The first part of a history of the use of storm surveillance radars by operational military and civil weather services in the United States is presented. The history of radar meteorological research is long and distinguished but already well described. Hence, this paper and its companion focus on the history of operational radar meteorology from its birth in World War II through the introduction of the first two operational Doppler weather radars. This part deals with the pre-Next-Generation Weather Radar era. An aaappendix to this part contains what is known by the authors about the principal technical characteristics of most of the radars discussed in both parts.
This and the companion paper describe the history of the operational use of storm surveillance radars by U.S. weather services. The papers are based on the experience of some of those who, at various times, have participated in or led operational weather radar programs.
The use of radar to observe the weather developed as an outcome of the intensive work on radar technology during World War II. The history of those early developments, and of the research aspects of radar meteorology, is well described in Hitschfeld (1986), Atlas (1990a), and Rogers and Smith (1996). Bigler et al. (1962) and Bigler (1981) summarize the history and status of the weather radar program conducted by what was then called the U.S. Weather Bureau; that material is updated and expanded in this paper. Research conducted by operational weather agencies is discussed here, as are research threads that have found their way into operational use or have been of great benefit to operational radar meteorology. Here we concentrate on the history of application of storm detection radar for operational purposes, such as severe storm identification. Length restrictions prevented addressing the history of the operational use of cloud detection radars, wind profilers, and most other clear-air applications, the exception being the widely used single-Doppler clear-air wind measurement technique. Commercial applications of weather radar could not be covered in the space available; Jorgensen and Gerdes (1951) present a good example.
Initially the security classification attached to radar systems of all kinds limited their use to the military weather services. Later the cost and complexity of these systems limited their operational use to government agencies, principally the military and civil weather services; however, remote displays from weather radar systems became affordable at airline weather offices, commercial weather services, and broadcast weather facilities. In the 1960s, increased availability of lighter-weight, solid-state electronics made it practical to manufacture a storm avoidance radar for use in commercial and eventually private aircraft. Some of these systems were adapted for use on the ground. The capabilities of aircraft weather radars have steadily grown, and they are now widely available. By 1969, a few television stations in the midwestern United States and along the southeast coast had installed radars for use in the weather segments of their news broadcasts; the trend broadened through the 1970s, as ground-based weather radars became more capable and more affordable. From the late 1960s through the present time, demand continued for remote radar weather information, originating from radars not under local control. The sophistication of the remote information provided in response to this demand grew significantly, from the simple, facsimile-based systems of the 1960s to the computer-based techniques used today.1
In the late 1980s and 1990s, responding to the successful development of techniques for employing single-Doppler weather radar observations, the Department of Commerce, Department of Defense (DOD), and Department of Transportation (DOT) jointly fielded two highly sophisticated, ground-based Doppler weather radar systems, the Next-Generation Weather Radar (NEXRAD), now called the WSR-88D, and the DOT’s Terminal Doppler Weather Radar. Remote single-radar and multiradar composite data service is provided by value-added NEXRAD Imagery Dissemination System (NIDS) vendors. Recently, the lower cost of reasonably sophisticated ground-based weather radars, some Doppler capable, has made them affordable by a broader range of commercial weather services and broadcast weather facilities. Some types of weather radar data are available, although not in real time, on the World Wide Web.
The letter designators used to designate electromagnetic frequency ranges originated because of the need for secrecy during World War II. The band designators were finally standardized by the Institute of Electrical and Electronics Engineers in 1984. Those of interest to meteorology (some are used in this paper) are contained in Table 1.
A table in the appendix of this paper contains what is known by the authors about the principal technical characteristics of most of the radars discussed.
2. Early origins
The earliest origins of radar meteorology are difficult to discern because the secrecy surrounding radar in World War II delayed reporting of important findings until 1945 and later. At the outbreak of the war, the maturity of the combatants’ radio-location technology differed among themselves only by about two or three years. The British work was more advanced than the others, largely due to the efforts of Sir Robert A. Watson-Watt. A Scottish physicist and meteorologist, Watson-Watt was a fellow of the Royal Meteorological Society by 1915, published a paper on sferics by 1922, and delivered the Symons Memorial Lecture in 1929 on“Weather and Wireless.” In the first of a number of positions he held in the British government from 1915 to 1952, Watson-Watt developed crude radio-location, direction-finding devices that could locate thunderstorms based on the sferics they emitted. By 1935, as head of the radio department of the National Physical Laboratory, he turned to the problem of radio location of military targets by measuring the distance between the transmitter and those targets. In 1935, he started investigating the use of electromagnetic waves to locate aircraft, work that influenced the design of Britain’s and the world’s first operational radar system, the Chain Home radars. That system was in place before the Battle of Britain and is credited with being one of the most important factors enabling the outnumbered Royal Air Force to turn back the Luftwaffe over the skies of England early in the war.
Beginning in July 1940, a radar of 10-cm wavelength was operated at the General Electric Corporation Research Laboratory in Wembley, England, where Dr. J. W. Ryde worked (Doviak and Zrnić 1993). It is likely that the first weather echo was seen on this radar or another like it in England, probably in late 1940 or possibly as late as February 1941. Perhaps to explain these weather echoes, which might interfere with detection of aircraft, Ryde was asked to investigate the attenuation and backscattering properties of clouds and rain (Probert-Jones 1990). Ryde reported this wartime work in the open literature later (Ryde 1946). Similar studies conducted from 1942 to 1944 at the Massachusetts Institute of Technology’s (MIT) Radiation Laboratory (the Rad Lab, as it was called), largely by Bent (1946), showed that weather could be detected on certain types of radars out to ranges of 150 mi at 3- and 10-cm wavelengths.
