The U.S. Environmental Protection Agency (USEPA), in conjunction with the U.S. Army Corps of Engineers (USACE), Michigan Department of Environmental Quality (MDEQ) and Eastern Michigan University, is developing a desktop GIS system for managing, analyzing, and visualizing contaminated sediments in tributaries, harbors, and coastal zones of the Great Lakes. The system is designed to meet challenges for a common database and a set of versatile analytical tools to support quality assurance, regulatory and enforcement activities, and in-field sampling operation for State and Federal government agencies. The system is built on top of Esri's ArcView 2.1, customized Avenue scripts, and specialized environmental analytical modules.
The system applies the technology of expert system to integrate administrative and professional intelligence with the technical capacity of ArcView 2.1. It has a built-in machine learning mechanism to enable users to accumulate knowledge or obtain insights from past system experience for formulating analytical tasks. With this added functionality, the system provides powerful analytic tools and flexible query builders for examining pollutant issues under various scenarios in priority geographic areas. The system also includes a customized field-data-entry module to facilitate in-field sampling activities. This module imports GPS readings, updates sampling site maps, and links with other environmental factors in real time. The module can be employed to locate in- field sampling sites for pollutant hot spot search through integration with the USEPA statistical package. This module also supports data entry of in-field laboratory results, which recognized the database structure, transfers common information (key table fields) to relevant tables, and automatically leads users to next appropriate table on the hierarchy after data entry is done for a table. Compulsory data entry fields must be entered to create a record, while others fields can be filled out in greater detail at a later time as warranted. The system is an all purpose visualization tool box that supports large scale (currently up to 1:1,200 scale) multi-media display of maps, images, photos, graphics and drawings depicting the environmental impacts of contaminated sediments. In addition to the contaminated sediment data, data on sources, hydrology, transportation, hypsography, and land use are being developed to enhance system capabilities.
Introduction
Beginning in 1993, the U.S. Army Corps of
Engineers - Detroit District (USACE-Detroit) developed
a relational database to manage sediment chemistry data
collected in the major waterways in Southeast Michigan.
This work was performed to support the U.S
Environmental Protection Agency - Region 5 (USEPA-
Chicago), and the Michigan Department of Environmental
Quality (MDEQ) for quality assurance, regulatory and
enforcement activities in and out of the field with the
use of applied GIS technology, under the South East
Michigan Initiative (called SEMI Project). The SEMI
project is expected to ultimately provide each of the
three agencies with a comprehensive geographic
information system (GIS) as well as a compatible
database for importing, analyzing, modeling,
visualizing, and reporting sediment sampling and
contaminant information for the southeast Michigan
region. Eastern Michigan University (EMU) has been
called in to provide GIS technical support to the SEMI
system. The SEMI project area includes the following
counties in Michigan: Lenawee, Livingston, Macomb,
Monroe, Oakland, St. Clair, Washtenaw, and Wayne, and
the watersheds: St. Clair, Black, Clinton, Detroit,
Rouge, Huron, and Raison Rivers, and Lake St. Clair.
Recently, the urgency for integrating decision
support capacity in the SEMI system becomes
overwhelming. As a result, the SEMI system has been
formally replaced by the Fully Integrated Environmental
Location Decision Support (FIELDS) System. The FIELDS
system is designed to be an innovative and cost
effective desk-top GIS system to support objective
decision making in and out of the field for managing,
analyzing, modeling and visualizing environmental
information within priority areas and sites (Williams,
Lin, and Graettinger, 1996). The completion of a
three-phase FIELDS system prototype in Southeast
Michigan (the original SEMI system) is scheduled to go
on-line by September, 1996. This paper introduces
major technical innovations and capacities of the
FIELDS system.
A Desktop GIS for Multiple Applications
The FIELDS system provides a set of dynamic and
powerful GIS tools for importing, analyzing, modeling,
visualizing, and reporting environmenal information to
assist decision-making in channel dredging, quality
assurance, regulatory and enforcement activities. In
addition to popular GIS functionalities of spatial data
management, mapping and analysis, the FIELDS is
designed to fulfill specific agency analytical tasks,
including,
(1) identification of pollutant hot spots;
(2) query of hot-spot laboratory results and
related environmental information;
(3) volumetric calculation of contaminated
sediment;
(4) estimate of engineering cost of sediment
removal and dredging; and
(5) customized visualization tools to communicate
decisions.
