, pro- et al. 2017). Furthermore, the project impact area of 7.4 km vided by Border Roads Organization. LULC in ‘pre-project’ was demarcated by merging 50 m buffers around the rivers and ‘project implementation’ scenarios was estimated using and the highway. The rationale of considering the project satellite images, whereas such an estimation was not pos- impact area was based on the accessibility of the river water/ sible for ‘post-project’ scenario. Five water pollutants were road runoff by the wildlife and humans living near the river/ considered for the study (Table 1) mainly based on the abil- highway (Antunes et al. 2001; Geneletti 2004). The study ity of the empirical model to predict their concentration in area has steep elevations, which is predominated by subtrop- the road runoff, and second, on the availability of a com- ical vegetation, interspersed with small human habitations, plete data set of historical water quality of the rivers near traditional farming areas, and towns like Rangpo, Singtam, the highway. Keeping in view of the ecological and cultural and Ranipool. The highway closely follows river Teesta sensitivity of the local water bodies, drinking water quality and Rani Khola (Fig. 1). It is also worth noting that Sikkim standards of US Public Health Service (USPH 1962) were has high biodiversity and it is home to a large number of considered, except pH, where Bureau of Indian Standards endemic species (Arrawatia and Tambe 2011). Moreover, (BIS 2012) standard was considered for the present study. it has a unique culture which gives high value to its natural Various inputs for SWQIA model were prepared, as men- resources. Therefore, unabated water pollution can severely tioned in Table 2. affect the ecological and cultural sanctity of this area. AHP model Data collection A structured questionnaire on pairwise comparison of water Based on the changes in Annual Average Daily Traffic pollutants and project alternatives was administered to a panel (AADT) and landuse & landcover (LULC), three time of experts, based on a numerical scale having values ranging frames were considered for the study, viz. the year 2004 from 1 to 9 as suggested by Saaty (2000) (Table 3). The expert as pre-project scenario, 2014 as project implementation choice software was used for the preparation of comparison Fig. 1 Study area 1 3 72 Page 4 of 17 Applied Water Science (2018) 8:72 1 3 Table 1 Description of water quality parameters Name Meaning Relation with road runoff Environmental impact Chemical oxygen demand (COD) Indicative measure of amount of organic matter Oil, dirt and grease generated by vehicles and A high COD causes, anaerobic condition in the present in the water areas close to the road water body, suffocation of aquatic organisms and loss of biodiversity pH Indicative measure of acidity or alkalinity of water Release of oxides of Sulphur and Nitrogen from At low pH, bioavailability of heavy metals causes incomplete combustion of fuel from vehicles metal toxicity of aquatic organisms. In contrast, a high pH causes loss of aquatic organisms due to ammonia toxicity Total dissolved solids (TDS) Aggregate measure of organic and inorganic ions Dissolved materials generated from traffic, rainfall A high concentration in water causes hardness, present in water and construction sites settle down on the road toxicity and eutrophication Total suspended solids (TSS) Aggregate measure of sediment particles sus- Atmospheric dust fall, dust carried by vehicles A high TDS causes, turbidity or cloudiness of water. pended in water from construction sites, untarred roads and dam- This hampers aquatic photosynthesis and respira- aged fraction of the road tion of aquatic organisms Heavy metals (Zn) Relatively dense metals or metalloids, example Generated as traces from fuel combustion, galvani- Biomagnification of heavy metals like zinc in food Zinc zation of steel, preparation of negative plates in chain. At high concentration, zinc interrupts the electric batteries and vulcanization of rubber. A normal metabolism of aquatic organisms and strong correlation has been found between aver- causes birth defects age daily traffic and zinc concentration in road runoff (Venner 2004) Doamekpor et al. (2016) Applied Water Science (2018) 8:72 Page 5 of 17 72 Table 2 Data types, source, and processing method for SWQIA model Data type Data source Data processing AADT BRO, MoRTH Direct input DEM bhuvan.nrsc.gov.in Direct input LULC Sikkim State Remote Sensing Applications Centre Supervised image classification Meteorological parameters Rahman et al. (2012) Direct input Slope Percent DEM from bhuvan.nrsc.gov.in ArcGIS Spatial Analyst Soil Texture CISMHE (2008b) Georeferencing, vectorization, rasterization Water quality profile of Teesta and Rani CISMHE (2008a) Direct input Khola rivers for year 2004 Water quality profile of Teesta and Rani Bhutia (2015), Gurung et al. (2015) Direct input Khola rivers for year 2014 Courtesy: MoRTH, Traffic data of NH 31A (nearest town, Singtam) collected on July 2004 and December 2004 by Ministry of Road Trans- port and Highways, Govt. of India. Downloaded from http://morth .nic.in/write readd ata/subli nkima ges/sikki m2987 77224 7.htm accessed on 18/05/2014. BRO, Traffic data of NH 31A (nearest town, Rangpo) collected from 29/06/2012 to 06/07/2012 by Border Roads Organization, Govt. of India. The AADT was projected for 2014 and 2039 based on annual growth rates (in percent) for traffic estimated by BRO LISS III accessed from http://bhuva n.nrsc.gov.in/data/downl oad/index .php under Resourcesat-I satellite image on 18/12/2014 Table 3 Importance scale used Scale of importance Description in AHP 1 Both decision elements are equally important 3 First element is slightly more influenced than the second 5 First element is stronger than the second 7 First element is significantly stronger than the second 9 First element is extremely significant than second 2,4,6,8 Judgement values between equally, slightly, strongly, very strongly and extremely Reciprocals When the ith criterion is compared to the jth criterion, a , then 1∕a is the judge- ij ij ment value when the jth criterion is compared with the ith, i.e., a = 1∕a . ij ij matrix and calculation of the weight of the