Groundwater qanat potential mapping using ... - Semantic Scholar
Earth Sci Inform DOI 10.1007/s12145-014-0145-7
RESEARCH ARTICLE
Groundwater qanat potential mapping using frequency ratio and Shannon’s entropy models in the Moghan watershed, Iran Seyed Amir Naghibi & Hamid Reza Pourghasemi & Zohre Sadat Pourtaghi & Ashkan Rezaei
Received: 25 August 2013 / Accepted: 15 January 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract The purpose of current study is to produce groundwater qanat potential map using frequency ratio (FR) and Shannon's entropy (SE) models in the Moghan watershed, Khorasan Razavi Province, Iran. The qanat is basically a horizontal, interconnected series of underground tunnels that accumulate and deliver groundwater from a mountainous source district, along a water- bearing formation (aquifer), and to a settlement. A qanat locations map was prepared for study area in 2013 based on a topographical map at a 1:50,000-scale and extensive field surveys. 53 qanat locations were detected in the field surveys. 70 % (38 locations) of the qanat locations were used for groundwater potential mapping and 30 % (15 locations) were used for validation. Fourteen effective factors were considered in this investigation such as slope degree, slope aspect, altitude, topographic wetness index (TWI), stream power index (SPI), slope length (LS), plan curvature, profile curvature, distance to rivers, distance to faults, lithology, land use, drainage density, and fault density. Using the above conditioning factors, groundwater qanat potential map was generated implementing FR and SE models, and the results were plotted in ArcGIS. The predictive capability of frequency ratio and Shannon's entropy models were determined by the area under the relative operating characteristic curve. The area under the curve (AUC) for Communicated by: H. A. Babaie S. A. Naghibi : A. Rezaei Department of Watershed Management Engineering, College of Natural Resources and Marine Sciences, Tarbiat Modares University (TMU), Noor, Mazandaran, Iran H. R. Pourghasemi (*) Young Researchers and Elite Club, Nour Branch, Islamic Azad University, Nour, Iran e-mail: [email protected] Z. S. Pourtaghi Department of Environment Management Engineering, College of Natural Resources, Yazd University, Yazd, Iran
frequency ratio model was calculated as 0.8848. Also AUC for Shannon's entropy model was 0.9121, which depicts the excellence of this model in qanat occurrence potential estimation in the study area. So the Shannon's entropy model has higher AUC than the frequency ratio model. The produced groundwater qanat potential maps can assist planners and engineers in groundwater development plans and land use planning. Keywords Qanat potential mapping . Frequency ratio . Shannon’s entropy . GIS . Iran
Introduction Groundwater is clarified as water in the saturated zones (Fitts 2002) which crams all the pore space of soil and geologic formations below the water table (Freeze and Cherry 1979). It is created by rainwater or snowmelt water which seeps down through the soil and into the underlying rocks (Banks and Robins 2002). Groundwater is looked upon as one of the most vital natural resources (Todd and Mays 2005) because of various characteristics such as consistent temperature, low development cost, widespread availability, and drought dependability (Jha et al. 2007). Another advantage of groundwater over surface water is that it can be tapped when required, on a stage- by- stage basis, and is less affected by catastrophic events. The fast increase in the human population has increased the use and demand for groundwater resources for agricultural, industrial, and drinking objects (Manap et al. 2012). As know the occurrence of groundwater at any place on the earth is not an issue of accidental but a consequence of the interaction of the climatic, geological, hydrological, physiographical, and ecological factors. So, the occurrence and movement of groundwater is controlled by topography, lithology, geological structures, slope and many other factors (Greenbaum 1992; Mukherjee 1996; Oh et al. 2011).
Earth Sci Inform
Considering these factors, we are capable to predict the groundwater potential in a basin implementing different methods. There are several methods for groundwater mapping such as frequency ratio, weight of evidence, and logistic regression methods (Ozdemir 2011a, b). A frequency-ratio model can provide a simple geospatial assessment tool to calculate the probabilistic relationship between the dependent variable and independent variables, including multi classified maps. The FR model has been successfully applied in many studies of landslide susceptibility assessment (Lee and Pradhan 2007; Akgün et al. 2008; Yilmaz 2009; Mohammady et al. 2012; Regmi et al. 2013; Jaafari et al. 2014) and forest fire (Pradhan et al. 2007). More recently, several researchers such as Oh et al. (2011), Ozdemir (2011a), Manap et al. (2012), Davoodi Moghaddam et al. (2013) and Pourtaghi and Pourghasemi (2014) used GIS-based frequency ratio approach for groundwater mapping. Ozdemir (2011a) found that frequency ratio and weights of evidence models are relatively good estimators compared to logistic regression in groundwater spring potential mapping in the Sultan Mountains (Konya, Turkey). Manap et al. (2012) investigated the application of probabilistic-based frequency ratio model in groundwater potential in the Langat basin (Malaysia). The results showed 84.74 % accuracies with standard error of 0.063 for the frequency ratio model which is quite satisfactory. Also, Shannon’s entropy is another model that were used in some researches such as landslide susceptibility mapping (Bednarik et al. 2010; Constantin et al. 2011; Pourghasemi et al. 2012a, b; Devkota et al. 2013; Jaafari et al. 2014). Qanat technology exists in more than 34 countries in the worldwide, but most are concentrated in the Iran, which has about 32164 active systems with a total discharge of about nine billion cubic meters (m3). The first recorded qanats were dug in the north-western regions of Iran and date back to 800 Before Christ (Salih 2007). The qanat is basically a horizontal, interconnected series of underground tunnels that accumulate and deliver groundwater from a mountainous source district, along a water- bearing formation (aquifer), and to a settlement (Perrier and Salkini 1991). The tunnels have several kilometers length and are approximately horizontal (Ahmadi et al. 2010). Fig. 2 demonstrates a profile of qanat and its components (Nazari Samani and Farzadmehr 2006). The main objective of current study was to assess and compare the results of groundwater qanat potential maps using the FR and SE models in the Moghan Watershed, Khorasan Razavi Province, Iran. Groundwater qanat potential maps obtained from spatial data modeling by the mentioned models are useful for water resources management. The output of this paper may provide technical support to government agencies as well as private sectors for groundwater exploration and assessment.
Study area The study area is located in the west of Mashhad, Khorasan Razavi Province, Iran. Moghan watershed is located between upper left (59°19' E and 36° 9′N) and lower right (59°28' E and 36° 6'N) in the upper part of Torogh Dam, with an area of 43.95 km2 (Fig. 1). Elevation in the study area ranges from 1,410 to 2,580 m above sea level, with an average of 1,893 m. The mean annual precipitation is recorded as 245 mm with almost equally spatial and temporal distributions (according to 3 meteorological stations data) (Ostad et al. 2013). The region exhibits distinctly mountainous topographical features. The major land use of the study area consists of rangeland and covers almost 76 % of Moghan basin. Also there are four other land uses including range land, field crops, orchard, residential, and cliff.
Data production In order to generate a qanat potential map for the study area, a qanat locations map was prepared for study region in 2013 based on a topographical map at a 1:50,000-scale and this was proved by the extensive field survey according to their qanat outlet (Fig. 1). By randomly partition (Lee et al. 2012; Oh et al. 2011; Pourghasemi et al. 2013a, b), 38 (seventy percent) of the qanat locations were used in groundwater potential mapping and 15 (30 %) were used for validation. For conducting a groundwater qanat potential map (GQPM), it is important to evaluate several qanat-related factors with the qanat inventory map. The main factors considered in the present study and those influential in the occurrence of a qanat are described below. Fourteen qanat impressive factors were considered in this investigation. These factors are slope degree, slope aspect, altitude, topographic wetness index (TWI), stream power index (SPI), slope-length (LS), plan curvature, profile curvature, distance to rivers, distance to faults, lithology, land use, drainage density, and fault density. Vector-type spatial database were extracted by transforming these factors using the ArcGIS 9.3 (ESRI 2008). The digital elevation model (DEM) with 20×20 m grid size has been made using the elevation contours and survey base points exhibiting the elevation values which were extracted from the 1:50,000-scale topographic maps. The DEM is used as the input to extract thematic maps of slope degree, slope aspect, altitude, topographic wetness index (TWI), stream power index (SPI), slope length (LS), plan curvature, and profile curvature (Fig. 3 (a - h)). For classification of effective factors we made use of different methods such as equal interval, natural break, normal or common standards based on literature reviews and expert knowledge (Manap et al. 2012; Pourghasemi et al. 2012a). On the other hands, the classification of factors was performed
Earth Sci Inform
Fig. 1 Qanat locations with digital elevation model (DEM) map of the study area
based on author’s experiences of study area and its condition. Slope plays a significant role in infiltration and runoff (Sarkar et al. 2001) and also in groundwater potential mapping, so the slope degree map of study area was prepared and classified into 4 classes according to equal interval scheme including 0–5, 5–15, 15–30 and >30° classes (Fig. 3a). The slope aspects were classified into 9 groups including eight directions and flat based on normal or common standard classification (Fig. 3b). Also the altitude map was prepared and classified into 5 classes based on equal interval classification scheme (Fig. 3c).
There is a vital role for topography in the spatial variation of hydrological conditions such as soil moisture and groundwater flow. Therefore, the secondary topographic indices have been used for describing spatial soil moisture patterns (Moore et al. 1991). The topographic wetness index (TWI) is a secondary topographic factor within the runoff model which is clarified according to (Moore et al. 1991): β TWI ¼ ln tanα
ð1Þ
Earth Sci Inform
where, ß tan α
LS ¼
is the cumulative upslope area draining through a point (per unit contour length). is the slope angle at the point.
