Quality control for land base and facilities conversion
Mohammad Tariq
Program Manager (GIS), Boston Water and Sewer Commission
425 Summer Street, Boston, MA 02210
David Koehler
Sr. Research Analyst,Pkmgraphics, Inc.
1300 Spring Street, Silver Spring, MD 20910
Abstract
One of the key factors in timely and successfi,d implementation of an AM/FM/GIS is the establishment
of quality control procedures for reviewing and accepting products from conversion contractors. The
Boston Water and Sewer Commission implemented uniform quality control (QC ) procedures for
inspection of hard copy and digital products to ensure confidence in the accuracy of the land base and
facilities GIS databases being developed. The hard copy and digital data are checked to confirm content
and positional accuracy of graphic features, adherence to the database design, correctness of attribute
data, annotation, topology and overall consistency and completeness of the databases.
Introduction
The Commission is responsible for providing water and sewer and storm drainage services in the City of
Boston. The City encompasses approximately 45 square miles of area and according to the 1990 US
census has a residential population of 574,283, plus an additional daytime population of 480,000. The
Commission owns and maintains approximately 1,340 miles of sewers, including 535 miles of sanitary
sewers, 490 miles of storm drains, and 315 miles of combined sewers, and a number of other facilities,
including pumping stations, tidegates and regulators. All wastewater generated in the city is transported
to the Massachusetts Water Resources Authority (MWRA) treatment facilities located at Deer Island in
Boston and Nut Island in Quincy. The Commission also owns and maintains approximately 1,182 miles
of water mains. Water for the City of Boston is purchased from the MWRA at an average of about 90
million gallons per day. There are currently 85,500 customer accounts.
The majority of the Commission’s maps and data regarding water and sewer facilities are in paper
format. Existing water and sewer-related information is maintained on 100’and 50’ scale maps. These
maps, drawings and related data are updated and revised manually.
The Commission’s major objectives in establishing an AM/FM/GIS are to:
- Develop an efficient means of maintaining up-to-date water and sewer system maps and data
bases.
- Develop and maintain a single, centralized source of facilities information which is easily
accessible to all departments.
- Develop a means of sorting, processing, manipulating and storing facilities information.
The Commission is implementing AM/FM/GIS in a phased approach. Phase 1 involved a feasibility
study to investigate current and fiture operations dealing with geographic data which would benefit from
the use of AM/FM/GIS technology. During Phase 2, Graphic Data Systems (GDS) software was selected
by the Commission as its GIS, and a data conversion pilot project was conducted for two neighborhoods
of the city. As part of the pilot project, the water and sewer facilities data was filly developed and
prototype applications were implemented. Based on the results of the pilot, a full-scale automated
mapping, facilities management and geographic information system is currently being implemented as
the final phase. During this phase a new land base is being developed from aerial photography taken
during the spring of 1995 and water and sewer facilities maps are being converted to a digital format.
The Commission has contracted for the development of a new land base map and production of digital
orthophotos and for conversion of existing water and sewer facilities maps which are registered to the
new land base data. The Commission is using a very comprehensive quality control (QC) process, which
includes review of hard copy plots of land base and facilities data and subsequent reviews of digital data
provided by the conversion contractors. This paper summarizes the quality control (QC) process used by
the Commission to check the quality of products delivered by the conversion contractors.
Database Development Activities
Land Base Mapping
new land base is being created from aerial photography taken during the spring of 1995. The land base
data includes development of 1“= 100’scale digital planimetric and topographic data as well as digital
orthophotography. The production of land base data required the acquisition of color aerial photography
at 1”= 600’ scale taken from an altitude of 3,600 feet, and establishment of horizontal and vertical photo
control and aerotriangulation.
The planimetric features such as transportation, hydrography, buildings, wetlands and other land features
have an accuracy of +/- 2’. These land base features are stereocompiled from the aerial photography
described above. The digital orthophotos have a pixel resolution of 1 foot. The topographic data
includes 2’ contours and spot elevations derived from a Digital Terrain Model (DTM).
