open rda build 7.0 training · in other cases the opera-tional impact is a significant improvement...

Version: 0511

Open RDA Build 7.0 Training

Presented by the Warning Decision Training Branch

ORDA

ORDA

RDA

RDA

2

Warning Decision Training Branch

Overview The Open RDA (ORDA) is a significant upgrade tothe Radar Data Acquisition (RDA) functional areaof the WSR-88D. The ORDA design allows for theremoval or replacement of many of the legacyRDA hardware components, such as the signalprocessor. This training presents the purpose ofthe ORDA upgrade, an overview of the hardwareand software, the ORDA-related windows at theMaster System Control Function (MSCF), and theoperational impacts.

Differences The Base Data generated by the ORDA has somedifferences in appearance as compared to the leg-acy RDA. In some cases these differences havelittle operational impact. In other cases the opera-tional impact is a significant improvement in thequality of the Base Data. By design the ORDA willproduce Base Data quality that is as good as orbetter than the legacy RDA. The information in thisdocument reflects the pre-deployment state ofknowledge of the operational impacts of the ORDABuild 7.0.

Resources Additional copies of this document, as well as allrelevant training materials are available from theORDA Build 7.0 Training web site:

http://wdtb.noaa.gov/buildTraining/ORDA/

Any new information that becomes available asthe deployment progresses will also be posted onthis site.

Deployment The ORDA Deployment is scheduled for Novem-ber 2005 through September 2006. The date forany particular office is available from the ORDAWeb Site:

3


http://www.orda.roc.noaa.gov/deployment/sched-ule/schedule.asp

The ORDA and RPG Build version that will beused for the initial part of the ORDA deployment isBuild 7.0. ORDA/RPG Build 8.0 is scheduled to bedeployed in the Spring of 2006. At the time whenORDA/RPG Build 8.0 is released, those sites thatalready have ORDA installed will be upgraded toBuild 8.0 on both the RPG and ORDA. For allremaining Single Channel sites, the initial ORDAinstallation will be ORDA Build 7.0, followed by anupgrade to ORDA/RPG Build 8.0. Only NWSRedundant and FAA sites will have an initialORDA installation with Build 8.0. The material inthis document is based on ORDA and RPGBuild 7.0.

PurposeThe requirements for converting the legacy RDAto an open systems design (ORDA) are the samerequirements that led to the conversion of the leg-acy RPG to the ORPG. An Open System design isflexible with respect to hardware, software andcommunications. The conversion to the ORDAdesign significantly improves processing speeds incomponents such as the signal processor. It offersimprovements, such as a better calibrated radarwith simpler calibration procedures, that benefitboth weather decision makers and system main-tainers.

New Science and Applications

The flexible design will allow for planned upgradesand provide a necessary foundation for futureenhancements. These include advanced tech-niques for mitigating velocity and range ambigu-ities, and Dual Polarization.

4


Hardware Overview The ORDA system is essentially a new set ofhardware components, along with the necessarysoftware and communications equipment. Thecomponents that are of operational interest arepresented here.

RVP8 - Radar VideoProcessor

The Radar Video Processor (RVP8) is the digitalreceiver/signal processor of the ORDA system(Figure 1). The RVP8 is a SIGMET Corporation,industrial grade PC with LINUX as the operatingsystem. The processing speed and memory of theRVP8 is sufficient for current and future complexalgorithms such as Dual Polarization.

The RVP8 replaces the legacy Hardwired SignalProcessor (HSP) and the Programmable SignalProcessor (PSP). Though it is a commercial prod-uct, it does contain some features that were cus-tomized by SIGMET for the WSR-88D. Forexample, the use of Batch mode (alternating lowand high PRFs at middle elevation angles) isunique to the WSR-88D. SIGMET modified thecode in the RVP8 to accommodate Batch mode.

RCP8 - Radar ControlProcessor

The Radar Control Processor (RCP8) is a replace-ment for the Concurrent Micro 5 computer (Figure2). As with the RVP8, the RCP8 is also a commer-

Figure 1. The RVP8, ORDA’s digital receiver and signal processor.

5


cial product from the SIGMET Corporation, run-ning with LINUX as the operating system.