Fig. 2 Multi-Step Search and Query is a successful
example of using the FIELDS system to search for hot
spots of contaminated sediment and to query laboratory
measurement results and related environmental
information in the Monroe Harbor, Michigan. This
customized analytical function involves multi-step
queries through dynamic join/link among a number of
data sets (see Fig. 3).
We first search for hot spots among all sampling
sites by linking and querying the pollutant laboratory
measurement results, and create a ArcView hot-spot
shapefile (the red dots in Fig. 2). Then we link the
newly created hot spots with the laboratory results and
other environmental information to conduct further
queries for subjective decision making (the yellow dot
in Fig. 2).
Fig. 3 demonstrates an important analytical task for
engineering estimate of sediment removal. However, due
to the nature of field sampling and laboratory
procedures, this is a tough programming job. For
example, each hot spot may have tens of lab samples and
each lab sample may have hundreds of chemical tests.
So changeable are the number of hot spots in a
designated area, the number of lab samples for each hot
spot, and the number of chemical tests for each lab
sample. Therefore, the calculation of the average
depth of lower section samples and the average
concentration of pollutants becomes tremendously
complicated. Fortunately, with the innovative
scripting in the FIELDS, we streamline this analytical
task to a button and a few dialogue browsers.
In addition to meet specialized analytical tasks
for technical staff in USEAP, USACE and MDEQ, the
FIELDS system also provides the broadest and simplest
means for novice system users to acquire immediate
visualization or query of priority areas of great
environmental concerns.
Fig. 4 provides an illustration that industrial activity
near the estuary of a river has polluted a fish sprouting
site.
Moreover, the FIELDS system handles spatial data
at very large scales (normally 1:12,000), covering
engineering details for maintaining navigation channels
by the USACE. Moreover, the USEPA prefers more
extensive sampling programs as well as more specific
chemical sampling than is currently available, both
requiring more accurate base maps. As a result, the
USEPA funded a Trenton Channel remapping projects at a
scale of 1:6,000. Maps at such a larger scale can show
detailed shoreline, hydrology, transportation and
structure features, including individual mapping of
residential, commercial, industrial, and institutional
constructions. The features were collected directly
from the digital orthophotography generated from the
project with a horizontal accuracy of +/- 1 meter. All
digital vector data are stored in a schema derived from
the Department of Defense Tri-Service Spatial Data
Standards (USACE, 1995; USGS, 1993, 1995). The FIELDS
System provides a set of display tools, taking full
advantage of the detailed base maps, sampling points,
orthophotos, and laboratory results as well as GIS
visualization power, and generating very informative
multi-media representation of contaminated sediments
and related environmental factors (see most figures in
the paper).
An Expert System Integrating with User s Knowledge and
Agency s Experiences
The FIELDS database contains laboratory results of
more than 200 chemicals, which are currently required
to be examined by the USEPA, and in addition a dozen
environmental factors, including hydrology,
hypsography, transportation, land use/cover, soil,
groundwater, and pollution sites. Numerous ways can be
employed to explore the interactions between pollution
chemicals and environmental factors. Moreover,
governments, agencies and organizations involved with
this project have shown varied interests in pollution
issues or prevention actions, and thus have different
requirements concerning query tools for examining the
contaminated sediment and related environmental data.
It is very challenging to accommodate users of various
political, operational, and technical backgrounds with
a common system, which signifies a critical need to
develop a flexible query tool box. This module is
called Query Knowledge Database (QKD), based on the
concepts and techniques of expert knowledge, machine
learning, and knowledge-base management. QKD enables
end-users to integrate their specific experience or
knowledge when performing a particular analytical task.
Explained in non-technical (popular) terms, the sub-
module On-Screen Query allows users to formalize
analytical criteria interactively in the customized
windows.