During the first half of 1943, Major J. Fletcher of the Army Air Forces Weather Service worked at the Rad Lab and, about a year later, established a program for use of weather radar within the Army Air Forces Weather Service (Wexler and Swingle 1947). Early efforts to use radar on the ground for operational meteorological purposes were of two kinds (Fletcher 1990). On the one hand, operational use was made of radars installed for nonweather missions, such as point and area defense, bombing, navigation, gunnery, and aircraft control and warning. On the other hand, some radars were dedicated to or converted for use principally in operational weather support, such as the radars in weather reconnaissance aircraft and ground-based radars at weather stations.
3. First radar operations at individual stations
Beginning in September 1943, air traffic control and harbor defense radars already installed on the Atlantic and Pacific sides of the Isthmus of Panama were used, on a noninterference basis, for weather surveillance purposes (Best 1973). These were single-station operations. By 1943, scientists at MIT’s Rad Lab, where most radar research and development in the United States was conducted during World War II, had completed a series of visits to many of these radar sites. The visits were performed to determine the effects of the atmosphere on radar propagation (today called radio meteorology) and to assess the usefulness of these radars in observing and to some extent forecasting atmospheric phenomena (today called radar meteorology).
In most of these single-station operations and in the first weather radar networks, an expert trained in both meteorology and radar served as a radar weather officer and had operational, technique development, and research duties. During the course of World War II, weather officers received 15 months of intensive training in meteorology and other subjects (nine months of meteorology training preceded by six months of preparatory studies for those who did not have two years of college mathematics and physics). One hundred graduates from that program were sent to Harvard University for four months of intensive training in electrical engineering and basic radar theory and then to MIT for three months of training on specific radar systems (Atlas 1990b; Fletcher 1990). Some were given special familiarization at the Rad Lab. Many of the early leaders of research in radar meteorology had this same education and training. Until 1947, Air Weather Service (AWS), formerly called the Army Air Forces Weather Service, had an explicit research mission: if in the conduct of operations it was found necessary to advance the state of meteorological knowledge or engineering practice or develop new techniques to apply that knowledge to customers’ weather support problems, the on-site personnel often had the education, training, and ability to do so locally. When a problem exceeded the depth of their understanding or capabilities to solve, these radar weather officers had the background to recognize that and the contacts at universities and laboratories, such as the Rad Lab, to help them solve the problem. In this way, problems could be solved expeditiously and wartime operational needs met quickly. Just as important, radar weather officers’ education and training enabled them to follow and understand research in radar meteorology and implement promising results locally. Under these circumstances, local weather radar programs had a high technical quality even by today’s standards (see, e.g., Air Weather Service 1945). To some extent, the requirement for training in both meteorology and radar has continued from World War II until the present day, although the mix of educational degrees and technical training has varied.
4. First radar networks used for weather surveillance
The first radar network used for weather surveillance was formed in Panama in April 1944, when weather observing and reporting began at two Harbor Defense Cristobal installations, facing the Atlantic Ocean (Best 1973). In May 1944, the network was augmented by using two higher-power aircraft-warning radars on Taboga Island, near Balboa, on the Pacific side and Fort Sherman on the Atlantic. The network took radar weather observations regularly, encoded them in a special radar reporting code, and transmitted them on Teletype communications. These reports were called RAREPs (radar reports), much as present-day observations are. The operations and research activities of the Panama radar network were managed by Lieutenant Myron G. H. “Herb” Ligda, who was assigned in February 1944 as the 6th Weather Region’s Radar Weather Officer. Under study at the time were relationships between echo“intensity” and surface visibility, differences between echoes over land and those over water, effects of the land–sea boundary on the movement of storms, the life cycle of convective storms, the effects of topography on storm movement and intensity, steering of storms by winds aloft, and detection of lightning by radar. An early radar climatology study was completed, showing storm genesis regions and storm motion patterns in Panama. These investigations were conducted to improve the operational value of radar weather information to weather forecasters and their customers.
The second weather radar network (located in India) grew by unifying several operations at individual stations; that network used radars assigned principally to performing a weather surveillance function. In the summer of 1944, the 2nd Weather Reconnaissance Squadron’s B-25 aircraft were modified to carry the “radio set” AN/APQ-13,2 which was actually an X-band radar. The APQ-13 was developed jointly by the Bell Telephone Laboratories and the MIT Rad Lab. It was manufactured in great numbers by Western Electric, the manufacturing arm of the Bell System of the day. It was soon realized that the APQ-13, although originally designed as a bombing and navigation system for the B-17, B-24, B-25, and B-29, could readily detect weather. Thereafter, the APQ-13 was often used for storm detection. In the fall of 1944, the second Weather Reconnaissance Squadron was deployed to a location near Calcutta, India. The aircraft radar worked well for storm detection, but it was difficult to use the heavily tasked weather reconnaissance aircraft and aircrews to stand watch for fast-moving, squall-line-type systems called nor’westers, for which operational warnings were desired. To meet this challenge, some APQ-13s were modified for ground use and installed on towers at weather stations. A single-station operation was tried successfully at Guskara, India. Radars were added at Chabua (June 1945) and Tezgoan (July 1945) near the Burma–India border to form a network covering the Assam Valley (Best 1973). Radar weather observations were taken, recorded, and transmitted so that weather forecasters could synthesize storm data from multiple radars and brief aircrews flying in the “little hump” area of the China–Burma–India theater.
Early use of radars individually and in networks for weather detection led to recognition of many basic features of storm structure and organization and of the value of the information for operational purposes (e.g., Maynard 1945). These wartime successes set the stage for postwar use of radar in meteorology and the accompanying growth of the new science of radar meteorology.
