Chapter 2

Literature Review

2.1 INTRODUCTION
The purpose of this literature review is to provide the reader with a general overview of
environmental modeling as it pertains to GIS. Recently, this concept has become a popular
subject in many science and engineering fields. Within the past five years, many conferences
have been hosted which primarily discuss research in environmental modeling and engineering
practices using GIS. The first part of this chapter gives a brief description of past and current
work that integrates modeling into GIS. Next, examples of existing models which have been
connected to or utilize GIS in some fashion is presented. Finally, a brief overview of
connections between GIS and the WASP5 water quality model is given.

2.2 ENVIRONMENTAL MODELING USING GIS

When examining previous research, attention should be given to the type of connection
established between the environmental model and the GIS software. In addition, the type of
GIS software used is a concern. Tim and Jolly (1994) present a good overview of three types
of model interfaces possible with GIS (Figure 2-1). Their paper describes the three levels of
integration as 1) ad hoc integration, 2) partial integration, and 3) complete integration. In the
first level, the GIS data structure and environmental model are developed independently. The
data is extracted from GIS, the model run separately, and the output analyzed at the user's
discretion. The second level results in GIS playing more of an integrated role in the modeling.
GIS supplies the data and then accepts the modeling results for processing and presentation.
The third level consists of complete model development within the GIS software. The user has
a single operating environment, where the data stored in the GIS is structured to meet the
demands of the model and vice versa (Tim and Jolly, 1994). It should also be noted that there
are numerous types of GIS software with which a model link can be accomplished. Most
literature reviewed for this project utilized

13
 Process
GIS Models
File
Exchange
Interface

GIS
User Interface

Pre- Post-
Processor Processor
User
Process
Models
Level 1: Ad hoc Integration

Process
GIS Models
Interactive User Interface
File
Exchange
Interface

User
Common User Interface

Level 3: Complete Integration

User

Level 2: Partial Integration

Figure 2-1 Levels of GIS/model integration.

Source: Tim and Jolly (1994)

14
the GIS software, Arc/Info and ArcView; although, some articles discussed below did
accomplish links to other GIS software programs.

Environmental Modeling Within GIS -- Level 3: Complete Integration

For the past decade, the integration of GIS with environmental modeling has become an
important research topic. The use of GIS for modeling provides ease and accuracy in the
management and spatial representation of data. Recent projects which have conducted
environmental modeling directly in the GIS have included studies in simulating hydrologic
processes, water balancing a river basin, predicting chemical concentrations in rivers, and
assessing non-point source loading over a watershed (Maidment, 1992; Stuart and Stocks, 1993;
Olivera and Maidment, 1996; Ye, 1996; Maidment, et al., 1996; Mizgalewicz, 1996; Saunders,
1996; Newell, et al., 1992).

The area of hydrologic processes is one of the primary areas of study in the literature
reviewed for this project. One application of hydrologic modeling in GIS incorporated a set of
generic modeling tools within an unspecified GIS software to implement a semi-distributed
hydrological model (Stuart and Stocks, 1993). A set of pre-established hydrologic equations,
which explain surface saturation, were integrated into GIS with the macro language of the GIS
software utilized. Spatial information needed for the equations, such as land slope, rainfall
intensity, and soil transmissivity, was stored in the form of GIS coverages. By using a GIS tool
interface, which incorporated the needed equations, along with the data coverages, the spatial
distribution of the surface saturation was calculated over a study area (Stuart and Stocks,
1993).

Besides hydrologic processes related to land saturation, numerous sources were found
pertaining to the concept of flow and transport modeling over the land surface. Maidment
(1992) investigated concepts in hydrologic modeling with GIS, using the Arc/Info software, such
as grid based watershed delineation, geographic representation of the stream network, and
dynamic segmentation for hydrologic feature locations on this stream network.

