Potter T.D., Colman B.R. (co-chief editors). The handbook of weather, climate, and water: dynamics, climate physical meteorology, weather systems, and measurements
Подождите немного. Документ загружается.
cially available data logging systems. The discussion below summarizes an experi-
ence with an audit of a program that identified a flaw in a widely accepted algorithm
for calculating the scalar wind direction using a single pass method. More extensive
details on the findings can be found in Baxter (1995).
Wind speed and direction data were collected as part of a dust monitoring
program at a construction site with the data used to assess the contribution of the
construction activities to the downwind par ticulate matter concentrations. The site
was located in the middle of a densely populated urban area with a number of tall
buildings surrounding it. This made it virtually impossible to meet the EPA siting
criteria for exposure of wind sensors (U.S. EPA, 2000). While subject to building
wake turbulence, the measurements were deemed adequate to assess the general
wind direction and aid in the evaluation of the dust-producing activities.
As part of the overall program an audit was performed that identified unusual
patterns in the collected data. The site was located in southern California, and the
location was strongly influenc ed by the afternoon sea breezes. These late morning to
early evening flow patterns produce a very consistent southwest wind. The data
collected by the meteorological system, however, showed frequent interruptions of
the afternoon southwest flow with winds from a variety of other directions, including
those from the northeast. Fur ther investigation into the data logging system identi-
fied the logger used a single pass scalar average algorithm generally accepted and
described in the EPA guidelines. This algorithm corrects for the rotation through
north, allowing the proper interpretation of wind direction when winds vary from
northwest (270
–360
) to the northeast (0
–90
).
With the identification of the anomalies in the afternoon wind direction data,
several tests were performed to determine whether the problems were in the physical
instruments or whether it was in the calculation methodology. The first test involved
programming the existing data logger with an additional unit vector algorithm and
comparing the two sets of wind direction calculations. The unit vector algorithm is
also described in the EPA guidelines. Figure 3 shows the comparison of 20 days of
wind direction data when wind speeds were 1 m=s or greater. The comparison
shows some agreement, but for a significant number of values there is no obvious
relationship.
The second test placed an identical wind sensor near the first one and logged
wind direction data on an independent data logger. This second system collected
data using the unit vector algorithm. Figure 4 shows unit vector comparison data
between the first and second systems when the wind speeds were 1 m=s or greater. It
is clear there is good agreement between the two systems when the unit vector
algorithm was used.
On the basis of the first two tests it was obvious there were questions about the
calculation resul ts of the singl e pass scalar algorithm. The third test performed used
a model to generate test wind data and perform wind direction averaging calculations
with a variety of methods. A simulated 1-s interval wind data set was generated and a
simple rotation introduced in the middle of the 3600-point hourly data set. This data
set was then evaluated using three averaging techniques: simple arithmetic, unit
vector, and the single pass scalar algorithm. The data set and results are shown in
856 INDEPENDENT AUDITING ASPECTS OF MEASUREMENT PROGRAMS
Figure 3 Comparison of 20 days of wind direction data using the single pass scalar and unit
vector averaging algorithms.
Figure 4 Comparison of 20 days of wind direction data using the unit vector averaging
algorithm programmed into two different data loggers.
3 CASE STUDIES 857
Figure 5. It was obvious from the results that there was something wrong with the
single pass scalar method technique. Further investigation into the algorithm
revealed that a complete circular rotation of the wind direction anytime during the
hour would result in erroneous values with the magni tude of the error effectively
being random. If there was not a complete rotation during the averaging period, then
there was excellent agreement between the methods. However, introducing one or
more complete 360
rotations during the averaging period resulted in an unrecover-
able error in the reported average.
The results of this audit were extremely important in unders tanding potential
errors in the collected data and resolving the ambiguous values obtained during
what should have been very consistent wind directions. Further discussi ons with
individuals who developed the scalar algorithm recognized the problem but also
carried the purpose of the algorithm one step further. Part of the derivation of the
algorithm was to develop a method that would correctly quantify the variability of
the wind direction expressed as the standard deviation or sigma theta. That calcula-
tion is performed correctly using the scalar technique, but at the expense of the
reported wind direction. Given these results, the optimum technique may be to use a
vector method for the wind direction (either a unit vector for non-wind-speed
weighted or straight vector for wind-speed weighted) and the scalar method for
the standard deviation of the wind direction. Each of these methods are described
in the EPA guidelines (U.S . EPA, 2000).