The tendency of water to accumulate at any point in the catchment (in terms of α) and the tendency of gravitational forces to move that water down slope (indicated in terms of tan b as an approximate hydraulic β gradient) are considered by the ln tanα index. Primarily, the water infiltration depends upon material properties such as permeability and pore water pressure on the soil strength. So, TWI was regarded as a contributing factor in this study (Fig. 2d). The stream power index (SPI) (Fig. 3e) is defined by Moore et al. (1991) as: SPI ¼ Bs tan α
ð2Þ
where, Bs a
is the specific catchment area (m2) the local slope gradient measured in degrees
Slope–length (LS) is the combination of the slope steepness (S) and slope length (L). The LS factor in the Universal Soil Loss Equation (USLE) is a measure of the sediment transport capacity of overland flow (Wischmeier and Smith 1978) (Fig. 3f). The combined slope length (LS) factor can be defined by an equation suggested by Moore and Burch (1986)
Fig. 2 A profile of qanat and its components (Nazari Samani and Farzadmehr 2006)
Bs 22:13
0:6
sin α 0:0896
1:3 ð3Þ
These can be estimated a function of basic terrain characteristics and can be quickly applied in integrated land and water information system (ILWIS 3.3) (ITC 2005). In the matter of plan curvature, positive curvature shows convex, zero curvature represent flat and negative curvature depicts concave (Mohammady et al. 2012). Using system for automated geoscientific analyses (SAGA 2.8), the plan curvature map was prepared (Fig. 3g). Also profile curvature was calculated, and classified into 3 groups based on common standard classification (Pourghasemi et al. 2013b) including < (−0.001), (−0.001)–(0.001), and > (0.001) (Fig. 3h). Furthermore, employing the topographic and geological database, the distance to rivers and faults were calculated, respectively. These factors provided by Spatial Analyst extension in ArcGIS 9.3 (ESRI 2008). The river buffer was calculated in 50 m intervals and the fault buffer was calculated in 100 m intervals, as shown in according to references (Fig. 4 (a-b)). One of the most important parts in forecasting groundwater potential zones is lithology. Using a 1:100,000-scale geological map, the lithology map was prepared and classified based on lithological units (type) into 3 groups including Limestone, recrystallize, dolomitic, Shale, Phylitic, dark grey-phylite, and Terraces (Fig. 5). The land use map was created using Landsat images by Iranian forest, range land, and watershed management (http://www.frw.org.ir/pageid/34/language/
Earth Sci Inform
Fig. 3 Topographical parameter maps of the study area; (a) slope degree, (b) slope aspect, (c) altitude, (d) topographic wetness index (TWI), (e) stream power index (SPI), (f) slope-length (LS), (g) plan curvature, (h) profile curvature
en-US/Default.aspx). Five land use classes were detected such as rangeland, field crops, orchard, residential and cliff (Fig. 6). The drainage density map shows the flow of water throughout the study area and defined as the total length of all the streams
and rivers in a drainage basin divided by the total area of the drainage basin (Sarkar and Kanungo 2004). The drainage density map was generated considering a 20×20 m grid cell and was classified into three classes (Fig. 7). Similarly, fault density map was prepared and
Earth Sci Inform
Fig. 3 (continued)
Fig. 4 a Distance to Rivers (buffer), b Distance to Faults (buffer) maps
Earth Sci Inform Fig. 5 Lithology map of the study area
classified into 4 groups (Fig. 8). Finally, all the mentioned qanat leading factors were converted to a raster grid with 20× 20 m cells for application of FR and SE models.
Methodology Application of frequency ratio model According to Bonham-Carter (1994), the frequency ratio is the probability of occurrence of a certain attribute. Frequency ratio approach is based on the observed relationships between distribution of groundwater qanat locations and each Fig. 6 Land use map of the study area
groundwater-related factor, to reveal the correlation between groundwater qanat locations and the factors in the study area. The calculation steps for an FR for a class of the qanatinfluencing factor are below (Eq. 4): FR ¼
A C
! B
ð4Þ
D
where, A is the number of pixels with qanat for each factor; B is the number of total qanats in study area; C is the number of pixels in the class area of the factor; D
Earth Sci Inform Fig. 7 Drainage density map of the study area
is the number of total pixels in the study area; and FR is the frequency ratio of a class for the factor. To obtain Groundwater Qanat Occurrence Potential Index (GQPI) ratings of the factors were summed as (Eq. 5): GQPI ¼
X
ðFRÞi ði ¼ 1; 2; …; nÞ
ð5Þ
where, the GQPI represents the groundwater qanat occurrence potential index; FR is the frequency ratio of a factor; and n is the total number of factors. Fig. 9 shows the flowchart of applied methodology in this study. Fig. 8 Fault density map of the study area
Application of Shannon’s entropy model The entropy index is a measure of “evenness”-the extent to which groups are evenly distributed among organizational units (Massey and Nancy 1988). More precisely, Theil (1972) defined entropy index as a measure of the average difference between a unit’s group proportions and that of the system as a whole. There is a one-to-one relationship between the quantity of entropy of a system and the degree of disorder called Boltzmann principle and has been used to represent the thermodynamic status of a system (Yufeng and Fengxiang 2009). Shannon improved upon the Boltzmann principle and established an entropy model for information theory. The equations implemented to calculate the information coefficient (Vj) representing the weight value for the parameter as
Earth Sci Inform Fig. 9 Flow chart of methodology
Data Used
Dependent Factor (Qanat)
Effective Factors Topographic Factors
Random Partition Distance to Rivers Drainage Density
Digital Elevation Model (DEM)
Training Qanats
Validation Qanats
(70% or 38 Points)
(30% or 15 Points)
Slope Degree Slope Aspect Altitude Application of Frequency Ratio and Shannon's Entropy Models
TWI SPI Slope Length
Qanat Potential Maps in West of Mashhad area
Plan Curvature Profile Curvature
Land use ROC curve and validation of models
Geology Factors
Lithology Selection of the best model
Distance to Faults Fault Density
a whole (Bednarik et al. 2010; Constantin et al. 2011) are given as following (Eqs. 6–12): FR E ij ¼ XM j
ð6Þ
FR j¼1
where, FR is the frequency ratio and Eij is the probability density. Hj ¼ −
Mj X i¼1
E ij log2 Eij ; j ¼ 1; …; n
H j max ¼ log2 M j; M j−number of I j¼ðH j max−H
j =H j max
classes
Þ;I¼ð0;1Þ; J ¼1;…
ð7Þ
ð8Þ ð9Þ
V j ¼ I j FR
ð10Þ
where, Hj and Hj max are entropy values; Ij is the information coefficient and Mj is the number of classes. Also, Vj depicts the resultant weight value for the parameter as a whole. The result ranges between 0 and 1. The closer the value is to the number 1, the greater the imbalance is. The complete calculation of weight determination for individual parameters is presented in Table 1. Finally, the relative operating characteristics (ROC) (Ozdemir and Altural 2013; Akgün 2012; Mohammady et al. 2012; Pourghasemi et al. 2012a) was employed to determine the accuracy of qanat potential maps produced in this research implementing frequency ratio and Shannon’s entropy models. ROC curve analysis is a common
Earth Sci Inform
methodology to evaluate the accuracy of a diagnostic test (Swets 1988). ROC curve is a helpful approach for representing the quality of deterministic and probabilistic inspection and estimating systems (Swets 1988). The ROC curve could be a represented illustration of the trade-off between the false-positive (M) and false-negative (Z) rates for every possible cutoff value. The ROC curve plots the false positive rate (1- specificity) on the X (Eq. 11) axis and the true positive (S) rate (sensitivity) on the Y axis (Eq. 12) (Negnevitsky 2002). TN X ¼ FPR ¼ 1− ð11Þ TN þ FP
TP Y ¼ TPR ¼ TP þ FN
between qanat occurrence and distance to rivers, for distances to rivers between 0 and 50 m the ratio is 2.13, indicating a high probability of qanat occurrence. Relationship between qanat occurrence and distance to faults shows that qanats are most abundant in the class 0–100 m with FR value of 2.72.In case of lithology, the highest value of FR (1.25) is for group of 1 (Shale, phylitic, dark grey phylite). Considering the case of land use, it can be seen that classes of field crops and orchard have values of 2.59, 2.47, respectively. So, indicating that the probability of occurrence of qanat in these land use types is very high. The drainage density 4–6 km/km2 has the largest frequency ratio value (FR=2.30), which means the attributes of this class have the strongest relationship with qanat occurrence. Regarding the fault density, results displayed that class 3–4.5 has the highest value of 3.15.