Facilities Conversion
The Commission’s existing 1“= 100’scale water and sewer maps are used as the primary source for
conversion of facilities data to digital format. The facilities data is registered to the new land base map
developed from the aerial photography, as described above. The facilities data includes graphic data as
well as attribute data stored in an Oracle database.
Delivery of Products
The city is divided into 12 areas for the delivery of data. The map sheets associated with these areas are
called delivery groups. The delivery groups generally contain 23 map sheets. Deliveries occur on a
monthly basis. The planimetric, topographic, and digital orthophoto data is being completed prior to the
start of the facilities conversion effort. The same delivery groups apply for both the land base data
deliveries and the facilities data deliveries.
Acceptance criteria of Delivered Products
A QC review is performed by the Commission for each delivery of products by the contractor. After QC
review of the check plots, confirmation plots or digital data, the Commission determines the status of the
product according to the following list:
All the products which have systematic errors that impede the review are rejected. The cause of
rejections are typically:
- A class of features is missing from plot
- Plot quality is poor
- Overall line quality is unacceptable
- An error, such as miscoding of features, is repeated more than 15 times
Rejected products are not completely reviewed by the Commission. The Contractor is required to
resubmit the rejected products in same form.
Corrections Required
The QC reviewers make comments for corrections during QC review. These comments maybe on check
plots, confirmation plots, or digital data, depending on the stage of the review. In case of check or
confirmation plots, the comments for corrections are placed directly on the plots. In the case of digital
QC, comments are annotated on a digital GDS drawing which overlays the relevant digital drawing files.
In response to these comments, the contractor is required to submit a confirmation plot or new digital
data at the Commission’s discretion.
The Commission accepts the submittal of the product. This authorizes the contractor to submit digital
data. If the acceptance is of digital data, then the product is filly accepted.
Accepted/BWSC Edit
The Commission determines that required corrections are minor enough to accept the submitted product,
however any errors noted will be corrected by the Commission staff once the entire digital delivery
group is accepted.
Quality contorl procedures
The QC process includes review of hard copy plots of land base and facilities data and subsequent
reviews of digital data provided by the conversion contractors. An overview of the QC process is
outlined in Figure 1 and described below. Hard copy review is completed first for a delivery group and
digital review is completed after acceptance of hard copy plots.
Review of Hard COPYCheck plots
Hard copy review is primarily performed to verify feature content, identi~ missing features, review
quality of linework, check annotation placement and content, and determine overall map quality and
appearance. Digital orthophotos are checked for image clarity, contrast, distortion or any other defects
which could impact the visual quality of the image. Check plots are used for the hard copy QC. The
Contractor is required to submit check plots for each map sheet in the delivery group. There are check
plots for each type of digital product:
- Digital orthophoto plot
- Planimetric features plot
- Planimetric and topographic features plot
- Water facilities plot
- Sewer facilities plot
Check plots are produced on a map sheet basis at a scale of 1“= 100’. The check plots are produced in
color to assist reviewers in identi~ing features. The check plots are then marked by reviewers with
comments that are marked on the outside boundary of the check plot in an orderly fashion. The hard
copy review is performed as follows:
1. Preliminary Review
Upon receipt of check plots, the Commission reviews the plots to ensure that they are sufficiently
legible and complete, and are plotted according to the acceptable standards using the required scale
and sheet format. Once legibility has been determined, the check plots are logged into the GIS
quality control status tracking application and assigned for review.
2. Digital Orthophoto Review
The digital orthophoto plots are checked primarily for image clarity, contras$ image distortion or any
defects that could impact the visual quality of the image, such as cloud shadows, misalignments, or
dirt. Adjacent orthophoto plots are checked for uniformity of color representation, contrast, and
clarity. If there are any problems with image quality, the digital orthophoto plot(s) are returned to
the contractor with written documentation of the problems. The contractor is responsible for the
correction of digital orthophotos and is required to submit another series of hard copy check plots for
review. If any problems with the rectification of the orthophotos are evident, the entire delivery
group is rejected.
