|
| UNITED
STATES DEPARTMENT OF COMMERCE Barbara Franklin, Secretary |
National Oceanic and
Atmospheric Administration John A. Knauss Under Secretary and Administrator |
National Weather
Service Elbert W. Friday Assistant Administrator |
|
|
TABLE OF CONTENTS I. INTRODUCTION .........................................................................................................................1 II. PROFILER NETWORK .............................................................................................................. 2
III. AFOS APPLICATIONS PROGRAMS ....................................................................................... 4
IV. SELECTED OPERATIONAL APPLICATIONS ...................................................................... 11
V. DOCUMENTATION AND ASSESSMENT ............................................................................. 31 REFERENCES ..................................................................................................................................... 33
|
|
I.
INTRODUCTION A
network of 29 wind profilers (the Wind Profiler Demonstration Network,
or WPDN) in the central United States provides tropospheric wind data to
National Weather Service (NWS) field offices. The data are characterized
by better horizontal resolution than that provided by the existing
rawinsonde network and provide much better vertical and temporal
resolution. Twelve of the WPDN profilers are located within the NWS
Southern Region. The goal of the WPDN is to provide an opportunity to
assess the utility of wind profiler data in an operational setting. The
results of that assessment may help determine whether the WPDN will be
followed by a national network, and if that is the case, the assessment
results will provide the basis for designing and establishing such a
network later in the decade. Ten NWS offices in the Southern Region have
a formal role in the assessment. They are WSOs Amarillo and Tulsa, and
WSFOs Albuquerque, Fort Worth, Lubbock, Norman, Little Rock, Memphis,
Jackson, and New Orleans. Fourteen Central Region WSFOs are also taking
part in the formal assessment. The
scheduled end of the formal assessment after Fiscal Year 1992 gives some
offices less than a year to provide input on the utility of
real-time wind profiler data. Nevertheless, judicious application
of profiler data at the field level will enhance the forecast and
warning program for as long as this technology is available. Four
training manuals and associated videotapes have been developed by the
Program for Regional Observing and Forecasting Services (PROFS) in
Boulder, CO. The first training manual (van de Kamp 1988) deals with
principles of wind profiler operation, while the second manual (Brewster
1989) discusses quality control of profiler data. The third (Brady and
Brewster 1989) deals with warm- season applications of profiler
data to analysis and forecasting, and the fourth (Jewett and Brady 1989)
discusses cool-season applications. General
Sciences Corporation (GSC) and the NWS Techniques Development Laboratory
developed a comprehensive applications software system (Battel et al.
1991) for displaying profiler winds and derived products on AFOS (see
Section III).
The purpose of this memorandum is to combine
important information from the training manuals, the AFOS profiler
applications software, and some of the relatively few research papers
dealing with profilers, and to help establish guidelines for NWS field
offices to integrate wind profiler data into local forecast operations.
The importance of producing solid documentation (by forecasters) of the
utility of profiler data is also discussed. The data and the software
for displaying them on AFOS are still relatively new and unfamiliar to
the majority of forecasters. The procedures outlined on the following
pages are intended to help smooth the transition to routine utilization
of this new data source and introduce its tremendous potential. 1 |
|
II. THE
PROFILER NETWORK
Wind
profilers are relatively low-power Doppler radars that measure the
atmospheric winds (including vertical motion) above a profiler site.
They are highly sensitive "clear-air" radars that operate with
wavelengths from about 33 cm to 6 meters. The WPDN profilers operate
with a wavelength near 74 cm, which corresponds to the profiler
frequency of 404.37 MHz (van de Kamp 1988). Stable
pulses emitted by the profiler continually scatter small amounts of
energy back to the antenna as they encounter fluctuations in the radio
refractive index (caused by turbulent mixing of volumes of air with
slight differences in temperature and moisture content). The profilers
are most sensitive to turbulent eddies with dimensions around one-half the wavelength of the emitted signal, or about 37 cm.