ð12Þ
Results Frequency ratio The results of spatial relationship between occurrence of qanat and conditioning factors implementing frequency ratio model is displayed in Table 1. The ratio of the area where groundwater qanat occurred to the total area is the relation analysis; so a value of 1 shows an average value. There would be a high correlation if the value is larger than 1, a lower correlation equals to the value lower than 1 (Oh and Lee 2010). The analysis of FR for the relationship between qanat occurrence and slope degree indicate that slope degree class 0–5 has the highest value of FR (3.11) followed by 5–15 class (2.65). The lowest value of FR (0.15) is for slope degree class of>30. In the study area, was observed that when slope gradient is increasing, frequency ratio is decreasing. According to the relationship between qanat occurrence and slope aspect, qanats are most abundant on south and east-facing slopes and in the west-facing slopes have the lowest abundant. In the Table 1, for the altitude between 1,600 and 1,800 m, the FR was 2.55, which implies a very high probability of qanat occurrence. Proportional to the other altitude classes qanats are concentrated in above a topographic wetness index of 12 (value of 5.42), and assessment of stream power index showed that the class of >600 and 0–200 have the highest FR values (1.43) and (1.11) respectively. In the case of slope-length, most of the qanats occurred in 0–20 class with FR value of 1.90. Considering the case of plan curvature, results show that flat shape had highest value of 1.78 and concave shape had the lowest value of 0.25. In the case of profile curvature, most of the qanat happened in class of 2200 17253 Topographic wetness index (TWI) 12 10134 Stream power index (SPI) 0–200 18293 200–400 18086 400–600 14892 >600 58570 Slope-length (LS) 0–20 18289 20–40 22376 40–60 23701
13.37 28.86 24.25 17.82 15.71
0.00 28.00 9.00 1.00 0.00
0.00 73.68 23.68 2.63 0.00
0.00 2.55 0.98 0.15 0.00
0.00 0.69 0.27 1.06 2.32 0.04 0.00
0.54 0.40
35.71 55.06 9.23
2.00 17.00 19.00
5.26 44.74 50.00
0.15 0.81 5.42
0.02 0.13 0.70 1.58 0.85
0.56 1.18
16.65 16.47 13.56 53.32
7.00 1.00 1.00 29.00
18.42 2.63 2.63 76.32
1.11 0.16 0.19 1.43
0.38 0.06 0.07 1.52 2.00 0.50
16.65 20.37 21.58
12.00 4.00 8.00
31.58 10.53 21.05
1.90 0.52 0.98
0.44 0.12 1.85 2.00 0.23
45475
41.40
14.00
36.84
0.89
0.21
38982 24367 46492
35.49 22.18 42.33
19.00 15.00 4.00
50.00 39.47 10.53
1.41 1.78 0.25
0.41 0.52 1.29 1.58 0.07
41405 21248 47188
37.70 19.34 42.96
26.00 8.00 4.00
68.42 21.05 10.53
1.82 1.09 0.25
0.58 1.27 1.58 0.35 0.08
47708 34621 16290 11222
43.43 31.52 14.83 10.22
36.00 0.00 3.00 0.00
92.31 0.00 7.69 0.00
2.13 0.00 0.52 0.00
0.80 0.00 0.20 0.71 2.00 0.00
Slope degree 0–5 5–15 15–30 >30 Slope aspect Flat North Northeast East Southeast South Southwest West Northwest Altitude (m)
>60 Plan curvature (100/m) Concave Flat Convex Profile curvature (100/m) < (−0.001) (−0.001)-(0.001) > (0.001) Distance to Rivers (m) 0–50 50–100 100–150 >150
0.30 0.47
0.24 0.17
1.07 0.08
0.18 0.21
0.20 0.21
0.64 0.42
Earth Sci Inform Table 1 (continued) Distance to Faults (m) 0–100 100–200 200–300 >300 Lithology Shale, phylitic, dark greyphylite Limestone. recrystalizd, dolomitic Terraces Land Use Cliff Field Crops Orchard Range Land Residential Drainage Density (km/km2) 0–2 2–4 4–6 >6 Fault Density (km/km2) 0–1.5 1.5–3 3–4.5 >4.5
Fig. 10 Qanat potential susceptibility mapping by frequency ratio (FR) model
9562 9505 10142 80632
8.71 8.65 9.23 73.41
9.00 5.00 4.00 20.00
23.68 13.16 10.53 52.63
2.72 1.52 1.14 0.72
0.45 1.83 2.00 0.25 0.19 0.12
0.08 0.13
87919
80.04
38.00
100.00
1.25
1.00 0.00 1.58
1.00 0.42
21660
19.72
0.00
0.00
0.00
0.00
262
0.24
0.00
0.00
0.00
0.00
8186 6693 10543 83607 812
7.45 6.09 9.60 76.12 0.74
0.00 6.00 9.00 23.00 0.00
0.00 15.79 23.68 60.53 0.00
0.00 2.59 2.47 0.80 0.00
0.00 1.44 2.32 0.44 0.42 0.14 0.00
0.38 0.45
60753 41318 7536 234
55.31 37.62 6.86 0.21
15.00 17.00 6.00 0.00
39.47 44.74 15.79 0.00
0.71 1.19 2.30 0.00
0.17 1.43 2.00 0.28 0.55 0.00
0.39 0.30
94695 2008 2754 10384
86.21 1.83 2.51 9.45
25.00 0.00 3.00 10.00
65.79 0.00 7.89 26.32
0.76 0.00 3.15 2.78
0.11 0.00 0.47 1.40 2.00 0.42
0.30 0.51
Earth Sci Inform Table 2 The distribution of the Qanat potential values and areas with respect to the groundwater qanat occurrence potential zones
Qanat potential zoning
Shannon’s entropy Model
Frequency Ratio Model
Range
Area (%)
Range
Low
(1.82)–(4.49)
33.7
(4.93)–(11.33)
26.6
Moderate High Very High
(4.49)–(6.24) (6.24)–(9.23) (9.23)–(14.99)
39.6 17.9 8.8
(11.33)–(14.87) (14.87)–(19.47) (19.47)–(32.31)
38.8 24.8 9.8
QPMSE ¼ ððSlope degreeFR 0:470Þ þ ðSlope aspectFR 0:144Þ þ ðAltitudeFR 0:400Þ þ TWIFR 1:180 þ ðSPIFR 0:170Þ þ ðLSFR 0:820Þ þ ðPlan curvatureFR 0:211Þ þ Profile curvatureFR 0:205 þ ðDistance to riversFR 0:425Þ þ ðDistance to faultsFR 0:126Þ þ ðLithologyFR 0:416Þ þ Land useFR 0:446 þ ðDrainage densityFR 0:302Þ þ ðFault densityFR 0:506ÞÞ
ð14Þ
The obtained pixel values were then classified based on natural break classification scheme (Bednarik et al. 2010; Constantin et al. 2011; Erner et al. 2010; Falaschi et al. 2009; Ram Mohan et al. 2011; Xu et al. 2012a, b) into low, moderate, high and very high potential classes. The qanat potential map achieved from the FR method, which covered 38.8 % of the total area, was designated to be a moderate QPM class. On the other hand, the area related to low, high and very high QPM zones are 26.6 %, 24.6 %, and 9.8 %, respectively (Fig. 10 and Table 2). Also, results of Shannon’s entropy model showed that low, moderate, high and very high QPM Fig. 11 Qanat potential susceptibility mapping by Shannon’s entropy (SE) model
Area (%)
zones including 33.7 %, 39.6 %, 17.9 %, and 8.8% of the total area, respectively (Fig. 11 and Table 2).
Validation of qanat potential maps The area under the curve (AUC) of the ROC depicts the quality of a forecasting system by showing the system’s ability to expect the correct occurrence or non-occurrence of predefined “events”. In the ROC method, AUC values, ranging from 0.5 to 1.0, are used to evaluate the accuracy of the model. The ideal model represents an AUC value close to 1.0, whereas a value close to 0.5 indicates inaccuracy in the model (Fawcett 2006). The qualitative relationship between AUC and prediction accuracy can be classified as follows: 0.9–1, excellent; 0.8–0.9, very good; 0.7–0.8, good; 0.6–0.7, average; and 0.5–0.6, poor (Pourghasemi et al. 2013a). The results of the prediction curves are exhibited in Fig. 12 and 13. According to Fig. 12, it is clear that in the qanat potential map using frequency ratio model, the AUC is 0.8848, which corresponds to the prediction accuracy of 88.48 %. Also, Fig. 13 showed that AUC of Shannon's entropy model was calculated 91.21 %. Therefore, it is seen that the model
Earth Sci Inform 1
Table 3 The sensivity analysis and factor analysis for FR model
True Positive Rate
0.8
0.6
0.4
"AUC=88.48%" "SE=0.625"
0.2
0 0
0.2
0.4 0.6 False Positive Rate
0.8
1
Fig. 12 Prediction rate curve for the qanat potential map(frequency ratio Model)
produced using Shannon’s entropy exhibited better performance than the frequency ratio model in the study area. Also, in current research, was attempted to assess influence factors using of sensitivity and factor analysis for FR model. The effect analysis researches indicate how a solution changes when the input factors are changed. If the selected factor results in a relatively large change in the outcome, then the outcomes it is can be effective to that factor. Effect analysis quantifies the unreliability of each factor. The factors that have the greatest impact on the computed spring occurrence map can therefore be identified using effect analysis (Lee and Talib 2005). The results of sensitivity analysis are given in Table 3. The area under the curve which includes all factors is equal to 88.48 %. By removing altitude, the area under the curve is 84.38 %, without slope aspect 85.35 %, without distance to faults 86.52 %, without distance to rivers 87.11 %, without drainage
AUC
Factors
Included all factors Altitude Slope Aspect Distance to Faults Distance to Rivers
88.48 % 84.38 % 85.35 % 86.52 % 87.11 %
Drainage Density Fault Density Landuse Lithology Slope-length (LS) Plan Curvature Profile Curvature Slope degree SPI TWI
85.94 % 87.11 % 85.16 % 85.94 % 87.11 % 84.18 % 85.94 % 88.28 % 85.55 % 88.28 %
density 85.94 %, without fault density 87.11 %, without land use 85.16 %, without lithology 85.94 %, without slope-length (LS) 87.11 %, without plan curvature 84.18 %, without profile curvature 85.94 %, without slope degree 88.28 %, without SPI 85.55 %, and finally area of curve without TWI is 88.28 %. As presented in Table 3, plan curvature and altitude are the most important on qanat potential analysis (AUC ratio=84.18 % and 84.38 %, respectively) in study area. After their land use (AUC=85.16 %), slope aspect (AUC=85.35 %) and SPI (AUC=85.55 %) had the most effect on spring analysis. Meanwhile, the effective factors such as lithology, drainage density, and profile curvature had the same effect on QPM. The other factors show lower and positive effects on qanat potential analysis, as well.
1
Conclusion
True Positive Rate
0.8
0.6
0.4
"AUC=91.21%" "SE=0.053"
0.2
0 0
0.2
0.4 0.6 False Positive Rate
0.8
1
Fig. 13 Prediction rate of the curve for the qanat potential map(Shannon's entropy model)
Various methods have been applied for regional groundwater potential assessment globally. The FR and SE methods are used in this study to map groundwater qanat potentials. Qanats locations were determined and mapped based on extensive field surveys and topographical map. Fifty three qanats were identified in the study area; 14 qanat-related factor maps of the entire study area such as topographical factors, land use map, geological factors, and thematic maps of digital elevation model (DEM) were extracted from the spatial database and used to produce QPM map using the FR and SE methods. In addition, the predictive capability of the model was determined by the area under the ROC curve. The area value under the ROC curve for the FR model was 0.8848. Also area under the curve for Shannon’s entropy model was 0.9121, which depicts the excellence of this model in qanat occurrence
Earth Sci Inform
potential estimation in the study area. So the Shannon’s entropy model has higher AUC than the frequency ratio model. According to Shannon’s entropy results, it can be concluded that TWI and fault density have the strongest relationships with qanat occurrence. Also, factors such as slope-length (LS), distance to faults, and slope aspect had the lowest importance on qanat potential map. As a final conclusion, the produced groundwater qanat potential maps can assist planners and engineers in groundwater development plans and land use planning. Acknowledgement The authors would like to thank of Dr. Hassan Ali Babaie “Editor-in-Chief” and two anonymous reviewers for their helpful comments on the previous version of the manuscript.
