Introduction

The integration of remote sensing (RS), geographic information systems (GIS) and global positioning system (GPS) (3S) has received considerable attention in the field of groundwater hydrology in recent times. Groundwater is a subterranean natural resource, having multidimensional facets. The popular technique i.e. remote sensing from different platforms (e.g. aircraft, satellite and others) has become a valuable tool in developing better understanding of subsurface water conditions (Todd 1980; Barrett and Kidd 1987). Remote sensing includes geophysical surveys of gravity, magnetics and electromagnetics (Brunner et al. 2006). Geophysical survey has offered the possibility of exploring underneath information of the earth (Brunner et al. 2006). Remote sensing technique has advantage over traditional/conventional techniques in terms of spatial, spectral, radiometric and temporal data availability. It offers acquisition of real- or near-real-time data also from inaccessible or remote areas within very short span of time. Therefore, it is an efficient and powerful technique used in assessment, exploration, evaluation, analysis, monitoring and management of groundwater for the long-term societal benefits. It is a rapidly changing domain with the advent of new improved sensors, platforms and new data processing algorithms/techniques which supports new forms of data and newer views of the landscapes through which hydrologist/hydraulic engineer could find better way to evaluate the earth’s surface and other specific features (Running et al. 1994). Satellite data provides quick and useful baseline information about the parameters and/or variables which controls the occurrence and movement of groundwater, such as macro and micro topography, geology, lithology, stratigraphy, structural controls (palaeo and neo-tectonics), geomorphology, soil types, land use/cover and geological lineaments (Das 1994). With the advent of new fine spatial (hyper spatial) and hyperspectral resolution satellite and aircraft imagery, new applications for large-scale mapping and monitoring of aquifers have become possible with fine details (Fortin and Bernier 1991). In arid and semi-arid environment, the major and minor geological feature can easily be interpreted because of very little vegetative and other category of land use/cover to hide the structural and stratigraphic information. Vegetation shows a close association with geology (Engman and Gurney 1991), which offers both indirect and direct ways for quantitative and qualitative valuable information about groundwater. The vegetative parameter and variables extracted from remotely sensed images helps in the identification of subsurface manifest. Wilkinson (1996) summarized the major challenges and problems to overcome for effectual use of the new form of satellite data. Gahegan and Flack (1999) presented a way to use modern computing tools with these data and different GIS data themes to resolve some of the identified problems. In the early 1960–70s, mostly hydrologists and hydrogeologists achieved the higher success rates when sites for drilling or detailed geophysical surveys were guided by lineament mapping (Teeuw 1995), based on both aerial photography and remotely sensed images (Teme and Oni 1991). Many researches and studies have demonstrated the potentiality of satellite remote sensing in hydrological modeling, groundwater mapping, exploration and management. GIS has ability to store, arrange, retrieve, classify, manipulate, analyze and present huge spatial data and information in intelligible way (Howari et al. 2007). GIS offers a common ground upon which people, pixel and their data could easily interact (Lakhtakia et al. 1993). In current scenario, GIS is regarded as an essential and efficient technique to study extended and complex groundwater systems. The utilities and potentialities of GIS in hydrogeology and hydrologic modeling are only at its beginning, but many successful applications have already started to develop (Bhasker et al. 1992; Gossel et al. 2004) and GIS has more scope in groundwater hydrologic modeling to develop better plans and pragmatic policy. There are few very good reviews on remote sensing and GIS applications in groundwater hydrology (Engman and Gurney 1991). These reviews highlighted the crucial role of remote sensing applications in groundwater hydrology. Out of these reviews, one excellent review was done by Jha et al. (2007) entitled “Integrated Remote Sensing and Geographic Information Systems: Prospects and Constraints” which highlighted Remote Sensing and GIS technologies potential utilities in groundwater hydrology. The paper revealed six major areas of Remote Sensing and GIS applications in groundwater hydrology (1) exploration and assessment (2) selection of artificial recharge sites (3) GIS based sub-surface flow and pollution modeling (4) groundwater pollution hazard assessment and protection planning (5) estimation of natural recharge distribution (6) hydrological data analysis and process monitoring. The paper also dealt with the constraints of remote sensing and GIS applications in developing nation.