4 CONCLUSION
In summary, independent audit s have become an integral part of many measurement
programs. The extent of the audit scope and freque ncy of performance depends on
Figure 5 Simulated hourly wind direction data showing a 360
wind shift in the middle of
the averaging period. The results of three different averaging techniques are plotted.
858
INDEPENDENT AUDITING ASPECTS OF MEASUREMENT PROGRAMS
the specific data needs and any applicable requirements. Not only do the audits aid in
the improvement of the data quality, but they can in many instances identify potential
problems that could invalidate collected data. Additionally, most regulatory driven
programs such as Prevention of Significant Deterioration (PSD), Nat ional Air Moni-
toring and State and Local Air Monitoring Stations (NAMS and SLAMS), and
Photochemical Assessment Monitoring Stations (PAMS) require independent
audits as part of the normal measurement program. So not only are audits good
for the data quality, they are also a regulatory requirement that must be fulfilled in
order to use the data in modeling or analysis projects.
REFERENCES
Baxter, R. A. (1995). Evaluation of the Wind Data Collected Using Different USEPA Approved
Calculation Algorithms. Paper presented at the Ninth Symposium on Meteorological
Observations and Instrumentation, March 27–31, Charlotte, North Carolina.
U.S. EPA (2000). United States Environmental Protection Agency, Meteorological Monitoring
Guidance for Regulatory Modeling Applications EPA-454=R-99-005, Office of Air Quality
Planning and Standards.
U.S. EPA (1995). United States Environmental Protection Agency, Quality Assurance Hand-
book for Air Pollution Measurement Systems, Vol. IV, Meteorological Measurements,
Document EPA=600=R-94=038d, Atmospheric Research and Exposure Assessment
Laboratory.
U.S. EPA (1998). United States Environmental Protection Agency, EPA Guidance for Quality
Assurance Project Plans EPA QA=G-5, Document EPA=600=R-98=018, Office of Research
and Development.
REFERENCES 859
CHAPTER 45
REGULATORY APPROACHES TO
QUALITY ASSURANCE AND QUALITY
CONTROL PROGRAMS
PAUL M. FRANSOLI
1 BACKGROUND
Meteorological data are often obtained to support environmental and engineering
studies. Such data are evaluated by regulatory bodies for suitability in various risk,
compliance, and design analyses. Information used in the regulatory arena is subject
to intense scrutiny by all involved parties, including those attempting to challenge
the credibility of work performed. To ensure that meteorological data are acceptably
correct and complete, and that the regulations are being applied consistently to all
applicants, some regulatory groups have promulgated quality assurance and quality
control (QA=QC) requirements and guidelines. Meeting the monitoring require-
ments and following the guidelines helps to ensure acceptance of the data. It is
usually preferable to agree on a monitoring plan with the regulatory group prior
to proceeding with the monitoring program. It also helps to protect proposed projects
from time-consuming demonstrations that the monitoring equipment and methods
were satisfactory for the intended purpose.
Site characterization in engineering design or regulatory environmental studies
differs from typical weather data collection programs. Site characterization programs
seldom require real-time information flow, which is an important element in routine
weather observations made for forecas ting purposes. Instead, data from remotely
operated stations can be stored on-site for long time intervals prior to collection.
Some on-site processing is to obtain some statistical properties of the measurements,
such as means, extremes, and standard deviations.
Handbook of Weather, Climate, and Water: Dynamics, Climate, Physical Meteorology, Weather Systems,
and Measurements, Edited by Thomas D. Potter and Bradley R. Colman.
ISBN 0-471-21490-6 # 2003 John Wiley & Sons, Inc.
861
The primary quality facto rs appearing in most regulatory measurement programs
are accuracy, precision, validity, completeness, and representativeness.
Accuracy and precision are addressed in depth in another chapter in this part.
The accuracy and precision requirements contained in regulatory guidance are
based on both application needs and capability of the equipment used. Early
atmospheric dispersion models had very simple data input requirements;
typical synoptic weather observations were adequate model input. The use of
site-specific (also known as on-site) data, that is, data intended to be
representative of the source and=or receptor areas, helped reduce the uncer-
tainty from the dispersion model ing portion of environmental studies. Leap-
frog improvement advances between model and measurement capabilities
continue to challenge both the measurement and the modeling fields.