5. Post-war use of World War II radars at weather stations
In 1945, with development of the first true weather radar already under way (see next section), it was decided to test whether the widely available APQ-13, or an alternative, the AN/APS-10, would be more effective as an interim radar until the first weather radars could be deployed. Tests conducted in 1946 favored the APQ-13 (Miller 1947), which was then fielded as the first widely distributed ground-based radar used for storm detection. As is often the case with interim solutions, the APQ-13 continued in operation at many military base and post weather stations until a follow-on weather radar was deployed in the 1960s. At the height of the APQ-13 program, AWS had more than 60 in operation at base and post weather stations worldwide. In October 1977, the last operational APQ-13 was removed from the Fort Sill, Oklahoma, post weather station. It was shipped to the Air Force Museum at Wright-Patterson Air Force Base (AFB), Ohio, which intended to display it in its original configuration as a bombing and navigation radar. The museum indicated that it would describe, in lineage notes, not only the radar’s history as a bombing and navigation system but also its much longer history as an operational weather radar, even though not built for the purpose.
Meanwhile, the Weather Bureau, now called the National Weather Service (NWS), obtained 25 AN/APS-2F aircraft radars from the navy in 1946. These radars had S-band wavelengths, so attenuation by rain was almost entirely avoided (Atlas and Banks 1951); however, detection of light rain and snow was minimal due to system performance limitations. The radars were modified for meteorological use and put into operation at a rate of about five per year. The modifications were performed by the Weather Bureau, which called the modified APS-2F radars WSR-1s, -1As, -3s, and -4s. These early WSR-series systems were all basically the same radar, differing among themselves principally in their indicators and controls. The WSR-1 had a plan position indicator (PPI, a horizontal, maplike display of echoes) and an A-scope (display of echo amplitude vs range or time), along with receiver and indicator elements, arranged vertically in a tall, narrow rack. The WSR-3 and -4 indicators were arranged in a three-panel horizontal console, and both these radars had a PPI, an A-scope, and a range–height indicator (RHI, a display of echoes in vertical cross section). The WSR-1A consisted of the receiver–indicator assembly of the WSR-3 plus the A-scope of the WSR-1, arranged in the vertically oriented rack of the WSR-1. The WSR-4 was essentially a WSR-3 with a traveling wave tube assembly that improved the system sensitivity. Modifications included replacing the small, aircraft-mountable antenna with a larger one and adding a power converter to permit operation on conventional power (V. Rockney 1997, personal communication). The first of these radars was commissioned at Washington, D.C. (Washington National Airport), on 12 March 1947 (National Climatic Data Center 1997, personal communication). On 1 June 1947, the second such radar installed at a Weather Bureau office was commissioned at Wichita, Kansas, in the heart of “tornado alley.” In May 1949, reports from the Wichita radar were used to help guide an aircraft in distress and threatened by surrounding severe weather to an area free of severe thunderstorms, where it could be landed safely. Only three months after its commissioning in August 1947, the WSR-1 at Norfolk, Nebraska, had paid for itself in cost-avoidance actions taken by the Elkhorn Valley power system based on warnings of approaching electrical storms. These radars formed a part of the fledgling U.S. Basic Weather Radar Network (see later section on that topic). In the 1950s, the Weather Bureau added to the number of such radars in service for storm detection purposes (see section 7).
6. Acquisition, deployment, and operational use of the first weather radar
The AN/CPS-9 Storm Detection Radar was the first radar designed, developed, and deployed specifically for use by meteorologists as a weather observing and short-range forecasting tool. Commensurate with the army origins of today’s air force weather organizations, the Army Signal Corps was asked to acquire the CPS-9. In 1943–45, the Signal Corps Engineering Laboratories established a special radar weather section at Evans Signal Laboratory, Belmar, New Jersey. That group designed the CPS-9 to meet requirements developed by the Army Air Forces Weather Service. Some of those requirements placed limits on the size and cost of the system; for example, one requirement was that the radar should be able to travel in a single C-54 aircraft. Studies were conducted to determine the utility of wavelengths from 1 to 10 cm in detecting precipitation and the availability of microwave components at those wavelengths. It was found that, within the desired wavelength range, components were available only in the X band and S band (Air Weather Service 1955). Developing components at other wavelengths would have protracted development of the radar for an additional two to three years, so the choice of wavelength for the CPS-9 was reduced to two. While it was possible to meet the technical requirements for radar resolution using an S-band radar, such a system, if produced, would have exceeded the limits on size, weight, and power. Accordingly, an X-band wavelength was selected for the CPS-9, thus determining the other characteristics of the radar and its cost.
Development models of the CPS-9 were produced by the Raytheon Manufacturing Company, Waltham, Massachusetts, the same company that would later produce the Weather Bureau’s WSR-57 radar and the DOT FAA’s Terminal Doppler Weather Radar. One of the development models was put in place at MIT’s Weather Radar Research Project (Austin and Geotis 1990). Other development copies received engineering testing at the Signal Laboratory and service testing by the Air Proving Ground of the air force. The Signal Laboratory also used its development copy of the CPS-9 to conduct research on operational use of the system from 1950 to 1953. Design of the CPS-9 was refined based on the results of testing (Williams 1953).
The CPS-9, an X-band system, had a 1° beamwidth, a 5-μs pulse duration, and good sensitivity to hydrometeors. It also had a 0.5-μs short-pulse mode for higher resolution of nearby targets. Its receiver was linear, with a limited dynamic range. Distance was measured in statute miles (mi) rather than nautical miles (n mi). The antenna, a little less than 8 ft in diameter, required no radome, and the entire radio frequency (RF) transmitter–receiver package rode on the back of the antenna. These features precluded the need for trouble-prone rotary joints and reduced the waveguide and radome losses. Figure 1 shows the console of a CPS-9 radar.
Production models were fabricated by Raytheon from 1953 to 1954 and installed at military bases worldwide. CPS-9s were also installed at laboratories such as the Air Force Cambridge Research Center [later renamed the Air Force Cambridge Research Laboratories (AFCRL)], the Air Force Geophysics Laboratory (AFGL), and the Phillips Laboratory (PL)], and all weather training facilities and universities. The original plan for employment of the CPS-9 called for a closely spaced network of radars at almost every military weather station in the continental United States and a limited number at military bases overseas. In fact, 56 CPS-9s were produced for all services combined (Williams 1953), and less than 50 went into operational use in the Air Force; APQ-13s had to be kept in operation at facilities that did not receive a CPS-9. The first operational CPS-9 was installed at Maxwell AFB, Alabama, on 20 June 1954; that radar remained operational for 30 yr before finally being replaced on 14 July 1984 by a more modern radar, the AN/FPS-77 (Fuller 1990a).
