1. Overview

2. Detailed Engineering Discussion

3. Maintenance Perspective

4. NCAR Recommendations and OSF Responses

5. Conclusions and Observations

6. OSF Plans

Calibrate (American Heritage Dictionary) - to check, adjust, or systematically standardize the graduations of a quantitative measuring instrument.
 

The entire radar system is calibrated in two steps.

• The transmitter and receiver are calibrated by an "end-to-end" calibration referenced to a calibration port behind the antenna: i.e., excluding the antenna system.
• The antenna system parameters (gain and pointing accuracy) are verified by sun flux measurements.

 

• Translating the terms in the radar equation for return power, P_r, to the calibration port
• Solving the radar equations for Z_e in terms of P_r
• Injecting a known receiver input power, P_c, into the calibration port
• Adjusting the output Z_e to conform to the value predicted by the radar equation for the given calibration power

 

Contractor - Deliver a system CAPABLE of being calibrated to less than 1 dB. NTR 3.7.1.2.3.2

Establish initial site calibration.

Not responsible for network-wide calibration.
 

Delivered - Two fully automated self-calibration procedures (online)

1) Every VCP during antenna retrace

2) At cold start and every 8 hrs

- SUNSCAN

Suitable for pointing accuracy but not suitable for transmitter/receiver waveguide loss and antenna parameter measurement

Later expanded and corrected to include flux or excess noise measurement

- MOMENT (offline)

Provides end-to-end calibration during INCO

- Diagnostics (offline)

RDASOT

- Tech manuals

Insufficient to assure continued calibration during field operation
 

Government - Witness INCO site calibration

Establish calibration processes and procedures for the network
 

Implemented calibration improvement efforts based on OSF report "Calibration of the WSR-88D" dated Sept 30, 1992
 

Conceived a practical method of implementing internal calibration check without use of test software Nov 7, 1992 (First report objective)

- Proved concept on KTLX radar in October 1993

- Initial Tech manual release Oct 1995
 

Discovered and corrected 3dB software design error in sun flux measurement

- Software error in RDASOT discovered September 1994

- Temporary fix to tech manual October 1995

- RDASOT software corrected in Build 10

- This fix accomplishes second major objective of Calibration report
 

Online and Offline calibration tech manual procedures improved (incl. in Oct 95 version, automated in Build 9.0 software, Oct 96)

- INCO shortcomings corrected

- New procedures conceived and incorporated
 

Network calibration-related analytical tools developed

- PC based software does statistical analysis of Adaptation Data

- PC based long term plotting and statistical analysis of site delta SYSCAL
 

Calibration seminars provided, (from 1992 to present)

- OSF Hotline El Techs

- NWSTC instructors

- Keesler instructors

- NRC technicians

- AF Special Maintenance Team
 

CCR for Reflectivity Error Estimate submitted (Sept 97)
 

RDA Maintenance Manual (EHB 6-510)

Original dated 15 Aug 1992

- Receiver Signal Processor Calibration Section 6-6.28 was particularly deficient

- Change 6 dated 6 Oct 1995 introduced complete rewrite of 6-6.28

- Maintenance Note 15 dated 29 Nov 1995 required performance of offline calibration within 60 days. Deadline could not be met because NWS test equipment needed calibration.
 

Complete manual revision dated 1 May 1996

- EHB 6-6.28 developed further
 

- Change 1 dated 1 Oct 1996 added provisions for redundant sites
 

- Change 2 dated 15 Aug 1997 updated to Build 9 software
 

- Change 3 targeted for 1 Apr 1998 updates to Build 10
 

Test Equipment

- HP436A and HP8481A provide power measurement uncertainty of 0.06 dB

- Frequency generation has an accuracy of 0.00004 dB
 

Test Equipment Certification Required

- Established for NWS starting in 1996, with contract options for 1997 and 1998

- (Note, NWS sites deferred Maintenance Note 15 until equipment was calibrated)

- Calibration is routine at DOD and FAA sites
 

RETURN TO TOP OF DOCUMENT

Return to main page.