15
This research proposed a “hybrid grid network” which consisted of the delineated watershed
grid and its connected stream network created from the delineation process. The modeling of
the spatially distributed hydrologic processes over the grid was combined with a model of flow
and transport processes in the stream network to result in a network of linearly connected flow
systems. The overall model was developed in GIS, while using basic hydrologic concepts and
equations to explain the flow and transport parameters.

In an ongoing study, an application of flow process modeling uses the concepts of

spatially distributed land surface parameters to calculate runoff over a watershed area (Olivera
and Maidment, 1996). This research applies the concept of spatially distributed unit
hydrographs over an area of cells. These unit responses, which are each independent of the
surrounding responses, are convoluted along a watershed flow path to produce a total runoff
response at the watershed outlet. This entire process is executed within the GIS software,
Arc/Info and Arc/Info’s subprogram, Grid (Olivera and Maidment, 1996). In contrast, a slightly
different approach to GIS surface water flow simulation is being conducted by Ye (1996). This
model, which utilizes the newer GIS programming language, Avenue through the software,
ArcView, is designed to simulate surface flow runoff based on the precipitation or soil moisture
contents defined on the subwatersheds of a river basin. This research is not only investigating
time varying flow processes over a watershed, but is also incorporating a groundwater flow
model with the surface water flow model. Both of these models interact with the user through
tool buttons in ArcView which activate Avenue programs (Ye, 1996).

Another environmental concept which has recently been integrated into GIS is water
balancing over a river basin. Specifically, a water balance model has been developed for the
Niger Basin in North Africa (Maidment, et al., 1996). This model used a grid-based analysis to
explain a relationship between flow, precipitation, evapotransporation, water demand, and water
storage. Similar water balancing efforts using GIS have been performed on Texas and the
Souss Basin in Morocco (Olivera and Maidment, 1996).

16
 Besides hydrologic processes, GIS has been used to assess pollutant loadings entering
a water body and to explain the transport of chemicals in surface water. Various studies have
investigated the concept of non-point source (NPS) loadings from watershed areas. Two
projects in particular used GIS to develop projected areal loadings of different chemical
constituents. Saunders (1996) developed a grid-based model which assessed NPS loadings of
nitrogen, phosphorous, cadmium, and fecal coliforms into a small coastal bay from the 6000
km2 San Antonio-Nueces Basin in South Texas. The method used a grid of land use-based
estimated mean concentrations (EMCs) multiplied by spatially distributed runoff volumes to
obtain an annual areal loading over the watershed (Saunders, 1996). A similar study also
utilized the concept of EMCs and runoff volumes to develop an assessment of NPS loads into
Galveston Bay, Texas (Newell, et al., 1992). Newell, et al.(1992), also utilized GIS to
spatially distribute runoff volumes, land use characteristics, EMCs, and final loading values. A
slightly different pollutant study currently in progress is applying GIS to project chemical
concentrations in the Upper Mississippi River Basin (Mizgalewicz, 1996). Using data collected
in the United States Geological Survey (USGS) toxic chemical program throughout the
Midwest, this GIS model is meant to explain the relationship between chemical concentrations
in a stream and parameters such as chemical application, runoff, precipitation, season, and
watershed characteristics. In addition, this research aims to describe chemical losses due to
transport downstream using GIS as the ultimate modeling interface for these processes
(Mizgalewicz, 1996).

Model Connections to GIS -- Level 1: Ad hoc Integration and Level 2: Partial Integration
As discussed in the previous section, various concepts have been developed within the
GIS framework to assist in traditional environmental modeling. Of more concern for this
research, though, is the establishment of a connection between an existing environmental model
and the GIS software. Many research endeavors have investigated the feasibility of linking
various models to GIS to assist in data management, manipulation, and output processing. Of
particular interest for this project were those previous studies which concentrated on water
quality and quantity model links. These projects have ranged from

17
incorporating an entire model into the GIS software, to concentrating on a subprogram of the
model to connect to the interface.