3. Detailed Planimetric Data Review
The orthophotos are used to check the planimetric data using a light table overlay where the
planimetric plot is registered to the digital orthophoto to veri~ the general feature content. Precise
registration is verified during the digital QC checks. When planimetric features on the photo image
that have been designated in the database design are not captured, corrections are noted on the check
plot. To streamline the feature checking procedure, the QC reviewer checks features on a block by
block basis. Small check marks are placed next to each feature to keep track of areas on the check
plot as they are examined. Street centerlines are checked for their continuity and to ensure that they
are joined at intersections and represent a schematic geometric center of streets. Symbology, general
appearance and quality of graphic features are reviewed, specifically smoothness of lines and
absence of unnecessary spikes, jags, or wobbles. Finally, annotation is examined to ensure that
features requiring annotation according to the database design have the correct character style and
size. Annotation content is compared with a known source to ensure that features are labeled
accurately. Street and place names are checked using sources such as existing 100’scale maps and
the City of Bostons’s Street Index.
4. Detailed Topographic Data Review
The topographic plots are examined for crossing contours, contours out of sequence, unusual spikes
that protrude from contour lines, and correct Iinestyles for the various types of contour lines (e.g.
index contours and depression contours). Areas that are checked carefhlly are large flat areas such as
parking lots, or large building footprints. These are generally areas that have been leveled off during
construction and should be represented as such by the contour lines. When the contour
representation across an area is suspect, the representation can be checked by referring to the original
aerial photos. By using a stereoscope to examine the stereo pair associated with the area in question,
topographic representation on the check plot can be evaluated. Further evaluation can be
accomplished by reviewing the DTM data. Plots of DTMs can also be requested, if necessary. Edits,
which the contractor subsequently makes to the contours, should be reflected in the DTM data.
Topographic annotation is also checked for index contours. For its placement, labels should be
“stacked” at appropriate points, and they should not interfere with planimetric features. As with the
planimetric plot, the QC reviewer examines the topographic check plot block by block to efficiently
cover the entire map sheet and ensure that all features are checked.
5. Water and Sewer Facilities Data Review
A light table overlay method is used to compare the water and sewer system check plots with the
original source maps. As with the pkmimetric check plot, small check marks are placed next to each
feature to keep track of the pipes and appurtenances that have been checked. The plots are checked
to ensure that all features on the source maps are captured in the digital data, and the alignment of
pipes and appurtenances are checked against the source maps. There is a significant amount of
annotation on the facility maps and it must be captured correctly. The annotation check is done with
the check plot and the source map side by side, and each annotation block is also marked with a
small check as it is compared to the corresponding annotation on the source maps.
6. Designation of Check Plots
The Commission determines the designation of each completed check plot in accordance with the
four possible acceptance criteria described above. It is possible that an orthophoto can be rejected
for aesthetic reasons and the planimetric and topographic features still be accepted for the same map.
If either the planimetric or the topographic plot are rejected, this constitutes an overall rejection of
the planimetric or topo data. Both pkmimetric and topo data must be accepted to designate the map
sheet as accepted. As long as the digital orthophoto is adequate to support planimetric compilation,
then the orthophoto can be rejected and the planimetric and topographic features can still be
accepted. The complete transmittal back to the Contractor is prepared with the result and required
action indicated for each map sheet in the delivery group.
Review of Hard Copy Confirmation Plots
The Contractor corrects all the errors indicated on the check plots and submits confirmation plots when
requested to do so. A confirmation plot is sent to the Commission along with the original check plot so
that QC reviewers can specifically verifj that the indicated errors have been corrected.
