National Severe Storms Laboratory
Norman, Oklahoma USA
As NOAA’s network of weather surveillance radars (officially designated as Weather Surveillance Radar 1988 Doppler, WSR-88D) is improving it is also approaching a mid point (fifteen years) in its life cycle. Latest accomplishment is deployment of the new modular Open Radar Program Generator (ORPG) which is based on network and distributed workstation technology. All real time algorithms for processing and interpreting weather radar data reside in the RPG. Thus, the new open system allows rapid transition from concept to operational implementation of latest advances in atmospheric science. But, much more can and will be done within the next few years. This is because NOAA has recognized the need for an orderly program to enhance and evolve the network as well as to explore future technology options for possible replacement of the network. Therefore an R&D radar facility has been established at the National Severe Storms Laboratory in Norman OK. This facility consists of a test bed WSR-88D (designated KOUN) meant to support further evolution of the network and an agile beam Phased Array Radar (PAR) to explore benefits of this versatile technology.
II. NEXRAD Evolution
Evolutionary changes envisioned for the next 10 years are explored and tested on the KOUN radar. These include a) new frontiers in weather observations such as the utility of polarimetric measurements for discriminating and quantifying precipitation, b) demonstrating algorithms and novel procedures for weather forecasting and warning, c) gaining additional insights in weather phenomena, and d) serving as a test bed for technological innovations pertinent to the WSR-88D.
This radar has some unique capabilities. It served as a proof of concept for the Open Radar Data Acquisition (ORDA) system of the WSR-88D network; implementation will start in 2005. Thus the KOUN has a powerful signal processor, a versatile control of transmitter phase from pulse to pulse, as well as control of spacing between transmitted pulses. Moreover, the radar has two modes of dual polarization capability. One termed Simultaneous transmission and reception of Horizontally and Vertically polarized waves (SHV) does what its name says. The other termed Linear Depolarization Ratio mode consists of transmitting horizontally polarized waves and receiving both horizontally and vertically polarized waves. Also, sampling of returned signals can be done at a rate five or ten times higher than the reciprocal of pulse length. Thus the radar can collect unlimited amounts of oversampled time series data, in dual polarization, with phase coded, and staggered PRT sequences. Why such versatility? To test signal processing ideas and algorithms on actual data from a variety of meteorological situations. Already a large archive of time series data exists and these are being processed and evaluated off line (http://cimms.ou.edu/rvamb/Mitigation_R_V_Ambiguities.htm) to either validate a new technology of choose the best solution among competing approaches. Significant system enhancements will be achieved once the choice technologies, discussed next, are introduced to the network.
a) Range and velocity ambiguities /software and hardware enhancements
Mitigation of range and velocity ambiguities has been one of the highest priorities for the WSR-88D system. An interim solution just implemented on the network consists of three successive scans at the lowest elevations with a different PRT in each scan. The velocities and powers from these scans are combined to increase the unambiguous velocity and minimize obscuration by range overlaid echoes.
Meanwhile, researchers at NSSL have devised a systematic phase code for transmitted pulses so that overlaid echo returns can be separated. The weaker echo, up to 40 dB bellow the stronger one, can be recovered. This technique has been tested with the KOUN radar on several storms (http://cimms.ou.edu/rvamb/SZ/SZ-2_Algorithm.htm) and is scheduled to be implemented on the network in 2006.
Moreover, staggered PRT has also been tested and recommended for higher elevation scans (http://cimms.ou.edu/rvamb/Staggered/Stag_Algorithm.htm). Until recently the drawback of staggered PRT was difficulty in clutter filtering. This was overcome, however, by NOAA scientists who developed spectral ground clutter filter for staggered PRT sequences. Thus in the very near future the volume coverage pattern on the WSR-88D will likely have phase coded sequences at the lowest elevation scans and staggered PRT sequences at the higher elevations. Further on, we envision adaptive (weather dependent) selection of staggered PRT and phase coded sequence so that ambiguities will all but disappear.