With the legacy system, when the RDA was con-trolled locally at the RDA shelter, the commandswere going to the Micro 5. Similarly with the ORDAsystem, when the RDA is controlled locally at theRDA shelter, the commands are going to theRCP8. When the RDA is controlled remotely fromthe MSCF, the commands are also going to theRCP8.

The RCP8 is connected to the Digital Control Unit(DCU), which drives the pedestal and implementsthe current VCP. When a VCP has been changedor downloaded, the RCP8 sends the VCP informa-tion to the DCU. The DCU is essentially telling thepedestal and antenna “what to do” in order to exe-cute that particular VCP.

RCP8 and VCP UsageThe local versions of the VCPs are stored at theRCP8. As with the legacy RDA, the local VCPs willbe limited to 11, 21, 31, and 32. The Change com-mand is used at the RPG in order to invoke one ofthe local VCPs, while the Download command isused to invoke the remote VCPs. The remoteVCPs are stored at the RPG and are 11, 12, 21,121, 31 and 32. With ORDA Build 7.0, there will be

Figure 2. The RCP8, ORDA’s radar control processor.

6


no need to adjust local procedures for changing ordownloading VCPs. There is no change to theVCP Control window at the RPG (Figure 3).

The RCP8 also relays information to the transmit-ter through the RVP8. For example, different PRFsare implemented throughout a VCP, and the trans-mitter needs to know how rapidly to transmitpulses. The RCP8 sends the PRF information tothe RVP8. The RVP8 triggers the transmitter at thecorrect rate for the current PRF.

KVM At the RDA shelter, the Keyboard Video Mouse(KVM) and Monitor provide access to both theRCP8 and the RVP8. The KVM is the interface forlocal user access only (Figure 4).

GPS The ORDA has a GPS server that sets the clockson the RVP8 and the RCP8 within very close toler-ances (Figure 5). The GPS servers will increasethe accuracy for every WSR-88D with respect to

Figure 3. VCP Control window. Note VCPs stored at the RDA vs. RPG.

7


time. Since it is GPS-based, the location of eachradar will also be highly accurate.

RDA HCI at the MSCFThe MSCF (Figure 6) is used to launch the RPGHuman Computer Interface (HCI). The RPG HCI isthe set of windows where weather decision mak-ers perform radar control tasks such as download-ing VCPs or changing the Doppler PRF. Thewindows within the RPG HCI that are impacted bythe ORDA will be presented where appropriate inthe Operational Impacts section (page 8).

With ORDA/RPG Build 8.0, the RDA HCI will beaccessible from the MSCF by clicking on the RDAHCI button. The RDA HCI main page is the sameinterface that is available to technicians locally at

Figure 4. The KVM and monitor, for local user access.

Figure 5. The GPS server.

8


the RDA shelter. The design is similar to the RPGHCI, with dynamic status information such as

• the current VCP and elevation angle• the state of the wideband connection• RDA state (operate or standby)• power source (utility or generator)

OperationalImpacts

This section presents the operational impacts ofthe ORDA:

1. Improvements in calibration2. Improved sensitivity in long pulse (VCP 31)3. New clutter suppression technique: Gaussian

Model Adaptive Processing (GMAP)4. Differences in Batch elevation Reflectivity data

Figure 6. MSCF and the RPG HCI.

9


5. Improvements to the quality of Spectrum Widthestimates

6. Differences in elevation angle settling 7. End of first trip Velocity less noisy with ORDA8. Differences in types of RDA Alarms9. Differences in RDA Performance Data Menus10.False Alarm at AWIPS

1. Improvements in Calibration

The ORDA design offers a better calibration proce-dure which is both more accurate and simpler toimplement. Compared to the legacy design, thereare only about 1/3 the number of components thatneed to be calibrated. In addition to fewer compo-nents that require calibration, this shortens the“path” that the signal must follow.

Calibration is performed by injecting a test signalinto the receiver and comparing the actual result toan expected result. With the legacy design, thetest signals that were used were at two fixedpower levels. With the ORDA, the test signal usedfor calibration varies over the entire dynamic rangeof the received signal, i.e. from the weakest to thestrongest. Testing has shown that the dynamicrange for ORDA is 96 dB, which is 3 dB better thanthe NEXRAD specifications.