In 1966 the CPS-9 was modified by addition of the Calibrated Echo Intensity Control (CEICON). This device allowed insertion of known amounts of attenuation into the receiver amplification chain until the apparent radar echo amplitude matched a power reference line on the A-scope. This procedure enabled estimation of the average power backscattered from weather targets and thus the radar reflectivity factor of those targets. The method was adapted from a similar technique developed for Weather Bureau radars by Bigler and Brooks (1963) and based upon concepts first described in Langille and Gunn (1948). The quality of measurements taken using the CEICON was limited by the coarseness of the attenuator steps, receiver saturation, and other problems. In 1970, two AWS CPS-9s were modified by addition of the NWS’s video integrator and processor (VIP) (see section 8). It was too late to include the VIP in the acquisition of AWS’s next storm detection radar, the AN/FPS-77. In 1966, AWS still had 40 CPS-9s in operation. By 1974, the number was reduced to 11. None are in the operational inventory today.
Applied research was directed toward the development of techniques for using the CPS-9 operationally. Technical reports emerging from these efforts were distributed by AWS to CPS-9 sites from 1952 to 1955. An operator’s manual for the CPS-9, written under the direction of Pauline Austin of MIT, was made available in 1955.
The CPS-9 proved particularly adaptable to investigations of the properties of synoptic-scale precipitation and cloud systems and development of early short-range forecasting techniques. It was known during design of the system that attenuation by intervening rainfall would place serious limitations on use of the radar for quantitative measurement of precipitation. It was suspected also that hail, lying outside the Rayleigh scattering region for an X-band wavelength, would cause underestimation of the backscattering cross section of such targets compared to the Rayleigh cross section. The underestimation factor would be much greater for an X-band radar than for an S-band system at typical hailstone diameters. Oddly, these characteristics later proved useful to Donaldson (1961), who analyzed profiles of the radar reflectivity factor with height, with knees in these profiles indicating the potential for hail and tornadoes.
7. Expansion of Weather Bureau radar and warning capabilities
The 1950s brought not only an expansion of military weather radar capabilities, but also a major expansion of the Weather Bureau’s radar systems. On 9 April 1953, a major tornado occurred in central Illinois north of Champaign–Urbana. An Illinois State Water Survey radar was being operated for maintenance and test purposes by its electronics technician, D. Staggs, during passage of the storm about 25 miles to the north. Staggs turned on a 35-mm camera and recorded a remarkable echo with a hooklike appendage extending southward from a strong thunderstorm cell. The storm was reported by Stout and Huff (1953) and later analyzed in detail by Huff et al. (1954). Within two months, two more hook-shaped echoes associated with tornadic storms were observed and photographed, one at Waco, Texas, the other at Worcester, Massachusetts. These three events provided an answer to the oft-asked question,“Would a tornado be identifiable on a radar scope?” The prevailing opinion at the time was no, because the tornado itself is small in horizontal dimension.
These three severe weather events led to the formation of a Texas Tornado Warning Network in which communications between Weather Bureau offices and local public officials were established. Major cities in Texas were approached for funds (some from the private sector and some from the public sector) to modify and install the APS-2F, designated the WSR-1, -1A, -3, or -4, in Weather Bureau offices. The Weather Bureau agreed to operate and maintain the radars and provide warnings to the public when confirmed sightings were made. Volunteer spotter networks were established. In some cases, a spotter would report a tornado before it was identified by radar, and sometimes identification was made from radar alone (Bigler 1956). The Texas Agricultural and Mechanical (A&M) Research Foundation handled the funds, arranged for the radar modifications to be performed in the laboratories of the Texas A&M University Department of Electrical Engineering, and ensured that antenna towers were erected and cables installed in Weather Bureau offices. Formation of the network began at a kickoff meeting held on 24 June 1953 (Kahan 1953);approximately six years were required before the network attained full strength. About 17 radars were modified and installed under this joint effort by local government, state, and federal agencies, and a university. Modifications made to the radars included a new antenna pedestal to support a 6-ft parabolic reflector, a rack-mounted PPI and A-scope, and a fiberglass radome to protect the antenna and allow operation without wind and ice loading.
The modified APS-2F at Texas A&M University, although not formally a part of the Texas Tornado Warning Network, was used at least once for warning purposes (Bigler 1956). On 5 April 1956, a tornado that produced damage in Bryan and College Station, Texas, was detected by the Texas A&M University radar. At noon that day, the Weather Bureau Forecast Center at Kansas City, Missouri, had issued what we would today call a tornado watch for an area just to the north of Bryan. The Texas A&M University radar observed strong, tall, hook-shaped echoes with V-notch signatures after 1400 LT. At 1445 LT, Texas A&M University meteorologists called the Bryan Police Department and forecast that a tornado would touch down 30 min. later. Actual damage started at 1509 LT. Texas A&M University also warned the College Station Consolidated School System, which decided to keep the children in their school buildings instead of releasing them at the scheduled time of 1500 LT. This is probably the first warning based solely on interpretation of radar data and is a good example of effective interaction between warning meteorologists and the local community. Today, with improved warning dissemination methods, increased community preparedness, and better radar capabilities and coverage, it would be less likely that a research team would be issuing warnings to communities directly.
After 1956, the task of modifying the APS-2F radars so they could be fielded as WSR-1s, -1As, -3s, and -4s was transferred to Weather Bureau headquarters, which had to relocate some antennas that had been mounted in locations where they were difficult to maintain. At the height of the program in April 1975, 82 of the WSR-1s, -1As, -3s, and -4s were operational. A few were replaced by the WSR-57, but most continued in service until replaced by the WSR-74C over an extended period from 1976 to 1980. None are in service today.