Validity of the data implies compliance with the accuracy and precision
requirements and goes beyond to ensure that erroneous measurements have
been removed from the validated data set. A data validation protocol often
begins with removing data from known periods such as quality control
checking and maintenan ce work that interrupts the measurements and
continues with identifying periods of sensor or on-site data processing and
recording equipment failures. While modern sensor and data system reliability
has increased immensely, all equipment is subject to error. Some sensor
failures can start as intermittent problems that may introduce a slight error
that is difficult to distinguish from normal variations. Other problems can occur
with sensors becoming temporarily incapacitated by external forces such as ice.
Many measurement programs have found that a meteorologist knowledgeable
of the measurement process and the specific program should be included in the
data review team.
Completeness means attaining an adequate amount of data for the intended
purpose. Modeling programs intending to identify short-term phenomena that
produce unacceptable hazards or risks need to be based on a sufficiently long
monitoring program time period to identify the high- risk meteorological
episodes. The most typical completeness requirement is 90% data recovery,
though 80% applies to some remote measurement stations. Modern equipment
reliability makes this requirement attainable, though frequent site checks by
knowledgeable operators are important to ensure that sensors have not been
damaged.
Representativeness refers to measurements being made in a location that is
representative of the area being characterized. The area could be the source
itself, or the potential receptors of an airborne effluent, or significant points
along a potential airflow pathway. Representativeness of the measurement
location is addressed in siting guidelines. In addition to the measurement
location being representative of an intended area, siting requirements also
include instrument exposure. Wind measurements can be affected by obstacles,
including the structure supporting the sensors. Temperature measurements can
862 REGULATORY APPROACHES TO QUALITY ASSURANCE AND QUALITY CONTROL
be affected by nearby heat sources or sinks, such as parking lots and cooling
towers.
2 STANDARDIZATION AND REGULATORY GUIDANCE
Regulatory monitoring requirements and guidance are based on the requirements set
by the governing bodies, such as the U.S. Nuclear Regulatory Commission (NRC)
and the U.S. Environmental Protection Agency (EPA). Others include the American
National Standards Institute (ANSI) working in conjunction with the American
Nuclear Society (ANS) or the American Society for Quality Control (ASQC). Of
these, the EPA guidelines on prevention of significant deterioration (PSD) monitor-
ing made the most significant recent advances in quality assurance and quality
control applied to meteorological monitoring. The primar y EPA monitoring
guidance document is Meteorological Monitoring Guidance for Regulatory Model-
ing Applications (U.S. EPA, 2000). Volume IV (U.S. EPA, 1994) in a series of
quality assurance handbooks for air pollution measurement systems contains some
of the most detailed technical guidelines on monitoring techniques. Volume IV
became the ‘‘how-to’’ book that explained many of the ‘‘why’’ questions behind
the guidelines and guided technicians through the rigorous detail of correctly
performing the equipment tests.
Early work in the nuclear power industry to site, license, and operate large
nuclear-powered electric-generating stations triggered some guidelines such as
Meteorology and Atomic Energy (Slade, 1968) and the Nuclear Regulatory Commis-
sion Safety Guide 23, which became Regulatory Guide 1.23. Technical guidance in
these documents established the equipment and network design specifications for
numerous nuclear-power-related projects. The 60-m tall meteorological tower with
wind and temperature measurements at the 10- and 60-m levels above the ground
became synonymous with RG 1.23 programs. Another important nuclear-related
guidance document appeared in 1984 as ANSI=ANS-2.5: Standard for Determining
Meteorological Information at Nuclear Power Sites (American Nuclear Society,
1984). This document helped to update the outdated RG 1.23 and was used for
many safety plans at nuclear power plants. This standard became less useful by
advances in monitoring technology. The replacement for this document was recently
approved as ANSI=ANS-3.11: Determining Meteorological Information at Nuclear
Facilities (American Nuclear Society, 2000).
Regulatory guidance is increasingly focused on using voluntary consensus stan-
dards, rather than having separate regulatory agencies promulgating independent
requirements. The American Society for Testing and Materials (ASTM) subcommit-
tee D22.11 (Meteorology) has produced consensus standards and practices related to
meteorological monitoring equipment and application methods. The standards
describe equipment design and testing considerations; the practices describe techni-
ques to use the equipment effectively in operational programs. Many of the standards
and practices address wind measurement, though atmospheric pressure, temperature,
2 STANDARDIZATION AND REGULATORY GUIDANCE 863
and humidity are also covered. These standards and practices are continually being
reevaluated, and new material is being developed.