Sander et al. (1996) demonstrated the combined use of various remotely sensed images acquired from different sensors such as Landsat TM, SPOT and infrared aerial photographs along with GPS, for improving spatial accuracy and reduction in redundancy, as well as used GIS as an interoperable platform for integration multi-source data to develop well sitting strategies with improved spatial accuracy and minimized cost. Travaglia and Dainelli (2003) had applied an integrated approach including remote sensing, GIS and field data. The result of their study indicated groundwater movement along fault. Local and regional flow of groundwater can also be detected with the help of remotely sensed images (especially thermal imagery) and simple to complex, i.e. one dimensional (1D), two dimensional (2D) and three dimensional (3D) hydrologic modeling of local and regional groundwater flow can be performed through GIS. GIS has capability of geo-visualization for improved understanding. Various researches and studies have shown the positive influence of space tool and techniques in targeting the artificial recharge. When natural recharge from rainfall events cannot meet continuous increasing groundwater demands, the balance is disrupted which calls for need of artificial recharge on local and regional basis (Shultz 1994). To understand the nature of aquifer systems, many experts are now using digital techniques to derive geological, structural and geomorphologic details (Humes et al. 1994). This is necessary in order to estimate and target the regional artificial recharge sites for recharging groundwater in order to balance withdrawal and availability. The rapid delivery of spatial data can be coupled with GPS and current field computer technologies to bring the imagery into the field for cover type validation. GPS technology renders an excellent basis for x, y, z location measurements (Case 1989; Lunetta et al. 1991).

GPS technology has greatly enhanced the ease and versatility of spatial data acquisition and also diversifies the approaches through which it is integrated with RS, GIS and GPS (Gao 2002).

Hydrological modeling

In general, there are two idealized uses for the simulation in groundwater hydrology. The first is the prediction (or forecasting) of future events based on calibrated and validated model (Loague and Freeze 1985). The second object is to develop the conceptual methods for the designing of future experiments to improve the understanding of the process (Loague 1988). Remote sensing methods are the most assuring source of spatially distributed data for both calibration and model input parameters. Topography, channel positions, aquifer thickness, evapotranspiration and precipitation data are all based on remote sensing (Milzow et al. 2008). Numerous lumped-parameter models (e.g. HEC-1, HEC-2, MODFLOW, SHE and SWAT) have been linked to GIS in these ways to predict surface and groundwater flows. Orzol and McGrath (1992), for example, described how the structure of MODFLOW was altered to promote its integration with ArcInfo. They demonstrated that the results were the similar as if the model was run as a stand-alone product. Likewise, Hellweger and Maidment (1999) automated a procedure to illustrate and connect hydrologic elements in ArcInfo and ArcView and drafted the results to an ASCII file that is readable by the Hydrologic Engineering Center’s-Hydrologic Modeling System (HEC-HMS). These lumped models simulate broad spectrum of processes (e.g. surface and subsurface water flow, sediment and pollutants transport) with continuous time simulation (e.g. SWAT-Arnold et al. 1993). These watershed-based models have been connected to GIS for many years and at present, several online versions are at hand (e.g. SWAT- Srinivasan and Arnold 1994 and L-THIA 2- Lim et al. 1999). The methods used to hookup GIS and simulation models are also vary extremely from one application to another (Wilson and Gallant 2000). Watkins et al. (1996) compared the advantages and disadvantages of various GIS/Model interfaces and showed how the spatial analysis and visualization potential of GIS could be used to enhance parameter estimation/determination, scale effects, grid design, access the responsiveness of model outputs to parameter uncertainty and model discretization.

Uses of remote sensing in hydrological modeling

There are numerous forms of remote sensing among which the most familiar type is the satellite remote sensing which is cost effective application as compared to geophysical survey but, it requires ground-truthing for validation of the results to meet the proper standard. Hence, groundwater models need spatio-temporal distribution of input and calibration of data (Brunner et al. 2007a, b). If such data is available, then the models can play an integral role in improved decision-making and to minimize the chance of uncertainties. The pertaining entities such as water flux, transmissivity or head cannot be observed directly by optical remote sensing but the manifest can. Becker (2006) did the extensive review on the potential of satellite remote sensing for groundwater, the ability of remote sensing to measure groundwater potential, storage and fluxes.

Optical remote sensing in groundwater hydrology

Remote sensing can expose the information that is not observable at ground in remote areas (Blumberg 1998; Dabbagh et al. 1997). Optical remote sensing has no or very limited ground penetrating capabilities ranging from upper few centimeters to very limited depth inside the earth surface because it consists of visible and thermal domain. The optical remote sensing lies between the bands ranging from 0.4 to 12.45 μm, which behaves differently when it is interacted with different matters in different conditions. Visible domain (VD) offers information regarding state and flux variables for selecting groundwater recharge sites, water quality, simulation of groundwater systems and other variables and parameters. Thermal domain (TD) deals with land surface temperature which offers valuable information in terms of thermal aberrations that can easily be detected in thermal infrared remotely sensed data. Thermal data (near infrared) can also give information on potential groundwater recharge sites, parameters for simulation of groundwater systems, water quality, etc.