In the area of groundwater, literature reviewed included links to the USGS model,
MODFLOW, and a European simulation program entitle MICRO-FEM (Biesheuvel and
Hemker, 1993. Brown, et al., 1996; Rindahl, 1996). Two studies included investigations with
MODFLOW. Rindahl (1996) established an "easy to use interface" through GIS to display
drawdowns, stream flow, and aquifer elevations simulated from the groundwater modeling
program. The research, which developed the link primarily for ease in output presentation,
utilized the GIS software, ArcView 2.1 and ArcView's programming language, Avenue. The
study established a polygon coverage of the modeled grid, attributed with the information
resulting from a typical MODFLOW model run. The attributes were in tabular format and
joined to the polygon coverage through a model "identification number" termed a "Loc-Tag".
Once joined to the corresponding coverage, the model results could be spatially displayed
through ArcView. Avenue scripts were also compiled which assisted the user in output
display. Another study investigating a GIS link to MODFLOW centered on the use of Arc/Info
to provide an efficient means of data preparation and visualization of simulation results (Brown,
et al., 1996). ModelGIS, an interface using FORTRAN 77 and Arc Macro Language (aml),
generated model grids, model layer elevations, aquifer properties, surface water data, and
model output. The options in ModelGIS were executed from a customized menu developed in
Arc/Info's subprogram, ArcTools. The different choices converted coverages to model based
data, assembled the data into MODFLOW input, executed the MODFLOW program, and
assisted in the evaluation of the modeling results. Throughout these processes, the user
interacted with the interface in the creation of the model grid and data inputs. This link also has
the capability of establishing a three dimensional modeling surface (Brown, et al., 1996).

The final GIS/groundwater model connection reviewed utilized the GIS software,
Integrated Land and Water Information System (ILWIS) (Biesheuvel and Hemker, 1993). This
study was primarily concerned with investigating the pre- and post-processing ability of

18
the GIS software with the European groundwater model, MICRO-FEM. The final product
resulted in the establishment of a GIS/model link by allowing ILWIS to perform the following:
1) build a model network, 2) determine values for the model input, 3) combine results of the
MICRO-FEM model run with other types of data, and 4) create background maps which can be
used during the modeling process (Biesheuvel and Hemker, 1993).

More closely related to this current research are past studies which incorporated links
to water quality models. Literature shows many different types of models and levels of
connections being established, including a link for the widely used Hydrological Simulation
Program (HSPF), as well as a connection with a number of selected equations, adapted from
existing models (Al-Abed and Whiteley, 1995; Chen, et al., 1995).

The HSPF model is used for simulation of watershed hydrology and water quality for a
variety of pollutants. The model requires many parameters which describe watershed
characteristics, including surface terrain, soils, land use, and vegetation. In addition, time series
data for rainfall, temperature and other climatic attributes are needed to run HSPF. Al-Abed
and Whiteley (1995) developed a loose connection to this watershed management model by
using information stored in GIS coverages as the HSPF input. The research used the concept of
look up tables and macros in Arc/Info to create spatial representations and tables of watershed
characteristics. This spatial information was then entered into the HSPF model by the user (Al-
Abed and Whiteley, 1995).

Two different articles relating to water quality modeling of agricultural chemicals were
reviewed. One project discussed a GIS interface to four Agricultural Research Service (ARS)
pollutant loading models: Agricultural Non-Point Source (AGNPS), A Basin Scale Simulation
Model for Soil and Water Resources Management (SWRRBWQ), Erosion Productivity Impact
Calculator (EPIC), Ground Water Loading Effects of Agricultural Management Systems
(GLEAMS) (Geter, et al., 1995). The goal of this research was to develop an interface which
resulted in standardized and consistent input data to all of the water quality models, while
providing a platform to interpret the model results through

19
tables, graphs, and maps. The user first enters the necessary model data in the form of
attributed coverages within the GIS software, Geographic Resource Analysis System (GRASS).
The total connection requires five raster based maps linked to sixteen attribute tables. This
base information is then interpreted by the GRASS interface and consistent model input is
determined. The link established actually writes the derived input into the formatted file
necessary for the models' input, and the connection did provide a means for the user to view the
model output through charts, tables, and raster maps (Geter, et al., 1995).