pollutants. Two where A is the comparison matrix. The elements of w must project alternatives were considered for the AHP model, viz. fulfill the condition, w = 1 , and under ideal condition, i=1 ‘with project’ and ‘without project’, for comparison of the impacts. The ‘with project’ alternative assumed that the high- = n . The reliability of the AHP model is assessed by max consistency ratio, CR = CI∕RI , where Consistency Index, way had been broadened and traffic volume had increased, while ‘without project’ alternative assumed no change in the CI =( − n)∕(n − 1) , and Random Consistency Index, max RI, that is obtained by a large number of simulation runs. It highway width and the traffic volume remains unchanged. In AHP, the elements of the comparison matrix, a > 0 , express varies upon the order of the comparison matrix (Saaty 2000; ij Taha 2010). An inconsistency value not more than 0.1 is the expert’s evaluation of the preference of the ith criterion in relation with the jth. It is worth noting that a = 1 whenever acceptable for an AHP model. ij i = j and a = 1∕a for i ≠ j . The total number of pairwise ij ji comparisons by expert is n(n − 1)∕2 , where ‘n’ is the total Modelling of seasonal peak storm runoff number of criteria under consideration. The eigenvector, w, matching the maximum eigenvalue, , of the comparison Rainfall occurs almost the entire year in Sikkim (IMD max 2014). However, there is a substantial drop in rainfall in the matrix is the preferred solution of the AHP model, that is non-monsoon months, which is from November to March. = . (1) max While April–October gets a relatively high proportion of or annual average rainfall (Rahman et al. 2012). Thereby, the non-monsoon months were considered as antecedent dry ⎛ a ⋯ a ⎞⎛ w ⎞ ⎛ w ⎞ 11 1n 1 1 period and the highest daily rainfall was considered as maxi- ⎜ ⎟⎜ ⎟ ⎜ ⎟ ⋮⋱⋮ ⋮ = ⋮ , (2) max mum intensity rainfall. The drainage area and micro-catch- ⎜ ⎟⎜ ⎟ ⎜ ⎟ a ⋯ a w w ⎝ ⎠⎝ ⎠ ⎝ ⎠ n1 nn n n ments feeding the road runoff/rivers in the study area were 1 3 72 Page 6 of 17 Applied Water Science (2018) 8:72 Fig. 2 Digital elevation model of the study area demarcated using Digital Elevation Model (DEM) (Machado et al. 2017; Siqueira et al. 2017) (Fig. 2). HEC-GeoHMS, a geospatial hydrological extension of ArcGIS, was used to prepare Soil Conservation Service-Curve Number (SCS- CN) maps for ‘pre-project and project implementation’ scenarios (Merwade 2012; Flemming and Doan 2013). For this, satellite images from LISS III were converted to LULC rasters using maximum likelihood method under image clas- sification extension of ArcGIS (Fig. 3a, b). LULC rasters were reclassified into water, agriculture, forest, and medium residential areas. Furthermore, soil texture map of the drain- age area was prepared from secondary sources (CISMHE 2008b) (Fig. 4). It was further reclassified into Hydrologic Soil Groups (HSG), based on the soil texture types (USDA 2007). CN maps were used to prepare Maximum recharge capacity maps (S Maps) based on the relation: Fig. 3 Landuse and Landcover map of a pre-project scenario (2004) and b project implementation scenario (2014) S = − 254. (3) CN Runoff from each micro-catchment was estimated using concentration in the highway runoff (Eq. 5). It is reliable in predicting road runoff concentration of conventional water the relation: pollutants like COD, pH, TSS, and TDS, while it is unable to 0 if P < 0.2 × S predict turbidity and dissolved oxygen: Q = , (4) (P−0.2×S) if P ≥ 0.2 × S P+0.8×S 6 C = exp b + a x , (5) i j j where Q is the runoff and P is the maximum intensity rain- j=1 fall (Vojtek and Vojteková 2016). where C is the concentration in the highway H and b is the y-intercept of the ith water pollutant, a is the proportional- Multiple linear regression model for traffic‑induced ity coefficient, and x is value of the jth predictor variable. water pollution The predictor variables include Event Rainfall as x , Maxi- mum Intensity Rainfall as x , Antecedent Dry Period as x , 2 3 The empirical model developed by Kayhanian et al. (2003) Cumulative Seasonal Rainfall as x , watershed area as x , was used in calculating the traffic-induced water pollutants 1 3 Applied Water Science (2018) 8:72 Page 7 of 17 72 pollutants estimated by the mass balance model was used to assess their nature of association. Preparation of water quality status index maps The project impact area map was overlaid upon the micro- watershed map and 100 random points were created within the project impact area. These points were populated with concen- tration of water pollutants of various years derived from mass balance model as attributes. The attributes of these points were based on their position with respect to the micro-watershed feeding their runoff to the rivers. These points were considered as known points for spatial interpolation of pollutant concen- trations over the project impact area. Empirical Bayesian Kriging (EBK) is a robust and straight- forward spatial interpolation technique. Unlike other types of Fig. 4 Soil texture map kriging, EBK considers uncertainty in spatial parameters. The algorithm behind EBK generates several semivariogram and AADT as x . a is the coefficient of x . However, for the models to minimize the prediction error generated from the 6 j j year 2039, except for AADT, the values of all other predictor uncertainty of model parameters. Each semivariogram gets a variables were not available. As a result, the most reliable weight, based on Bayes’ rule, which predicts how likely the estimate of water pollutants given by Kayhanian et al. (2003) observed data can be generated from a semivariogram (Baner- for AADT > 30,000 was considered for the year 2039. jee et al. 2016; Cui et al. 1995; Krivoruchko 2012; Pilz and Spöck 2008). Hence, EBK was used for spatial interpolation Estimation of water pollutant concentration of the water pollutants. Cross-validation criteria were used in the project impact area using mass balance to assess the quality of spatial model made from the spatial model interpolation. Mean Standardized Error and Standardized Root Mean Square Error were used as cross-validation criteria for The concentration of water pollutants due to traffic-induced the interpolation of the year 