Satellite remote sensing data are available from Landsat multi spectral scanner (MSS) with 80 m spatial resolution in early 1972. Before 1972, single broad spectrum aerial photographs were used for mapping of hydrogeological units and geomorphological features. Smith (1997) has discovered that MSS band 7 (0.8–1.1 µm) was precisely suitable for distinguishing water or moist soil from dry surface due to strong absorption of water near infrared band.

Kaufman et al. (1986) have manifested the potential application of high resolution (Landsat TM) data with the combination of TM bands 1, 4 and 7 to identify different lithological units based on textured properties and vegetation densities. They have also reported main flow paths and submarine springs near the coast with the aid of TM thermal band 6. Teeuw (1995) has used Landsat Thematic Mapper (TM) to examine groundwater exploration sites in the region of fine-grained sediments in the west of Tamale in northern Ghana and to locate fracture zone with the help of Landsat TM using simple contrast enhanced (histogram- stretched) for images of band 4 (Near InfraRed, 0.76–0.90 µm), band 5 (Mid InfraRed, 1.55–1.75 µm), and band 6 (Thermal InfraRed, 10.4–12.45 µm), to produce appropriate images. Band 4 can be used in best way to distinguish different types of vegetation and soil, band 5 provides information about shallow depressions, valleys and other associated features and band 6 records thermal anomalies of the earth’s surface, which is useful for identifying major lineaments. The lineaments are the potential sites for artificial recharge and exploration. Travaglia and Dainelli (2003) have used Landsat enhanced thematic mapper (ETM) data in digital format due to the availability of three near to mid infrared bands, they extremely useful for the terrain and lineament mapping and analysis. Additionally as Landsat ETM provided eight co-registered spectral channels, this permitted large spectrum of band combinations which are useful in visual interpretation of different features. In the view of the hydrological objectives, Landsat ETM data were selected and acquired in the dry season to verify features (vegetation, soil moisture) related to the occurrence of water and to avoid overshadowing by too much vegetation. Drainage, which is conveniently visible in remote sensing imagery, thus it reflects the lithology and structure of a given area to varying degrees and can be of higher value for groundwater resources evaluation (Singh et al. 2013).

Tensional faults, that is those orthogonal to the direction of crustal extension or parallel to the direction of the tectonic stress, may be believed to open (containing water) and somewhat wider than compressive/shear faults, which are orthogonal or inclined with respect to the direction of tectonic stress and therefore tend to be tighter (containing no or very less amount of water). Thus, it should be much easier to recognize tensional faults in satellite scene than shear faults and this should be reflected in the lineaments frequency histogram.

Sener et al. (2004) have demonstrated the role of integration of GIS and remote sensing in groundwater investigation in Burdur, Turkey. They have derived informations about the geology, land use and lineament with the help of Landsat TM data composed of various analyses on the TM 7–4–1 band. In addition, contours, roads, creeks and springs were digitized from a topographic map of 1/100,000 scale to produce a drainage density map. They have produced groundwater potential map by integrating thematic maps such as maps representing annual rainfall, geology, land use, lineament density, slope, topography and drainage density. Thermal imagery has described long wave radiation which are emitted from the surface and used to examine temperature anomalies in water bodies, though it’s more complex interpretation that could be used in groundwater modeling. Sibliski and Okonkwo (2007) have used airborne MSS, air borne thermal remote sensing and ground resistivity. The results from image analysis indicates that darker shades represent cooler, deeper water areas due to subsurface and underground cavities whereas lighter shades indicates surface or near surface warmer water along with sources of contamination, direction of flow of such contamination, the position of interceptive measures and the analysis of corresponding contamination. Remote sensing and geophysical techniques together can be used for differentiating the fracture zones for well site location in Lohardaga and Gumla distract of Bihar, India (Sinha et al. 1990). Mukherjee et al. (2007) have used remote sensing and geophysical techniques for determining the facture zones for well site location in Aravali Quartzite, Delhi, India. They have used IRS 1C LISSIII data and vertical electrical sounding (VES) techniques for identification groundwater prospect zones in hard rock terrain. They have conducted VES survey using Schlumberger electrode configuration to identify the detailed variations of groundwater prospect zones. The results of these studies have showed the significance of integrated approaches of remote sensing, GIS and GPS with other geophysical techniques.