The module Knowledge Acquisition records down the
analytical statements specified in the sub-module On-
Screen Query, and stores them as a file called
Knowledge File for later references. The sub-module
also prompts users to answer a set of predefined
questions (Fig. 5),
which are appended to the knowledge file as "metadata" to
facilitate knowledge database management. Clearly this
phase performs the function of machine learning. The
sub-module Knowledge Database shows users all existing
knowledge files sorted by the involved agencies or
projects and stored in the FIELDS System, allowing
users to load, read, and modify these knowledge files
as new query statements (Fig. 6).
The module Expert Query actually launches query actions.
A Real Time GIS System Supporting USEPA/USACE In-field
Activities
One of the objectives of the USEPA is for the
desktop GIS to be capable of providing user access,
updates, and views in the field. With recent
advancements in portable laboratory analysis in the
field, the USEPA intends to use the FIELDS system on
notebook-PCs to integrate with USEPA Mud Puppy GPS
system to evaluate sampling site design plans and
contaminant levels while on-site. These tasks require
that the FIELDS System be a real-time portable GIS
system.
We have developed several Avenue Script modules to
provide these technical capacities.
Figure 7 describes the GPS Loader Module, which
reformats GPS reading export files into a format
supported by ArcView and converts the geographical
coordinates from the longitude-latitude to the state-
plane system. Then the loader module creates a point
shape file (the red dots in Fig. 7) based on the newly
converted GPS readings by launching an Add-XY-Event
function in ArcView. This loader module also supports
dynamic link or join with other environmental
information or laboratory results, assuring that the
spatial information derived from GPS readings be part
of the real time portable FIELDS system.
A field data-entry module has been designed to
allow the USEPA staff and contractors to update the
FIELDS database on-site and to visualize laboratory
results in the field. The FIELDS database is a
relational database for contaminated sediments in the
major waterways of Southeast Michigan developed by an
USACE contractor. The work involved development of an
Oracle database structure and loading routine, and a
dBase counterpart along with data dictionaries and
accompanying documentation. Population of the database
has been completed with all available sediment data
sets from the Detroit River, as well as from the
watersheds of the Clinton, Rouge, Huron, and Raisin
Rivers. The FIELDS sediment database structure has
been designed to hold all data attributes generated
from historic sampling and laboratory analysis (Fig. 8).
It includes tables for project names, station locations
and political references, sampling methods and related
information, and laboratory analysis results. The
database also contains watershed and river reach
designators to allow specific geographic look-ups and
related political or hydrologic-based queries. The
database is designed in a manner to facilitate
execution of (1) structured query for pollution
chemicals; (2) spatial query for environmental
impacts; and (3) the combination of the structured and
spatial queries for exploring joint effects of
pollutants and environmental factors.
The field data-entry module (Fig. 9) recognizes the
FIELDS database structure, transfers common information
(key table fields) to relevant tables, and automatically
leads users to next appropriate table on the hierarchy
after data entry is done for a table, Moreover, some
of these data entry fields are compulsory, while others
can be filled out in great detail at a later time, as
warranted.
Another innovation of the FIELDS System is the
integration of statistics for designing sampling plans
for sediment studies. The integration of statistics
assures that the selection of sampling sites and the
location of these sites in field can obtain an
acceptable and desirable level of accuracy and
precision at minimum costs (Lubin, Williams, and Lin,
1995). The data collected from these sample sites can
then be used in contaminant and mass/volume analysis,
sediment risk assessment, and cost and remediation
scenarios to make well-informed decisions (Fig. 10).
References
Lubin, A. N., M. H. Williams, and J. C. Lin, 1995.
Statistical Techniques Applied To Sediment Sampling
(STATSS) , USEPA Region 5, Water Division, 77 West
Jackson Blvd, Chicago, IL 60604-3590
U.S. Army Corps of Engineers, CORPSMET, April, 1995
U.S. Geological Survey, The SDTS Mapping of the DLG-E
Model, October, 1993
U.S. Geological Survey, Metadata Definition, June, 1995
Williams, M. H., J. Lin, and G. D. Graettinger, 1996.
The FIELDS System, USEPA Region 5, Water Division, 77
West Jackson Blvd, Chicago, IL 60604-3590