2014 (Chang 2017; Lloyd 2009). pollution at various locations of the rivers within the project Pollutant maps prepared from spatial interpolation were con- impact area was estimated using the mass balance model (Bar- verted to Single Factor Pollution Index (SFPI) maps using thwal 2012; Davie 2008): Eq. 6: Q C + Q C D j ij ijk i j=1 C = , P = ∑ (6) (7) ijk Q + Q D j j=1 where P is the SFPI value and C is the measured concen- where C is the downstream concentration of the ith water ijk ijk tration at the ith location for the jth water pollutant of the pollutant in the river, Q and C are the upstream discharge j ij kth year. S is the standard value of the jth water pollutant. rate in l/s and concentration of the ith water pollutant in The SFPI maps were further reclassified based on Table 4. mg/l for the jth stream or river. The runoff from the micro- P < 1 is an indication of low pollution level, while catchment area, Q , was calculated using SCS-CN method ijk P > 1 indicate moderate-to-high pollution level depending (Eqs. 3 and 4). The concentration of the water pollutant ijk on how low or how high the SFPI value is from one (Li et al. in the highway runoff, C , was calculated using empirical 2009; Yan et al. 2015). The reclassified SFPI maps were used model (Eq. 5). The concentration of water pollutant C at to prepare WQSI maps for various years using the relation: Rangpo was taken as the model output and it was compared with the observed data using model validation criteria. (Pal- W P j ijk j=1 iwal and Srivastava 2014). A correlation matrix of water WQSI = � � , ijk ∑ (8) W P j ijk j=1 max Table 4 Standard of SFPI P ≤ 0.4 0.4 ∼ 1.0 1.0 ∼ 2.0 2.0 ∼ 5.0 > 5.0 Pollution levels Not-polluted Slight polluted Medium polluted Heavy polluted Serious polluted 1 3 72 Page 8 of 17 Applied Water Science (2018) 8:72 where W is the weight of the water pollutant calculated from (MACR), a summary sensitivity index, was used to assess j � � ∑ the overall sensitivity of the entire study area with change AHP model and P is calculated from Eq. 7. W P ijk j ijk in water pollutant weight: j=1 max is the maximum value in the set of W P . The WQSI varies j ijk WQSI − WQSI CR it it0 it MACR = × 100 = , from 0 to 1. A WQSI value close to zero indicates no impact, N WQSI N it0 i=1 while a value close to 1 implies high adverse impact. The (12) WQSI was further reclassified using natural break classifica- where MACR is the mean absolute value of change rate of tion (Table 5) (Mushtaq et al. 2015). WQSI value due to change in the weight of water pollutant and N is the total number of pixels. Equation 11 was also Spatially explicit sensitivity analysis used to assess the temporal change in WQSI over various project scenarios. MACR ≥ indicate that the SWQIA SA of project alternatives was done with respect to change in model is sensitive to the ith water pollutant weight at the water pollutant weight for the AHP model. ‘One-At-a-Time’ is αth step size, while MACR <𝛼 implies an insensitivity. In i𝛼 a relatively simple SA method, which mainly involves chang- other words, if a change of say ± 10% of a model input brings ing one input variable at a time to see its effect on the model a ≥ 10% change in model output, the MACR curve slope will output. Its major limitation is that it does not capture the effect be ≥ 45 . In such cases, the model will be considered as sen- of simultaneous variation of input variables on the model out- sitive to the model input (Longley et al. 2010). The overall put (Murphy et al. 2004). OAT-based SESA was performed on methodology is illustrated in Fig. 5. WQSI of the year 2014 as a case study, by changing the water pollutant weight. The water pollutant weight was changed for Results a range of ± 20% with a step size of ± 2% for each water pol- lutant considering a uniform probability distribution within a AHP weight of water pollutants range of 0–1. The WQSI run maps were generated using Eq. 9: � � n j The order of AHP weight was found to be: W P + 1 − W ∑ P t it t ij j≠t j≠t Zn > COD > TSS > TDS > pH . The inconsistency value of WQSI = � � , (9) � � ∑ W the AHP model was within acceptable limit (CR = 0.01 < 0.1) n j W P + 1 − W P t it t ij j≠t j (Table 6). The analysis showed that the ‘with project’ alterna- j≠t max tive had a higher priority of 0.727 as compared to the ‘without project’ alternative priority score of 0.273. Moreover, the pro- Subject to the condition: ject alternatives, viz. ‘with project’ and ‘without project’, did not change the order of priority level at various OAT weight W = 1 (10) change combinations in a range of ± 20%. This proved that j=1 the alternatives were insensitive to the water pollutant weight, as each instances of weight change did not yield an equal or where WQSI is dependent on the tth water pollutant and � � j above change in the project alternatives (Banerjee and Ghose step size, . W is the changed weight, and 1 − W ∑ is t t j≠t 2017; Longley et al. 2010). Thereby, the AHP model was the adjusted weight for the jth water pollutant. Other varia- robust. bles hold the same meaning, as given in Eqs. 6, 7, and 8 (Chen et al. 2011; Xu and Zhang 2013). To evaluate the Runoff estimation, outputs of mass balance model, and thematic maps of water pollutants change in the WQSI value per pixel per step size, a change function was used: CN maps for pre-project and project implementation scenar- WQSI − WQSI it it0 CR = × 100, ios were used to estimate runoff from the micro-catchments (11) it WQSI it0 (Fig. 6a, b). Water pollutants generated at various sections of the highway intersecting with various micro-catchments were where CR is the change rate of WQSI at the ith location for it estimated using Empirical model (Supplementary Figure S1, the tth WQI at the αth step size. Mean Absolute Change Rate Table 5 Impact category of Range ≤ 0.55 0.55 ∼ 0.70 0.70 ∼ 0.85 > 0.85 WQSI Pollution levels No impact Slight adverse impact Moderate adverse impact High adverse impact 1 3 Applied Water Science (2018) 8:72 Page 9 of 17 72 Meteorological data River discharge rate LULC River water pollutants Traﬃc data DEM proﬁle Drainage area Soil texture Empirical model Mass balance model HEC-GeoHMS Water Known points pollutant concentraon CN maps Spaal interpolaon Expert opinion