References Ahmadi H, Nazari Samani A, Malekian A (2010) The qanat: a living history in Iran. Water and sustainability in arid regions. Springer, Netherlands, pp 125–138 Akgün A (2012) A comparison of landslide susceptibility maps produced by logistic regression, multi-criteria decision, and likelihood ratio methods: a case study at Ýzmir. Turkey Landslides 9(1):93–106. doi:10.1007/s10346-011-0283-7 Akgün A, Dag S, Bulut F (2008) Landslide susceptibility mapping for a landslide prone area (Findikli, NE of Turkey) by likelihoodfrequency Ratio and weighted linear combination models. Environ Geol 54(6):1127–1143 Banks D, Robins N (2002) An introduction to groundwater in crystalline bedrock. Norges geologiske undersøkelse, Trondheim Bednarik M, Magulová B, Matys M, Marschalko M (2010) Landslide susceptibility assessment of the Kralˇovany–Liptovsky’ Mikuláš railway case study. Phys Chem Earth Parts A/B/C 35(3):162–171 Bonham-Carter GF (1994) Geographic information systems for geoscientists: modeling with GIS. Pergamon Press, Ottawa Constantin M, Bednarik M, Jurchescu MC, Vlaicu M (2011) Landslide susceptibility assessment using the bivariate statistical analysis and the index of entropy in the Sibiciu Basin (Romania). Environ Earth Sci 63:397–406 Davoodi Moghaddam D, Rezaei M, Pourghasemi HR, Pourtaghi ZS, Pradhan B (2013) Groundwater spring potential mapping using bivariate statistical model and GIS in the Taleghan watershed Iran, Arabian. J Geoscience. doi:10.1007/s12517-013-1161-5 Devkota KC, Regmi AD, Pourghasemi HR, Yoshida K, Pradhan B, Ryu IC, Dhital MR, Althuwaynee F (2013) Landslide susceptibility mapping using certainty factor, index of entropy and logistic regression models in GIS and their comparison at Mugling–Narayanghat road section in Nepal Himalaya. Nat Hazards 65(1):135–165 Erner A, Sebnem H, Duzgun B (2010) Improvement of statistical landslide susceptibility mapping by using spatial and global regression methods in the case of More and Romsdal (Norway). Landslides 7: 55–68 ESRI (2008) ArcGIS 9.3 ESRI Inc., Redlands, California Falaschi F, Giacomelli F, Federici PR, Puccinelli A, D’Amato Avanzi G, Pochini A, Ribolini A (2009) Logistic regression versus artificial neural networks: landslide susceptibility evaluation in a sample area of the Serchio River valley, Italy. Nat Hazards 50(3):551–569 Fawcett T (2006) An introduction to ROC analysis. Pattern Recogn Lett 27(8):861–874 Fitts CR (2002) Groundwater science. Academic, San Diego
Freeze RA, Cherry JA (1979) Groundwater. Prentice-Hall, Englewood Cliffs Greenbaum D (1992) Structural influences on the occurrence of groundwater in SE Zimbabwe. Geol Soci London Special Pub 66:77–85 International Institute for Aerospace Survey and Earth Sciences (ITC) (2005) ILWIS 3.3 user guide, www.itc.nl Jaafari A, Najafi A, Pourghasemi HR, Rezaeian J, Sattarian A (2014) GIS-based frequency ratio and index of entropy models for landslide susceptibility assessment in the Caspian forest, northern Iran. Int J Environ Sci Technol. doi:10.1007/s13762-013-0464-0 Jha MK, Chowdhury A, Chowdary VM, Peiffer S (2007) Groundwater management and development by integrated remote sensing and geographic information systems: prospects and constraints. Water Resour Manage 21(2):427–467 Lee S, Pradhan B (2007) Landslide hazard mapping at Selangor, Malaysia using frequency ratio and logistic regression model. Landslides 4(1):33–41 Lee S, Talib JA (2005) Probabilistic landslide susceptibility and factor effect analysis. Environ Geol 47(7):982–990 Lee S, Song KY, Kim Y, Park I (2012) Regional groundwater productivity potential mapping using a geographic information system (GIS) based artificial neural network model. Hydrogeol J 20(8):1511– 1527. doi:10.1007/s10040-012-0894-7 Manap MA, Nampak H, Pradhan B, Lee S, Soleiman WNA, Ramli MF (2012) Application of probabilistic-based frequency ratio model in groundwater potential mapping using remote sensing data and GIS. Arab J Geosci. doi:10.1007/s12517-012-0795-z Massey DS, Nancy AD (1988) The Dimensions of Residential. Social Forces 67(2):281–315 Mohammady M, Pourghasemi HR, Pradhan B (2012) Landslide susceptibility mapping at Golestan Province, Iran: A comparison between frequency ratio, Dempster–Shafer, and weights of-evidence models. J Asian Earth Sci 61:221–236 Moore ID, Burch GJ (1986) Sediment transport capacity of sheet and rill flow: application of unit stream power theory. Water Res 22(8): 1350–1360 Moore ID, Grayson RB, Ladson AR (1991) Digital terrain modeling: a review of hydrological, geomorphological and biological applications. Hydro Process 5:3–30 Mukherjee S (1996) Targeting saline aquifer by remote sensing and geophysical methods in a part of Hamirpur_Kanpur, India. Hydro J 19:1867–1884 Nazari Samani A, Farzadmehr J (2006) Qanat as a traditional and advantageous approach for water supply in Iran. In: Proceedings of the International Symposium on Waterand Management for Sustainable Irrigated Agriculture, Adana, Turkey Negnevitsky M (2002) Artificial Intelligence: a guide to intelligent Systems, Addison– Wesley/Pearson Education. Harlow, England Oh HJ, Lee S (2010) Assessment of ground subsidence using GIS and the weights of evidence model. Eng Geol 115:36–48 Oh HJ, Kim YS, Choi J-K, Park E, Lee S (2011) GIS mapping of regional probabilistic groundwater potential in the area of Pohang City, Korea. J Hydrol 399:158–172 Ostad M, Mosaedi A, Mesdaghi M, Samadi SZ, Sepehr A (2013) Evaluation of the relation between vegetable cover and precipitation using remote sensing in Torogh basin, Mashhad. First national conference of strategies for sustainable development in agriculture, natural resources and environmental sections. Tehran, Iran Ozdemir A (2011a) GIS-based groundwater spring potential mapping in the Sultan Mountains (Konya, Turkey) using frequency ratio, weights of evidence and logistic regression methods and their comparison. J Hydrol 411:290–308 Ozdemir A (2011b) Using a binary logistic regression method and GIS for evaluating and mapping the groundwater spring potential in the Sultan Mountains (Aksehir, Turkey). J Hydrol 405:123–136. doi:10. 1016/j.jhydrol.2011.05.015
Earth Sci Inform Ozdemir A, Altural T (2013) A comparative study of frequency ratio, weights of evidence and logistic regression methods for landslide susceptibility mapping: Sultan Mountains, SW Turkey. J Asian Earth Sci 64:180–197 Perrier E, Salkini AB (1991) Supplemental Irrigation in the Near East and North Africa. Kluwer Academic Publisher, Norwel Pourghasemi HR, Pradhan B, Gokceoglu C (2012a) Remote sensing data derived parameters and its use in landslide susceptibility assessment using Shannon’s entropy and GIS, AEROTECH IV–2012. Appl Mech Mater 225:486–491. doi:10.4028/www.scientific.net/AMM. 225.486 Pourghasemi HR, Mohammady M, Pradhan B (2012b) Landslide susceptibility mapping using index of entropy and conditional probability models in GIS: Safarood Basin, Iran. Catena 97:71–84 Pourghasemi HR, Moradi HR, Fatemi Aghda SM (2013a) GIS-based landslide susceptibility mapping with probabilistic likelihood ratio and spatial multi-criteria evaluation models (North of Tehran, Iran). Arab J Geosci. doi:10.1007/s12517-012-0825-x Pourghasemi HR, Goli Jirandeh A, Pradhan B, Xu C, Gokceoglu C (2013b) Landslide susceptibility mapping using support vector machine and GIS. J Earth Syst Sci 122(2):349–369 Pourtaghi ZS, Pourghasemi HR (2014) GIS-based groundwater spring potential assessment and mapping in the Birjand Township, southern Khorasan Province. Iran, Hydrogeology. doi:10.1007/s10040013-1089-6 Pradhan B, Suliman MDH, Awang MA (2007) Forest fire susceptibility and risk mapping using remote sensing and geographical information systems (GIS). Disaster Prevention and Management 16:344– 352. doi:10.1108/09653560710758297 Ram Mohan V, Jeyaseelan A, Naveen Raj T, Narmatha T, Jayaprakash M (2011) Landslide susceptibility mapping using frequency ratio method and GIS in south eastern part of Nilgiri District, Tamilnadu, India. Int J Geomatics & Geosci 1(4):951–961 Regmi AD, Yoshida K, Pradhan B, Pourghasemi HR, Khumamoto T, Akgun A (2013) Application of frequency ratio, statistical index and
weights-of-evidence models, and their comparison in landslide susceptibility mapping in Central Nepal Himalaya. Arab J Geosci. doi: 10.1007/s12517-012-0807-z Salih A (2007) Qanats a Unique Groundwater Management Tool in Arid Regions: The Case of Bam Region in Iran. UNESCO Water eNewsletter No. 186: QANATS Sarkar S, Kanungo DP (2004) An integrated approach for Landslide Susceptibility Mapping using remote sensing and GIS. Photogram Eng & Remote Sens 70(5):617–625 Sarkar B, Deota B, Raju P, Jugran D (2001) A geographic information system approach to evaluation of groundwater potentiality of Shamrimicro watershed in the Shimla Taluk, Himachal Pradesh. J Indian Soc Remote Sens 29(3):151–164 Swets JA (1988) Measuring the accuracy of diagnostic systems. Sci 240: 1285–1293 Theil H (1972) Statistical decomposition analysis. North-Holland Publishing Company, Amsterdam Todd DK, Mays LW (2005) Groundwater hydrology. Wiley, New York Wischmeier WH, Smith DD (1978) Predicting Rainfall Erosion Losses: A Guide to Conservation Planning, U.S. Department of Agriculture, Agriculture Handbook. No. 537 Xu C, Xu X, Dai F, Xiao J (2012a) Landslide hazard mapping using GIS and weight of evidence model in Qingshui River watershed of 2008 Wenchuan earthquake struck region. Journal of Earth Sciences 23(1):97–120 Xu C, Dai F, Xu X, Lee YH (2012b) GIS-based support vector machine modeling of earthquake-triggered landslide susceptibility in the Jianjiang River watershed, China. Geomorphology 145–146:70–80 Yilmaz I (2009) Landslide susceptibility using frequency ratio, logistic regression, artificial neural networks and their comparison: a case study from Kat landslide (Tokat-Turkey). Comput Geosci 35(6):1125–1138 Yufeng S, Fengxiang J (2009) Landslide Stability Analysis Based on Generalized Information Entropy. 2009 International Conference on Environmental Science and Information Application Technology: 83–85
RESEARCH ARTICLE
Groundwater qanat potential mapping using frequency ratio and Shannon’s entropy models in the Moghan watershed, Iran Seyed Amir Naghibi & Hamid Reza Pourghasemi & Zohre Sadat Pourtaghi & Ashkan Rezaei
