244 WEATHER AND FORECASTING VOLUME 13 History of Operational Use of Weather Radar by U.S. Weather Services. Part II: Development of Operational Doppler Weather Radars ROGER C. WHITON* AND PAUL L. SMITH1 Air Weather Service, Scott Air Force Base, Illinois STUART G. BIGLER National Weather Service, Washington, D.C. KENNETH E. WILK National Severe Storms Laboratory, Norman, Oklahoma ALBERT C. HARBUCK# Air Weather Service, Scott Air Force Base, Illinois (Manuscript received 14 March 1997, in ®nal form 19 February 1998) ABSTRACT The second part of a history of the use of storm surveillance radars by operational military and civil weather services in the United States is presented. This part describes the genesis and evolution of two operational Doppler weather radars, the Next-Generation Weather Radar and Terminal Doppler Weather Radar. 1. Advances made by Doppler radar operational application of this capability. Thus, opera- meteorological research tional use of Doppler weather radar had to await the development of pulse-Doppler technology (that provid- The wartime Rad Lab investigators recognized the ed the range capability) for the extraction of moments possibility that radar systems could employ the Doppler such as mean radial velocity and spectrum width from effect to measure target velocities. This offered a po- tential for remote measurement of wind speeds. The pulse Doppler spectra and techniques for interpreting Weather Bureau followed up on this potential beginning the velocity patterns observable with a single radar. in fall 1956 and continued through 1960. This early Rogers (1990) reviews the history of early efforts to effort involved conducting tests on an experimental, 3- apply Doppler techniques in radar meteorology. cm, continuous-wave (CW) Doppler weather radar sys- In 1961, the Air Force Cambridge Research Labo- tem at Wichita Falls, Texas, and Wichita, Kansas. Al- ratories (AFCRL) put into operation a 5-cm pulsed though plagued by noisy magnetrons and attenuation by Doppler radar called Porcupine that was adapted for rain, these early systems were capable of detecting 205- meteorological measurements. A signal and data pro- mph (;30 km h21) winds near a tornado vortex (Holmes cessor called the plan shear indicator (PSI) (Armstrong and Smith 1958; Rockney 1960; Smith and Holmes and Donaldson 1969) was developed in connection with 1961). The inability of a CW radar system to determine the Porcupine Doppler to enable the Doppler data to be the range to the target was a serious impediment to displayed in real time. The PSI employed a coherent memory ®lter (CMF) to make coarse, real-time Doppler *Current af®liation: Science Applications International Corpora- spectral analyses over the entire range of the radar (Chi- tion, O'Fallon, Illinois. mera 1960; Atlas 1963; Groginsky 1965, 1966). Using 1 Current af®liation: Institute of Atmospheric Sciences, South Da- the PSI, the ®rst mesocyclone detected by Doppler radar kota School of Mines and Technology, Rapid City, South Dakota. # Current af®liation: Amherst Systems, Inc., Warner Robins, Geor- was recorded on 9 August 1968 (Donaldson et al. 1969). gia. Donaldson (1970) investigated Doppler radar's ability to resolve vortices of different sizes and showed that the mesocyclone, which Fujita and others had linked Corresponding author address: Dr. Roger C. Whiton, SAIC, 619 W. Hwy 50, O'Fallon, IL 62269. with tornadoes, could be identi®ed at more distant rang- E-mail: [email protected] es than the smaller tornado vortex signature could. q 1998 American Meteorological Society Unauthenticated | Downloaded 09/27/21 09:54 AM UTC JUNE 1998 WHITON ET AL. 245 Based on this, Donaldson (1970) developed a set of the radar. Use of a second Doppler radar with a different minimum values of shear, persistence, and vertical ex- viewing aspect permits determination of the full wind tent as requisites for identi®cation of a mesocyclone vector (with some restrictions). Dual-Doppler radar ob- signature. Kraus (1973) identi®ed the Brookline, Mas- servations were conducted by Peace and Brown (1968) sachusetts tornado of 1972 based on its vortex signature. as early as 1967, by Lhermitte (1970) and Lhermitte Investigations by the Weather Bureau to determine and Miller (1970) in 1969, by the National Hail Re- desirable characteristics of a Doppler weather radar and search Experiment (Knight and Squires 1982) in the its utility in meteorology were continued in the 1960s 1970s, and by others. Scan procedures and methods for at the National Severe Storms Laboratory (NSSL) (e.g., deriving two- and three-dimensional velocity structures, Lhermitte and Kessler 1964). The initial experiments for regions containing precipitation echoes, were de- used an X-band system with a comb ®lter for spectral veloped. It was evident, however, that for operational analysis of the echoes; this system used the same trailer purposes improved means for displaying and interpret- and some of the hardware employed in the Smith and ing the velocity patterns