Max. intensity rainfall Map reclassiﬁcaon AHP model SCS Curve Number SFPI map Method Water pollutant Overlay analysis weight Catchment Runoﬀ SESA = Input Box = Intermediate Output Box = ProcessBox WQSI map MACR = Final Output Box Fig. 5 Flowchart of spatial water quality impact assessment model Table 6 HP weight significantly to the water quality along the Rani Khola impact area as compared to nominal contribution along the Teesta Parameter COD pH TDS TSS Zn area (Supplementary Tables S8–S10). The pollutants concen- Weight 0.197 0.161 0.150 0.163 0.329 tration estimated by mass balance model was compared with the downstream water pollutant profile at Rangpo for model validation (Supplementary Table S11). The model validation criteria showed satisfactory results (Table 7). The correlation Tables S1–S3). The outputs of the Empirical model, rivers discharge rates and their pollutant profiles (Supplementary matrix of water pollutants for 2004 showed a significant rela- tionship between all the water pollutants. TDS had a moder- Table S4) were fed into the Mass balance model. The con- tribution of pollution load from the highway runoff along ate correlation with TSS and Zn. In 2014, except for Zn, the remaining water pollutants, namely, COD, pH, TDS, and TSS, each micro-catchment was then estimated (Supplementary Tables S5–S7). Traffic-induced water pollution contributed showed strong association amongst each other (Table 8). 1 3 72 Page 10 of 17 Applied Water Science (2018) 8:72 Ranipool, the entire impact area was seriously affected due to COD pollution (Fig. 7a). pH maps of 2004 showed slight pollution all over the project impact area. In contrast, in 2014 and 2039, the impact area along Rani Khola showed slight pollution, while moderate pollution along Teesta (Fig. 7b). TDS in all scenarios remained at a Not-polluted level (Fig. 7c). On the other hand, TSS levels under 2004 and 2014 scenarios remained largely under seriously pol- luted level except for an upper fraction of the impact area up to Martam. Furthermore, under 2039 scenario, almost the entire impact area was seriously polluted under TSS pollu- tion (Fig. 7d). Zn level under all the scenarios was within the not-polluted level (Fig. 7e). WQSI maps High adverse impact due to water pollution was observed from Martam all the way to Rangpo under 2004 and 2014 scenarios. While the remaining fraction of the impact area above Martam suffered a moderate adverse impact. Under 2039 scenario, almost the entire impact area had a high adverse impact, except for the impact area above Ranipool which had a moderate adverse impact (Fig. 8). Change in WQSI value from 2004 to 2014 showed marginal change of 5% occurred along Teesta, while no change occurred along Rani Khola area. In contrast, a substantial change in WQSI value from 2004 to 2039 occurred in the entire impact area, except for few pockets of no change in WQSI value. The changes were most prominent along the Rani Khola impact area. Comparing WQSI value from 2039 with 2014, signifi- cant change in WQSI was observed along Rani Khola, with 19% change from Ranipool to Martam, while 12% change from Martam to Singtam (Fig. 9). Fig. 6 a SCS-CN map of pre-project scenario (2004). b SCS-CN map Spatially explicit sensitivity analysis of project implementation scenario (2014) MACR of WQSI over the change in water pollutant weights Table 7 MBM validation criteria showed an approximately linear curve with varied slopes and intercept at zero. Second, MACR curves showed Criterion Fractional bias Normalized Correlation Index of symmetry over y-axis, implying that the absolute value mean square coefficient agree- error ment of change rate is the same for equal and opposite change in water pollutant weights. Slopewise the order of water Ideal value 0 0 1 1 pollutants was found to be, Zn > COD > TSS > TDS > pH, Result 0.003 0.041 0.997 0.961 implying that the sensitivity of WQSI to water pollutant weights was in harmony with the order of AHP weights (Fig. 10). Moreover, the slopes of sensitivity were much Cross-validation criteria showed satisfactory results for flatter indicating a low sensitivity of WQSI to change in EBK interpolation (Table 9). SFPI maps of water pollut- water pollutant weights. ants showed a sharp change in concentration at three loca- Figure 11 illustrates the locationwise change rate of the tions viz. Ranipool, Martam, and Singtam. Under 2004 and WQSI value of project implementation scenario at 16% 2014 scenarios, COD maps showed heavy pollution along increase of all the water pollutant weights. Change in COD Rani Khola and seriously polluted condition along Teesta. and TSS weights led to a slight rise in WQSI value all along However, for 2039 scenario, except for impact area above the project impact area. A greater change was observed 1 3 Applied Water Science (2018) 8:72 Page 11 of 17 72 Table 8 Correlation matrix of water pollutants under ‘without’- and ‘with’-project scenarios (n = 14) COD (2004) COD (2014) pH (2004) pH (2014) TSS (2004) TSS (2014) TDS (2004) TDS (2014) Zn (2004) Zn (2014) COD 1 1 0.900 0.921 0.951 0.952 0.663 0.695 0.994 − 0.675 pH 1 1 0.888 0.904 0.866 0.862 0.845 − 0.392* TSS 1 1 0.550 0.572 0.937 − 0.734 TDS 1 1 0.589 0.062* Zn 1 1 Correlation is significant at the 0.05 level (2-tailed) *p > 0.05. No significant p value Table 9 Cross-validation criteria of empirical Bayesian kriging The major contributors to water pollution were COD, TSS, and to some extent pH. Map name Mean standardized Standardized root It was interesting to note that, as per the experts’ error mean square error opinion, heavy metal had the highest water pollutant Ideal value 0 1 weight, but actually, its contribution to water pollution COD 2014 0.003 0.769 was nominal in the impact area. This contradiction was pH 2014 0.012 0.884 also observed while comparing with other studies (Bing- TSS 2014 0.002 0.850 ham et al. 2002). Estimates from the Empirical model TDS 2014 0.016 0.963 partly support the previous studies. It showed that road Zn 2014 0.007 0.819 runoff had a relatively higher concentration of COD and TSS, and it was relatively alkaline than the nearby rivers (USEP 1996; Gan et al. 2008). It is worth mentioning here along Teesta area in case of COD, while it was greater