Hyperspectral remote sensing in groundwater hydrology

Hyperspectral sensors are the most advanced remote sensing systems which have capabilities of detecting and recording the object, phenomenon and processes in more than hundreds narrow spectral bands across the visible and medium infrared portions of the spectrum (typically 0.4 to about 2.5 μm). This technology has been used globally in different fields to detect discharge of surface water pollution, map sensitive vegetation distribution and map the disturbance of natural drainage adjacent to canals, etc. Determination of discharge-recharge zones by utilizing hydrologic models requires large volume of data from various sources. Integration hyperspectral remotely sensed data, GIS and GPS could act as an effective tool in characterizing groundwater flow systems (Singh et al. 2013) and discharge-recharge relationships. Hyperspectral data is used to detect the subtle changes in vegetation, water, soil and mineral reflectance (Navalgund et al. 2007). Calculating soil water balance as a function of time requires various data in addition to average evapotranspiration (ET) and precipitation (P) to account the water storage in the soil. A soil water balance model (SWBM) requires some data on field capacity of soil, which could be estimated based on soil type (Brunner et al. 2007a, b). Hyperspectral satellite information can aid (Ben-Dor et al. 2004; Brunner et al. 2007a, b) required data sets in the appropriate hydrologic modeling and management of groundwater.

However, hyperspectral remote sensing has enormous potential to extend satellite remote sensing beyond what has been possible with aerial photography and other broadband multispectral imagers. In the near future, it should be of great interest to the hydrologists, hydrogeologist/hydraulic engineers as a way to create timely and reliable information about groundwater resources. The current common limitation of hyperspectral remote sensing lies in the image processing because it provides enormous amount of data, which require different approaches and methodologies for processing and analysis.

Microwave remote sensing in groundwater hydrology

A radar image interpretation is arduous in nature when compared with visual or near-infrared image because a radar image shows the backscatter signal of emitted pulses. The launch of SeaSat satellite in 1978, the first ever civilian spaceborne imaging radar instrument (SAR) has opened a new era to the utility of radar data to track changes in Earth's oceans, land and ice (Winokur 2000). Estimating landscape surface roughness using radar helps in modeling of the land surface. Surface soil moisture data has derived from passive microwave data that provide information about the surface and subsurface processes along with the hydrologic fluctuations. The existence of clouds that appears as the single most important breakage for optical remote sensing satellite to capture the image in bad weather conditions. The development of microwave remote sensing, particularly radar imageries, solved the problem because the radar pulse can penetrate cloud cover. Various groundwater exploration and management projects incorporated both synthetic aperture radar (SAR) imagery and optical remote sensing imagery simultaneously in a single project (Chen et al. 1999). Radar remote sensing has offered the capability to deal with the problems pertaining to rainfall estimation and the utilization of microwave tomography (new imaging method based on contrast in dielectric properties of materials) as an alternative to weather radar measurement in limited areas. Various researchers have used the optical and SAR data as the basis for glaciological modeling and snow melt monitoring (Haefner et al. 1996). Apart from its all-weather capability, the most vital advantage of using SAR imagery lies in its ability to sharply distinguish boundary between land and water. Single broadband of electromagnetic spectrum has very confined capabilities in discriminating the process and spatial features present on the earth’s surface (Teeuw 1995). Various scientific researches have identified and validated the importance of microwave satellite images for groundwater assessment, exploration, management and hydrologic modeling. Edet et al. (1998) have used black and white radar imagery and aerial photographs to define hydrologic and hydro-geologic features in parts of the study area. The explicit results of the study have showed the areas in the form of high, medium and low groundwater potential zone.

Subtle changes in land elevation may be precisely measured using interferometric synthetic aperture radar (InSAR). InSAR analysis includes the imaging of the same point from two viewing angles at different times (Becker 2006). Elevation changes can be calculated from the change in phase of the reflected signal (Galloway et al. 1998). The precession of elevation measurements will depend on the climatic region. In a very humid region, accuracy of measurement is in the order of 10 cm and in very dry regions, in the order of 1 mm (Galloway et al. 1998). The capability of InSAR has been illustrated in the study of groundwater storage changes in semi-arid regions such as Southern California (Galloway et al. 1998). The utility of InSAR is augmented through its combination with numerical models of groundwater withdrawals at the basin scale (Galloway et al. 1998). The strength of InSAR is that it offers greater spatial extent and resolution of land subsidence that can be achieved through ground-based measurements (Becker 2006).

Errors in remote sensing

Here in this review paper, the common errors and their types in remote sensing are briefly highlighted. For more details, readers are referred to consult the research paper written by Lunetta et al. (1991).

  1. 1.

    Data acquisition error: Geometric aspects, sensor systems, platforms, ground control, scene consideration.

  2. 2.

    Data processing error: geometric rectification, data conversion.

  3. 3.

    Data analysis error: quantitative analysis, classification systems and data generalization.

  4. 4.

    Data conversion error: raster to vector and vector to raster.

  5. 5.