P_r = average return power, watts

P_t = transmitter power, watts

G = antenna gain, dimensionless

= radar wavelength, m

= antenna half-power beamwidth, radians

= equivalent energy pulse width, s (4 db pulse width here)

c = electromagnetic propagation constant (3 X 10⁸ m/s)

R = range to pulse volume, m

K = complex index of refraction; K² is conventionally taken to be 0.93 for water and 0.2 for ice

Z_e = effective radar reflectivity factor, m³ (Often Z_e is expressed in mm⁶/m³ for use in empirical rainfall rate equations such as Z = 200r^1.6 with rate (r) in mm/hr. This requires a units conversion factor of 10^-18.)

L = loss factors associated with propagation and receiver detection
 

Reflectivity, Z_e, is calculated from the return power by solving the radar equation for Z_e in terms of P_r.

L_p = range-dependent propagation loss, L_t = transmitter waveguide loss

L_r = receiver waveguide loss, L_d = receiver detection loss
 
 
 

In the WSR-88D, it is desirable to express Z in mm⁶/m³, P_r in mW, and range in Km, requiring unit conversion factors of
 
 
 

Equation (2) may thus be recast as where the first three terms are independent and range-dependent variables, and the last term is a form of the radar constant.

In logarithmic form, reflectivity is given by where Z_e is normalized to 1 mm⁶/m³ and C is the radar constant.
 

The terms of Equation (3) can be determined with accuracy more than adequate for calibration to +/- 1 dB.

@ measured at field site

* taken from vendor measurements, with dry radome included
 

The factor SYSCAL contains all the quantities in the last term of equation (3) plus the constant converting receiver output power (the digital quantization factor, a^2, in digits squared per mW) to receiver power in dBm and the receiver gain, g, relating input to output power.

The WSR-88D calculation for dBZ_e becomes 
 

where P₀ = receiver output power and SYSCAL is typically in the range of 8 dB to 12 dB.
 

Two terms are monitored actively - P_t (continuously) and a²g (during online calibration):

- Changes in these parameters appear as DELTA SYSCAL, S.

- After system grooming and Reflectivity accuracy of +/- 1dB is established, delta SYSCAL is baselined.

- Significant departures of delta SYSCAL from this baseline (zero) value indicate a change from the initial conditions (not necessarily reflectivity error).
 

RETURN TO TOP OF DOCUMENT

Return to main page.

The reflectivity calibration is accomplished by injecting 4 test signals (simulated test targets, 1 cw, 3 pulses) into the front end every volume scan.
 

The EXPECTED reflectivity of each of the test targets is calculated in software using the following equation:

During self-calibration all of the terms in the equation are considered constants with the exception of P_r, the test target power referred to the reference point and P_t, the measured transmitted power. We may, therefore, rewrite the equation as:

Actually, since P_r is not measured directly but is obtained from stored power and path loss data, only the measured transmitted power changes from volume scan to volume scan. While in operation, this is the only parameter which will change Z_e(Expected).

The MEASURED reflectivity of the test targets is determined as:

dBZ_e(Measured) = ECHO POWER + 20LogR + R*Atmos + SYSCAL.
 

Since the range R of the test targets is fixed as programmed in the software calibration routine, this equation may be rewritten as:

dBZ_e(Measured) = ECHO POWER + K2 + SYSCAL.
 

In the calibration process SYSCAL, the system calibration variable, is automatically adjusted so that Z_e(Measured) = Z_e(Expected). Since only one value of SYSCAL is produced, the result of the calibration process is that the average difference between expected and measured reflectivity of the four test targets is set to zero by establishing the baseline value of SYSCAL.
 

During the active volume scan, the resultant value of SYSCAL is applied to ECHO POWER to produce accurate reflectivity measurements of weather returns.
 