b) Increase in capability – Reduction of errors and faster volume coverage
Rapid update of observation and maintaining small errors of estimates are conflicting requirements in weather radars. Recently a new method for estimation of spectral moments had been proposed and successfully tested. The scheme operates on oversampled echoes in range; that is, samples of in-phase and quadrature phase components are taken at a rate several times larger than the reciprocal of the transmitted pulse length. The spectral moments are estimated by suitably combining weighted averages of these oversampled signals (in range) with usual processing of samples (spaced at pulse repetition time) at a fixed range location. The weights in range are derived from a whitening transformation, hence, the oversampled signals become uncorrelated and consequently the variance of the estimates decreases significantly. Because the estimates’ errors are inversely proportional to the volume scanning times, it follows that storms can be surveyed much faster than is possible with current processing methods, or equivalently, for the current volume scanning time, accuracy of the estimates can be greatly improved. This massive improvement is achievable at large Signal-to-Noise Ratios (SNR). Extensions of the method so that the estimates do not degrade at low SNR have also been developed. After full deployment of the ORDA (sometime in 2007) these new techniques will become part of the signal processing algorithms.
c) Increase in capability – Dual polarization
As part of the evolution and future enhancement of the NEXRAD radars, the National Severe Storms Laboratory recently upgraded the KOUN WSR-88D radar to include polarimetric capability. In 2003 the Joint POLarization Experiment (JPOLE) was conducted in central Oklahoma to test the practicality and operational utility of a polarimetric WSR-88D radar (http://cimms.ou.edu/~schuur/jpole/).
The experiment lasted one year and provided a large dataset that demonstrates advantages of dual-polarization. Notable are significant improvements in point and areal rainfall estimates (1.7 and 3.7 times smaller rms errors than for estimates from reflectivity factor) and measurements of heavy precipitation. Also, confirmed was the unique classification capability to identify nonmeteorological echoes (ground clutter/anomalous propagation, insects, birds, and chaff) and superior hail detection. Other achievements are discrimination between rain and snow, very precise identification of the melting zone, and even detection of tornadoes. Yet, there are problems such as determination of hail size, presence of icing conditions, or separation of dry aggregated snow from light rain that challenge radar polarimetry.
The highly successful operational validation of the polarimetric method was followed by a cost benefit study which suggests that $ 17838 million (in 2003 dollars) would be saved over twenty years if the network acquires the polarimetric capability. And this is a conservative estimate as the study considers gains in water management and reduction in flood damages but excludes mitigation of effects from snow storms. With such clear justification a decision has been made (by DOC, DOT, and DOD) to initiate the upgrade and start the retrofit in about 2008.
Detection of phenomena, determination of their strength or hazard potential, tracking of these, and quantitative measurements of precipitation and winds are at the essence of various algorithms in the RPG. Evolution of algorithms is an ongoing process and will continue to occupy practitioners and scientists for years to come. Presently, much work remains to enhance the existing algorithms. Most of these use procedures that are based on decision trees. Progress has been made in classification of echoes by applying fuzzy logic schemes to polarimetric radar data. Similar schemes and neural networks are proposed for detecting vortices. New more efficient methods are sought. Modern pattern recognition and image processing techniques need to be explored together with hybrid approaches that blend numerical models with data. Continuous assimilation of high quality polarimetric weather radar data into numerical models might increase lead time for detection of hazards such as tornadoes, flash floods, hail, and wind shear (microbursts).
III New systems
There are at least three desirable features that can not be achieved with the improved WSR-88D technology. These are a) update of volume scans at intervals of 1 minute or less, b) observations near ground level over large areas, and c) multipurpose use to sample weather, control air traffic, and track non cooperative airplanes. The agile beam phased array technology can deal with issues a) and c) whereas a dense network of small, short wavelength radars is a contender to alleviate b). I anticipate that the time for operational implementation of these is beyond 2015.
a) The National Weather Radar Testbed (Phased-Array)
The National Weather Radar Testbed (Phased-Array) facility has been established in Norman. It is based on the SPY-1A antenna and it is undergoing engineering evaluation. Although the testbed does not have polarimetric capability it is well suited for evaluating and quantifying all the benefits of agile beam. The ultimate advance in technology is the polarimetric agile beam phased array radar and a natural next step is to build one and evaluate it.