Received: 25 August 2013 / Accepted: 15 January 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract The purpose of current study is to produce groundwater qanat potential map using frequency ratio (FR) and Shannon's entropy (SE) models in the Moghan watershed, Khorasan Razavi Province, Iran. The qanat is basically a horizontal, interconnected series of underground tunnels that accumulate and deliver groundwater from a mountainous source district, along a water- bearing formation (aquifer), and to a settlement. A qanat locations map was prepared for study area in 2013 based on a topographical map at a 1:50,000-scale and extensive field surveys. 53 qanat locations were detected in the field surveys. 70 % (38 locations) of the qanat locations were used for groundwater potential mapping and 30 % (15 locations) were used for validation. Fourteen effective factors were considered in this investigation such as slope degree, slope aspect, altitude, topographic wetness index (TWI), stream power index (SPI), slope length (LS), plan curvature, profile curvature, distance to rivers, distance to faults, lithology, land use, drainage density, and fault density. Using the above conditioning factors, groundwater qanat potential map was generated implementing FR and SE models, and the results were plotted in ArcGIS. The predictive capability of frequency ratio and Shannon's entropy models were determined by the area under the relative operating characteristic curve. The area under the curve (AUC) for Communicated by: H. A. Babaie S. A. Naghibi : A. Rezaei Department of Watershed Management Engineering, College of Natural Resources and Marine Sciences, Tarbiat Modares University (TMU), Noor, Mazandaran, Iran H. R. Pourghasemi (*) Young Researchers and Elite Club, Nour Branch, Islamic Azad University, Nour, Iran e-mail: [email protected] Z. S. Pourtaghi Department of Environment Management Engineering, College of Natural Resources, Yazd University, Yazd, Iran
frequency ratio model was calculated as 0.8848. Also AUC for Shannon's entropy model was 0.9121, which depicts the excellence of this model in qanat occurrence potential estimation in the study area. So the Shannon's entropy model has higher AUC than the frequency ratio model. The produced groundwater qanat potential maps can assist planners and engineers in groundwater development plans and land use planning. Keywords Qanat potential mapping . Frequency ratio . Shannon’s entropy . GIS . Iran
Introduction Groundwater is clarified as water in the saturated zones (Fitts 2002) which crams all the pore space of soil and geologic formations below the water table (Freeze and Cherry 1979). It is created by rainwater or snowmelt water which seeps down through the soil and into the underlying rocks (Banks and Robins 2002). Groundwater is looked upon as one of the most vital natural resources (Todd and Mays 2005) because of various characteristics such as consistent temperature, low development cost, widespread availability, and drought dependability (Jha et al. 2007). Another advantage of groundwater over surface water is that it can be tapped when required, on a stage- by- stage basis, and is less affected by catastrophic events. The fast increase in the human population has increased the use and demand for groundwater resources for agricultural, industrial, and drinking objects (Manap et al. 2012). As know the occurrence of groundwater at any place on the earth is not an issue of accidental but a consequence of the interaction of the climatic, geological, hydrological, physiographical, and ecological factors. So, the occurrence and movement of groundwater is controlled by topography, lithology, geological structures, slope and many other factors (Greenbaum 1992; Mukherjee 1996; Oh et al. 2011).
Earth Sci Inform
Considering these factors, we are capable to predict the groundwater potential in a basin implementing different methods. There are several methods for groundwater mapping such as frequency ratio, weight of evidence, and logistic regression methods (Ozdemir 2011a, b). A frequency-ratio model can provide a simple geospatial assessment tool to calculate the probabilistic relationship between the dependent variable and independent variables, including multi classified maps. The FR model has been successfully applied in many studies of landslide susceptibility assessment (Lee and Pradhan 2007; Akgün et al. 2008; Yilmaz 2009; Mohammady et al. 2012; Regmi et al. 2013; Jaafari et al. 2014) and forest fire (Pradhan et al. 2007). More recently, several researchers such as Oh et al. (2011), Ozdemir (2011a), Manap et al. (2012), Davoodi Moghaddam et al. (2013) and Pourtaghi and Pourghasemi (2014) used GIS-based frequency ratio approach for groundwater mapping. Ozdemir (2011a) found that frequency ratio and weights of evidence models are relatively good estimators compared to logistic regression in groundwater spring potential mapping in the Sultan Mountains (Konya, Turkey). Manap et al. (2012) investigated the application of probabilistic-based frequency ratio model in groundwater potential in the Langat basin (Malaysia). The results showed 84.74 % accuracies with standard error of 0.063 for the frequency ratio model which is quite satisfactory. Also, Shannon’s entropy is another model that were used in some researches such as landslide susceptibility mapping (Bednarik et al. 2010; Constantin et al. 2011; Pourghasemi et al. 2012a, b; Devkota et al. 2013; Jaafari et al. 2014). Qanat technology exists in more than 34 countries in the worldwide, but most are concentrated in the Iran, which has about 32164 active systems with a total discharge of about nine billion cubic meters (m3). The first recorded qanats were dug in the north-western regions of Iran and date back to 800 Before Christ (Salih 2007). The qanat is basically a horizontal, interconnected series of underground tunnels that accumulate and deliver groundwater from a mountainous source district, along a water- bearing formation (aquifer), and to a settlement (Perrier and Salkini 1991). The tunnels have several kilometers length and are approximately horizontal (Ahmadi et al. 2010). Fig. 2 demonstrates a profile of qanat and its components (Nazari Samani and Farzadmehr 2006). The main objective of current study was to assess and compare the results of groundwater qanat potential maps using the FR and SE models in the Moghan Watershed, Khorasan Razavi Province, Iran. Groundwater qanat potential maps obtained from spatial data modeling by the mentioned models are useful for water resources management. The output of this paper may provide technical support to government agencies as well as private sectors for groundwater exploration and assessment.
Study area The study area is located in the west of Mashhad, Khorasan Razavi Province, Iran. Moghan watershed is located between upper left (59°19' E and 36° 9′N) and lower right (59°28' E and 36° 6'N) in the upper part of Torogh Dam, with an area of 43.95 km2 (Fig. 1). Elevation in the study area ranges from 1,410 to 2,580 m above sea level, with an average of 1,893 m. The mean annual precipitation is recorded as 245 mm with almost equally spatial and temporal distributions (according to 3 meteorological stations data) (Ostad et al. 2013). The region exhibits distinctly mountainous topographical features. The major land use of the study area consists of rangeland and covers almost 76 % of Moghan basin. Also there are four other land uses including range land, field crops, orchard, residential, and cliff.
Data production In order to generate a qanat potential map for the study area, a qanat locations map was prepared for study region in 2013 based on a topographical map at a 1:50,000-scale and this was proved by the extensive field survey according to their qanat outlet (Fig. 1). By randomly partition (Lee et al. 2012; Oh et al. 2011; Pourghasemi et al. 2013a, b), 38 (seventy percent) of the qanat locations were used in groundwater potential mapping and 15 (30 %) were used for validation. For conducting a groundwater qanat potential map (GQPM), it is important to evaluate several qanat-related factors with the qanat inventory map. The main factors considered in the present study and those influential in the occurrence of a qanat are described below. Fourteen qanat impressive factors were considered in this investigation. These factors are slope degree, slope aspect, altitude, topographic wetness index (TWI), stream power index (SPI), slope-length (LS), plan curvature, profile curvature, distance to rivers, distance to faults, lithology, land use, drainage density, and fault density. Vector-type spatial database were extracted by transforming these factors using the ArcGIS 9.3 (ESRI 2008). The digital elevation model (DEM) with 20×20 m grid size has been made using the elevation contours and survey base points exhibiting the elevation values which were extracted from the 1:50,000-scale topographic maps. The DEM is used as the input to extract thematic maps of slope degree, slope aspect, altitude, topographic wetness index (TWI), stream power index (SPI), slope length (LS), plan curvature, and profile curvature (Fig. 3 (a - h)). For classification of effective factors we made use of different methods such as equal interval, natural break, normal or common standards based on literature reviews and expert knowledge (Manap et al. 2012; Pourghasemi et al. 2012a). On the other hands, the classification of factors was performed
Earth Sci Inform
Fig. 1 Qanat locations with digital elevation model (DEM) map of the study area
based on author’s experiences of study area and its condition. Slope plays a significant role in infiltration and runoff (Sarkar et al. 2001) and also in groundwater potential mapping, so the slope degree map of study area was prepared and classified into 4 classes according to equal interval scheme including 0–5, 5–15, 15–30 and >30° classes (Fig. 3a). The slope aspects were classified into 9 groups including eight directions and flat based on normal or common standard classification (Fig. 3b). Also the altitude map was prepared and classified into 5 classes based on equal interval classification scheme (Fig. 3c).
There is a vital role for topography in the spatial variation of hydrological conditions such as soil moisture and groundwater flow. Therefore, the secondary topographic indices have been used for describing spatial soil moisture patterns (Moore et al. 1991). The topographic wetness index (TWI) is a secondary topographic factor within the runoff model which is clarified according to (Moore et al. 1991): β TWI ¼ ln tanα
ð1Þ
Earth Sci Inform
where, ß tan α
LS ¼
is the cumulative upslope area draining through a point (per unit contour length). is the slope angle at the point.