    Error assessment: sampling, spatial autocorrelation, error matrix, locational accuracy, discrete multivariate statistical techniques and reporting standards.

  6. 6.

    Final product presentation error: geometric (spatial) error, thematic (attribute) error.

Uses of GIS in hydrological modeling

In the past few years, the areas of researches based on GIS modeling have seen in elevated concern (Shamsi 2005). The use of GIS for simple two-dimensional modeling using standard overlay procedures is now widespread (Berry 1987). Extension to three-dimensional spatial and dynamic modeling is crucial for applications in many disciplines (e.g. groundwater, surface water, watershed management and modeling, climatology, marine science, geology and soil modeling) (Davis and Davis 1998; Ehlers et al. 1989). This approach involves programming the model with the available tools of GIS. Modeling within the GIS is effective for models like Universal Soil Loss Equation, DRSTIL, DRASTIC, SWAT or TR55 and others. These are mathematically and conceptually simple models, requiring least extent of developer expertise. This approach also works well for home-grown models and for components of bigger data systems (DePinto et al. 1993). It is the easiest approach because the only software one needs to know is GIS. As GIS software has emerged to include more hydrology-specific tools, this level of integration has increased in popularity.

Data bridge

Certainly, the most common approach to connecting models and GIS has been through data conversion. People have been really resourceful and creative in writing custom programs to pass spatial data from GIS to a model, later covert the results back to display and further examined in GIS, using models such as AGNPS, HEC-2, WASP4, SHE and many others.

Embedded code

This is the complex and tightest method of integration. It requires extreme programming resources and reduces the redundancy. Usually this includes embedding of the input/output routines of the GIS into the model, permitting the model to read and write GIS data in its native format. The inadequacy of intermediate conversion steps develop an application with a speed that grants development of interactive applications which was not previously possible, with models such as SWMM and MODFLOW. Some developers have utilized the graphic user interface (GUI) tool kit of the GIS to develop turnkey applications that launch the model from within the GIS so that it is hidden from the user who simply communicates through a menu. While great for the end user, this type of interface can be a nightmare for the developer to develop and maintain.

Linkage issues

Even though the methods of integration change, the issues approximately remain the same. Complex calculations like differential equations or series approximations could be done with the help of model. If the mathematics is within the realm of GIS, the modeling within the GIS can be considered doing. Some models can be conveniently canned into an intuitive GUI, but for a model with a complex interface, or one that is easily accepted like AGNPS, it is apparently best to help the model to understand data from GIS. Some models receive regular or significant updates, which can be tough to ahieve if the model is embedded inside a large application. Some models may be needed by law, or have legal implications, that could be nullified by an unauthorized developer making adjustments to the model. Additional discussions of methods and considerations for linking models and GIS can be found in Maidment (1993) and Fedra (1993).

Problems

According to Howari et al. (2007), there are three types of common problems: the existing problems, inherited problems and the computational problems. The hydraulic model usually has the capacity to analyze, predict and solve engineering problems without taking into consideration the geographical prospective (McKinney and Cai 2002). Under these circumstances, GIS becomes a valuable tool (Pullar and Springer 2000). It is noteworthy that there are strong grounds for believing that GIS has an important function to play since groundwater has the multi-dimensional aspect which has a spatial component. Furthermore, for the past two decades, many GIS integrated modeling applications have capitalized GIS as database manager and visualization tool (Westervelt and Shapiro 2000). Moreover, once data are available in GIS, it can be extracted, combined with other data, reformatted as necessary for various modeling processes and even used to generate other inputs needed by the models (Robbins and Phipps 1996).

Djokic et al. (1994) have developed a tight coupling procedure, namely ARC/HEC-2, which exports the terrain data from Arc/info into HEC-2 and converts HEC-2 water surface elevations into GIS coverage in Arc/info. Evans (1998) has created avenue scripts to import cross-section location as XYZ coordinates from terrain models to develop channel and reach geometry to be used in HEC-RAS calculations. Upon completion of the calculation, GIS was used to visualize the results. In 1998, ESRI, a leading GIS software provider has enhanced Evan’s (1998) codes and created AVRAS which is a more user-friendly feature. The latest version of AVRAS is HEC-GeoRAS 3.1.1.

