Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 118
APPENDIX E
HMCFS Sampling and Scheduling
E.1 HMCFS Sampling and Scheduling
Identifying exactly how much HMCFS data needs to be collected can be challenging. Although
more good quality data is not a negative by itself, collecting much more data than is necessary may
result in misdirected use of scarce resources, such as funding or personnel time. Traditional sur-
vey sampling designs may identify the number of units to be sampled (e.g., 1,000 persons), while
traffic sampling procedures identified in FHWA's Traffic Monitoring Guide (14) recommend
24-, 48-, or even 72-hour samples. Neither of these are practical for most local HMCFS because
· Traffic levels may not be known or highly variable by time of day, day of week, or season of year;
· Long duration samples can be very difficult to achieve in practice when conducted manually,
as is the case for HMCFS data collection of UN/NA placard IDs; or
· A high level of sampling may not be required to develop a general characterization of hazmat
traffic, which may be sufficient for some HMCFS objectives.
The following sections describe six different types of sampling and their scheduling: conven-
ience, representative, cluster, stratified and proportional, random, and census. Several of the
descriptions include examples of how that sampling technique might be implemented for a
hypothetical HMCFS.
E.2 Convenience Sample Scheduling
For a convenience sample, data are collected at opportune times and locations. For example,
data collectors might conduct truck counts before work, during lunch breaks, and after work at
an intersection or location between their home and workplace, or some other location when they
have time to do so on any given day. These data may provide a general sense of traffic levels at
certain times and locations, but are unlikely to give a reliable estimate of traffic patterns in the
area. However, as the number of data collectors and range of times and locations for which data
are collected increases, the quality and reliability of data for some locations may improve. With-
out a very large pool of convenience sample data it will be difficult to determine traffic patterns
across a jurisdictional area at different times (aside from chance). However, convenience sam-
pling can be used to provide a very general idea of hazmat transportation in certain areas of the
community. Moreover, some routes or route segments are likely to be well represented, but others
are likely to be left unobserved.
For example, three health professionals from a local hospital located on an Interstate bypass
in a rural county's main city (the county seat, located at the center of the county) volunteer to
participate in HMCFS data collection. One volunteer occasionally has some extra time for data
collection on Monday and Tuesday mornings before work, one during lunch break on Mondays
E-1
OCR for page 119
E-2 Guidebook for Conducting Local Hazardous Materials Commodity Flow Studies
and Wednesdays, and one after work on Thursdays. Whenever they have some extra time, the
volunteers conduct truck and placard counts from the hospital parking lot that overlooks the
roadway. Because of how the roadway is constructed, they can only collect data for westbound
traffic. These data can provide only a very general indication of hazmat traffic patterns for the west-
bound traffic on the roadway throughout the week. Note that if the volunteers collected a lot of
data (say, at least five different data counts) for each of those days and times, that could provide a
better picture of traffic patterns but only for those particular days and times for that roadway.
E.3 Representative Sample Scheduling
With representative sampling, the data collection locations are selected to represent major
types of hazmat transport corridors in the community. For example, data collection might be
conducted at one location on an Interstate, one location on a bypass loop, one location on a
major urban arterial, and one location at a downtown intersection of primary roads. The data
collection would be scheduled at each location at different times during the morning, daytime,
and evening but not on any particular day of the week or month of year. The collected data can
be used to establish general traffic patterns for these particular locations throughout the day (e.g.,
lower traffic levels during morning/evening and higher traffic levels during the day). The data
also can be used to generally characterize the type of traffic on similar roads, but they cannot be
used to accurately describe traffic characteristics on other roads or determine patterns of truck
transport throughout an area. Without a very large pool of representative sample data, it will be
difficult to determine differences in traffic patterns across different days of the week or months
of the year.
For example, a volunteer fire department is located in a community near an Interstate high-
way. Three firefighters from the department participate in HMCFS data collection. Over the
course of several months, the volunteers conduct truck and placard counts on each direction of
the Interstate during weekdays. They make sure that they have at least a half-hour of collected
data for each daytime hour (e.g., 89 A.M.) and for each direction. They also coordinate to col-
lect data during the daytime on Saturdays--on one Saturday they count in the morning and on
another Saturday they count in the afternoon. The LEPC assumes that these traffic counts rep-
resent traffic on the Interstate at the other end of the county and that the truck and placard traf-
fic is similar for all weekdays at other times of the year for the weekday counts and for all weekend
days at other times of the year based on the Saturday counts.
E.4 Cluster Sample Scheduling
Cluster samples expand representative samples and are often best suited for situations where
the goals and objectives are focused on very specific routes and route segments. For example,
data locations are selected on an Interstate on both sides of a community, on major highways
and arterials, and at key intersections. Data are collected at multiple times for each day of the
week, throughout each day, at all locations. Data collection may be expanded to represent dif-
ferent months or seasons of the year. Although data may not be usable to characterize traffic flow
patterns for an entire transport network, the traffic levels for the individual major components
of a transportation network can begin to be identified for different days of the week and differ-
ent times of the year, assuming that the observed traffic patterns apply to other times for which
traffic is not observed.