The tendency of water to accumulate at any point in the catchment (in terms of α) and the tendency of gravitational forces to move that water down slope (indicated in terms of tan b as an approximate hydraulic β gradient) are considered by the ln tanα index. Primarily, the water infiltration depends upon material properties such as permeability and pore water pressure on the soil strength. So, TWI was regarded as a contributing factor in this study (Fig. 2d). The stream power index (SPI) (Fig. 3e) is defined by Moore et al. (1991) as: SPI ¼ Bs tan α
ð2Þ
where, Bs a
is the specific catchment area (m2) the local slope gradient measured in degrees
Slope–length (LS) is the combination of the slope steepness (S) and slope length (L). The LS factor in the Universal Soil Loss Equation (USLE) is a measure of the sediment transport capacity of overland flow (Wischmeier and Smith 1978) (Fig. 3f). The combined slope length (LS) factor can be defined by an equation suggested by Moore and Burch (1986)
Fig. 2 A profile of qanat and its components (Nazari Samani and Farzadmehr 2006)
Bs 22:13
0:6
sin α 0:0896
1:3 ð3Þ
These can be estimated a function of basic terrain characteristics and can be quickly applied in integrated land and water information system (ILWIS 3.3) (ITC 2005). In the matter of plan curvature, positive curvature shows convex, zero curvature represent flat and negative curvature depicts concave (Mohammady et al. 2012). Using system for automated geoscientific analyses (SAGA 2.8), the plan curvature map was prepared (Fig. 3g). Also profile curvature was calculated, and classified into 3 groups based on common standard classification (Pourghasemi et al. 2013b) including < (−0.001), (−0.001)–(0.001), and > (0.001) (Fig. 3h). Furthermore, employing the topographic and geological database, the distance to rivers and faults were calculated, respectively. These factors provided by Spatial Analyst extension in ArcGIS 9.3 (ESRI 2008). The river buffer was calculated in 50 m intervals and the fault buffer was calculated in 100 m intervals, as shown in according to references (Fig. 4 (a-b)). One of the most important parts in forecasting groundwater potential zones is lithology. Using a 1:100,000-scale geological map, the lithology map was prepared and classified based on lithological units (type) into 3 groups including Limestone, recrystallize, dolomitic, Shale, Phylitic, dark grey-phylite, and Terraces (Fig. 5). The land use map was created using Landsat images by Iranian forest, range land, and watershed management (http://www.frw.org.ir/pageid/34/language/
Earth Sci Inform
Fig. 3 Topographical parameter maps of the study area; (a) slope degree, (b) slope aspect, (c) altitude, (d) topographic wetness index (TWI), (e) stream power index (SPI), (f) slope-length (LS), (g) plan curvature, (h) profile curvature
en-US/Default.aspx). Five land use classes were detected such as rangeland, field crops, orchard, residential and cliff (Fig. 6). The drainage density map shows the flow of water throughout the study area and defined as the total length of all the streams
and rivers in a drainage basin divided by the total area of the drainage basin (Sarkar and Kanungo 2004). The drainage density map was generated considering a 20×20 m grid cell and was classified into three classes (Fig. 7). Similarly, fault density map was prepared and
Earth Sci Inform
Fig. 3 (continued)
Fig. 4 a Distance to Rivers (buffer), b Distance to Faults (buffer) maps
Earth Sci Inform Fig. 5 Lithology map of the study area
classified into 4 groups (Fig. 8). Finally, all the mentioned qanat leading factors were converted to a raster grid with 20× 20 m cells for application of FR and SE models.
Methodology Application of frequency ratio model According to Bonham-Carter (1994), the frequency ratio is the probability of occurrence of a certain attribute. Frequency ratio approach is based on the observed relationships between distribution of groundwater qanat locations and each Fig. 6 Land use map of the study area
groundwater-related factor, to reveal the correlation between groundwater qanat locations and the factors in the study area. The calculation steps for an FR for a class of the qanatinfluencing factor are below (Eq. 4): FR ¼
A C
! B
ð4Þ
D
where, A is the number of pixels with qanat for each factor; B is the number of total qanats in study area; C is the number of pixels in the class area of the factor; D
Earth Sci Inform Fig. 7 Drainage density map of the study area
is the number of total pixels in the study area; and FR is the frequency ratio of a class for the factor. To obtain Groundwater Qanat Occurrence Potential Index (GQPI) ratings of the factors were summed as (Eq. 5): GQPI ¼
X
ðFRÞi ði ¼ 1; 2; …; nÞ
ð5Þ
where, the GQPI represents the groundwater qanat occurrence potential index; FR is the frequency ratio of a factor; and n is the total number of factors. Fig. 9 shows the flowchart of applied methodology in this study. Fig. 8 Fault density map of the study area
Application of Shannon’s entropy model The entropy index is a measure of “evenness”-the extent to which groups are evenly distributed among organizational units (Massey and Nancy 1988). More precisely, Theil (1972) defined entropy index as a measure of the average difference between a unit’s group proportions and that of the system as a whole. There is a one-to-one relationship between the quantity of entropy of a system and the degree of disorder called Boltzmann principle and has been used to represent the thermodynamic status of a system (Yufeng and Fengxiang 2009). Shannon improved upon the Boltzmann principle and established an entropy model for information theory. The equations implemented to calculate the information coefficient (Vj) representing the weight value for the parameter as
Earth Sci Inform Fig. 9 Flow chart of methodology
Data Used
Dependent Factor (Qanat)
Effective Factors Topographic Factors
Random Partition Distance to Rivers Drainage Density
Digital Elevation Model (DEM)
Training Qanats
Validation Qanats
(70% or 38 Points)
(30% or 15 Points)
Slope Degree Slope Aspect Altitude Application of Frequency Ratio and Shannon's Entropy Models
TWI SPI Slope Length
Qanat Potential Maps in West of Mashhad area
Plan Curvature Profile Curvature
Land use ROC curve and validation of models
Geology Factors
Lithology Selection of the best model
Distance to Faults Fault Density
a whole (Bednarik et al. 2010; Constantin et al. 2011) are given as following (Eqs. 6–12): FR E ij ¼ XM j
ð6Þ
FR j¼1
where, FR is the frequency ratio and Eij is the probability density. Hj ¼ −
Mj X i¼1
E ij log2 Eij ; j ¼ 1; …; n
H j max ¼ log2 M j; M j−number of I j¼ðH j max−H
j =H j max
classes
Þ;I¼ð0;1Þ; J ¼1;…
ð7Þ
ð8Þ ð9Þ
V j ¼ I j FR
ð10Þ
where, Hj and Hj max are entropy values; Ij is the information coefficient and Mj is the number of classes. Also, Vj depicts the resultant weight value for the parameter as a whole. The result ranges between 0 and 1. The closer the value is to the number 1, the greater the imbalance is. The complete calculation of weight determination for individual parameters is presented in Table 1. Finally, the relative operating characteristics (ROC) (Ozdemir and Altural 2013; Akgün 2012; Mohammady et al. 2012; Pourghasemi et al. 2012a) was employed to determine the accuracy of qanat potential maps produced in this research implementing frequency ratio and Shannon’s entropy models. ROC curve analysis is a common
Earth Sci Inform
methodology to evaluate the accuracy of a diagnostic test (Swets 1988). ROC curve is a helpful approach for representing the quality of deterministic and probabilistic inspection and estimating systems (Swets 1988). The ROC curve could be a represented illustration of the trade-off between the false-positive (M) and false-negative (Z) rates for every possible cutoff value. The ROC curve plots the false positive rate (1- specificity) on the X (Eq. 11) axis and the true positive (S) rate (sensitivity) on the Y axis (Eq. 12) (Negnevitsky 2002). TN X ¼ FPR ¼ 1− ð11Þ TN þ FP
TP Y ¼ TPR ¼ TP þ FN
between qanat occurrence and distance to rivers, for distances to rivers between 0 and 50 m the ratio is 2.13, indicating a high probability of qanat occurrence. Relationship between qanat occurrence and distance to faults shows that qanats are most abundant in the class 0–100 m with FR value of 2.72.In case of lithology, the highest value of FR (1.25) is for group of 1 (Shale, phylitic, dark grey phylite). Considering the case of land use, it can be seen that classes of field crops and orchard have values of 2.59, 2.47, respectively. So, indicating that the probability of occurrence of qanat in these land use types is very high. The drainage density 4–6 km/km2 has the largest frequency ratio value (FR=2.30), which means the attributes of this class have the strongest relationship with qanat occurrence. Regarding the fault density, results displayed that class 3–4.5 has the highest value of 3.15.