that, traffic composition, high rainfall, and dense vegeta- tion can be major players in mitigating traffic-induced along Rani Khola (from Singtam to Martam) in case of TSS. Change in pH weight led to a marginal change in heavy metal pollution in the study area (Hwang and Weng 2015; Schiff et al. 2016). WQSI value with almost 1% drop in WQSI value between Singtam and Martam. The slight decline in WQSI occurred The mass balance model estimates were fairly close to the observed data. However, variations in the pollution pro- all through the impact area with the rise in TDS weight. Moderate fall in WQSI value occurred with an increase in files can be attributed to the higher instances of landslides, landuse change, or change in rainfall, which mass balance Zn weight. A fall of 5% occurred along the Teesta area, while 4% fall occurred in the remaining impact area. On model is not equipped to predict. A significant correlation was observed between the water pollutants. This observa- comparing Figs. 7 with 11, it can be inferred that a greater positive change rate was observed for areas with higher tion is in harmony with the previous studies (Bhandari and Nayal 2008; Waziri and Ogugbuaja 2010). The contrast of concentration of water pollutants like COD, while a greater negative change was observed for areas with a lower con- water pollutants in Teesta as compared to Rani Khola can be due to the construction of a number of small-to-medium centration of WQP like Zn. These arguments are further emphasized using Fig. 12. It can be seen that an increase hydel projects in Teesta such as in Dikchu area (DFEWM 2012). Satisfactory results of model validation criteria or decrease of Zn, COD, and TSS weight by 18% led to equal but opposite change in WQSI value. For instance, an reinstate the reliability of mass balance model in SWQIA (Agrawal 2005). COD, pH, and TSS showed a higher con- increase in COD weight led to increase in change rate in areas with higher COD concentration, while a decrease in centration in Teesta. In terms of water quality, Teesta was much more polluted than Rani Khola. The abrupt transi- COD weight caused lower change rate in areas with higher COD concentration. tion of water quality at Martam and Singtam can be partly explained by relatively greater runoff contributed by large micro-catchments at specific points along the river. Moreo- ver, the change in water quality at Singtam area was mainly Discussion due to the addition of large amount of COD and TSS into the impact area from Teesta. Cross-validation criteria of One of the main objectives of this study was to perform an SWQIA for road-broadening-induced vehicular traffic EBK showed satisfactory results validating the reliability of EBK. increase. The study revealed that the traffic volume had a major effect on water pollution in the project impact area. 1 3 72 Page 12 of 17 Applied Water Science (2018) 8:72 Fig. 7 a Change in COD level under various project scenarios. b Change in pH level under various project scenarios. c Change in TDS level under various project scenarios. d Change in TSS level under various project scenarios. e Change in Zn level under various project scenarios Fig. 8 Water quality status index maps of various project scenarios 1 3 Applied Water Science (2018) 8:72 Page 13 of 17 72 Fig. 9 Percent change in WQSI values a from pre-project to project implementation sce- nario. b From pre-project to the post-project scenario. c From project implementation to the post-project scenario 6.0 Fig. 10 MACR of WQSI over the change in water pollutant weights 5.0 4.0 3.0 2.0 1.0 0.0 -20-15 5-10 -5 05 10 15 20 Weight change (in percent) COD pH TSS TDS Zn 45 deg. slope The set of water pollutants considered in this study weight of AHP and slope of MACR, and gentle slopes of are widely accepted for the analysis of highway runoff MACR (≤ 45 ) showed the robustness of SWQIA model. (Agrawal et al. 2005; Venner 2004). However, a limited Our study emphasized on SESA of criteria weight. How- number of water pollutants and a relatively small stretch ever, it undermined the importance of SA of an attribute of highway were considered in the present study. A wider on the model (Chen et al. 2011). The present model used set of water pollutants including nutrients, metals, salts, OAT, which is essentially a deterministic method of SA. BOD, and oil and grease can improve SWQIA model. Use of Monte Carlo simulation-based AHP can further Moreover, a wider acceptance of SWQIA model demands add to the validity of the SESA method (Xu and Zhang its validation for a wider study area like a longer road sec- 2013; Qi et al. 2013). tion or a network of roads. Although OAT is a relatively At present, hardly, any highway induced water pollution simple SA method, its application in the present study spatial models are available to serve for highway projects does highlight the role of criteria weight in overall spatial of developing countries, especially for hilly areas (Banerjee impact assessment. Such an attempt has not been made and Ghose 2016). Subjected to its validation in larger study in earlier SWQIA studies. The MACR outcomes were in areas and comprehensive water quality analysis, SWQIA harmony with SESA method suggested by Xu and Zhang model can be considered as a decision support tool for (2013). The similarity between priority order of criteria stakeholders in highway projects. Therefore, as to spatially 1 3 MACR (in percent) 72 Page 14 of 17 Applied Water Science (2018) 8:72 Fig. 11 Change rate maps of WQSI due to increase in WQP weight by 16% visualize and interpret, the impacts of project-induced traf- showed that all the project scenarios had moderate-to-high fic volume change on the water quality in the vicinity of the adverse impact due to water pollution. SESA showed that highway. Moreover, SESA can be used as a reliable tool by the overall water quality summarized by WQSI was most the impact analyst to visualize the role of criteria weights sensitive to heavy metal weight followed by COD. Model and thereby incorporate people’s and experts’ perceptions validation and cross-validation criteria suggested that the into impact measurement and mitigation measures. With model has a good predictive capability and spatial