According to Sui and Maggio (1999), GIS functionalities embedded in the hydrological modeling packages were primarily adopted by hydrologic developers. This approach has and advantage in terms of maximum freedom for the system design. This approach provides the ability to incorporate the latest development in hydrological modeling. The problem in this approach lies in the form of data management and visualization. The developers of the latest version of RiverCAD, HEC-HMS 2.0, RiverTools, MODFLOW and SUTRA have basically taken these approaches (Djokic et al., 1995). On the other hand, GIS software developers in recent years have made extra efforts to upgrade the analytical and modeling capabilities of their products. Hellweger and Maidment (1999) have accomplished a vital lead in the integration of hydrologic models and GIS, with the advancement of a GIS-based tool, named HECPREPRO. The tool includes the compilation of Arc/Info Language scripts (AMLs) utilized to pre-process and export spatial data to HEC-HMS, a broadly used hydrologic package with various options for simulating rainfall-runoff processes. Research and development of the tools for watershed hydrology are continuing to improve land form descriptions, surface interpolation and flow routing algorithms (Moore et al. 1993). GIS has also included tools to solve two-dimensional groundwater flow problems. These tools allow the generation of a particle tracking, Darcian flow field and Gaussian dispersion (Tauxe 1994). They have showed assuring advancement in the quick assessment of local, large-scale groundwater problems and in illustrating capture zones for well head protection.

Existing problems

GIS has supported hydrologist and hydraulic engineers with the ideal computing platform for data inventory, mapping, parameter extraction, visualization, surface modeling and interface development for hydrological modeling, thus enormously facilitating the design, calibration and implementation of several hydrologic and hydraulic models (Lanza et al. 1997; Su and Troch 2003). The technical hurdles related to the database integration are well documented (Adam and Gangopadhyay 1997). Very fewer papers in the literature have discussed the broad conceptual issues involved in the integration of GIS with hydrological modeling. Many problems in both hydrological models and the current generation of GIS are usually noticed. These problems must require to be addressed before we can make the integration of GIS with hydrological modeling theoretically consistent, scientifically rigorous and technologically interoperable (Lanza et al. 1997; Su and Troch 2003).

Inherited problems

According to Chow et al. (1988), hydrological models could possibly be classified according to the conceptualization and assumptions of three key parameters: randomness, space and time. In current practices, deterministic lumped models are dominated in GIS-based hydrological modeling. There is the availability of various deterministic lumped modeling packages such as the US Army Corps of Engineers HEC-RAS and HEC-HMS, the US Soil and Conservation Service’s TR-20 and TR-40, USDA’s SWAT, DoT WSPRO, EPA’s WASP and BASINS and USGS’s DRM3 and PRMS, etc. (Singh and Frevert 2002a, b). The future of these deterministic models has been challenged by various researches (Grayson et al. 1992; Smith and Goodrich 1996) and many researchers have been active in developing spatially distributed and stochastic models (Beven and Moore 1992; Romanowicz et al. 1993). Nevertheless, these newly developed models are widely used in practice.

Computational problem

The development of GIS till date has depended upon a limited map metaphor (Burrough and Farnk 1995). The majority of GIS databases are presently represented in vector format, which is appropriate due to storage efficiency but difficult to manipulate analytically (Howari et al. 2007). The process of vectorization or rasterization has peculiar errors that manifest as representational errors in a given GIS system (Howari et al. 2007). As a consequence, the representation schemes and analytical functionalities in GIS are equipped to map layers and geometric transformations. The layer approach inevitably forces a segmentation of geographic features (Raper and Livingstone 1995). This representation scheme is not only temporarily fixed but is also incapable of handling overlapping aspect (Hazelton et al. 1992).

Hence, to complete the seamless integration of GIS and hydrological models, more studies are required at higher level to create and incorporate novel approaches to conceptualize space and time that are interoperable together within GIS and hydrological models (Su and Troch 2003). Certainly, the current practices of integrating GIS and hydrological modeling are essentially technical in nature and have not touched more fundamental aspects in either of hydrological modeling or GIS (Howari et al. 2007). Simply being able to run a HEC-RAS or HEC-HMS model in Arc Info or a CAD system improves neither the theoretical foundation nor the performance of the model. GIS-based hydrological modeling has resulted in tremendous representational compromise (Gan et al. 1997). Tackling such problem paves the way for a fresh look at the integration of GIS with hydrological modeling.

Issues

Some of the common modeling concerns that are to be managed as described by Noman et al. (2001) are the following:

  1. 1.

    Data structure of the digital terrain model (DTM).

  2. 2.

    Hydraulic model integration.

  3. 3.

    Delineation.

  4. 4.

    Accuracy of maps.

  5. 5.

    Acceptability, flexibility and expandability and uncertainties.