Off-line CalibrationThe off-line calibration process is better automatedand can be viewed graphically. The technician isable to see the actual dynamic range that wasgenerated, and can also see the linearity of thetransition from lower to higher power (the more lin-ear, the better).

On-line CalibrationAs with the legacy RDA, an on-line calibration pro-cedure is performed at the end of each volumescan as the antenna drops back down to 0.5°. A

10


number (Delta Sys Cal at AWIPS or Calib at theRPG HCI), will continue to be generated each vol-ume scan, with ±4 as a maintenance mandatorycondition and ±2 as a maintenance required condi-tion. With the calibration improvements that theORDA brings, this number is expected to showless variation over time as compared to the legacyRDA.

With fewer components as compared to the legacyRDA, this on-line calibration procedure is com-pleted much faster with the ORDA.

Calibration and RadarProducts

A better calibrated radar produces better qualityBase Data, which positively impacts all of theBase and Derived Products. Rainfall estimates areparticularly sensitive to calibration errors. With theZ=300R1.4 relationship, a calibration error of +4 dBdoubles the rainfall rate, while an error of -4 dBhalves the rainfall rate. Calibration errors that arenot this extreme still result in significant rainfallestimation errors over time. The simpler and moreaccurate calibration procedures with ORDA areexpected to reduce the impact of calibration errorson the accuracy of rainfall estimates.

2. Improved Sensitivity inLong Pulse (VCP 31)

One benefit of the ORDA hardware design is animprovement in the sensitivity, especially whenusing long pulse (VCP 31). Having some tasksperformed digitally with ORDA eliminates lossesdue to hardware that were present with the legacyRDA. This results in an increase in long pulse sen-sitivity of about 3 dB, which will improve detectionof weak returns such as boundaries and snowbands.

Figure 7 depicts VCP 31 on the left with VCP 32(just 10 minutes later) on the right. Both imagesare from an ORDA system and there is no weather

11


present. Note the improved areal extent of weakreturns with VCP 31. This improvement in arealextent will be greater with an ORDA VCP 31 ascompared to a legacy RDA VCP 31.

3. New Clutter Suppression Technique: GMAP

Reference: A First Look at the Operational (DataQuality) Improvements Provided by the OpenRadar Data Acquisition (ORDA) System, J. Chris-man, C. Ray, Radar Operations Center, Norman,OK

The new clutter suppression technique with ORDAis called Gaussian Model Adaptive Processing(GMAP). A Gaussian curve is a good approxima-tion of the power spectrum for weather targets aswell as clutter. Gaussian is part of this titlebecause GMAP uses this approximation. Adaptiveis part of this title because the routine adapts tothe weather and clutter that is present in eachrange bin. Signal removal is based on the spec-trum width of the clutter signal, which GMAP iden-tifies in an iterative process.

Figure 7. Comparison of VCP 31 (left) and VCP 32 (right) with ORDA.

12


There are similarities and differences between thelegacy RDA clutter filtering technique and GMAPwith respect to both design and implementation.The most significant difference between GMAPand the legacy RDA clutter filtering technique isthat with GMAP most of the weather signal thatis removed with the clutter will be restored.

Legacy RDA filtering With legacy RDA clutter suppression, a level ofsuppression (low, medium, or high) was selectedfor each clutter region defined. Each of these lev-els of suppression had an associated velocitynotch width, though the actual notch width varieddepending on the antenna rotation rate and thewaveform. Clutter filtering was applied to the sig-nal that fell within the particular notch width, cen-tered on zero velocity. Velocity notch widthsranged from ±1.7 kts in the lower (Split Cut) eleva-tions to ±10 kts or more in the middle (Batch) ele-vations.

This technique worked well provided that theentire clutter signal was within the notch width andthe entire weather signal was outside of the notchwidth. Any signal that fell within the notch widthwas removed, whether it was composed of clutteror weather. Once the signal was removed by thelegacy clutter filters, it was no longer recoverable.In Figure 8, both the clutter signal and the weathersignal are centered at zero velocity and clutter fil-tering has not yet been performed. The blackcurve is the Gaussian estimate of the weather sig-nal.