Postulated advantages of agile beam phased array that will be demonstrated are.
- Faster volume coverage by using beam multiplexing, processing of oversampled signals, and/or pulse compression.
- Adaptive scanning to match local obstruction.
- Adaptive scanning to match spatial distribution of storms.
- Adaptive change in transmitted waveforms to match the phenomena under investigation. For example transmission of longer uniform sequences in directions of suspect vortices for spectral analysis and short bursts of staggered PRT elsewhere.
- Improved ground clutter canceling due to elimination of spectrum spread caused by beam smearing.
- Tracking small aircraft while simultaneously making weather radar observations.
- Better retrieval of hazardous cross-beam winds from data with temporal separation of one minute or less.
Partners in the exploration of phased array for weather and related applications are NOAA’s National Severe Storms Laboratory and National Weather Service, Lockheed Martin Corporation, Office of Naval Research, University of Oklahoma, Federal Aviation Administration, and Basic Commerce Industries.
b) Short wavelength (3 cm) radars
These radars offer two possibilities both of which might benefit NWS. One is to serve needs of hydrology in areas far from the WSR-88D or where the WSR-88D beam is blocked by mountains. Two is to be connected in a dense network for viewing storms close to the ground. In either case polarimetric capability is essential as it allows correction of attenuation and differential attenuation at relatively close range. An inexpensive network hinges on development of affordable dual polarization antennas with some beam agility necessitating a relatively large ~ 3 deg beamwidth. Issues of clutter canceling, quality of polarimetric measurements, attenuation correction, and mitigation of ambiguities need to be resolved.
An engineering center funded by NSF and lead by University of Massachusetts has been established to work, among other things, on various aspects of networking 3 cm wavelength radars. CASA, short for Center for collaborative Adaptive Sensing of the Atmosphere, partners will be testing concepts in various locations with a small network of about four radars. The ultimate goal is to install affordable radars on cellular phone towers. These would have dual polarization, some beam agility, and would be operating in monostatic and bystatic modes.
The evolutionary and revolutionary advancements of the WSR-88D are well defined and will be implemented on the network by about 2010. Thereafter much effort will be expended on automatic interpretation of radar data to pinpoint storm initiation, to detect weather hazards, and to quantify precipitation as well as winds. Better radar observations might help solve the enigma of tornado genesis; search for definite precursors is ongoing and if successful it would increase lead time for tornado detection. Also forthcoming are other sophisticated uses of radar data, such as retrieval of near ground humidity or measurements of winds in non-precipitating clouds and clear air. Possibly the greatest benefit to short term (up to couple of hours) prognosis of spatially small but intense convective hazards might ensue from skillful assimilation of polarimetric weather radar data into numerical prediction models.
- Less certain and very challenging is the future of polarimetric agile beam phased array radar. Technical characteristics, design, and cost of these systems are being established. Plans are to explore and quantify the advantages for weather observations compared to the current system. Undisputable trump of agile beam phased array radars is the versatility and adaptability for multipurpose application within the meteorological community as well as outside of it, all this with a significant decrease of time between successive observations.
- Role of the 3 cm wavelength radars for hydrologic application and as dense sub-networks for observations close to the ground, where it matters, will be evaluated within the next ten years.
- In the future the Nation may have a multiple purpose long range network of agile beam phased array radars to serve diverse needs, one of which is weather observations. Imbedded within might be dense sub-networks of short wavelength radars and radars for short range hydrologic applications. Further, radar information from airborne radars and other private proprietors (e.g., TV stations, municipalities) might become part of the integral coverage. Time will tell, and I hope I live to see it