Since the eddies are embedded in the wind flow, a slight Doppler shift
of the emitted signal is produced and the wind speed can be measured. The
WPDN profilers use a phased array antenna to emit signals in three
directions, or beams: a vertical beam and two off-vertical beams
pointing generally north and east. The actual directions have been
slightly altered to minimize interference with the operational
geostationary satellites. The elevation angle of the off-vertical beams
is 73.7 degrees (i.e., 16.3 degrees off the vertical beam). The vertical
beam is used to measure the w-component of the wind (vertical
velocity), while the off-vertical beams are used to measure the u-
and v-components. The
profiler operates in a low mode and a high mode, measuring winds every
250 m in the vertical from a minimum sampling height of 500 m. Internal
electronic constraints prevent the WPDN profilers from measuring winds
below that height. The height resolution is 350 m in the low mode and
1,000 m in the high mode. Therefore, winds measured by the profiler are
an average within each resolution "volume" (350 m or 1,000 m),
centered every 250 m vertically. Winds are measured in the low mode from
500 m to 9.25 km AGL, while high mode sampling extends from 7.5 to 16.25
km AGL (note the overlap between 7.5 km and 9.25 km). A
critical assumption in the use of wind profiler data is that the
horizontal wind field is uniform, or homogeneous, across all three
beams. This is not necessarily true at any given moment, since the
horizontal separation between the "north" and "east"
beams is equal to four-tenths of the sampling height. For example,
at a height of 15,000 feet, these beams are separated by 6,000 feet, or
more than one mile. Obviously, variations in the wind field often exist
across horizontal distances this large, especially if there is
convection in the area.
2 |
|
B. The Data The
profiler produces a vertical wind profile every six minutes, sampling
for two minutes (about one minute in the low mode and one minute in the
high mode) in each of the three beams.
Non- uniformity in the wind field across the beams can be a
problem for such short averaging times, but the assumption of uniformity
is usually valid for longer periods of time. Hence, the six-minute
measurements are used to produce an hourly measurement through a process
known as "consensus averaging," designed to filter out any
unrepresentative six-minute measurements (Brewster 1959). Each
profiler sends data (via satellite or dedicated phone lines) to the
central Hub computer in Boulder at the end of each six-minute
averaging period. The data are in the form of "spectral
moments," which consist of estimates of radial velocity, returned
power, and spectral width for each height, beam position and resolution
mode. The spectral width represents the spread in velocity values and is
related to shear and turbulence. The
Hub computes the wind profiles from the six-minute data, using
basic quality control routines consisting of consensus averaging, median
filter and vertical consistency checks. The profiles are transmitted to
Suitland, MD, and placed on the NWS Gateway for distribution over the
AFOS system (in WSFO Norman's case, they are transmitted via ISPAN). The
WPDN profilers produce high-quality wind profiles with an accuracy
of better than 1 m/s. In a one-month reliability test with a
research profiler, over 96 percent of the wind measurements were judged
to be good measurements. The WPDN profilers are expected to equal or
exceed that performance. Nevertheless, it is important to recognize the
limitations of the data and some of the more common sources of problems. Spurious
measurements represent the greatest data problem. They can be produced
by a number of things. The profiler radar, like other radars, produces
side lobes of emitted energy in addition to the primary focused beam. If
a side lobe encounters highly reflective targets such as aircraft,
trucks, buildings, mountains, rain shafts, etc., the energy returned to
the radar can overwhelm the clear-air return in the main beam.
Returns from fixed objects (ground clutter) can be removed, but ground
clutter algorithms cannot screen out moving objects. Returns from
unwanted objects can also be received through the main beam (aircraft,
birds, etc.). One
of the most significant causes of spurious returns is precipitation
contamination. If precipitation is falling in all three beams at a
uniform fall speed, the geometric relationships allow for removal of the
vertical contribution due to the fall speed, and the u-, v- and w-components can be successfully retrieved. In the case of precipitation
falling in only one or two of the beams, or if the fall speed is highly
variable, the correct wind values cannot be retrieved.
Quasi-stationary wave activity can also produce highly variable
vertical velocities across a profiler sampling domain. 3 |
|
Velocity
aliasing (Brewster 1989) is another source of spurious wind
measurements. In the case of the WPDN profilers, the Nyquist velocity
(the maximum radial velocity that can be measured) is about 30 kt in the
low mode and 45 kt in the high mode. Radial velocities greater than
these values are "folded" or "aliased." In
practice, aliased velocities are so vastly different from correct
measurements that they are usually easy to recognize. This is evident in
Fig. 1, which shows folded wind data from the Haskell, OK, profiler
above 470 mb under a very strong jet. In early 1992, a "velocity
unfolding" algorithm was implemented to correct this problem. Other
problems arise when the data are not representative of surrounding
conditions; i.e., the measurement is valid but the measured phenomenon
is small and/or short-lived. Wind fields in the vicinity of
thunderstorms will usually be significantly different from the
prevailing flow. The measurement of such localized wind fields may be
perfectly accurate, but unrepresentative of surrounding conditions, and
the forecaster may have a hard time deciding whether the wind
measurements are valid or "bad." The
Hub's quality control routines are designed to flag the spurious
measurements, while leaving the decision concerning representativeness
to the user. Spurious and unrepresentative data often look similar
(large changes over a small height or time difference). Consequently,
"good" data are occasionally judged erroneous by the quality
control algorithms (although an improvement in the vertical shear check
quality control algorithm in early 1992 has reduced the number of good
data points being flagged as bad, especially in the lower few thousand
feet). On the other hand, allowing unrepresentative data to pass may
lead to some spurious data passing the checks. Hence, the user must
develop skill in recognizing the difference if the data are to be used
to their maximum utility. Brewster
and Schlatter (1986) describe automated quality control of wind
profiler data in detail. The second profiler training manual
(Brewster 1989) also discusses the subject. In addition to tropospheric wind data, limited surface data are available from the 14 profiler sites equipped with a Profiler Surface Observation System (PSOS). The PSOS sites provide hourly-averaged temperature, relative humidity, rainfall rate and wind measurements. III.
AFOS APPLICATIONS PROGRAMS A.
Overview
Profiler
products for AFOS are generated by the GSC/TDL applications software,
and are described in the accompanying user's manual (Battel et al.
1991). Programs have also been developed by ERL (PROFS) for displaying
profiler data on the Pre-AWIPS system at Norman. Periodic updates
to both of these packages will continue for some time. 4 |
|
The
profiler data are transmitted over AFOS in a format called Binary
Universal Form for Data Representation (BUFR) via the product identifier
NMCWPDERL. The data include station information, surface (PSOS) data,
and tropospheric wind data. The AFOS programs decode and process
NMCWPDERL,
then display the data using time-section (time
versus height), cross-section (horizontal distance versus height),
and plan view (plots of data from several stations at various
atmospheric levels) formats. With
all 29 WPDN profilers operational, the product NMCWPDERL has to be split
into three parts in order to handle the volume of data. Therefore, the
message will be sent three times each hour, and in order to save 24
hours of data, it is necessary that 72 versions of the product be saved
in the data base. The programs can be run as stand -alone programs with appropriate switches, or they can be run automatically after user-specified options have been inserted into an AFOS preformat. Automatic AFOS schedulers such as WATCHDOG can be used for running the decoder and other parts of the software system. Detailed instructions, along with a description of the various options available to the user, are outlined in the user's manual which accompanies the software. The software system is quite diversified and offers a great many options for creating and utilizing the products. The user's manual contains specific instructions for loading and using the software.Following
is a brief overview (incorporating ideas gained from actually using
these programs) of important steps necessary for generating some of the
more commonly used, or potentially useful, profiler graphics. The reader
should consult the user's manual for the finer details. Most of the files necessary for running the programs should reside in an RDOS directory other than SYSZ. For convenience, a little -used APPL or USER directory is best. It is essential that one file -- WPDATA (the file that contains the decoded data) -- not reside on SYSZ, since that file can grow to a size of 1200 RDOS blocks if 24 hours' worth of data from all 29 profilers are stored. Many offices might very well want to store that much data. While no single office would want (or be able, for that matter, due to the AFOS time required) to routinely create time-sections for all 29 stations, most offices will probably want the capability to generate plan view plots of data from the entire network on a routine basis.One
of the first steps in tailoring the system to a specific office's needs
is to decide what stations to decode and how many hours' worth of data
to store. If plan view plots covering the entire network are desired,
the default value (i.e., decode all stations) should be retained. Most
offices will likely want to retain the default value of 24 hours of
stored data. These options can be changed through the WPSET program.
However, it only takes about 75 seconds for the WPDEC program to decode
all data from all 29 profilers for one hour. If WPDEC is scheduled (via
WATCHDOG) to run every hour, this amount of time should not be a
problem. Obviously, if WPDEC hasn't been run for a while, decoding 29
stations could require lots of time -- up to half an hour to
decode 24 hours of data. 6 |
|
In creating a basic set of routinely-generated products, the AFOS preformat CCCMCPWPD (Fig. 2) is used to specify the type of output desired (products, levels, hours, contour intervals, station selections, etc.). The preformat is three pages long on AFOS, with each menu (time-section, cross-section and plan view) on a separate page. Type M: WPD, and fill out the header block CCCWPDxxx. After the desired options have been selected and have been saved in the product CCCWPDxxx, the program CWPCF can be executed to create a corresponding "options file" -- WPCF.nn, where nn can range from 00 to 99. Therefore, up to 100 different WPCF.nn files, representing 100 different sets of option selections, can be created. EXAMPLE:
RUN: CWPCF 03/e |
|
The
first command takes the most recent CCCWPDxxx (output from the preformat
screen) and creates the options file WPCF.03. The second command
"points" the system to that specific options file when
subsequent programs are run. How
does one keep track of the different options that have been selected in
the various option files (created from the AFOS preformats)? Each time
CWPCF is run to create a WPCF.nn file, a corresponding file called
SAVPF.nn is also created. The contents of a particular WPCF.nn file can
be seen by displaying the SAVPF.nn file on AFOS: For
example, the command DSP:SAVPF.99 at an ADM will reveal the contents of
WPCF.99 (actually, the contents of the CCCWPDxxx from which WPCF.99 was
created). Or, simply run the program PWPCF.SV as described in the user's
manual. Most
offices will probably need only a limited number of different option
files, depending on the forecast situation. The desired products for the
desired set of stations can be selected by simply running WPSET (RUN:WPSET
nn/e) at an ADM. Forecasters
at most offices will likely find that there is not enough AFOS time to
locally produce all the products they would like to see. The software
was designed in modules so that different offices could share the
workload. For example, a WSFO may wish to run only time-sections
and plan views, while letting one of its WSOs produce cross-section
products. There is great flexibility in how the program load may be
distributed among offices. A local switch enables data base products to
be routed between offices on the same SDC.
|
|
B.
Product Description Time-sections: These
products provide the forecaster with a vertical profile of the wind
every hour. Time increases from right to left to simulate a west to east
cross-section (more correctly, an "upstream" to
"downstream" cross-section). The height (y-axis) is labeled in
both km and kft, as well as in corresponding pressure levels for a
standard atmosphere. The user can specify the base height (default is
mean sea level), then up to 7,250 m or 24,000 feet of data are plotted.
The number of hours can also be selected (default is 16 hours). The
following time-section plots are available: -horizontal
wind velocity -horizontal
wind speed -u,
v, w wind components -thermal
wind -perturbation
wind -wind
direction and wind speed shear -returned
power -relative
vorticity -horizontal
divergence -derived
vertical velocity Cross-sections: The
background format for cross-section displays (from stations located
along a vertical plane) is similar to that used for time-sections.
It should be remembered that, when selecting a base height, elevation
varies from station to station. One hour's worth of data (selected by
the user) are displayed for each station for each product. The default
hour is the most recently decoded hour. The following
cross-section plots are available: -horizontal
wind velocity -orthogonal
and parallel wind components -w
wind component -thermal
wind -spectral
peak power Plan
Views: Plan view plots, similar to standard-level plots of rawinsonde winds, are of great value in supplementing information supplied by the time-section products. In fact, the use of the two together can be critical to gaining an understanding of more subtle phenomena that are not obvious on one type of display alone. 9 |
|
The following plan view plots are available: -horizontal
wind velocity -streamlines -horizontal
wind speed -thermal
wind -relative
vorticity -horizontal
divergence -w
wind component -time
difference fields -returned
power C.
Utility of the Products This subject is described in more detail in the user's manual. The following show just a few examples of how profiler data -- and derived products -- may be used for analysis and forecasting. Some specific examples and cases are described in the next section (Section IV).
wind
velocity -fronts
(location, passage, slope) -troughs
-ridges
-jet
streams -low
level jets -mesoscale
circulations -enhancing
constant-level NMC charts streamlines
wind speed & wind direction shear thermal
wind
-warm/cold advection -location of warm or cold air
perturbation
wind -small
features exaggerated (enhanced) -passage of vorticity maxima
10 |