Figure 9 depicts the result after legacy RDA clutterfiltering has been applied. The clutter signal fallswithin the notch width and has been removed, butthe weather signal within the notch width is also

13


removed. Since the computed returned power isthe area under the curve, legacy RDA clutter filter-ing in Figure 9 results in a loss of returned powerfor that range bin and an underestimate of reflec-tivity.

Figure 8. Before legacy RDA filtering is applied, a clutter signal (blue) and a weather signal (red) are both centered on zero velocity.

Figure 9. Legacy RDA clutter filtering has been applied. Though the clutter signal has been removed, the weather signal that fell within the notch width has also been removed.

14


GMAP clutter filtering GMAP still removes a portion of the signal cen-tered around zero velocity. However, the width ofwhat is initially removed is not fixed, but adaptive.This width is also significantly narrower than thenotch widths used by the legacy RDA filter. GMAPuses 1.33 kts (± .67 kts) as an initial width, then asneeded it can iteratively narrow this width to betterisolate the clutter signal. GMAP processing alsohas steps that rebuild the weather signal using theGaussian estimate of that signal which existedprior to clutter removal.

The performance improvement with GMAP will bemost noticeable where there is both clutter andweather signal in the same range bin with veloci-ties close zero. See Figure 10 for the result ofGMAP suppression in the case where the clutterand weather signal are both centered on zero(same initial condition as in Figure 8).

Weather signals are not usually centered at zerovelocity. GMAP can rebuild any portion of theweather signal that falls within the clutter signal

Figure 10. GMAP clutter filtering has been applied. Though the clutter signal has been removed, the weather signal has been mostly restored.

15


that was initially removed. In Figure 11, theweather signal is offset from zero and no clutter fil-tering has yet been applied.

In Figure 12, the legacy RDA filtering has beenapplied. A portion of the weather signal has beenremoved along with the clutter signal. This resultsin a bias of the base data induced by the clutter fil-tering process. There is a loss of power and thusan underestimate of reflectivity. Since velocity esti-mates are power weighted, the power loss wouldalso bias the velocity estimate away from zero.

In Figure 13, GMAP filtering has been applied tothe power spectrums (clutter and weather) in Fig-ure 11. A small portion of the weather signal wasinitially removed along with the clutter signal.GMAP then rebuilt most of the weather signalacross the gap. There is significantly less clutterfilter induced bias with GMAP as compared to thelegacy RDA filtering technique.

Figure 11. In this example, the weather signal is not centered at zero velocity.

16


Radar example of GMAPperformance

In Figure 14, the velocity image shows the clutterpattern near the RDA when there is no weatherpresent. For range bins with only clutter present, aprocess called censoring assigns no data. Notethe areas where clutter has been removed andthere is no data, particularly a ridge at short rangeto the southwest.

Figure 12. Legacy RDA clutter filtering has been applied. Though the clutter signal has been removed, the weather signal that fell within the notch width has also been removed.

Figure 13. GMAP clutter filtering has been applied. Though the clutter signal has been removed, the weather signal has been mostly restored.

17


In Figure 15, the reflectivity image on the leftshows that a squall line has passed through and isjust east of the RDA (note the scale of the rangerings). On the associated velocity image which hasbeen zoomed in over the RDA, note that there areweather returns available over most of the areas ofclutter. Since GMAP can restore the weather sig-nal in areas of clutter, the ridge to the southwest isno longer apparent.

Differences in Clutter Censoring

Clutter censoring is a technique used by both thelegacy RDA and ORDA after clutter filtering hasbeen performed. Censoring is necessary sincethere are limits to the amount of power that can beremoved for both legacy and ORDA filtering.Some clutter targets (e.g. mountains at closerange) return significantly greater power than thefilters can remove. Censoring is a technique toremove residual clutter. If the residual clutter doesnot represent any weather signal, it is censored,which means it is not displayed.