8. First Weather Bureau weather radars
Hurricanes became a major factor in Weather Bureau planning and budgeting in the mid-1950s. Hurricanes Carol and Edna struck the U.S. Atlantic coast within 11 days of each other in 1954. In 1955, three more hurricanes struck the East Coast, but this time the Weather Bureau was better prepared. Long-range, high-powered (by 1950s’ standards) SP-1M S-band radars were installed on Nantucket Island, Massachusetts; Hatteras, North Carolina; and San Juan, Puerto Rico. The SP-1M radars were navy search systems modified for meteorological use by addition of a traveling wave tube assembly. These radars proved to be a major asset for tracking hurricanes.
The extensive damage caused by hurricane-force winds and heavy flooding in two consecutive years reversed the declining budgetary fortunes of the Weather Bureau in Congress. Moving quickly, the Weather Bureau senior staff developed a major budget proposal for fiscal year 1956 to improve hurricane and tornado warning services. A sympathetic Congress approved the funding increases, and the bureau launched a major effort to improve its warning services. The budget package included funds for design, procurement, installation, and staffing of what would eventually become the Weather Bureau’s flagship radar, the WSR-57. The original budget package provided for acquisition of 31 radars, including one system set aside for electronics technician training.
In order to minimize the effects of attenuation by rainfall, the Weather Bureau specified an S-band wavelength for the WSR-57. Design was complete by 1957. The Weather Bureau selected the Raytheon Manufacturing Company, which earlier had produced the CPS-9, as prime contractor for the WSR-57 and ordered an initial quantity of 31 radars in 1958. The navy ordered eight radars and applied to the radar its military nomenclature, AN/FPS-41 (Rockney 1958). A staff of six (five radar meteorologists and one electronics technician) was funded at each radar station. Since the WSR-57s were installed in existing Weather Bureau offices, in almost all cases one or two electronics technicians were already assigned there. All of the electronics technicians received a complete course of instruction in radar maintenance, so they were able to help each other solve difficult problems and ensure continuous daily coverage even during periods of temporary absence. The radar meteorologists and electronics technicians quickly became a team to ensure high-quality radar operations.
The WSR-57 retained some features of the CPS-9 design, including an off-center PPI and an RF package mounted on the antenna pedestal. The WSR-57 had to use a larger antenna than the CPS-9 in order to achieve a 2° beamwidth at an S-band wavelength. The antenna was enclosed in a fiberglass radome for protection from the weather elements and to permit continuous operation in high wind, freezing rain, and hail. The WSR-57 had a somewhat higher peak transmitted power than the CPS-9 and similar pulse durations. The receiver chain included a choice of linear or logarithmic response and an adjustable step attenuator similar to the CPS-9’s CEICON. The WSR-57’s wider beamwidth and longer wavelength made it less sensitive to hydrometeors than the CPS-9, but this was not considered a serious limitation. The WSR-57’s improved ability to detect storms behind intervening rainfall and to observe hurricanes at great distances were considered more important design objectives. Its main PPI had a reflection plotter that mirrored onto the PPI without parallax any annotations drawn by the operator on a faceplate above the PPI. This feature made it easy to outline the areas and lines constituting a RAREP and measure the associated azimuths and ranges that defined those echo features. The reflection plotter could also be used to track significant individual echoes, areas, lines, and other features as they formed, moved, and dissipated over time. The tracks could be used to obtain movement data needed for transmission in the RAREP and to make short-term forecasts of echo movement for severe storm warning and flash-flood forecasting purposes. Figure 2 shows the console of a WSR-57 radar.
Included in the WSR-57 equipment package was a repeater PPI display, used exclusively for radarscope photography. A 35-mm and a Polaroid camera were permanently mounted over the PPI scope. Included in the photographic field were indicators and text showing station identifier, date and time, attenuator control settings, range marker settings, pulse duration, linear or logarithmic receiver selection, and a frame number. The antenna elevation angle was displayed by a strobe line on the PPI (Rockney 1958). Thousands of feet of film, produced by these cameras for almost four decades, are on file at the National Climatic Data Center (NCDC). The films served as a component of aircraft accident and incident investigations when it was suspected that thunderstorm-associated aviation hazards might have been a contributing factor. The scope photography capabilities built into these early radars led to the multilevel digital data archiving systems of today’s WSR-88D.
An emergency power system was installed at all the WSR-57 and SP-1M stations to ensure uninterrupted operation if commercial power failed. Back-up voice radio communications were also installed at the coastal stations. The Weather Bureau started making fixes in trouble-prone components such as the sensitivity time control (STC) circuits soon after deployment of the radars. In 1963, the Weather Bureau took steps to standardize performance of the WSR-57s. Starting in 1964, sunrise and sunset observations of the sun were used to calibrate the antenna azimuth position at sites lacking suitable, surveyed ground targets. By 1982, the solar procedure was applied also to the WSR-74Cs and -74Ss but limited to times when the sun was more than 10° above the horizon.
