Page 207 - Biosystems Engineering

GIS-Based W atershed Modeling Systems 185

The most sophisticated approach for interfacing GIS with a pre-
dictive model is termed “integration” or “embedding.” In this
approach, functional components of one system are incorporated
within the other system, thus eliminating the need for intermediate
transfer software (Liao and Tim 1997). In this approach, a seamless
integration is developed through sharing of processes and data, thus
reducing redundancy (Martin et al. 2005) and increasing computa-
tional performance. A true integration, however, is difficult, as a lot of
communication is required between GIS programmers and model
developers. To the best of our knowledge, although limited attempts
have been made to integrate simpler water resources models, no
attempt to truly integrate a watershed model with GIS has been made
due to complexities involved in developing these systems.
As pointed out by a number of researchers (e.g., Tim et al. 1996;
Burrough 1997; Liao and Tim 1997) and cited by Martin et al. (2005),
interfacing strategies are limited by the lack of compatible data struc-
tures, software requirements, and model–GIS functionality require-
ments. Linking approach underutilizes the functional capabilities of
GIS by using is just as a display medium. Combining and integration
approaches are more sophisticated but are hindered by huge devel-
opment costs. Often, the interfacing strategy is limited by research
objectives, expertise of developers, and availability of resources. Most
current, state-of-the-art GIS-based WMSs can, at best, be described as
combined systems.

5.4.2 Challenges with Interfacing
Perhaps the biggest challenge to interfacing a GIS with a watershed
model is the lack of a time dimension within the GIS. The absence of
time dimension limits a user’s ability to readily model, within GIS,
spatial variability over time (Martin et al. 2005). The approach to
overcoming this is to visualize a time series of historic surveys, remote-
sensing data, or future time variations predicted by models using a
series of overlays that may be analyzed using statistical approaches
(Croft and Kessler 1996).
The relational database structure of GIS also limits the collusion
of GIS and some predictive models. The database relation, through a
common key item between two sets of databases, is a weak connec-
tion between the two entities. Martin et al. (2005) stated that “When
compared to the mathematical rigor of a hydrologic model, spatial
relationships do not effectively capture the governing hydrologic
algorithms. Differential equations utilized in a typical hydrologic
model thus have limited operability within a GIS data structure.
Accordingly, hydraulic models utilizing advanced algorithms or
complex mathematical structures are currently incapable of being
fully integrated into a GIS relational database.”