Figure 14. Example of GMAP clutter removal with no weather present. Note the ridge at close range to the southwest

18


The censoring technique for ORDA has some dif-ferences that will improve the availability of data inclutter prone areas. The censoring process withthe legacy RDA is performed on bins of .54 nm(1 km) range resolution. With ORDA, censoring isperformed on bins of .13 nm (.25 km) range reso-lution. The ORDA censoring process is alsodynamic, while the legacy process is hard codedand static. The result is less data loss and an over-all smoother appearance with ORDA (Figure 16).

Types of SuppressionUnchanged

Bypass Map The Bypass Map is still a “yes/no” map for whetheror not to apply filtering. The Bypass Map will con-tinue to be used to address contamination fromnormal clutter targets (e.g. terrain and buildings).

Figure 15. Example of GMAP clutter removal and weather recovery with a squall line passing through.

19


Bypass Map generation with ORDA is a signifi-cantly faster process and the resulting maps are ata higher resolution.

The resolution of the ORDA Bypass Map is 1° x.54 nm as compared to 1.4° x .54 nm with the leg-acy RDA. A “yes” to performing suppression forany particular 1° x .54 nm bin is dependent on thecharacteristics of the returned signal. Clutter char-acteristics are high power, low velocity and lowspectrum width.

Bypass Map GenerationThe conditions under which a new Bypass Mapshould be generated have not changed withORDA. The Bypass Map should be generatedunder conditions of no precipitation, no AP orother abnormal beam refraction. When theBypass Map is generated, two versions of the sig-nal characteristics for each range bin are com-pared:

Figure 16. Example of differences in clutter censoring with ORDA (left) vs. legacy RDA (right).

RDAORDA

20


1. unprocessed data (power, velocity, and spec-trum width)

2. data processed by GMAP (power, velocity, andspectrum width)

If there is a significant power difference betweenthese two versions, then the bin is flagged asrequiring clutter suppression. That’s why it is soimportant to generate new Bypass Maps underconditions where only “typical” clutter exists. Gen-eration of a new Bypass Map is still an off-line pro-cedure with ORDA, but will take about 20 minutes.

For the Build 7.0 ORDA, there will continue to betwo maps generated. The first map is for elevationangles below 1.65° and the second map is for ele-vation angles above 1.65°. These two maps corre-spond to each of the two segments, low and high.There may be additional segments (up to 5) avail-able in future builds, resulting in additional BypassMaps.

All Bins It will still be necessary to use All Bins suppressionfor areas of AP. Since AP is transient in space andtime, the Bypass Map cannot identify where theclutter targets are located. All Bins suppressionapplies suppression on every bin throughout thegeographic region that has been defined.

Applying All Bins suppression to areas of AP byGMAP has been found to be just as effective atremoving the AP as the legacy clutter suppressiontechnique. However, since GMAP is able torestore most of any weather signal that is initiallyremoved, there is less bias in the base data. WithAll Bins suppression in the legacy RDA applied toweather, reflectivities were often significantlyreduced and velocities were biased away from

21


zero. With GMAP, the base data will be less biasedwhen All Bins is used, as compared to the legacyRDA.

AP Removal ExampleIn the following example, a large area of AP existsover most of the radar display with small areas ofprecipitation to the east and south (Figure 17).

This data example was captured for both ORDAand a legacy RDA which was nearby. For bothradars, All Bins suppression was applied toaddress the AP contamination. In Figure 18, AllBins suppression is applied to the ORDA (left) andlegacy RDA (right) Reflectivity data.

In Figure 19, All Bins suppression is applied to theORDA (left) and legacy RDA (right) Velocity data.Note that the ORDA GMAP is just as effective atremoving the AP as the legacy RDA technique.

Figure 17. There is unfiltered AP over most of the display with areas of precipitation to the east and south.

22


Differences in the ZeroIsodop Area with All Bins

When All Bins suppression is applied to weathertargets, there is a difference in the appearance of

Figure 18. AP removal using All Bins for Reflectivity (ORDA left and legacy RDA right).

RDA

ORDA

Figure 19. AP removal using All Bins for Velocity (ORDA left and legacy RDA right).