The VIP (Shreeve and Erdahl 1968) was added to the WSR-57 in about 1968, and the Digital VIP (DVIP) became available for use at Digital Radar Experiment (D/RADEX) stations in 1974 (Shreeve 1974). A sample of the output of the WSR-57’s logarithmic amplifier was provided to the VIP, which had the ability to average, or integrate, the instantaneous backscattered power returned by distributed targets such as weather echoes. Thus it could obtain in a reproducible and automatic fashion an estimate of the average backscattered power and infer automatically from that and other quantities the radar reflectivity factor Z of weather targets. The VIP enabled the radar operator to display contours of Z directly on the radar’s indicators. The six VIP intervals, which from an engineering point of view are simply intervals of range-normalized average backscattered power, could have been set in terms of convenient increments of the logarithm of Z. However, the VIP also had to provide some compatibility with the existing five-category RAREP echo intensity code, which were defined using rainfall rates R, and also had to meet some specialized hydrometeorological needs. To meet those objectives, the six VIP intervals were defined by subdividing the existing five-category system in terms of R and then associating with each threshold a value of Z using the relationship of Marshall and Palmer (1948). Eventually, meteorologists became so familiar with the VIP levels they began to use them instead of the radar reflectivity factor in conversations and even in scientific papers. Initially, DVIPs were purchased only for D/RADEX and Radar Data Processor (RADAP) II sites; in 1978 DVIPs were bought and installed on all WSR-57 radars.
Some engineering details of the WSR-57’s design left much to be desired. For example, the elevation control was open-loop, in which the antenna followed a command signal but provided no feedback to indicate the true antenna position, making it impossible at the radar console to distinguish between commanded elevation angle and actual elevation angle. The WSR-57 had no tachometer feedback for rate control in the elevation drives. These shortcomings could easily have affected the accuracy of the WSR-57’s observations of radar echo tops. The WSR-57’s pedestal and antenna design were inadequate in a number of ways, including an offset center-of-gravity that caused preloading and additional wear on the bull-gear driving the antenna, and lubrication of the elevation drive that occurred only while the antenna was actually moving. The biggest problem was excess heat in the radome, which decreased the lifetime of receiver–transmitter-modulator components located there. The original linear receiver proved inadequate and was eventually taken out of service at most sites. The original logarithmic receiver was retained for many years but was eventually replaced by a solid-state logarithmic receiver as soon as reliable log receivers became available at an affordable price.
On fair-weather days, the WSR-57 staffs had ample time for analysis of data for local research studies (Hexter 1963). Progress reports were exchanged to cross-feed preliminary results between offices conducting similar studies. Since many of the research organizations of the time used X-band radars, it was important for all to learn of any differences in data interpretation caused by the WSR-57’s S-band wavelength.
The first operational WSR-57 was installed at Miami, Florida, in June 1959. The rest of the first 31 radars were installed during the early 1960s. All had to be located in existing Weather Bureau offices. The primary purpose of the network design was tracking hurricanes as they approached and crossed the Atlantic and Gulf Coasts (14 WSR-57s). The spacing between the radars was approximately 200 n mi. The three SP-1Ms were to remain in place at least temporarily. Cooperative agreements were reached with the Dow Chemical Company of Freeport, Texas, and the Copano Research Foundation of Victoria, Texas, for reporting weather observed by their radars. Subsequent experience demonstrated that the eye of a well-developed hurricane could reliably be observed to a distance of 200 n mi and, on at least one occasion, to 230 n mi away, no matter how much intervening rain was occurring.
Eleven of the WSR-57s were installed in the Midwest to detect severe local storms. Two radars were installed in the mountainous west. These installations were at Sacramento, California, principally for support to state and federal hydrologic efforts in water-thirsty California, and at Point Six Mountain (elevation about 8000 ft) near Missoula, Montana, to support state and federal efforts to fight forest fires by helping to locate areas of high probability of cloud-to-ground lightning strikes. Three WSR-57s were installed inland of the East Coast for tracking heavy rain areas as hurricanes decayed when passing over land.
Funding for 14 additional WSR-57 radars was obtained in 1966 and 1967 to expand the network east of the Rocky Mountains, staffed as the earlier stations were. Some of these additional sites were in locations where no Weather Bureau office already existed, establishing a concept of network design based on optimum spacing rather than availability of an existing office. For example, a site was established 60 mi east of Denver, in Limon, Colorado. This site ensured network continuity with the WSR-57 radars at Garden City, Kansas, and Grand Island, Nebraska. In addition, the Limon site was far enough east to avoid major ground clutter and occultation problems caused by the mountains southwest through northwest of Denver.
For use in Southeast Asia, the Air Force later ordered three FPS-41s, which were not returned to the United States (see section 13). The Weather Bureau agreed to operate or otherwise accept responsibility for the eight FPS-41s bought by the navy. At the height of the program, NWS owned or operated 53 WSR-57/FPS-41 radars, of which two were used for electronics technician training and the rest used operationally as primary stations on the Basic Weather Radar Network.
An important element of the new network design was near real-time telephone-line transmission of PPI data and handwritten alphanumeric annotations to nearby offices. The equipment was designated Radar to Telephone Transmission System (RATTS-65) and later as Weather Bureau Radar Remote (WBRR) (Hilton and Hoag 1966). Later, dial-in capability was added to permit access by a wide range of users, including military stations, airline offices, and television stations (Bigler 1969). Those early efforts at providing radar data remotely led to the dial-in or nonassociated principal user processor (PUP) and NIDS capabilities of today’s WSR-88D.
With improvements made over the years, despite its weaknesses, the WSR-57 remained NWS’s flagship radar until deployment of the WSR-88D in the 1990s. The last operational WSR-57 was removed from service at Charleston, South Carolina, on 2 December 1996.
9. Follow-on Weather Bureau radars
The 1960s were a period of great change in the electronics industry. Led by the space program, miniaturization of components made possible by the transistor caused a dramatic move away from the vacuum tube technology of the early years of radio, television, and the radar technology of the 1940s. Within the Weather Bureau, the need for replacement of the 1940s vintage WSR-1, -1A, -3, and -4 radars was becoming urgent. Spare parts left over from wartime stockpiles were vanishing. Manufacture of components using the now-outdated vacuum-tube technology was diminishing as companies changed their designs and production lines to remain competitive by using the new technology.
The 1967 Federal Plan for Weather Radars and Remote Displays, written by the Office of the Federal Coordinator for Meteorological Services and Supporting Research, identified a continuing need for weather radars of three basic types: synoptic weather radars, local-use or local-warning radars, and remote displays. The 1969 edition of that same plan indicated the Weather Bureau intended to buy modern local-warning radars to replace the aging WSR-1s, -1As, -3s, and -4s.