ð12Þ
Results Frequency ratio The results of spatial relationship between occurrence of qanat and conditioning factors implementing frequency ratio model is displayed in Table 1. The ratio of the area where groundwater qanat occurred to the total area is the relation analysis; so a value of 1 shows an average value. There would be a high correlation if the value is larger than 1, a lower correlation equals to the value lower than 1 (Oh and Lee 2010). The analysis of FR for the relationship between qanat occurrence and slope degree indicate that slope degree class 0–5 has the highest value of FR (3.11) followed by 5–15 class (2.65). The lowest value of FR (0.15) is for slope degree class of>30. In the study area, was observed that when slope gradient is increasing, frequency ratio is decreasing. According to the relationship between qanat occurrence and slope aspect, qanats are most abundant on south and east-facing slopes and in the west-facing slopes have the lowest abundant. In the Table 1, for the altitude between 1,600 and 1,800 m, the FR was 2.55, which implies a very high probability of qanat occurrence. Proportional to the other altitude classes qanats are concentrated in above a topographic wetness index of 12 (value of 5.42), and assessment of stream power index showed that the class of >600 and 0–200 have the highest FR values (1.43) and (1.11) respectively. In the case of slope-length, most of the qanats occurred in 0–20 class with FR value of 1.90. Considering the case of plan curvature, results show that flat shape had highest value of 1.78 and concave shape had the lowest value of 0.25. In the case of profile curvature, most of the qanat happened in class of 2200 17253 Topographic wetness index (TWI) 12 10134 Stream power index (SPI) 0–200 18293 200–400 18086 400–600 14892 >600 58570 Slope-length (LS) 0–20 18289 20–40 22376 40–60 23701
13.37 28.86 24.25 17.82 15.71
0.00 28.00 9.00 1.00 0.00
0.00 73.68 23.68 2.63 0.00
0.00 2.55 0.98 0.15 0.00
0.00 0.69 0.27 1.06 2.32 0.04 0.00
0.54 0.40
35.71 55.06 9.23
2.00 17.00 19.00
5.26 44.74 50.00
0.15 0.81 5.42
0.02 0.13 0.70 1.58 0.85
0.56 1.18
16.65 16.47 13.56 53.32
7.00 1.00 1.00 29.00
18.42 2.63 2.63 76.32
1.11 0.16 0.19 1.43
0.38 0.06 0.07 1.52 2.00 0.50
16.65 20.37 21.58
12.00 4.00 8.00
31.58 10.53 21.05
1.90 0.52 0.98
0.44 0.12 1.85 2.00 0.23
45475
41.40
14.00
36.84
0.89
0.21
38982 24367 46492
35.49 22.18 42.33
19.00 15.00 4.00
50.00 39.47 10.53
1.41 1.78 0.25
0.41 0.52 1.29 1.58 0.07
41405 21248 47188
37.70 19.34 42.96
26.00 8.00 4.00
68.42 21.05 10.53
1.82 1.09 0.25
0.58 1.27 1.58 0.35 0.08
47708 34621 16290 11222
43.43 31.52 14.83 10.22
36.00 0.00 3.00 0.00
92.31 0.00 7.69 0.00
2.13 0.00 0.52 0.00
0.80 0.00 0.20 0.71 2.00 0.00
Slope degree 0–5 5–15 15–30 >30 Slope aspect Flat North Northeast East Southeast South Southwest West Northwest Altitude (m)
>60 Plan curvature (100/m) Concave Flat Convex Profile curvature (100/m) < (−0.001) (−0.001)-(0.001) > (0.001) Distance to Rivers (m) 0–50 50–100 100–150 >150
0.30 0.47
0.24 0.17
1.07 0.08
0.18 0.21
0.20 0.21
0.64 0.42
Earth Sci Inform Table 1 (continued) Distance to Faults (m) 0–100 100–200 200–300 >300 Lithology Shale, phylitic, dark greyphylite Limestone. recrystalizd, dolomitic Terraces Land Use Cliff Field Crops Orchard Range Land Residential Drainage Density (km/km2) 0–2 2–4 4–6 >6 Fault Density (km/km2) 0–1.5 1.5–3 3–4.5 >4.5
Fig. 10 Qanat potential susceptibility mapping by frequency ratio (FR) model
9562 9505 10142 80632
8.71 8.65 9.23 73.41
9.00 5.00 4.00 20.00
23.68 13.16 10.53 52.63
2.72 1.52 1.14 0.72
0.45 1.83 2.00 0.25 0.19 0.12
0.08 0.13
87919
80.04
38.00
100.00
1.25
1.00 0.00 1.58
1.00 0.42
21660
19.72
0.00
0.00
0.00
0.00
262
0.24
0.00
0.00
0.00
0.00
8186 6693 10543 83607 812
7.45 6.09 9.60 76.12 0.74
0.00 6.00 9.00 23.00 0.00
0.00 15.79 23.68 60.53 0.00
0.00 2.59 2.47 0.80 0.00
0.00 1.44 2.32 0.44 0.42 0.14 0.00
0.38 0.45
60753 41318 7536 234
55.31 37.62 6.86 0.21
15.00 17.00 6.00 0.00
39.47 44.74 15.79 0.00
0.71 1.19 2.30 0.00
0.17 1.43 2.00 0.28 0.55 0.00
0.39 0.30
94695 2008 2754 10384
86.21 1.83 2.51 9.45
25.00 0.00 3.00 10.00
65.79 0.00 7.89 26.32
0.76 0.00 3.15 2.78
0.11 0.00 0.47 1.40 2.00 0.42
0.30 0.51
Earth Sci Inform Table 2 The distribution of the Qanat potential values and areas with respect to the groundwater qanat occurrence potential zones
Qanat potential zoning
Shannon’s entropy Model
Frequency Ratio Model
Range
Area (%)
Range
Low
(1.82)–(4.49)
33.7
(4.93)–(11.33)
26.6
Moderate High Very High
(4.49)–(6.24) (6.24)–(9.23) (9.23)–(14.99)
39.6 17.9 8.8
(11.33)–(14.87) (14.87)–(19.47) (19.47)–(32.31)
38.8 24.8 9.8
QPMSE ¼ ððSlope degreeFR 0:470Þ þ ðSlope aspectFR 0:144Þ þ ðAltitudeFR 0:400Þ þ TWIFR 1:180 þ ðSPIFR 0:170Þ þ ðLSFR 0:820Þ þ ðPlan curvatureFR 0:211Þ þ Profile curvatureFR 0:205 þ ðDistance to riversFR 0:425Þ þ ðDistance to faultsFR 0:126Þ þ ðLithologyFR 0:416Þ þ Land useFR 0:446 þ ðDrainage densityFR 0:302Þ þ ðFault densityFR 0:506ÞÞ
ð14Þ
The obtained pixel values were then classified based on natural break classification scheme (Bednarik et al. 2010; Constantin et al. 2011; Erner et al. 2010; Falaschi et al. 2009; Ram Mohan et al. 2011; Xu et al. 2012a, b) into low, moderate, high and very high potential classes. The qanat potential map achieved from the FR method, which covered 38.8 % of the total area, was designated to be a moderate QPM class. On the other hand, the area related to low, high and very high QPM zones are 26.6 %, 24.6 %, and 9.8 %, respectively (Fig. 10 and Table 2). Also, results of Shannon’s entropy model showed that low, moderate, high and very high QPM Fig. 11 Qanat potential susceptibility mapping by Shannon’s entropy (SE) model
Area (%)
zones including 33.7 %, 39.6 %, 17.9 %, and 8.8% of the total area, respectively (Fig. 11 and Table 2).
Validation of qanat potential maps The area under the curve (AUC) of the ROC depicts the quality of a forecasting system by showing the system’s ability to expect the correct occurrence or non-occurrence of predefined “events”. In the ROC method, AUC values, ranging from 0.5 to 1.0, are used to evaluate the accuracy of the model. The ideal model represents an AUC value close to 1.0, whereas a value close to 0.5 indicates inaccuracy in the model (Fawcett 2006). The qualitative relationship between AUC and prediction accuracy can be classified as follows: 0.9–1, excellent; 0.8–0.9, very good; 0.7–0.8, good; 0.6–0.7, average; and 0.5–0.6, poor (Pourghasemi et al. 2013a). The results of the prediction curves are exhibited in Fig. 12 and 13. According to Fig. 12, it is clear that in the qanat potential map using frequency ratio model, the AUC is 0.8848, which corresponds to the prediction accuracy of 88.48 %. Also, Fig. 13 showed that AUC of Shannon's entropy model was calculated 91.21 %. Therefore, it is seen that the model
Earth Sci Inform 1
Table 3 The sensivity analysis and factor analysis for FR model
True Positive Rate
0.8
0.6
0.4
"AUC=88.48%" "SE=0.625"
0.2
0 0
0.2
0.4 0.6 False Positive Rate
0.8
1
Fig. 12 Prediction rate curve for the qanat potential map(frequency ratio Model)
produced using Shannon’s entropy exhibited better performance than the frequency ratio model in the study area. Also, in current research, was attempted to assess influence factors using of sensitivity and factor analysis for FR model. The effect analysis researches indicate how a solution changes when the input factors are changed. If the selected factor results in a relatively large change in the outcome, then the outcomes it is can be effective to that factor. Effect analysis quantifies the unreliability of each factor. The factors that have the greatest impact on the computed spring occurrence map can therefore be identified using effect analysis (Lee and Talib 2005). The results of sensitivity analysis are given in Table 3. The area under the curve which includes all factors is equal to 88.48 %. By removing altitude, the area under the curve is 84.38 %, without slope aspect 85.35 %, without distance to faults 86.52 %, without distance to rivers 87.11 %, without drainage
AUC
Factors
Included all factors Altitude Slope Aspect Distance to Faults Distance to Rivers
88.48 % 84.38 % 85.35 % 86.52 % 87.11 %
Drainage Density Fault Density Landuse Lithology Slope-length (LS) Plan Curvature Profile Curvature Slope degree SPI TWI
85.94 % 87.11 % 85.16 % 85.94 % 87.11 % 84.18 % 85.94 % 88.28 % 85.55 % 88.28 %
density 85.94 %, without fault density 87.11 %, without land use 85.16 %, without lithology 85.94 %, without slope-length (LS) 87.11 %, without plan curvature 84.18 %, without profile curvature 85.94 %, without slope degree 88.28 %, without SPI 85.55 %, and finally area of curve without TWI is 88.28 %. As presented in Table 3, plan curvature and altitude are the most important on qanat potential analysis (AUC ratio=84.18 % and 84.38 %, respectively) in study area. After their land use (AUC=85.16 %), slope aspect (AUC=85.35 %) and SPI (AUC=85.55 %) had the most effect on spring analysis. Meanwhile, the effective factors such as lithology, drainage density, and profile curvature had the same effect on QPM. The other factors show lower and positive effects on qanat potential analysis, as well.
1
Conclusion
True Positive Rate
0.8
0.6
0.4
"AUC=91.21%" "SE=0.053"
0.2
0 0
0.2
0.4 0.6 False Positive Rate
0.8
1
Fig. 13 Prediction rate of the curve for the qanat potential map(Shannon's entropy model)
Various methods have been applied for regional groundwater potential assessment globally. The FR and SE methods are used in this study to map groundwater qanat potentials. Qanats locations were determined and mapped based on extensive field surveys and topographical map. Fifty three qanats were identified in the study area; 14 qanat-related factor maps of the entire study area such as topographical factors, land use map, geological factors, and thematic maps of digital elevation model (DEM) were extracted from the spatial database and used to produce QPM map using the FR and SE methods. In addition, the predictive capability of the model was determined by the area under the ROC curve. The area value under the ROC curve for the FR model was 0.8848. Also area under the curve for Shannon’s entropy model was 0.9121, which depicts the excellence of this model in qanat occurrence
Earth Sci Inform
potential estimation in the study area. So the Shannon’s entropy model has higher AUC than the frequency ratio model. According to Shannon’s entropy results, it can be concluded that TWI and fault density have the strongest relationships with qanat occurrence. Also, factors such as slope-length (LS), distance to faults, and slope aspect had the lowest importance on qanat potential map. As a final conclusion, the produced groundwater qanat potential maps can assist planners and engineers in groundwater development plans and land use planning. Acknowledgement The authors would like to thank of Dr. Hassan Ali Babaie “Editor-in-Chief” and two anonymous reviewers for their helpful comments on the previous version of the manuscript.