reli- reliable spatial and temporal data, the present model can be ability in pollution prediction. This model can be further used for a more refined SWQIA in similar areas. improved by more detailed meteorological and spatially distributed water quality data. The inclusion of stochas- ticity in criteria weighing along with attribute SA can Conclusion substantiate the spatial explicit SA presented here. This SWQIA methodology can also be applied to other project Vehicular traffic is a significant contributor to water pol- impact analysis by selecting appropriate theoretical models lution to its nearby areas. SWQIA model suggested that for water pollutants measurement. AHP-based capturing water bodies as far as 500 m away from the highway can of experts’ as well as people’s perception of impact crite- be affected by road runoff pollution. Spatial analysis of ria, geovisualization of impacts, temporal extrapolation of water quality impact due to highway projects in mountain- impacts, and SESA can substantially facilitate the decision ous areas involved weighing of impact criteria and spatial support system of the project stakeholders. impact classification. These are essential steps in com- Acknowledgements We thank Abhranil Adak, Assist. Prof., and Chan- bining and interpreting pollution impacts. These impacts drashekhar Bhuiya, Professor, Dept. of Civil Engineering; and Santanu showed high pollution level in the Teesta drainage area Gupta, Assist. Prof., School of Basic & Applied Sciences, SMIT, Sik- mainly due to COD, pH, and TSS. Temporal analysis kim; Nirmalya Chatterjee, Associate, and Sunita Pradhan, Associate, 1 3 Applied Water Science (2018) 8:72 Page 15 of 17 72 Fig. 12 Change rate maps of WQSI value due to change in Zn, COD, and TSS weights by ± 18% ATREE, Gangtok; Rakesh Ranjan, Associate Prof. Dept. of Geology, Aldheimer G, Bennerstedt K (2003) Facilities for treatment of Sikkim University; and Sujata Subba, TGT Environment Sc., TNA, stormwater runoff from highways. Water Sci Technol Journal Gangtok for extending their expertise in this project. Int Assoc Water Pollut Res 48(9):113–121 Antunes P, Santos R, Jordão L (2001) The application of Geographi- cal Information Systems to determine environmental impact sig- Compliance with ethical standards nificance. Environ Impact Assess Rev 21(6):511–535. https :// doi.org/10.1016/S0195 -9255(01)00090 -7 Conflict of interest No potential conflict of interest was reported by Arrawatia PRMLE, Tambe S (2011) Biodiversity of Sikkim: explor- the authors. ing and conserving a global hotspot. Gangtok: Sikkim: Informa- tion and Public Relations Department. http://dspac e.cus.ac.in/ Open Access This article is distributed under the terms of the Crea-jspui /handl e/1/3028. Accessed 10 Jan 2018 tive Commons Attribution 4.0 International License (http://creat iveco Arriaza M, Nekhay O (2008) Combining AHP and GIS modelling to mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- evaluate the suitability of agricultural lands for restoration. In tion, and reproduction in any medium, provided you give appropriate Modelling agricultural and rural development polices, Sevilla credit to the original author(s) and the source, provide a link to the Banerjee P, Ghose MK (2016) Spatial analysis of environmental Creative Commons license, and indicate if changes were made. impacts of highway projects with special emphasis on moun- tainous area: an overview. Impact Assess Project Appraisal 34(4):279–293. https://doi.or g/10.1080/14615517.2016.11764 03 Banerjee P, Ghose MK (2017) A geographic information system-based References socioeconomic impact assessment of the broadening of national highway in Sikkim Himalayas: a case study. Environ Dev Sustain Agrawal ML (2005) A spatial quantitative approach for environ- 19(6):2333–2354. https ://doi.org/10.1007/s1066 8-016-9859-7 mental impact assessment of highway projects. IIT, Khargpur Banerjee P, Ghose MK, Pradhan R (2016) GIS based spatial noise Agrawal ML, Maitra B, Ghose MK (2005) Spatial assessment of impact analysis (SNIA) of the broadening of national highway impacts on water quality due to highway development project. in Sikkim Himalayas: a case study. AIMS Environ Sci 3(4):714– Asian J Water Environ Pollut 2(1):85–92 738. https ://doi.org/10.3934/envir onsci .2016.4.714 Barthwal RR (2012) Environmental impact assessment, 2nd edn. New Age International Private Limited, New Delhi 1 3 72 Page 16 of 17 Applied Water Science (2018) 8:72 Bhandari NS, Nayal K (2008) correlation study on physico-chemical Flemming MJ, Doan JH (2013) HEC-GeoHMS Geospatial Hydrologi- parameters and quality assessment of Kosi River Water, Uttara- cal Extension User’s Manual Version 10.1. US Army Corps of khand [Research article]. https ://doi.org/10.1155/2008/14098 6 Engineers Bhutia TY (2015) Biochemical properties of Teesta river system in Forsyth AR, Bubb KA, Cox ME (2006) Runoff, sediment loss and Sikkim, M.Phil. Thesis, Dept. of Chemistry, School of Physical water quality from forest roads in a southeast Queensland coastal Sciences, Sikkim University, Gangtok plain Pinus plantation. For Ecol Manag 221(1):194–206. https :// Bingham RL, Neal HV, El-Agroudy AA (2002) Characterization doi.org/10.1016/j.forec o.2005.09.018 of the potential impact of storm runoff from highways on the Gajendra C (2011) Water quality assessment and prediction modelling neighbouring water bodies. Case-study: Tamiami trail project. of Nambiyar river basin, Tamil Nadu. PhD thesis, Faculty of Civil In: 7th conference, Biennial stormwater research and watershed Engineering, Anna University, Chennai 600 025 management, May 22–23, pp. 229–239 Gan H, Zhuo M, Li D, Zhou Y (2008) Quality characterization and BIS (2012) Indian Standard Drinking Water—specification (Second impact assessment of highway runoff in urban and rural area of Revision) IS 10500: 2012. Bureau of Indian Standards, Manak Guangzhou, China. Environ Monit Assess 140(1–3):147–159. Bhavan, 9 Bahadur Shah Zafar Marg, New Delhi 110002https ://doi.org/10.1007/s1066 1-007-9856-2 Brown AL, Affum JK (2002) A GIS-based environmental modelling Geneletti D (2004) Using spatial indicators and value functions to system for transportation planners. Comput Environ Urban Syst assess ecosystem fragmentation caused by linear infrastructures. 