RDA

ORDA

23


the zero isodop area with the legacy RDA vs. theORDA. Note from Figure 20 that the zero isodoparea with the legacy RDA (right) is very narrowcompared to the ORDA (left). This example is froma Batch mode elevation, where the difference willbe the greatest. Since the legacy RDA is unable torestore any weather signal that is removed, thevelocity estimates are often biased away fromzero. With GMAP, most of the weather signal isrestored, significantly reducing this bias. The resultis a wider, more realistic zero isodop area.

Implementing GMAP Clutter Suppression

Region and Select CodeThere are fewer operator decisions required withGMAP. There are no longer any notch widths, sothere are no longer any levels of suppression forthe Doppler and Surveillance channels. The geo-

Figure 20. Difference in zero isodop area when All Bins is applied in the Batch elevations.

ORDA

RDA

24


graphic area (range to range and azimuth to azi-muth) for a clutter suppression region still needs tobe determined, followed by the Select Code(Bypass Map, All Bins, or None), which defines thetype of suppression.

Elevation Segment Another difference with the Clutter Regions Editorwindow design (Figure 21) is the number of eleva-tion segments. With the legacy RDA, there weretwo segments, low (<1.65°) and high (>1.65°) andtwo Bypass Maps were built, one for each seg-ment.

Note that in Figure 21 there are the numbers 1through 5 listed as segments, with only 1 and 2enabled. GMAP allows for up to five different seg-ments. However, ORDA Build 7.0 matches the leg-acy design. Segment 1 will be the low segmentand Segment 2 will be the high segment. Addi-tional segments are expected to become select-able in later builds and the Bypass Map generationprocess will build one map for each segment.

Recommendations forGMAP Implementation

1. Legacy Bypass Maps cannot be transferred tothe ORDA. Bypass Maps will be generated bythe ORDA installation team as soon as theinstallation is complete. This is done to verifythat the map generation process is valid, and isunlikely to be conducted under ideal conditions.

2. After installation, sites can use GMAP All Binseverywhere until Bypass Maps can be gener-ated under ideal conditions.

3. Bypass Maps should be generated as soon asideal conditions permit.

At the time of this writing, additional instructionsconcerning the process of Bypass Map generationare being developed. Any documents produced by

25


the Radar Operations Center will also be availablefrom the ORDA Build 7.0 Training web site:


4. Once Bypass Maps are generated, sites shouldthen use GMAP Bypass Map or All Bins as theyjudge operationally appropriate. However,applying All Bins suppression everywherefor the high segment (Batch elevations)should be avoided.

Figure 21. Clutter Regions window at RPG with ORDA installed.

26


4. Differences in BatchElevation Reflectivity

Data

There have been problems with the legacy RDAwhen applying All Bins suppression in the Batchelevations, in some cases significantly biasing thebase data. With the ORDA, some of these prob-lems have been mitigated. However, there arestill reasons to avoid applying All Bins sup-pression in the Batch elevations with ORDA.

There are two differences with ORDA Reflectivitydata in the Batch elevations as compared to thelegacy RDA. The first difference is a ring of slightlyreduced reflectivity which corresponds to theboundary of the first trip in the associated velocity.This is a consequence of data processing at theboundary of the first trip in the velocity. It is alsoaccentuated by applying All Bins suppressioneverywhere in the high segment (Batch eleva-tions). The data processing technique will beimproved with ORDA Build 8.0. The second differ-ence is slightly reduced reflectivity in areas thatcorrespond to range folding in the velocity. Thissecond difference is entirely a consequence ofapplying All Bins suppression everywhere in thehigh segment (Batch elevations).

1. Ring of Slightly ReducedReflectivity

In Figure 22, the data examples come from the3.4° elevation angle. In the reflectivity product,there is a ring of slightly reduced reflectivity. Thisring corresponds to the boundary at the end of thefirst trip in the associated velocity product.

The first few range bins near the radar do not havedata assigned to them. This small data gap nearthe radar also makes data assignment difficult atthe boundary at the end of the first trip of thevelocity data. This problem was also present withthe legacy RDA, but was mitigated by a smoothingtechnique applied to the bins at the end the veloc-

27


ity first trip. ORDA Build 8.0 will apply a similarsmoothing technique to mitigate this problem. Thering is accentuated when All Bins suppression isapplied everywhere in the high segment (Batchelevations), but All Bins is not the primary cause.