A small group of engineers from the Sperry Rand Company and Vitro, Inc., formed a company of their own, Enterprise Electronics Corporation, to design and manufacture a new generation of C-band weather radars. By 1969, they delivered their first production model to a television station in Tampa, Florida. A year later, a second unit was installed in a station in Jackson, Mississippi. In fiscal year 1976, NWS received funding over 3 fiscal years to replace 82 aging local-warning radars with 66 modern, C-band local-warning radars manufactured by Enterprise Electronics; within NWS, the radar was called the WSR-74C. An additional two systems were purchased, one for use at the NWS Training Center (NWSTC) and the other for use at NWS headquarters, bringing the total buy to 68. The WSR-74Cs were purchased with an integral DVIP. The radar had a coaxial magnetron and a completely enclosed, oil-bath modulator, and it made maximum use of mid-1970s solid-state electronics. It used a lightweight pedestal because the antenna was shielded by a radome and had permanently lubricated antenna gears. NWS made arrangements for cooperative use of one forerunner of the WSR-74C, called the WR100-5, also built by Enterprise. Five of the more capable S-band WSR-74S radars (see following paragraph) were actually used as local-warning radars, bringing the total number of NWS local-warning radars to 71 in 1982.
In 1976, the Basic Weather Radar Network still had a few state-size gaps and a slightly larger number of smaller-size gaps. At that time, the WSR-57 production line was closed, and the WSR-57’s old, vacuum-tube design was no longer modern enough to justify buying more of them. At the time, Enterprise Electronics offered an S-band weather radar similar in many respects to the WSR-74C. To close the five remaining gaps in the network and to replace seven WSR-57s that failed between 1981 and 1985, NWS needed 12 “synoptic” radars that by network standards should be S-band systems. NWS also needed four more local-warning radars with special requirements such as hurricane and heavy precipitation detection, which indicated S-band radars (P. Hexter 1997, personal communication). To meet these needs, NWS decided to buy 16 operational S-band radars (none for the NWSTC) as the last purchase before NEXRAD. Enterprise Electronics was selected as the source of the radar known as the WSR-74S. The WSR-74Ss were purchased with an integral DVIP.
From the operator’s point of view, the WSR-74C and WSR-74S were quite similar. They had a single PPI that could be equipped either with a reflection plotter or an illuminated map overlay. Their RHI was earth curvature–corrected like that of its predecessor, the AN/FPS-77. The RHI was almost as large as the PPI, making it easy to use operationally. These radars had an A-scope and digital readouts of range, elevation, azimuth, and time. Thirteen WSR-74Cs still have not been decommissioned and, of those, eight remain in active use today. No WSR-74Ss are in the NWS inventory today, having been replaced by the WSR-88D. Some of these radars are in commercial use.
10. Acquisition, deployment, and employment of AN/FPS-77 storm-detection radar and interim replacement
In the early 1960s, the air force recognized the need for a new weather radar system to replace the APQ-13 (Paulsen and Petrocchi 1966), by then having exceeded its expected operational life. By 1966, AWS had decided to replace not only the APQ-13s but also the aging CPS-9s, which were increasingly difficult and expensive to maintain. At that time, air force weather units’ requirements for weather radars vastly exceeded the number available, and AWS allocated radars using a formula that included mission and weather.
C-band components, which had become much more prevalent by the time it was necessary to select the new radar’s wavelength, were a candidate. Experience with the CPS-9 had shown that for most applications, the radar wavelength should be longer than X band. Weather Bureau experience with the WSR-57 pointed toward use of the S band. S-band antenna and transmitting system components, being much larger than equivalent X- and C-band components, were considered too expensive for the advantage of less attenuation of radar energy by precipitation. For that reason, a compromise, the C band, was selected for the new radar, one of whose program objectives was to be cheaper than the CPS-9.
The air force procured 103 AN/FPS-77(V) Storm Detection Radars, with a C-band wavelength, from Lear-Siegler Electronics. These systems were purchased in two increments from 1964 to 1966. Test of the first production system was accomplished at Griffiss AFB, New York, in November 1964. Full-scale production began in 1965. The first operational FPS-77 was installed in March 1966. Almost all the FPS-77s had been installed by 1969. One was destroyed, found rusting on a loading dock at the long-abandoned Suffolk County AFB on Long Island, New York. Of the remaining 102 systems, 78 were in operation at base weather stations and test ranges by 1977; nine more were programmed for installation at similar operational facilities; nine were in use as mockups at maintenance shops, and six were located at weather training and weather equipment maintenance training facilities. At the height of the program, after installation of the contingency reserve, 87 would eventually be installed at operational base weather stations, the rest of the 102 radars going to maintenance facilities and technical training schools. Two FPS-77 radars remain in operational use today at Royal Air Force (RAF) Base Mildenhall, England, and Katterbach Air Base (AB), Germany.
Because the number of requirements far exceeded the number of available radars, AWS was considering the use of radar remoting options variously called RATTS and WBRR. These remoting systems had the capability of making available in weather stations an annotated copy of the PPI at nearby radar facilities. The plan included using not only NWS WSR-57s as the source radar, but also some AWS CPS-9s and FPS-77s. The remote radar thrust was driven by the economy of providing data remotely versus installing a local radar. Shortfalls in the operational capabilities of the WBRR in meeting the demanding needs of weather units providing operational weather support to customers, technical inadequacies of the WBRR itself, and concern over how long FPS-77 antenna pedestals would last if used continuously in the mode needed to update the WBRR, caused the airforce to terminate its WBRR program. The option was kept open for individual base weather stations to establish their own remote connections to NWS WSR-57 radars. Shortfalls in the RATTS and WBRR systems created a niche market that firms such as Kavouras, Inc., Alden Electronics, Inc., WSI, and others eventually filled by selling services that provided timely access to attractive, color, remote radar displays at an affordable price (E. Dash 1996, personal communication). The wide availability of radar data on television weathercasts created a demand among the operational customers of military weather services for the same sort of service. In the years between fielding the FPS-77 and NEXRAD, many military weather stations, National Guard and Reserve facilities, and weather centers acquired remote radar displays linked to NWS radars. Remote access to radar weather data was incorporated into the design of NEXRAD by making the data available to value-added resellers under the NIDS.