observed by a single Doppler Holmes (1961) experiments (E. Kessler 1997, personal radar would be necessary. The fortuitous advent of af- communication). In 1971, NSSL put into operation their fordable microprocessors for digital signal processing ®rst S-band Doppler weather radar designed speci®cally capable of operating in a real-time environment and the for severe storm studies (Sirmans and Doviak 1973). later availability of high-resolution, graphical, color dis- NSSL added a similar system at Cimarron Field, 42 km plays, greatly facilitated these Doppler radar develop- northwest of NSSL, in 1973, with full dual-Doppler ments. Fundamental research using dual-Doppler data operations beginning in 1974. Using these radars, Bur- produced the understanding needed to interpret the sin- gess, Brown, and others frequently observed mesocy- gle-Doppler velocity patterns. clone signatures and produced some of the ®rst real- The early interest focused on the storm-scale velocity time Doppler displays. The ®rst reported tornado vortex patterns, which would provide clues to storm severity. signature was associated with the Union City tornado However, the color plan position indicator (PPI) velocity (Burgess et al. 1975). Keystone research in the structure displays also revealed intriguing patterns even in wide- of mesocyclones (Burgess 1976) and tornado vortices spread stratiform precipitation (Kraus and Donaldson (Brown and Lemon 1976), combined with an abundance 1976). They were identi®ed as re¯ections of the vari- of severe storms in Oklahoma, helped NSSL set the ation of the wind velocity with height; sometimes the stage for the Joint Doppler Operational Project (JDOP) patterns contained sharp gradients indicating the pres- by providing the fundamental knowledge of storm dy- ence of frontal boundaries (Wilson et al. 1980). Such namics, based on dual-Doppler information, that was information is useful in weather analysis and forecast- necessary before single-Doppler schemes could be test- ing. Wood and Brown (1986) provide a convenient sum- ed in JDOP. mary of the characteristics of these velocity patterns and In signal processing, important mathematical under- their interpretation. pinnings had been provided by Cooley and Tukey The increased sensitivity of weather radar systems, (1965), developers of the fast Fourier transform, and by due in part to the Doppler processing performed on the the real-time Doppler velocity processing schemes of signal, meant that some of the velocity patterns could Rummler (1968a,b,c) and Miller and Rochwarger even be observed in clear air, adding to the forecasting (1972). In 1974, based on Rummler's scheme, Grogin- value of the data. The velocity azimuth display (VAD) sky (1972), Lhermitte (1972), and Novick and Glover technique earlier developed by Lhermitte and Atlas (1975) placed into operation the ®rst multichannel pulse (1961) had been improved upon by Browning and Wex- pair processor (PPP) at the AFCRL. Work at NSSL (e.g., ler (1968). That fundamental underpinning would later Sirmans and Bumgarner 1975) led to improvements in be extended by Rabin and Zrnic (1980) to provide wind the PPP technique; using different pulse repetition in- information in the clear air using data from a single tervals (Sirmans et al. 1976; Doviak et al. 1978) pro- Doppler radar. Rabin and ZrnicÂ's work would later serve vided a means of resolving some of the range-Doppler as the fundamental basis of one of the Weather Sur- ambiguities troublesome even with S-band radars. The veillance Radar-88 Doppler's (WSR-88D's) most im- improved PPP technique was implemented on the NSSL portant operating modes, the clear-air or VAD-wind radars in 1975 by Sirmans and at the National Center mode. for Atmospheric Research (NCAR) by Gray et al. Based in part on the early echo isolation algorithms (1975). After the PPP made possible real-time Doppler developed in the late 1960s, Captain D. Forsyth, then velocity calculations, color monitors, then becoming an Air Force of®cer assigned to the Air Force Geo- commercially available, permitted real-time color dis- physics Laboratory (AFGL), was asked by K. Glover plays of the radar re¯ectivity factor, mean radial veloc- to develop an echo-tracking algorithm. Working with ity, and spectrum width (Gray et al. 1975; Jagodnik et his co-investigator Captain C. Bjerkaas, another Air al. 1975). Force of®cer assigned to the laboratory, Forsyth pro- A single Doppler radar measures only one component duced the tracking algorithm that was used with some of the wind velocity, namely, the component radial to success in JDOP (see section 3). From 1980 to 1982, Unauthenticated | Downloaded 09/27/21 09:54 AM UTC 246 WEATHER AND FORECASTING VOLUME 13 Forsyth continued developing the algorithm after mov- WSR-57 without need for participation by the Depart- ing to NSSL. The improved algorithm was used in the ment of Transportation or DOD (Johannessen and Kes- Boston area Next-Generation Weather Radar (NEX- sler 1976). RAD) demonstration in 1983±84.