2. Areas of Slightly Reduced Reflectivity

Batch processing alternates between low and highPRFs, such that each radial is sampled by a num-ber of pulses from each of the low and high PRFmodes. In the Batch elevations, one of these twomodes is used for reflectivity processing, depend-ing on whether the associated velocity data isrange folded. In areas that are range folded, thereflectivity estimation comes from the low PRFdata, which has a low number (3-10) of pulses. Inareas that are not range folded, the reflectivity esti-

Figure 22. Batch elevation reflectivity with All Bin suppression applied to the high segment. Note the ring of reduced reflectivity which is associated with the end of the first trip in the velocity data.

28


mation comes from the high PRF data, which hasa high number (25-278) of pulses.

With ORDA, GMAP is applied only to the highPRF data in the Batch elevations where the num-ber of pulses is sufficiently high. For the low PRFdata, a simple cancellation filter is applied sincethe number of pulses is very low. These two differ-ent clutter filtering techniques and the distributionof range folding in the associated velocity datamay result in slight differences in the magnitude ofreflectivity. This use of differing filtering techniquesin the Batch elevations also existed with the legacyRDA, but differences in the data were less appar-ent.

In Figure 23, there is an example of reflectivityreduction beyond the first trip that is associatedwith range folded velocity data. All Bins suppres-sion was in effect everywhere within the high seg-ment (Batch elevations). To mitigate this problem,sites are encouraged to generate Bypass Maps atthe earliest opportunity after initial ORDA installa-tion and to avoid using All Bins suppressioneverywhere in the high segment.

5. Improvements to theQuality of Spectrum

Width Estimates

Though the same technique is used by ORDA forestimating spectrum width, ORDA design allowsfor a better implementation as compared to thelegacy RDA. There were errors with both hard-ware and software in estimating spectrum widthwith the legacy RDA.

The legacy RDA hardware device that contributedto spectrum width errors no longer exists withORDA. ORDA also provides the needed softwareupgrade. Spectrum Width estimates are particu-larly difficult in areas where the returned power is

29


weak, and ORDA design improves that process.Where the returned power is weak, it will be closeto the noise level, often resulting in erroneouslyhigh spectrum widths.

Spectrum width data generated by the legacy RDAhad a bias toward higher values and a more noisyappearance. Though there will still be high spec-trum widths with a weak power signal, the fre-quency of occurrence is lower with ORDA and theoverall appearance is less noisy. In addition toproduct appearance, these improvements to theSpectrum Width estimation also produce betterinput to the Radar Echo Classifier (REC) algo-rithm. The REC uses all three base moments todetermine areas of unaddressed clutter. The RECis used by the Precipitation Processing System

Figure 23. Batch elevation radar products with All Bins suppression applied everywhere in the high segment. Note the area of reduced reflectivity associated with range folding in the corresponding velocity data.

30


(PPS) to prevent clutter from contaminating pre-cipitation products.

In Figure 24, the Spectrum Width on the right wasgenerated by the legacy RDA, while the SpectrumWidth on the left was generated at the same timeby the ORDA on a radar very close by. Note thatthe legacy RDA image has regions of higher val-ues and is noisier in overall appearance.

6. Differences inElevation Angle Settling

As the antenna transitions from one angle to thenext, there is a short time required for the antennato settle on the given angle and to then begin col-lecting data. The NEXRAD system requirement isthat the antenna must be within 0.2° of the targetelevation angle before data collection begins. Theantenna settling process is the most challenging at0.5°, since the antenna has transitioned downfrom the highest elevation of the particular VCP

Figure 24. Spectrum width generated by the ORDA on the left and the legacy RDA on the right, with the two radars close together. Note that the legacy has more high values and the overall look is noisier.

RDA

ORDA

31


(19.5° or 4.5°). Occasionally, this settling effecthas been and will continue to be apparent in thedata, causing a discontinuity from the azimuthwhere data collection began compared to where itended.