Numerous modeling techniques used to integrate environmental models with GIS have been discussed and analyzed earlier by many researchers to find an optimum combination of various methods (Alaghmand et al. 2008). These involved individual pixel analysis to process model, remote and in situ sensing, data assimilation and state space estimation algorithms. In the present scenario, there are some general problems in hydrological research on passing information from point information to regionally distributed information. The correlations between ground measurements and remote sensing data are subjected to noise during collection, processing and analysis. Such stochastic relationships can, however, be used in the conditioning of stochastic models and data assimilation. Earth observation from airborne or space borne platform is the ideal observational approach capable of providing data at the relevant scales and resolution required to extrapolate findings of in situ (field) studies to larger areas, to document the heterogeneity of the landscape at the regional scale and to connect these findings into a global view (Schaepman 2006). Extrapolation can either be accomplished by statistical and/or GIS techniques (Guisan and Zimmermann 2000), as well as by process modeling of extended and complex system. Usually, only less number of point measurements are available, although groundwater models needs spatial and temporal distributions of input and calibration data. If such specific data are not available, models cannot play auxiliary role in decision making, as they are particularly undermined and uncertain. Current advancement in remote sensing has opened new sources of spatially distributed data. As the relevant entities such as heads or transmissivities, water fluxes, etc, cannot be measured directly by remote sensing, ways have to be discovered to connect the measured quantities as input data needed by the model (Brunner et al. 2007a, b). Regional hydrological models needs distributed input data. Classical hydrological measurements provide only point data, for example at a weather station, a gauging station or a borehole. In principle, the patterns from remote sensing can be translated to a deterministic distribution of input data on a cell-by-cell basis or in the form of zones. Even if absolute values of these data are uncertain, they still curtail the degree of freedom of the model and thus lead to a better-posed inverse problem and a robust solution. The integration of GIS with distributed parameters, hydrologic model is playing a raising role in designing, calibrating, modifying and comparing these models. Successful application of GIS technology in hydrologic modeling needs careful planning and extensive data manipulation. Three primary tasks have been identified in most hydrologic applications with sophisticated computerized numerical models: the spatial database construction, integration of spatial model layers and the GIS and model interface. The first task is generally time-consuming, but is becoming more justifiable with the ever-increasing volume of data available from many organizations. The second task can be generalized in a series of spatial overlaying and/or projecting procedures that yield the final modeling layer. This task is becoming less tedious with the rapid advancement of sophisticated GIS capabilities. The last task may require actual programming with complication varying from case to case. There is one article which is on both an introductory overview to GIS applications in hydrologic modeling and a review of what has been achieved in this rapidly growing field (Zhang et al. 1990). In the recent years, there has been increasing interest in the issue of scale in remote sensing (Dungan et al. 2002). With the aid of remote sensing and GIS, it is easier to predict changes in ungauged basins, perform modeling base on the data, etc. The hydrologic modeling utilize variety of models such as simple linear model to process-based model, also known as deterministic model which can be divided into single-event models and continuous simulation models. The more complex and realistic models are stochastic models, which have different conceptualization and assumption of randomness, space and time (Sui and Maggio 1999) in comparison to deterministic models. Stochastic models deals differently with uncertainties.

Wilson and Gallant (2000) have described issues concerning ‘‘scale’’ and they referred to the level of details at which information can be observed, represented, analyzed and communicated. Various researchers attempting to model hydrological processes at the scale of drainage basin have found difficulties extracting, synthesizing and aggregating data from limited number of locations (Stuart and Stocks 1993; Singh et al. 2013) and finding the suitable data for fully distributed models such as Institute of Hydrology Distributed Model (IHDM) or the Systeme Hydrologique European (SHE). These models operate in finer spatial resolution. For appropriate functioning, these models require many parameter values that are difficult to measure in the field which has limited their usefulness for practical purposes (Bathurst 1988; Stuart and Stocks 1993).

Models of integration and their need

Groundwater aquifer is a complex natural system which requires knowledge and expertise from different fields. In this era of unprecedented data proliferation, the close integration of remote sensing, GPS and GIS has expected greater importance for handling of disparate, yet intrinsically complimentary spatial data (Mesev 1997). Integration is necessary to equip the users with information in more significant way (Ehlers et al. 1991). According to Mesev (1997), the best RS–GPS–GIS integration can facilitate extended inventories, rapid database update, greater analytical flexibility and broader potential applications. Nevertheless, the worst side of RS–GPS–GIS integration may cause redundancy, analytical complexity, compounded errors (additive and multiplicative errors) and unfocussed objectives. These techniques, in recent times, provide data ranging from 100s of meters to centimeter in resolution. GPS is also a satellite-based platform utilized for identification of exact location, velocity and time, yet it is mostly used for identification of earth’s coordinates. GIS distinguishes itself from other two technologies in that it enables data from multi-sources to be collected, integrated, analyzed, retrieved and even modeled owing to its powerful analytical functionality (Gao 2002). GIS functionality cannot be fully met and realized without the fidelity in database.