As with the check plots, the confirmation plots are given a designation of one of the four acceptance
categories described earlier. Confirmation plots are only rejected if corrections are not made as noted on
the check plots. Confirmation plots are returned for corrections if the requested corrections are not made
satisfactorily. The plot is accepted if all the indicated errors have been corrected. If it is determined that
the contractor did not carry out the corrections to an acceptable level or new errors are introduced or new
errors are present that were not in the original check plot, the confirmation plot is returned to the
contractor for corrections.
Performing Digital QC
After the hard copy QC has been completed for an entire delivery group, digital QC is performed on the
delivexy group. The purpose of digital QC is to verifi digital orthophoto tone and contrast, planimetric
and facility feature coding errors, edgematching, correct topology, data structure, and compliance with
the database design. Two applications (one for land base data and another for facilities data) were
developed in GDS to check that all the features were captured according to the database design. Figure 2
shows an example of one of the GDS QC application menus. A description of the components of the two
QC applications follows:
1. Digital Orthophoto ReviewBoth color and monochrome versions of the digital orthophotos are checked visually on the screen
for tone and contrast. The edges are checked against the map sheet grid to ensure proper alignment
and the planimetric features are checked for accurate registration. The orthophotos are also checked
for image distortion and mosaicking with the adjacent orthophoto. The hardware zoom and pan
fi.mctions are used to zoom in on ground features such as pavement or water edges to check for
registration errors between the vector data and the digital orthophotos.
2. Specifications CheckAll the features which have coding errors are displayed and queried to determine the coding error. A
plot of the features with wrong codes can be generated, if so desired. A report that lists the features
that do not correspond to the database design in terms of coding, line style, or character style is
created, and the features that do not conform are assigned error codes, such as ERO (objects) or ERI
(items).
3. Predetermined DisplaysPredetermined feature displays are required for the more detailed feature checks to be performed.
All features are drawn in white on the screen. Subsequent choices from feature selection sets will
overlay the white background with colored line representations on a feature by feature basis. This
function of the QC application allows the users to query any features that do not get drawn in
response to the correct feature code, or features that do not get drawn at all. For example, if the user
specifies that “wetlands” features to be drawn in blue on top of white planimetric features, and some
“wetlands” remains white rather than turning blue, the user will recognize that those “wetlands”
features have received incorrect codes. This differs from the land base specifications check since a
feature code might be a valid code according to the database design, but the code is inappropriately
assigned to the individual feature (e.g. the miscoded “wetlands” features are queried and it is
discovered that they have been accidentally coded as “river”).
4. Data Structure and Topological Checks
The QC applications provide selections to check for required topology in the data and to highlight
any situations where the structure of the digital data is inappropriate. For instance by assigning a
special linestyle to line features, such as pipes, small ticks can be created at every vertex to show
where some vertices exist in the middle of a straight section of pipe or to illustrate sections where
vertices need to be added.
Another function of the QC application allows the user to check the connectivity of lines. This is
important for checking the water and sewer systems and street centerlines. Lines to be checked are
displayed, then the user is prompted for two points along the network of lines. The program traces a
line between the two points and beeps if no connecting path can be found.
Another selection is to shade all topologically structured polygons, such as buildings and parking lots
to ensure that these have been assigned topology as required.
5. Edit Calls
The QC applications have fimctions designed to indicate errors on a GDS drawing (called an ERR
drawing) that overlays the GIS data. The user is provided with a set of menu selections that provide
tools to quickly indicate Wd describe errors on the ERR drawing. First the user must specify which
752?drawing is being QC’d in order to set a label which refers to that drawing, then the user can choose
from an assortment of tools which indicates a feature or defines an area that needs to be corrected.
Another fimction produces an error report listing all the features that have been assigned wrong
codes. Both the error report and the ERR drawing are provided to the contractor for making
corrections to the digital data.