For example, a school complex (elementary, junior high, and high school) is located near an
Interstate highway. This section of Interstate has had several major truck accidents in the past
OCR for page 120
HMCFS Sampling and Scheduling E-3
decade. Community officials are concerned that their emergency warning and communication
system and shelter-in-place procedures are appropriate to the hazards that may be present, espe-
cially since the schools, including playground and outdoor athletic facilities, were constructed
on land near the Interstate. The LEPC schedules data collection for this section of the Interstate
over the course of 3 months during the spring (MarchMay). The schedule over the 3-month
period includes three half-hour counts during each daytime hour (e.g., 89 A.M.), on three week-
days (e.g., Monday, Wednesday, and Friday) during school and after-school hours (7 A.M.7 P.M.)
and on each direction of the Interstate. The schedule is repeated so that there are two datasets
per sampled weekday.
With the approval of their supervisors and senior administrators, four city firefighters, four
city police officers, and four school teachers participate in HMCFS data collection using truck
and placard ID counts. The firefighters take responsibility for the 711 A.M. period, the police
officers for the 11 A.M.3 P.M. period, and school teachers for the 37 P.M. period. With 72 hours
of data collection per group (0.5 hours per sample × 3 samples per hour of the day × 4 hours of
the day per period × 3 days per week × 2 directions of the roadway × 2 samples per weekday =
72 hours), and 4 data collectors per group, this works out to around 18 hours of data collec-
tion for each participant over 3 months. Assuming that the observed traffic represents the
overall traffic during this time period, this should provide the community with a very good
idea of the springtime, weekday, daytime hazmat transport hazards on that portion of the
Interstate.
E.5 Stratified and Proportional Sample Scheduling
Both stratified and proportional samples require prior knowledge of the sampled population
to determine the required data collection parameters. For example, previous data on traffic
counts might be used to identify average expected traffic levels on a daily basis at key transporta-
tion network locations. Previous information about traffic levels at each location may also be
available. For example, at one location it may be known that peak traffic during the day is three
times the level that is seen during the night, with mid-morning and mid-afternoon traffic lev-
els twice that seen during the night, on average. Based on this information, a stratified sample
determines the total number of vehicles that need to be counted in the morning, at peak hours,
in the afternoon, and at night. This calculation is completed for each network location. Data
are fully collected when the number of sampled vehicles is obtained at each location and each
designated time.
A proportional sample might separate the time periods into fixed length segments (e.g.,
30-minute or 1-hour slots), and sample them proportionally to the expected traffic in each time
period. The schedule of data collection at each location would then reflect the expected volume
of traffic in these locations. Given daily and seasonal variations in traffic patterns, either process
may need to be repeated for each location and time period. Overall estimates of average annual
daily traffic may be available from metropolitan and state planning agencies for major roadways
and combined with estimates of daily and seasonal traffic patterns. However, the statistical com-
putations associated with determining stratified and proportional sampling make this method
generally impractical for most hazmat traffic survey applications other than those that require
very in-depth knowledge of traffic patterns and have sufficient resources available for coordinating
and conducting the data collection.
Local entities whose HMCFS requires stratified and proportional sampling may consider ask-
ing a transportation professional, consultant, university faculty member, or other person with
statistical training in traffic analysis for assistance with sampling design.
OCR for page 121
E-4 Guidebook for Conducting Local Hazardous Materials Commodity Flow Studies
E.6 Random Samples
Traffic observations are made in a random manner, either by time of day/week/month or by
number of vehicles, throughout a transportation network. Random samples are most appropri-
ate when goals and objectives are focused on a limited number of routes or route segments, and
when the decision objectives require high degrees of reliability. Otherwise, random samples can
result in data collection that is expensive and time consuming. Random samples are usually
unnecessary except for all but the most extreme hazmat transport applications, especially since
other less expensive sampling procedures can yield adequate information for most objectives. A
data collector simply going out to different locations at different times as convenient (see Sec-
tion D.2) is not a random sample. Local entities whose HMCFS requires random sampling may
consider asking a transportation professional, consultant, university faculty member, or other
person with statistical training in traffic analysis for assistance with sampling design.
E.7 Census
A complete census of all traffic on a transportation network is nearly impossible to obtain
without automated data collection procedures such as tag readers or video-based systems that
collect data about vehicle locations and commodities carried. Although systems capable of con-
ducting a census of hazmat traffic have been conceptualized, none warrant serious consideration
in the immediate timeframe for local jurisdictions. As future technology development and data
collection procedures develop, collection of hazmat transport census data may become feasible.