References Ahmadi H, Nazari Samani A, Malekian A (2010) The qanat: a living history in Iran. Water and sustainability in arid regions. Springer, Netherlands, pp 125–138 Akgün A (2012) A comparison of landslide susceptibility maps produced by logistic regression, multi-criteria decision, and likelihood ratio methods: a case study at Ýzmir. Turkey Landslides 9(1):93–106. doi:10.1007/s10346-011-0283-7 Akgün A, Dag S, Bulut F (2008) Landslide susceptibility mapping for a landslide prone area (Findikli, NE of Turkey) by likelihoodfrequency Ratio and weighted linear combination models. Environ Geol 54(6):1127–1143 Banks D, Robins N (2002) An introduction to groundwater in crystalline bedrock. Norges geologiske undersøkelse, Trondheim Bednarik M, Magulová B, Matys M, Marschalko M (2010) Landslide susceptibility assessment of the Kralˇovany–Liptovsky’ Mikuláš railway case study. Phys Chem Earth Parts A/B/C 35(3):162–171 Bonham-Carter GF (1994) Geographic information systems for geoscientists: modeling with GIS. Pergamon Press, Ottawa Constantin M, Bednarik M, Jurchescu MC, Vlaicu M (2011) Landslide susceptibility assessment using the bivariate statistical analysis and the index of entropy in the Sibiciu Basin (Romania). Environ Earth Sci 63:397–406 Davoodi Moghaddam D, Rezaei M, Pourghasemi HR, Pourtaghi ZS, Pradhan B (2013) Groundwater spring potential mapping using bivariate statistical model and GIS in the Taleghan watershed Iran, Arabian. J Geoscience. doi:10.1007/s12517-013-1161-5 Devkota KC, Regmi AD, Pourghasemi HR, Yoshida K, Pradhan B, Ryu IC, Dhital MR, Althuwaynee F (2013) Landslide susceptibility mapping using certainty factor, index of entropy and logistic regression models in GIS and their comparison at Mugling–Narayanghat road section in Nepal Himalaya. Nat Hazards 65(1):135–165 Erner A, Sebnem H, Duzgun B (2010) Improvement of statistical landslide susceptibility mapping by using spatial and global regression methods in the case of More and Romsdal (Norway). Landslides 7: 55–68 ESRI (2008) ArcGIS 9.3 ESRI Inc., Redlands, California Falaschi F, Giacomelli F, Federici PR, Puccinelli A, D’Amato Avanzi G, Pochini A, Ribolini A (2009) Logistic regression versus artificial neural networks: landslide susceptibility evaluation in a sample area of the Serchio River valley, Italy. Nat Hazards 50(3):551–569 Fawcett T (2006) An introduction to ROC analysis. Pattern Recogn Lett 27(8):861–874 Fitts CR (2002) Groundwater science. Academic, San Diego
Freeze RA, Cherry JA (1979) Groundwater. Prentice-Hall, Englewood Cliffs Greenbaum D (1992) Structural influences on the occurrence of groundwater in SE Zimbabwe. Geol Soci London Special Pub 66:77–85 International Institute for Aerospace Survey and Earth Sciences (ITC) (2005) ILWIS 3.3 user guide, www.itc.nl Jaafari A, Najafi A, Pourghasemi HR, Rezaeian J, Sattarian A (2014) GIS-based frequency ratio and index of entropy models for landslide susceptibility assessment in the Caspian forest, northern Iran. Int J Environ Sci Technol. doi:10.1007/s13762-013-0464-0 Jha MK, Chowdhury A, Chowdary VM, Peiffer S (2007) Groundwater management and development by integrated remote sensing and geographic information systems: prospects and constraints. Water Resour Manage 21(2):427–467 Lee S, Pradhan B (2007) Landslide hazard mapping at Selangor, Malaysia using frequency ratio and logistic regression model. Landslides 4(1):33–41 Lee S, Talib JA (2005) Probabilistic landslide susceptibility and factor effect analysis. Environ Geol 47(7):982–990 Lee S, Song KY, Kim Y, Park I (2012) Regional groundwater productivity potential mapping using a geographic information system (GIS) based artificial neural network model. Hydrogeol J 20(8):1511– 1527. doi:10.1007/s10040-012-0894-7 Manap MA, Nampak H, Pradhan B, Lee S, Soleiman WNA, Ramli MF (2012) Application of probabilistic-based frequency ratio model in groundwater potential mapping using remote sensing data and GIS. Arab J Geosci. doi:10.1007/s12517-012-0795-z Massey DS, Nancy AD (1988) The Dimensions of Residential. Social Forces 67(2):281–315 Mohammady M, Pourghasemi HR, Pradhan B (2012) Landslide susceptibility mapping at Golestan Province, Iran: A comparison between frequency ratio, Dempster–Shafer, and weights of-evidence models. J Asian Earth Sci 61:221–236 Moore ID, Burch GJ (1986) Sediment transport capacity of sheet and rill flow: application of unit stream power theory. Water Res 22(8): 1350–1360 Moore ID, Grayson RB, Ladson AR (1991) Digital terrain modeling: a review of hydrological, geomorphological and biological applications. Hydro Process 5:3–30 Mukherjee S (1996) Targeting saline aquifer by remote sensing and geophysical methods in a part of Hamirpur_Kanpur, India. Hydro J 19:1867–1884 Nazari Samani A, Farzadmehr J (2006) Qanat as a traditional and advantageous approach for water supply in Iran. In: Proceedings of the International Symposium on Waterand Management for Sustainable Irrigated Agriculture, Adana, Turkey Negnevitsky M (2002) Artificial Intelligence: a guide to intelligent Systems, Addison– Wesley/Pearson Education. Harlow, England Oh HJ, Lee S (2010) Assessment of ground subsidence using GIS and the weights of evidence model. Eng Geol 115:36–48 Oh HJ, Kim YS, Choi J-K, Park E, Lee S (2011) GIS mapping of regional probabilistic groundwater potential in the area of Pohang City, Korea. J Hydrol 399:158–172 Ostad M, Mosaedi A, Mesdaghi M, Samadi SZ, Sepehr A (2013) Evaluation of the relation between vegetable cover and precipitation using remote sensing in Torogh basin, Mashhad. First national conference of strategies for sustainable development in agriculture, natural resources and environmental sections. Tehran, Iran Ozdemir A (2011a) GIS-based groundwater spring potential mapping in the Sultan Mountains (Konya, Turkey) using frequency ratio, weights of evidence and logistic regression methods and their comparison. J Hydrol 411:290–308 Ozdemir A (2011b) Using a binary logistic regression method and GIS for evaluating and mapping the groundwater spring potential in the Sultan Mountains (Aksehir, Turkey). J Hydrol 405:123–136. doi:10. 1016/j.jhydrol.2011.05.015
Earth Sci Inform Ozdemir A, Altural T (2013) A comparative study of frequency ratio, weights of evidence and logistic regression methods for landslide susceptibility mapping: Sultan Mountains, SW Turkey. J Asian Earth Sci 64:180–197 Perrier E, Salkini AB (1991) Supplemental Irrigation in the Near East and North Africa. Kluwer Academic Publisher, Norwel Pourghasemi HR, Pradhan B, Gokceoglu C (2012a) Remote sensing data derived parameters and its use in landslide susceptibility assessment using Shannon’s entropy and GIS, AEROTECH IV–2012. Appl Mech Mater 225:486–491. doi:10.4028/www.scientific.net/AMM. 225.486 Pourghasemi HR, Mohammady M, Pradhan B (2012b) Landslide susceptibility mapping using index of entropy and conditional probability models in GIS: Safarood Basin, Iran. Catena 97:71–84 Pourghasemi HR, Moradi HR, Fatemi Aghda SM (2013a) GIS-based landslide susceptibility mapping with probabilistic likelihood ratio and spatial multi-criteria evaluation models (North of Tehran, Iran). Arab J Geosci. doi:10.1007/s12517-012-0825-x Pourghasemi HR, Goli Jirandeh A, Pradhan B, Xu C, Gokceoglu C (2013b) Landslide susceptibility mapping using support vector machine and GIS. J Earth Syst Sci 122(2):349–369 Pourtaghi ZS, Pourghasemi HR (2014) GIS-based groundwater spring potential assessment and mapping in the Birjand Township, southern Khorasan Province. Iran, Hydrogeology. doi:10.1007/s10040013-1089-6 Pradhan B, Suliman MDH, Awang MA (2007) Forest fire susceptibility and risk mapping using remote sensing and geographical information systems (GIS). Disaster Prevention and Management 16:344– 352. doi:10.1108/09653560710758297 Ram Mohan V, Jeyaseelan A, Naveen Raj T, Narmatha T, Jayaprakash M (2011) Landslide susceptibility mapping using frequency ratio method and GIS in south eastern part of Nilgiri District, Tamilnadu, India. Int J Geomatics & Geosci 1(4):951–961 Regmi AD, Yoshida K, Pradhan B, Pourghasemi HR, Khumamoto T, Akgun A (2013) Application of frequency ratio, statistical index and
weights-of-evidence models, and their comparison in landslide susceptibility mapping in Central Nepal Himalaya. Arab J Geosci. doi: 10.1007/s12517-012-0807-z Salih A (2007) Qanats a Unique Groundwater Management Tool in Arid Regions: The Case of Bam Region in Iran. UNESCO Water eNewsletter No. 186: QANATS Sarkar S, Kanungo DP (2004) An integrated approach for Landslide Susceptibility Mapping using remote sensing and GIS. Photogram Eng & Remote Sens 70(5):617–625 Sarkar B, Deota B, Raju P, Jugran D (2001) A geographic information system approach to evaluation of groundwater potentiality of Shamrimicro watershed in the Shimla Taluk, Himachal Pradesh. J Indian Soc Remote Sens 29(3):151–164 Swets JA (1988) Measuring the accuracy of diagnostic systems. Sci 240: 1285–1293 Theil H (1972) Statistical decomposition analysis. North-Holland Publishing Company, Amsterdam Todd DK, Mays LW (2005) Groundwater hydrology. Wiley, New York Wischmeier WH, Smith DD (1978) Predicting Rainfall Erosion Losses: A Guide to Conservation Planning, U.S. Department of Agriculture, Agriculture Handbook. No. 537 Xu C, Xu X, Dai F, Xiao J (2012a) Landslide hazard mapping using GIS and weight of evidence model in Qingshui River watershed of 2008 Wenchuan earthquake struck region. Journal of Earth Sciences 23(1):97–120 Xu C, Dai F, Xu X, Lee YH (2012b) GIS-based support vector machine modeling of earthquake-triggered landslide susceptibility in the Jianjiang River watershed, China. Geomorphology 145–146:70–80 Yilmaz I (2009) Landslide susceptibility using frequency ratio, logistic regression, artificial neural networks and their comparison: a case study from Kat landslide (Tokat-Turkey). Comput Geosci 35(6):1125–1138 Yufeng S, Fengxiang J (2009) Landslide Stability Analysis Based on Generalized Information Entropy. 2009 International Conference on Environmental Science and Information Application Technology: 83–85