Figure 25 is a 0.5° Base Reflectivity product froman ORDA field test conducted at Corpus Christi,TX. Note the discontinuity to the west-southwest.Data collection began at the 247° azimuth(antenna rotating clockwise) and the antennaangle was actually at 0.65°. By the end of data col-lection, the antenna angle was at 0.48°.

Though this data discontinuity has occurred in thepast with the legacy RDA (Figure 26), it may bemore frequent with ORDA at 0.5°. Data collectionbegins at 0.5° once the antenna is within 0.2° of0.5° and the on-line calibration check is complete.With the legacy RDA, the on-line calibration pro-cess takes longer and thus allows more time forantenna settling at 0.5°. There are plans to

Figure 25. ORDA field test at CRP. Antenna settling results in a data discontinuity at 0.5°

32


improve antenna settling performance in a futuresoftware build.

7. End of First TripVelocity Less Noisy

With the legacy RDA, the transition from the first tothe second trip of the velocity data has often had anarrow ring of false, sometimes noisy, velocity val-ues. This is an artifact of the Range UnfoldingAlgorithm as it processes the first few bins of datain the first trip (close to the radar) vs. the first fewbins of data in the second trip.

The ORDA processes data close to the radar in away that has reduced the frequency of this prob-lem. For both the Split Cut and the Batch eleva-tions, the transition from the end of the first trip tothe beginning of the second trip with ORDA will bebe a ring of missing data (Figure 27). In futurebuilds, the ring of missing data will be replaced bya smooth transition from the first to the second trip.

Figure 26. Data discontinuity at 0.5° due to antenna settling on a legacy RDA.

33


8. Differences in Types of RDA Alarms

There are differences in the types of RDA Alarmswith the legacy RDA vs. the ORDA. With the leg-acy RDA, there were 8 types of alarms:

• ARC - Archive II (no longer done at RDA)

• UTL - tower utilities

• CTR - RDA controller (Micro 5 computer)

• WID - wideband

• PED - pedestal

• XMT - transmitter

Figure 27. Transition from 1st to 2nd trip in velocity data for legacy RDA (right) vs. ORDA (left).

ORDA

RDA

34


• RSP - receiver/signal processor

• USR - a device that was not used

With the ORDA, there are 7 types of alarms:

• RCV - receiver (RVP8 and other components)

• UTL - tower utilities

• CTR - RDA controller (RCP8)

• XMT - transmitter

• PED - pedestal

• COM - wideband and RDA LAN communica-tions

• SIG - signal processor (RVP8)

See Figure 28 for a comparison of the RDAAlarms windows for legacy RDA vs. ORDA.

9. Difference in RDAPerformance Data Menus

There are differences in the RDA PerformanceData menus between the legacy RDA and ORDA.From the RDA Performance Data application but-ton on the RPG HCI Main Page, the RDA Perfor-mance Data window is opened. The selections onthe RDA Performance Data window will vary fromthe legacy RDA to the ORDA (Figure 29).

One item that is often noted as part of a shiftchange checklist is the Generator Fuel Level. Withthe legacy RDA, the Generator Fuel Level waspart of the Tower Utilities submenu. With ORDA,the Generator Fuel Level is part of the EquipmentShelter window (Figure 29).

35


Figure 28. Types of RDA Alarms for legacy (top) vs. ORDA (bottom).

Figure 29. RDA Performance Data for legacy RDA (right) vs. ORDA (left).

36


10. False Alarm at AWIPS After ORDA installation, there is the possibility offalse alarms indicating an interruption in productflow. When the RDA is accessed through a dial upconnection, this alarm is falsely generated. It is ared banner alarm at the bottom of the D2D, butthere will be no interruption in radar products.

Summary andResources

This document presents the operational impacts ofORDA Build 7.0. There are differences in theappearance of the Base Data with the legacy RDAvs. the ORDA. In some cases, the differenceshave minimal operational impact. In other cases,the differences are the result of improvements inthe quality of the Base Data due to ORDA design.

All relevant training materials are available fromthe ORDA Build 7.0 Training web site:


For those NWS offices that wish to track trainingcompletion, the ORDA Build 7.0 Training stream-ing presentation can also be accessed from theLMS.

open rda build 7.0 training · in other cases the opera-tional impact is a significant improvement...

Documents