26(6):577–590. https ://doi.org/10.1016/S0198 -9715(01)00016 -3 Int J Appl Earth Obs Geoinf 5(1):1–15. https://doi.or g/10.1016/j. Canter L (1995) Environmental impact assessment, 2nd edn. McGraw- jag.2003.08.004 Hill, New York Glasson J, Therivel R, Chadwick A (2005) Introduction to environmen- Chang K-T (2017) Introduction to geographic information systems, 4th tal impact assessment, 3rd edn. Routledge, London edn. McGraw Hill Education, New Delhi Granato GE (2013) Stochastic empirical loading and dilution model Chen Y, Yu J, Khan S (2010) Spatial sensitivity analysis of multi- (SELDM) version 1.0.0: U.S. Geological Survey Techniques and criteria weights in GIS-based land suitability evaluation. Environ Methods, book 4, chap. C3, 112 p. U.S. Geological Survey, Reston, Model Softw 25(12):1582–1591. https ://doi.org/10.1016/j.envso Virginia http://pubs.usgs.gov/tm/04/c03/. Accessed 27 June 2016 ft.2010.06.001 Gurung S, Subba S, Jha S, Pandey M (2015) Analysis of physic-chem- Chen H, Wood MD, Linstead C, Maltby E (2011) Uncertainty analysis ical variations in water samples of river Teesta of Sikkim. Int J in a GIS-based multi-criteria analysis tool for river catchment Eng Technol Res 2(2):49–60 management. Environ Model Softw 26(4):395–405. https ://doi. Hwang C-C, Weng C-H (2015) Effects of rainfall patterns on highway org/10.1016/j.envso ft.2010.09.005 runoff pollution and its control. Water Environ J 29(2):214–220. CISMHE (2008a) Aquatic Environment and Water Quality in Carrying https ://doi.org/10.1111/wej.12109 capacity of study of Teesta Basin in Sikkim- Volume VI, Bio- Karbassi AR, Mohammad Hosseini MF, Baghvand A, Nazariha M logical Environment- Terrestrial and Aquatic Resources. Centre (2011) Development of Water Quality Index (WQI) for Gor- for Inter-Disciplinary Studies of Mountain & Hill Environment, ganrood River. Int J Environ Res 5(4):1041–1046. https ://doi. University of Delhi, Delhiorg/10.22059 /ijer.2011.461 CISMHE (2008b) Land Environment- Soil in Carrying capacity of Kayhanian M, Singh A, Suverkropp C, Borroum S (2003) Impact of study of Teesta Basin in Sikkim- Volume III, Biological Environ- annual average daily traffic on highway runoff pollutant concentra- ment- Terrestrial and Aquatic Resources. Centre for Inter-Disci- tions. J Environ Eng 129(11):975–990. https ://doi.org/10.1061/ plinary Studies of Mountain & Hill Environment, University of (ASCE)0733-9372(2003)129:11(975) Delhi, Delhi Kim L-H, Zoh K-D, Jeong S-M, Kayhanian M, Stenstrom MK (2006) Crosetto M, Tarantola S, Saltelli A (2000) Sensitivity and uncertainty Estimating pollutant mass accumulation on highways during dry analysis in spatial modelling based on GIS. Agr Ecosyst Environ periods. J Environ Eng 132(9):985–993. https ://doi.org/10.1061/ 81(1):71–79. https ://doi.org/10.1016/S0167 -8809(00)00169 -9 (ASCE)0733-9372(2006)132:9(985) Cui H, Stein A, Myers DE (1995) Extension of spatial information, Krivoruchko K (2012) Empirical bayesian kriging—implemented in bayesian kriging and updating of prior variogram parameters. Envi- ArcGIS Geostatistical Analyst. ArcUser. http://www.esri.com/ ronmetrics 6(4):373–384. https://doi.or g/10.1002/env.3170060406 news/arcus er/1012/empir ical-byesi an-krigi ng.html Davie T (2008) Fundamentals of hydrology, 2nd edn. Routledge, Kumar D, Alappat BJ (2009) NSF-water quality index: does it London represent the experts’ Opinion? Pract Period Hazard Toxic DFEWM (2012) Environmental Impact Assessment of Dikchu HE Radioact Waste Manag 13(1):75–79. https ://doi.or g/10.1061/ Project. Department of Forest, Environment and Wildlife Man- (ASCE)1090-025X(2009)13:1(75) agement, Government of Sikkim: Gangtok Li M-H, Barrett ME (2008) Relationship between antecedent dry Doamekpor L, Darko R, Klake R, Samlafo V, Bobobee L, Akpabli period and highway pollutant: conceptual models of buildup and C, Nartey V (2016) Assessment of the contribution of road run- removal processes. Water Environ Res 80(8):740–747. https://doi. offs to surface water pollution in the New Juaben Municipality, org/10.2175/10614 3008X 29645 1 Ghana. J Geosci Environ Prot 4:173–190. https://doi.or g/10.4236/ Li Z, Fang Y, Zeng G, Li J, Zhang Q, Yuan Q, Ye F (2009) Tem- gep.2016.41018 poral and spatial characteristics of surface water quality by an Fallahzadeh RA, Almodaresi SA, Dashti MM, Fattahi A, Sadeghnia M, improved universal pollution index in red soil hilly region of Eslami H, Taghavi M (2016) Zoning of nitrite and nitrate concen- South China: a case study in Liuyanghe River watershed. Environ tration in groundwater using Geografic Information System (GIS), Geol 58(1):101–107. https://doi.or g/10.1007/s00254-008-1497-4 Case study: drinking water wells in Yazd City. J Geosci Environ Lilburne L, Tarantola S (2009) Sensitivity analysis of spatial models. Prot 04(03):91. https ://doi.org/10.4236/gep.2016.43008 Int J Geogr Inf Sci 23(2):151–168. https://doi.or g/10.1080/13658 Feizizadeh B, Jankowski P, Blaschke T (2014) A GIS based spatially-81080 20949 95 explicit sensitivity and uncertainty analysis approach for multi-cri- Lloyd C (2009) Spatial data analysis: an introduction for GIS users. teria decision analysis. Comput Geosci 64(Supplement C):81–95. OUP Oxford, Oxford https ://doi.org/10.1016/j.cageo .2013.11.009 Longley PA, Goodchild M, Maguire DJ, Rhind DW (2010) Geographic information systems and science, 3rd edn. Wiley, Hoboken 1 3 Applied Water Science (2018) 8:72 Page 17 of 17 72 Machado ER, Junior do Valle RF, Sanches FLF, Pacheco FAL (2017) The Taha HA (2010) Operations research: an introduction, 9th edn. Pear- vulnerability of the environment to spills of dangerous substances son, Upper Saddle River on highways: a diagnosis based on multi criteria modeling. Transp Takyi SA (2012) Review of Environmental Impact Assessment Res Part D Transp Environ. https: //doi.org/10.1016/j.trd.2017.10.012 (EIA): approach, process and challenges. https ://www.acade mia. Merwade V (2012) Creating SCS curve number grid using HEC- edu/37036 29/R evie w_of_Envir onmen t al_Im pac t_Asses sment GeoHMS. https: //web.ics.purdue .edu/~vmerwa de/educat ion/cngri _EIA_Approa ch_Proces s_and_Challe nges. Accessed 13 Oct 2017 d.pdf. Accessed 12 Jan 2018 Thomson NR, McBean EA, Snodgrass W, Monstrenko IB (1997) High- Murphy JM, Sexton DMH, Barnett DN, Jones GS, Webb MJ, Collins way stormwater runoff quality: development of surrogate param- M, Stainforth DA (2004) Quantification of modelling uncertain- eter relationships. Water Air Soil Pollut 94(3–4):307–347. https ties in a large ensemble of climate change simulations. Nature ://doi.org/10.1007/BF024 06066 430(7001):768–772. https ://doi.org/10.1038/natur e0277 1 Tong STY, Chen W (2002) Modeling the relationship between land use Mushtaq F, Nee Lala MG, Pandey AC (2015) Assessment of pol- and surface water quality. J Environ Manag 66(4):377–393. https lution level in a Himalayan Lake, Kashmir, using geomatics ://doi.org/10.1006/jema.2002.0593 approach. Int J Environ Anal Chem. https://doi.or g/10.1080/03067 USDA (2007) Hydrologic soil groups (Chapter 7) in hydrology national 319.2015.10775 17 engineering handbook. 