3 PRIMARY ELEMENTS
The basic elements of meteorological monitoring programs that achieve the data
quality factors described above are summarized in this section. This material is
intended to be a ‘‘primer’’ on the topic; compliance with current requirements and
guidance should be based on the material effective during the time of a given
program. The primary information needs are input to atmospheric dispersion
models, with other applications of on-site measurements such as engineering and
hydrology.
Siting
Siting criteria for meteorological monitoring programs include the number of
stations, their locations, and the measurements to be made. Simple programs in
flat terrain may only require a single station. The primary wind measurements are
made at 10 m above ground level (agl), and temperature and atmospheric moisture at
2 m agl. Additional wind measurements may be necessary at higher levels to prop-
erly document airflow in complex terrain, near large manmade obstacles or bodies of
water, particularly if the source is located much higher than 10 m agl. Such ter rain
or obstacles may also require more stations to properly document trajectories of
airborne material or the locations of extreme wind or temperature conditions.
Siting criteria also include instrument exposure at the station itself. Wind,
temperature, moisture, and precipitation sensors are easily influenced by supporting
structures, such as towers and poles. Nearby natural and manmade obstacles can also
adversely influence a measurement, making it unrepresentative of the area intended.
Equipment Selection and Procurement
Equipment selection should be driven by guidance specifications, availability of
vendors to perform calibration services (which could include the manufacturers),
product and service reliability, and cost. When initial cost is one of the few factors
used in the selection process, the overall cost over the first few years of operation can
far exceed initial differences. The vendor exhibits at trade shows and discussions
with other program operators are valuable information resources.
Operator Training
Even the best equipment systems can produce u nusable data if the station operators
and data processing staff are inadequately trained and super vised. Some regulatory
groups and private companies offer valuable training programs. As with most quality
control and quality assurance activities, it is wise to document the training received.
864 REGULATORY APPROACHES TO QUALITY ASSURANCE AND QUALITY CONTROL
Installation (Testing, Location Documentation)
Proper equipment installation involves adequate field testing to ensure that the
system was not damaged in transit and has been properly installed. Equipment
checking is addressed in the next section. Another essent ial station startup step all
too easily overlooked is proper station location documentation. Seemingly obvious
local landmarks can easily change in time, particularly when a station is the first step
in a major development. The resolution and reference coordinate system should be
compatible with the purposes of the data.
Calibrations, Checks , and Corrective Actions
Three important quality control activities are equipment calibrations, simple checks,
and corresponding corrective actions. Calibrations imply formal comparisons of
equipment responses with known conditions produced by a standard. Standards
must have performance traceable to reliable sources, such as the National Institute
of Standards and Technology (NIST). Less complex testing and checks can be
accomplished in the field to ensure continued reliability of the measurement process.
Such checks can be comparative measurements made by the monitoring equipment
and a collocated standard, or by placing the equipment in a known operating condi-
tion, such as aiming a wind vane toward a known direction. Calibration and check
results can either be made to demonstrate operation within the required tolerance
limits or to allow for instrument adjustments that bring the monitoring equipment
response within specified limits of the known condition.
Equally important to the calibration or check itself is the corresponding corrective
action to be applied when the instrument response does not meet the desired
tolerance limit. In addition to performing adjustment or maintenance to bring the
response within the desired range, the nonconforming response of equipment oper-
ating on-line to collect data should be carefully documented and provided to the data
validation staff to ensure that the out-of-tolerance data can be removed from a
validated data set.
Routine Operations and Maintenance
Continuing the credibility of the data beyond the careful equipment selection, opera-
tor training, and checks requires vigilant routine operations and proper maintenance.
Procedures specific to the given operating program should be available to (and used
by) equipment operators and data validation staff. Site checks should be documented
on checklist forms or in logbooks. Many meteorological sensors used in monitoring
programs for regulatory purposes are designed for a balance of sensitivity to subtle
airflow conditions and reliability of operation. Such equipment can be susceptible to
damage from natural hazards, such as ice, blowing dust, birds, and other animals. It
is difficult to tell from data acquired by telemetry if one of the anemometer cups is
missing.
3 PRIMARY ELEMENTS 865