According to Gao (2002), these geospatial technologies are independent of one another in their basic functions. But these technologies are fundamentally complimentary in their secondary functions. When these technologies are implemented individually, they could work properly in certain cases, but functionality of these technologies can only be fully realized through their integration to manage the natural resources. Integration not only promotes their wide ranging applications in resource management and monitoring (Thakur et al. 2011), but also widens the scope to which they are applicable (Gao 2002). According to Gao (2002), the integration of these technologies in combination with ground monitoring systems has confirmed to be an efficient method for collecting, managing, analyzing, modeling and presenting output of spatial data for local, regional and global water resources development and management (Chen et al. 1999).

Numerous researches have done useful work on integration of RS–GPS–GIS in their studies to map, identify and explore recharging sites, drilling sites, well sitting sites and others. But, maximal researches have not focused and discussed the types of models used for the integration and level, at which they were used for the integration of RS-GPS-GIS. Nevertheless, in situations such as to meeting the increasing demand of fresh water supply, inventory of potential recharge sites, groundwater pollution, impact of urbanization and industrialization on groundwater studies, closer and deeper look on the types of models of integration and levels of integration is needed.

Examples of integration in practices

According to Gao (2002), there are various diverse methods for the integration of remote sensing, GIS and GPS. He conceptualized and summarized these methods into four models: linear model (LM), interactive model (IM), hierarchical model (HM) and complex model (CM). The adopted flow charts are shown below (Figs. 1, 2, 3, 4). The details can be found in Gao (2002).

Fig. 1
figure 1

The linear model of integration (Source: Gao 2002)

Full size image
Fig. 2
figure 2

The interactive model of integration (Source: Gao 2002)

Full size image
Fig. 3
figure 3

The hierarchical model of integration (Source: Gao 2002)

Full size image
Fig. 4
figure 4

The complex model of integration (Source: Gao 2002)

Full size image

Various researchers have used the concept of RS–GPS–GIS integration in linear model in the field of groundwater sitting sites, exploration, mapping lineaments, point pollution source, potential recharge sites and others. Interactive model has mostly been used by the agriculture scientists for the estimating the crop yield. However, very few scientific studies have focused on the discussion about the type of integration.

Integrating GIS with hydrologic model

The capability to implement model in a GIS environment has anticipated rise in the number of models, generally used to predict soil loss, sediment yield, nutrient loss, pollutant transport in watershed and groundwater movement. Some of these models are AGNPS (agricultural non point source), SWAT (soil and water assessment tool), ANSWERS (aerial non point source watershed response simulation) and HSPF (hydrologic simulation program-fortran). The integrated use of GIS and prediction models can be considered a powerful instrument to support decision makers in identifying areas at risk of pesticide contamination. The integration of models with GIS has a number of benefits including short running time and quick production of results. WebGIS has been used for integration and visualization of hydrological data in the model (UIZ 2015).

Methods of integration of GIS with hydrological modeling

According to Kopp (1996), three general approaches exist to GIS model integration with hydrological modeling, namely GIS-based modeling, Data Bridge and embedded code. The integration methods are constantly developing but still the general approaches and issues remains the same. Figure 5 demonstrates the methods of GIS integration with hydrological modeling.

Fig. 5
figure 5

Integrating GIS with Hydrological modeling: current practices (Source: Sui and Maggio 1999)

Full size image

Conclusion

Water, a vital natural and a potential resource necessary for all forms of lives, is itself in a great danger in terms of degrading quality and diminishing quantity. The integration of 3S tools and techniques works as a central concept in water resource management. However, when a question arises on whether the management has to be integrated, then the answer should be yes. Integration of any known scientific technology leads to better, improved understandings and helps different streams of peoples to solve the known as well as to predict the emerging new problems. Various researches have dealt with problems relevant to water such as waterborne diseases, sewage pollution, effluent and salt-water intrusion (Thakur et al. 2012). This paper gives an insight to the researchers working in the field of geospatial application in water resource.

In many developing countries, diverse GIS applications are still being used as advanced digital cartographic systems oriented towards the maintenance of digital geographic data (Densham 1991) and relatively inferior with respect to high-level analysis and modeling abilities that are necessary for sustainable management of resources. The development and assessment of topographic and hydrologic database that extend over large areas are the areas of active research. The future of distributed mobile GIS (DM GIS) will be invaluable for applications in the fields such as emergency water supply, groundwater pollution, scientific field studies, environmental monitoring and planning (Karimi et al. 2000; Gao 2002). The most challenging issue for the efficient implementation of DM GIS is the inspection of mathematical relationships among the spatial objects (e.g. topology) for the data in the database and reconciliation of topological inconsistencies when they occur (Gao 2002).