Should be initiated by large delta SYSCAL, alarms, or PMI's

(Alarm if delta SYSCAL exceeds 4 dB)
 

Can be analyzed by Reflectivity Error Estimate procedure
 

To eliminate Reflectivity Error

- test signal path must be calibrated

- Power Monitor consistency must be checked and power errors, if any, corrected by calibration
 

After calibration, accuracy is verified by existing tech manual procedure
 

Baseline value of SYSCAL then is established
 

Inaccuracies in Test Signal Path
 

Inaccuracies in reported transmitted power P_t
 

Inaccuracy in Antenna parameters

- Monitored by Sunscan Subtest 2
 

Inaccuracy in Test Equipment
 

- Sunscan checks antenna position, antenna, radome, and waveguide gain/loss
 

- Compares noise power of sun (as reported by solar observatory) with internal noise source
 

RETURN TO TOP OF DOCUMENT

Return to main page.

Make test equipment, sensors, generators, and techniques traceable to a defined level.

- Test equipment calibration routines and procedures with traceability to NBS are in place for the entire network, all three agencies
 

Develop the means for acquiring engineering databases for each site

- A limited capability is in place with Operations and Engineering remote site status monitoring and statistical analysis including delta SYSCAL. (Alaska and Pacific sites added July 97)

- This capability will be expanded with "Wiretap" and offline Comprehensive Calibration Verification
 

Execute rigorous cross-check manual verification on a sampling of sites

- Verifications are in progress

Devise a comprehensive calibration process management and control strategy which includes engineering, operations, training, logistics, and quality assurance, emphasizing statistical sampling and analysis

- Process management in place. This will be more comprehensive when "Wiretap" and Calibration Verification data become available.
 

Expand automation, data communication, and database aspects of the calibration process, including an on-line network calibration database

- Limited capability in place with Operations and Engineering site status monitoring, including delta SYSCAL

- Will be significantly expanded with "Wiretap"

- Calibration data included on level 2 tapes with Build 10 software

- Develop sub-process improvements, particularly in automation, traceability, and locating assignable causes of calibration uncertainty

Sufficient information is available from existing status monitoring to identify problem sites

CCR for calculation and display of reflectivity error estimate submitted

6 year modification plan proposes funding for calibration automation

Open RDA project provides opportunity for improvement!

Test equipment is now traceable to national standards
 

RETURN TO TOP OF DOCUMENT

Return to main page.

Calibration engineering design and hardware used in the online and offline calibrations is capable of system calibration to less than 1dB

Analysis of field data indicates earlier (before 1992) first order radar calibration problems were combinations of lack of test equipment certification and execution of necessary calibration procedures. Present first order problems are associated with execution of existing procedures

There are some known field RDA alarm and calibration problems

System calibration can be monitored regularly from a central point

- comprehensive calibration field check is possible

Completion of Maintenance Note 15 will establish baseline calibration
 

RETURN TO TOP OF DOCUMENT

Return to main page.

Operations and Engineering will continue to assist in correction of observed problems

The OSF will assist in development of on site maintenance follow-on training similar to the distance learning capability available for operators

OSF will incorporate into routine field procedures an "offline" comprehensive calibration verification (with minimum field staff impact)

OSF will expand and, eventually, incorporate routine monitoring and analysis of site performance and calibration parameters (all sites)

OSF will implement on line calculation and display of reflectivity error estimate
 

Long Term Plans:

OSF will propose extending "sun scan" to include antenna main lobe pattern measurement

Will investigate modification of on-line calibration routine from point check linear extrapolation over linear range to input/output regression over full dynamic range

Will investigate tightening boundaries on calibration alarm monitoring

OSF will study refinement of atmospheric loss correction routine

OSF Engineers will review code to identify changes that may increase precision
 

RETURN TO TOP OF DOCUMENT

Return to main page.