The second agricultural chemical pollutant article (Tim and Jolly, 1994) was interested
in demonstrating the concept of integrating the aforementioned water quality model, AGNPS,
with an Arc/Info interface. GIS provided the means to generate and spatially organize the data
needed for the non-point source modeling effort, while AGNPS was used to predict water
quality related parameters such as soil erosion and sedimentation. A partial integration link
was established (see Figure 2-1), by developing computer programs which provided "access
points" between the GIS database, the AGNPS model, and the user. The link read the model
input from raster coverages imported into Arc/Info's subprogram, Grid. Once the grid-based
data were converted to a readable format by AGNPS and the model executed, the output was
re-imported into Grid and displayed through ArcPlot (Tim and Jolly, 1994).

2.3 PREVIOUS GIS/WASP MODEL CONNECTIONS

Research was conducted to review any connections with the WASP5 model which have
been accomplished in the past. Findings indicated that one project utilized equations and
concepts from WASP5 in a GIS connection, while another study connected WASP4's
subprogram, EUTRO4, to the Arc/Info software (Chen, et al., 1995; DePinto, et al., 1993;
DePinto, et al., 1994).

The first study developed a model to evaluate the impact of land use and watershed
management practices on the water quality of a Reservoir in Taiwan (Chen, et al., 1995).

20
The article did not specify exactly which GIS software was utilized for the model development,
but it did use equations and subprograms from the WASP5 program to model nutrient cycles.
Geographic data was imported into the model from GIS and spatial attributes of some
watershed and reservoir characteristics are entered through dialog boxes. Since the article
discussed the use of the program in a Windows-based environment, ArcView with Avenue
programming may have been used, but the accuracy of this assumption is unknown. The
overall program, termed Integrated Watershed Management Model (IWMM), was run with
menus. The menu choices activated object-oriented programs to execute the equations which
modeled everything from reservoir hydraulics and hydrology, to water chemistry and nutrient
cycles (Chen, et al., 1995).

The second research project, conducted by the Great Lakes Program at the State
University of New York at Buffalo and a water quality modeling firm, Limno-Tech, Inc, was the
most developed connection to the WASP model found in this literature search (DePinto, et al.,
1993; DePinto, et al., 1994). The result of this project was termed Geographically-based
Watershed Analysis and Modeling System (GEO-WAMS). GEO-WAMS, which linked
WASP4 (version 4 of the WASP model) to Arc/Info, performs the following functions: spatial
and temporal exploratory analysis of system data; model scenario management; model input
configuration; model input data editing; model input conversion to the proper file format; model
output interpretation, reporting, and display; and model calibration, confirmation, and
application. The interface also considered time variance in the system, along with possible
three dimensional segmentation. In addition to the WASP4 connection, the system also
performed a loading analysis with a spatial watershed model. All of these processes were
conducted under a menu driven atmosphere created in Arc/Info and executed with Arc/Info's
macro language (DePinto, et al., 1993; DePinto, et al., 1994). Although this connection could
be termed a "partial integration" as discussed in Tim and Jolly (1994) (see Figure 2-1), the
overall link from the model to Arc/Info was rather complex. The data necessary for the
watershed loading model and WASP4 link included, but were not limited to, system geometry,
morphometry, hydrology, soil properties, land-use, and point-source loadings, all in the form of
GIS coverages, grids and tables. The research

21
used a portion of the Buffalo River as the study area for the prototype link and connected just
WASP4's subprogram, EUTRO4, in order to model dissolved oxygen in the reach (DePinto, et
al., 1993; DePinto, et al., 1994).

22