Lessons Learned
- A significant investment of time must be expended in the quality control process. Time estimates
used for planning quality control activities are (per map sheet):
| Digital Orthophoto Check Plot |
-30 minutes |
| Planimetric Check Plot |
-90 minutes |
| Topo Check Plot |
-30 minutes |
| Water System Check Plot |
-180 minutes |
| Sewer System Check Plot |
-180 minutes |
| Planimetric, Topographic, and Digital Orthophoto Digital QC |
-90 minutes |
| Water System Digital QC |
-120 minutes |
| Sewer System Digital QC |
-120 minutes |
- There area variety of options that an organization can consider when planning the logistics of
conducting quality control during a large data conversion project:
Contract out all quality control work
Complete in-house quality control
Split QC responsibility with a contractor
Contract out QC management, QC staff training, and QC application development
The choice depends on the amount of money that can be spent on such an endeavor and the
flexibility an organization has in terms of hiring. The more QC there is conducted in-house the less
expensive the QC will be, but there will be requirements for hiring a QC manager or assigning those
duties to an existing employee. The QC duties will take the QC manager away from most other work
until the period of data conversion is over (anywhere from 1 to 3 years or more). QC technicians
will also need to be hired. It is suggested that there be significant in-house GIS expertise if complete
in-house QC review is planned by an organization.
The other options are varying degrees of using contractors. If the organization feels it can provide
technical staff, but needs expertise, contracting out QC management, training and application
development can be effective. Splitting QC responsibility with a contractor is appropriate when an
organization feels it can only commit portions of their staff members time. Finally contracting out
all QC work, can be effective when an organization has restrictions on hiring and does not have the
existing personnel to carry out the necessary steps.
- Development of a database design is extremely important and should be understood by both the QC
reviewers and the conversion contractor. An initial pilot project should be conducted that tests the
adequacy of the database design. The pilot project should result in a database design that has been
updated to reflect all changes necessary to accommodate unforeseen circumstances. Also, a formal
procedure should be developed for proposing and accepting changes to the database design during
the compilation and conversion process.
- A large data conversion project results in the generation of a great quantity of hard copy plots that
need to be kept in an organized fashion. When planning for quality control, provisions must be
made for storing plots in a manner that allows el%cient retrieval by delivery group and status. Often
there will be check plot deliveries for a delivery group during the same time period that the
contractor is submitting confirmation plots with accompanying check plots from a previous delivery
group. It is important for the sake of efficiency to be able to store them all separately.
Several methods for tracking quality control must be implemented. A digital quality control
application that color codes map sheet tiles according to the acceptance status, scheduled delivery
date, delivery group, and staff member assignment is extremely helpfid. This provides a quick way
to generate graphic status reports. Other aspects of quality control tracking involve developing
transmittal sheets, and a quality control log.
Conclusions
Data conversion is a complex task and requires careful planning and management. A comprehensive
workpkm should be developed which outlines what features are to be captured, how they are to be
captured and represented and coded, and what attributes are to be recorded. A pilot project can resolve
these problems early in the conversion effort and can serve as a guideline for a full scale conversion
process. A mechanism for maintaining communication between the contractor and client for
identification of any anomalies in the source materials, situations not taken into account by written
specifications, or errors in data conversion throughout the conversion process should be developed.
Once a problem is identified, the contractor and the client should work together to resolve the problem.
The accepted resolution should then be used to modifi the workplan for the data conversion.
QC should not be thought of as merely “catching contractor mistakes”. While the primary purpose of
carrying out QC procedures is to identifi and correct errors in delivered products, the QC process serves
many fimctions:
- Quality control insures that the digital database is complete in its content of graphic features and
associated attributes.
- Quality control familiarized users at the Commission with their digital database and its uses.
Carrying out quality control procedures provides an opportunity to identi~ anomalies in source
documents and to correct them.
- Digital quality control familiarizes users at the Commission with their GIS software and its
capabilities.
- Quality control develops a level of confidence in the digital database among users at the
Commission.
Boston Water and Sewer Commission Land Base Conversion Quality Control Process
|