2007. https ://www .nr cs.usda.gov/Inter Ohimain EI, Imoobe TO, Bawo DDS (2008) Changes in water physic- ne t/FSE_DOCUM ENT S/nr cs1 42p2_05273 1.pdf. Accessed 12 chemical properties following the dredging of an oil well access Jan 2018 canal in the Niger Delta. World J Agric Sci 4(6):752–758 USEP (1996) Indicators of the environmental impacts of transportation Ostovari Y, Beigi Harchegani H, Davoodian AR (2012) Spatial varia- highway, rail, aviation, and maritime transport. U.S. Environmental tion of nitrate in the Lordegan Aquifer. Water Irrig Manag 2:55–67 Protection Agency, Policy, Planning, and Evaluation Pagotto C, Legret M, Le Cloirec P (2000) Comparison of the hydraulic USPH (1962) USPHS 956: Drinking Water Standards, US Public behaviour and the quality of highway runoff water according to Health Service. The executive director office of the federal regis- the type of pavement. Water Res 34(18):4446–4454. https ://doi. ter, Washington DC org/10.1016/S0043 -1354(00)00221 -9 Venner M (2004) Identification of research needs related to highway Paliwal R, Srivastava L (2014) Policy Intervention Analysis: environ- runoff management. Transportation Research Board mental impact assessment. The Energy and Resources Institute, Vojtek M, Vojteková J (2016) GIS-based approach to estimate surface TERI, New Delhi runoff in small catchments: a case study. Quaestiones Geographi- Pilz J, Spöck G (2008) Why do we need and how should we implement cae 35(3):97–116. https ://doi.org/10.1515/quage o-2016-0030 Bayesian kriging methods? Stoch Env Res Risk Assess 22(5):621– Waziri M, Ogugbuaja VO (2010) Interrelationships between physico- 632. https ://doi.org/10.1007/s0047 7-007-0165-7 chemical water pollution indicators: a case study of River Yobe- Qi H, Qi P, Altinakar MS (2013) GIS-based Spatial Monte Carlo analy- Nigeria. Am J Sci Ind Res 1(1):76–80 sis for integrated flood management with two dimensional flood Wu Jy S, Allan Craig J, Saunders William L, Evett Jack B (1998) simulation. Water Resour Manag 27(10):3631–3645. https ://doi. Characterization and pollutant loading estimation for highway org/10.1007/s1126 9-013-0370-8 runoff. J Environ Eng 124(7):584–592. https ://doi.org/10.1061/ Rahman H, Karuppaiyan R, Senapati PC, Ngachan SV, Kumar A (ASCE)0733-9372(1998)124:7(584) (2012) An analysis of past three-decade weather phenomenon Xu E, Zhang H (2013) Spatially-explicit sensitivity analysis for land in the mid-hills of Sikkim and strategies for mitigating possible suitability evaluation. Appl Geogr 45(Supplement C):1–9. https impact of climate change on agriculture in Climate Change in ://doi.org/10.1016/j.apgeo g.2013.08.005 Sikkim Patterns, Impacts and Initiatives. Information and Public Yan C-A, Zhang W, Zhang Z, Liu Y, Deng C, Nie N (2015) Assessment Relations Department, Government of Sikkim, Gangtok of water quality and identification of polluted risky regions based Saaty TL (1980) The analytic hierarchy process: planning, priority on field observations and GIS in the Honghe River Watershed, setting, resource allocation. Mcgraw-Hill, New York China. PLoS One 10(3):e0119130. https ://doi.org/10.1371/journ Saaty TL (1990) How to make a decision: the analytic hierarchy pro-al.pone.01191 30 cess. Eur J Oper Res 48(1):9–26. https ://doi.org/10.1016/0377- Yannopoulos S, Basbas S, Andrianos TH, Rizos CH (2004) Receiving 2217(90)90057 -I waters pollution investigation due to the interurban roads storm Saaty TL (2000) Fundamentals of decision making and priority theory water runoff. In: Proceedings from the International Conference with the analytic hierarchy process. RWS Publications, Pittsburg on Protection and Restoration of the Environment VII, Mykonos, Saaty TL, Vargas L (1994) Fundamentals of decision making and pri- Greece (CD) ority theory with the analytic hierarchy process: 6, 1st edn. Rws Yannopoulos S, Basbas S, Giannopoulou I (2013) Water bodies pollu- Pubns, Pittsburgh tion due to highways stormwater runoff: measures and legislative Sadat-Noori SM, Ebrahimi K, Liaghat AM (2014) Groundwater qual- frameworks. Glob NEST J 15(1):85–92 ity assessment using the water quality index and GIS in Saveh- Zhou F, Guo H, Liu Y, Hao Z (2007) Identification and spatial patterns Nobaran aquifer, Iran. Environ Earth Sci 71(9):3827–3843. https of coastal water pollution sources based on GIS and chemometric ://doi.org/10.1007/s1266 5-013-2770-8 approach. J Environ Sci 19(7):805–810. https ://doi.org/10.1016/ Schiff KC, Tiefenthaler LL, Bay SM, Greenstein DJ (2016) Effects of S1001 -0742(07)60135 -1 rainfall intensity and duration on the first flush from parking lots. Water 8(8):320. https ://doi.org/10.3390/w8080 320 Publisher’s Note Springer Nature remains neutral with regard to Siqueira HE, Pissarra TCT, Junior do Valle RF, Fernandes LFS, jurisdictional claims in published maps and institutional affiliations. Pacheco FAL (2017) A multi criteria analog model for assessing the vulnerability of rural catchments to road spills of hazardous substances. Environ Impact Assess Rev 64:26–36. https ://doi. org/10.1016/j.eiar.2017.02.002 Stenstrom MK, Silverman G, Bursztynsky TA, Zarzana AL, Salarz SE (1982) Oil and grease in stormwater runoff. Association of Bay Area Governments, Berkeley 1 3
Applied Water Science – Springer Journals
Published: May 3, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera