1. Background
In Spring 2002, the three NEXRAD
agencies[1],
through the Program Management Committee (PMC),
requested that
the Technical Advisory Committee (TAC) undertake the development of a strategic plan for the long-term evolution
of the total NEXRAD program. The tri-agencies’ request was based on the belief that in the WSR-88D, the nation
has a weather surveillance radar system that can be continuously upgraded to take advantage of previously
untapped capabilities, emerging technologies (particularly in the area of digital signal processing), and
improved scientific understanding of meteorological phenomena.
It was requested that the plan address both the radar system and the national radar network, and describe
possible enhancements to the radar, focusing on the 2007-2020
time frame[2]. The intent of the strategic
plan is to guide the evolution of the radar through the final 15 to 20 years of its service life. Special
attention was to be given to laying the foundation for a smooth transition to an as yet undefined follow-on
system[3] anticipated for deployment in the 2020-2025 time frame.
It was anticipated that research and
development work currently getting underway at the National Weather Radar Testbed to investigate phased-array
technology would be accomplished in parallel with many of the enhancements discussed here and that there
would be a continuous exchange of information between the two parallel efforts.
In reviewing the request, the TAC decided that strategic planning for the radar must be carried out
by the NEXRAD agencies themselves. The TAC is of the opinion that it does not have the breadth of knowledge
necessary to plan from the perspective of agency mission accomplishment, which should be fundamental to any
strategic plan for the WSR-88D.
The TAC does have the expertise necessary to identify current and future needs, to endorse current
research and development activities and point out opportunities to build on these, and to recommend
specific enhancements to the radar system that could be addressed in a strategic plan. The agencies
can then consider as part of their strategic planning effort how each of these recommendations might
contribute to mission accomplishment and decide whether or not to undertake the necessary research,
development, and implementation to realize the enhancement.
This white paper then provides “possible strategic directions” for the radar and the national network.
It describes in Section 2 how the TAC solicited input from individuals knowledgeable about the WSR-88D
and its potential for further development, synthesized that input, and developed the following list of
possibilities. Sections 3 and 4 present a synthesis of the input provided by this pool of experts together
with the ideas of the members of the TAC. It is hoped that this material will help guide the tri-agencies
in the long-term evolution of the NEXRAD program.
NEXRAD managers and engineers can use this synthesis in preparing future upgrades to radar
hardware/software and developing new operational strategies. Scientists can also use this
synthesis in planning basic studies in weather radar research. However, each point should be
evaluated in terms of agency mission and projected future needs.
2. Procedure
A select group of 25 weather radar experts was solicited to provide short (approximately 3
or so pages)
discussions addressing the development path they believe the NEXRAD program should follow during the
period 2007-2020. In providing their comments, these experts were asked to consider both enhancements
and upgrades to the WSR-88D system, and broader strategies for the national network as a whole. They were
encouraged to be speculative in their comments.
Fifteen responses were received to the solicitation. Appendix 1 gives a listing of the individuals who
provided input. Appendix 2 provides a verbatim tabulation of all input received. The responses were
analyzed by the members of the TAC and synthesized into the eight recommendations given in Section 3.
In preparing these recommendations, TAC members not only considered the suggestions of the external
experts, but also folded in their own perspectives and considered a number of radar enhancement
efforts currently underway.
3. Recommendations by the Technical Advisory Committee
1. Coordinate enhancements to the WSR-88D with parallel developments of
the follow-on system to NEXRAD.
As a first general principle, enhancements to the WSR-88D through the remainder of its service
life should be closely coordinated with the parallel development of the follow-on system. While
each is a distinct program, the two efforts offer many opportunities for synergy. The follow-on
system may be a completely new radar, or it could be a major retrofit and upgrade to the current
radar, replacement of the current mechanically scanned dish antenna with a phased array antenna
would be a good example.
Coordination will ensure that promising technologies developed in one program (e.g., phased-array
antenna, pulse compression, oversampling, etc…) are considered in the other and that the transition
to the follow-on system will be smooth from the point of view of the users of the radar data. While
the development of the follow-on system is only in its earliest stages at the National Weather Radar
Testbed, now is the time to develop the management structure to ensure this coordination.
2. Increase the density of radars in the national network.
As articulated in the National Academy study “Weather Radar Technology Beyond NEXRAD”, a significant
shortcoming of the present NEXRAD network is the limited coverage at low levels. This heavily
impacts useful ranges for precipitation estimation (snowfall as well as rainfall), severe storm
detection, convergence line detection, and boundary layer wind estimation. Several respondents
also note problems in detecting low-level phenomenon (initiation of convection, lake effect snow)
and phenomena passing over the radar (cone of silence). Part of this limitation is due to the inherent
geometry of the radar beam passing over the curved Earth, coupled with having a rather widely spaced
network of fixed station radars. However, the problem is exacerbated by programmatic decisions that
no radar is to be operated with the center of the beam below 0.5o elevation and that
RHI[4] scans
were not to be made. The only way to remedy this problem is to have more radars closer together.
In some instances, scanning at lower elevation angles would also help. In regard to the former,
the situation could be improved at relatively modest cost by assimilating data streams from other
radars, especially the FAA’s Terminal Doppler Weather Radars (TDWR) and Airport Surveillance Radars
(ASR). Commercial weather radars used by the media and mobile X-band radars also offer possibilities.
Along the U.S-Canadian border, joint sharing of Doppler weather radar data by the two nations would
contribute to improved coverage. These multiple data streams could be used to prepare national, regional,
and local radar mosaics.
It would be worthwhile to have a systematic evaluation of what might be done to address such problems
and to evaluate where significant gains might be made (e.g., what would be the impact on tornado forecasting
of having the ability to scan at elevations below 0.5°? Is there an optimum minimum elevation angle, which
might be different for each radar?). Serious consideration should be given to the possibility that the
optimum follow-on system might consist of a relatively few large radars providing regional coverage
supplemented by a large number of short range systems focused on the boundary layer. An Engineering
Research Center headed by the University of Massachusetts was recently funded by the National Science
Foundation to explore possible designs for such short-range systems. Experiments with such a strategy
could thus be carried out in 2007 to 2013 time frame with an enhanced WSR-88D network providing the
regional coverage and prototype small radars serving as boundary layer monitors.
The above discussion assumes that over the next decade forecasters and other users will shift
from working with data from single radars to utilizing products that are digital mosaics of multiple
radar data streams (perhaps incorporating data streams from other sources as well), a National 4-D Radar
Database. It is anticipated that such mosaics could provide a more reliable, detailed, accurate view of the
atmosphere on the mesoscale than could be obtained from the output from any single radar. Moving in this
direction represents a major change in the way the radar is utilized and managed. Originally sold to the
Congress as a local severe weather warning system for use by forecasters, the 140+ WSR-88D’s have
collectively become the nation’s primary mesoscale environmental observing tool (witness recent efforts
to extend its applications to Homeland Security). Not only is the radar data being used by forecasters,
it is being utilized to initialize numerical forecast models and to drive systems that automatically
produce products for use by non-meteorologists, e.g., FAA air traffic controllers and tower operators.
Moving in this direction – toward a true national radar network supplying a central database accessed by a
diverse group of users -- will require major policy decisions by the tri-agencies. These will need to
address how the individual radars are upgraded, how such a network (which may include TDWRs, ASRs, and
other radars, as well as the WSR-88Ds) is to be constituted and centrally operated (as opposed to the
current “centrally managed”), and the characteristics of the database of mosaic-based data and derived products.
3. Produce the best quality data possible from the WSR-88D throughout
the remainder of its service life.
Users are demanding not only high quality data – accurate, reliable, timely -- from the radar network
but also measures of the quality of the data to be included as part of the data stream. This demand
for data quality measures is a direct consequence of the increasingly more diverse and more sophisticated
uses being made of the data. In particular, applications being developed by the FAA are based on
processing the NEXRAD data using fully automated algorithms with the end product going directly to end
users such as controllers, supervisors, and traffic flow managers. Similarly, the NOAA National Weather
Service (NWS) and several university based research groups are now assimilating reflectivity and velocity
data directly into numerical forecast models. Private vendors are generating automatically value-added
radar products and transmitting them directly to non-meteorologist users, such as the media; consumer oriented
products (briefly noted in Recommendation 7) are in the near horizon. All of these applications require
that quality control/assurance be applied automatically, thus measures of data quality provided by the radar
greatly simply the process of flagging bad data, adjusting analysis schemes, etc.... Consequently, “the best
quality data possible” includes not only improving wherever possible the reliability and accuracy of the data
stream, but also associated “meta data”. As noted by Bumgarner (p. 2 in Appendix 2), this meta data consists
both of engineering data about quality of radar performance and estimates of the accuracy and reliability of
the meteorological data. The former is quantitative and provides measures of radar performance against
specifications; the latter would be of a more qualitative nature, for example, providing a first estimate
that the return was from a meteorological target or from a bird.
As a minimum, real-time meta-data regarding calibration, timing, beam position/pointing, and other
system settings could be provided to users. More extensive meta-data could include self-diagnostic
information on the radar, information on samples that are being eliminated during signaling process
(extremes and unusual returns are often of special interest), and a “confidence index” that estimates
the reliability of the resulting products from the radar system.
Procedures could be developed and implemented that ensure that radar system calibration is as close
to perfect as possible. Precise, synchronized timing (for example, based on GPS time signals) and angular
indexing of each radial of data could be implemented to facilitate integration of data from different radars.
Signal processing could be improved to almost completely mitigate ground clutter, and range and
velocity folding. As part of the deployment of the polarimetric retrofit, signal processing could be
enhanced to use this additional data for enhanced identification of non-meteorological targets (ground
clutter, sea clutter, birds, insects, chaff, etc) so that they can be removed from appropriate products.
It would be highly desirable to occasionally operate some radars at elevations below 0.5o. For example,
this would enhance use of Fabry’s moisture mapping
technique[5], providing forecasters detailed information
on low-level moisture. Such information would also have value in the initialization of small-scale numerical
models.
Bi-static technology (additional remote receivers associated with one NEXRAD emitter) could be explored
to increase the amount of data generated from one radar.
The scanning rate of the WSR-88D could be increased. Using the fast signal processing techniques now
available and over sampling, the WSR-88D should be able to scan at its rated maximum of 6 rpms, which
translates to approximately 3 minutes for a full Volume Coverage Pattern (VCP). Efforts could be
accelerated to determine the usability and limitations of the over sampling method with operational
testing performed on a test bed as described in Recommendation 8. Further, in partnership with the
effort to develop the follow-on system, efforts could be made to develop technology that would reduce
to less than 1 minute the time required for a full-coverage VCP.
As the network of WSR-88 radars (supplemented and extended by other radars as noted in
Recommendation 2) evolves to a hybrid, regionally or nationally controlled network, adaptive
network scanning strategies could be explored to maximize the information content in data obtained
in the areas of greatest interest. Optimal methods for scanning would consider the phenomena of
interest and the spacing and capability of the individual radars. Techniques could be devised and
tested to automatically select the optimal scanning strategy based on the weather scenario present
and tri-agency requirements. Implementation of such technology and operating procedures raises basic
policy issues for the tri-agencies.
4. Increase significantly the computing power and the telecommunications bandwidth.
All data streams – from in-situ measurements, satellites, and the radar network – will be
continuously expanding for the foreseeable future. As an example, the installation of polarimetric
capabilities on the WSR-88D will result in a 2- to 4-fold increase in the data stream. Several of the
appended suggestions from the radar experts would also increase significantly the amount of data that
will need to be handled. It is apparent that future successful operation of the radar system – either
individually or as part of a national network -- will require substantial increases in both the
computational power associated with the radar and the telecommunication bandwidth that is the backbone
of the network. While every effort should be made to optimize the use of existing bandwidth through
compression schemes, current trends indicate that a wider telecommunication bandwidth will become
essential by 2010.
The TAC recommends that if at all possible, the development of the technology to provide the
recommended growth in both computing power and bandwidth be done as a multi-agency effort, perhaps
even covering several types of radar. There appears to be numerous opportunities for synergy – and
cost savings -- in developing a common system to meet multiple agencies needs.
5. Develop, test, and implement advanced radar waveforms and signal processing techniques.
Phase coding, staggered pulse repetition time, pulse compression, and over sampling are possible
engineering advances that need to be vigorously pursued since they are likely to lead to improved
data quality, sensitivity, and increased scan rates.
Spectral processing of the radar data stream could be investigated since it is well suited for
adaptive removal of artifacts and interference from external radiation sources. Further a gain of
about 10 dB in effective signal to noise ratio can be achieved, which would extend coverage in clear
air and enable better cloud detection. Spectral processing schemes for identifying tornadic
circulations could be explored.
6. Support a coordinated effort to integrate radar data and other data into enhanced decision
support systems.
Enhancements to the radar must not be considered in isolation. Data fusion
systems[6], locally run
storm-scale numerical models, decision support systems, and fully automated processing systems
generating products for non-meteorologists are but a few examples of the technologies either
currently being deployed or under development in several institutions within the United States
and other countries. Many such systems are being deployed or developed by one or another of the
tri-agencies to support mission accomplishment. All this suggests that there needs to be a
well-articulated vision for the desired functionality of tri-agency systems using radar data
in the 2010 to 2020 period, a vision that will in turn drive radar enhancements and network development.
For example, it seems clear that the NWS forecaster workstation should provide a seamless
interface between 4-D assimilated observations and model results, providing a picture of the
immediate past, the present, and the near future on micro-, meso-, and synoptic scales. Articulating
what functionality such a workstation must provide (for example, the ability to select particular
enhancements of imagery) determines the requirements for products from the radar and for mosaic or
composited products, as well as the necessary assimilation and display tools. The mosaics could be
created from the National Radar 4-D Database, and so could include data from not only the WSR-88D
network but also all other appropriate radars as discussed above. The goal here is to achieve nearly
uniform, high-resolution data all across the nation, including in the boundary layer. As noted
previously, this approach, where user-forecasters interact with 4-D assimilated fields in addition
to the data streams directly, has implications for operational strategies and organization of the
forecast office.
7. Establish a coordinated program to develop products and display tools for the
various user communities.
There is a growing community of users who rely on the radar data. These users range from professional
and technical staff employed by the tri-agencies to private sector weather forecasters, to the media
(including the national media such as The Weather Channel), to emergency managers, and transportation
system managers. Each of these communities needs to be considered as changes are made to the way the
radar network is operated and as new products are developed. Indeed, as a general principle, the TAC
strongly recommends involving the user community, be it the tri-agency users or users outside the
tri-agencies, as early in the change or development process as possible.
There is increasing demand for radar products that flow seamlessly into Geographic Information
Systems. In the future, radar data products could be formatted to be compatible with standard
Geographic Information System (GIS) software. Indeed, the architecture of the future forecaster/user
workstation discussed in Recommendation 6 could be based on standard GIS software.
Similarly, there are strong indications from the commercial market that users will increasingly
demand high resolution radar data and products be made available in a wide range of formats,
including in vehicles and on Personal Data Assistants (PDAs). There appear to be many opportunities
for partnership with the private sector in this area.
Radar retrieval techniques could be developed to obtain high-resolution wind fields. The retrieval
of high-resolution near-surface water vapor fields from radar refractivity could be implemented on
all radars in the national network. The development of enhanced warning and nowcasting techniques
that utilize the high resolution multiple data sets could be accelerated.
Voice-activated intelligent
agents[7] can be developed to speed manipulation of data fields,
speeding analysis and so improving warnings and other time-sensitive applications. While 3-D
visualization tools have not yet found significant applications in meteorology, recent advances
in such technology plus the anticipated future increases in 4-D environmental data streams suggest
that it would likely be profitable to explore 3-D visualization tools for the operational forecaster.
4-D mosaics from multiple radars offer opportunities for such explorations.
Under the direction of the World Weather Research Program, efforts are being formulated to coordinate
these and similar activities to improve efficiencies and speed the transfer of technology to operations.
The TAC could monitor these activities and provide advice to the PMC/NEXRAD tri-agencies on the implications
and possibilities for the U.S. national radar network in the areas of technology (for both the radar
and the network) and applications of the data.
Further, mechanisms – workshops, public forums, electronic newsletters, and focused
websites – could be developed by which issues related to changes to the radar system and
evolution of network operations can be discussed with external user community in a timely manner.
Issues on the immediate horizon include fully automated use of the current NEXRAD products and the
development of new products based on polarimetric data. In the future, the rate of change is likely
to increase significantly, so these mechanisms could be institutionalized and not handled on an ad hoc basis.
8. Establish national test beds and prototyping sites for testing new radar technologies and
nowcasting/forecasting
capabilities in operational environments.
To properly develop many of the promising technologies now on the horizon, a number of test beds
and prototyping sites should be established. These could be located around the nation to explore
systematically various approaches to dealing with important meteorological phenomena and the needs
of important user communities, e.g., an eastern Great Lakes facility to explore both lake effect snows,
highly sheared thunderstorms with low ceilings and visibility, and critical aviation products, such
as winds aloft, for the highly congested air space over that region. There are many other issues
concerning wavelengths, calibration, resolution and communications that must be investigated. Further, it
is desirable that the activities under recommendations 6 and 7 be implemented first at the test beds.
In addition to a Northeast U.S./eastern Great Lakes facility, prototype sites could be considered
for the Gulf Coast (hurricanes), central Great Plains (severe thunderstorms, winter storms), the mountainous
Western states (winter storms, fire), and in the northwest or central California coast (heavy rainfall, winter
season storms). In many cases, this can be accomplished by deploying two Radar Data Acquisition (RDA)
units to an existing WSR-88D site. One RDA would provide normal signal processing for current operational
use, while the second would provide experimental signal processing, perhaps supporting an experimental
forecasting effort.
It is important that prototyping efforts be structured to ensure that the resulting products are
provided in real time to external users so that the user community can provide feedback early in the
development process on the value and utility of the products for their specific agency applications.
To develop means for increasing the density of radars in the national network, field experiments
could also be conducted with the aim of developing techniques for supplementing and extending the
national WSR-88D network using other deployed radars (TDWR, ASR, mobile X-band radars, and commercial
radars such as those used by the media).
Appendices:
- Individuals providing input
- Verbatim input received
[1] The NEXRAD agencies are the Department of Commerce/National Oceanic and Atmospheric
Administration (NOAA), Department of Transportation/Federal Aviation Administration (FAA), and Department
of Defense/Departments of the Air Force and Navy.
[2] The agencies have in place a formal NEXRAD Product Improvement Program
and a software enhancement program that address the near-term (0 to 5 years).
[3] The National Research Council recently released a report regarding possible
approaches for the replacement of the WSR-88D: National Research Council, 2002: Weather Radar
Technology Beyond Nexrad. National Academy Press, Washington, D.C.81 pp.
[4] RHI = “range-height indicator”, referring a display that shows a
vertical cross section through the target. In many radars this is accomplished by holding
the radar antenna fixed in azimuth and scanning in elevation.
[5] For details, see Fabry, F., C. Frush, I. Zawadzki and A. Kilambi, 1997:
On the extraction of near-surface index of refraction using radar phase measurements from ground
targets. J. Atmos. and Ocean. Technol., 978-987.
[6] Data fusion systems will assimilate and meld data from a variety of
radar types, satellite, rain gauges, Aircraft Communication Addressing and Reporting System
(ACARS), lightning, mesonets, soundings, and models as well as the WSR-88D radars.
[7] Intelligent agents, also called intelligent decision aides,
are a logical extension of today’s browsers and search engines. They are anticipated
to be only semi-autonomous, requiring direction and guidance from the user. This distinguishes
them from nearly autonomous (and much harder to construct) artificial intelligence systems.