The FPS-77’s C-band wavelength allowed attaining a 1.6° beamwidth with an antenna diameter of 8 ft, much smaller than that required by the WSR-57. It had a low-cost antenna design with loosely joined mechanical elevation drive linkages that introduced considerable uncertainty in measurement of radar echo heights. Its peak transmitted power was comparable to the CPS-9’s. The FPS-77 had only one pulse duration, a compromise about midway between the long pulse and short pulse of other weather radar systems. The FPS-77’s PPI used an interesting but not totally successful dark-trace storage tube that could be operated in lighted environments. The FPS-77’s console is shown in Fig. 3.
Operation and maintenance of the FPS-77 were troubled by shortcomings in the radar’s capabilities and design, lack of an operator’s manual, no initial operator training at installation time, lack of follow-on training, and limitations in maintenance test equipment and procedures. Although billed as a radar capable of taking quantitative measurements, the FPS-77 had significant shortcomings in the antenna design, logarithmic amplifier, iso-echo, and STC circuits. The radar had not been equipped with a signal integrator such as the VIP, making it necessary for operators to make estimates of the average backscattered power manually using the gain reduction technique to measure the radar reflectivity factor of weather echoes.
In the late 1960s, the AWS leadership noted a lack of readiness in operational use of radar. Initially, it was thought that simply providing an operator’s manual for the FPS-77 would do the job. An operator’s manual was published as AWS’ Part C of Federal Meteorological Handbook 7, the Weather Radar Manual, by 1973. AWS units with weather radar had appointed radar coordinators and increased their emphasis on radar training and certification. Nevertheless, the Weather Radar Manual proved difficult to implement. When improved radar reflectivity factor measurement procedures were put into operational use, technical difficulties were identified. By this time it was clear that an electrical engineer who was at the same time a radar meteorologist would be needed to address the most serious of these problems. One of the authors (PLS) took a year’s sabbatical and served as AWS Chief Scientist from summer 1974 to summer 1975. Together with a support team including two other authors (RCW and ACH), he visited a number of FPS-77 facilities at technical training schools, maintenance facilities, and operational sites, taking fundamental measurements at these radars. Analysis of the data showed lack of fidelity in the logarithmic amplifier, STC, and iso-echo circuitry. A separate set of problems was discovered in the antenna positioning system, as shown by angular measures made using a gunner’s quadrant and solar boresighting. No records could be found indicating that any FPS-77’s effective antenna system gain had been measured, so standard gain horn measurements were taken at every site that had the capability to radiate. Many of the calibration techniques described in Smith (1968, 1974) were applied to FPS-77 radars. This intensive period of interest in calibration of weather radars included the American Meteorological Society’s Weather Radar Calibration Workshop, hosted by one of the authors (KEW) at the National Severe Storms Laboratory (NSSL) in 1974. During the workshop, the views of all calibration experts could be considered and practical experiments conducted using the NSSL radars and a leased, transportable radar. A technique for solar boresighting, suggested by discussions at the workshop, was applied to the FPS-77 to calibrate antenna, beam- and display-positioning, and effective antenna system gain (Whiton et al. 1976); the technique was later improved and extended to multiparameter radars like the National Center for Atmospheric Research CP-2 Doppler system by Frush (1984). More recently, Rinehart (1991) provided an in-depth treatment of solar position as a function of time.
By the summer of 1975, a priority-ordered list of operational and maintenance actions was developed, including continuation of the calibration visits to radar sites. Despite the transfer of responsibility for maintenance of weather equipment to the Air Force Communications Command, replacement of the logarithmic amplifier with a reliable, solid-state version was performed in 1979. A further outcome of this work was the formation of an AWS radar calibration team, which visited radar sites and made improvements in the FPS-77’s meteorological measurement capability. The radar calibration visits continued until most of the FPS-77s were replaced by NEXRAD, whose technical requirements emphasized calibration both by self-test and externally when necessary and by a calibration visit team whose services can be requested when needed.
Over time, improvements were seen. Lessons learned from this undertaking are that radar, no less than a barometer, is a quantitative instrument by which critical atmospheric variables are measured; all radars should be equipped at least with a signal integrator such as the VIP or equivalent (Doppler radars require more signal processing capability); and solid, practical, and sustainable calibration technology and techniques must be a part of any radar purchased or developed. A final lesson, not necessarily learned as well, is that the best calibration ideas and most effective maintenance come about when the operators and maintainers work as a team in a collaborative environment.
To replace the remaining AN/FPS-103s (see section 14), to replace some FPS-77s after that radar was declared unsupportable and was being phased out of the inventory, and to meet a few additional weather radar requirements at a time when the NEXRAD procurement was slipped, AWS established the Operational Radar Replacement (ORR) program. Approved by the air staff in 1984, the ORR program procured a “gap-filler” radar, the AN/FPQ-21, from Enterprise Electronics; the contract award occurred in early 1986. Twenty-four systems were procured, of which five are still in operation today. The FPQ-21 was similar but not identical to the NWS’s WSR-74C. All the systems procured by the air force had an antenna reflector 12 ft in diameter, giving a beamwidth of 1.1°. All were equipped with an integral DVIP and had color displays. Only commercial manuals, not military-type technical orders and documentation, were procured with this system. No training was procured with the systems. The first FPQ-21 was installed at Fort Sill, Oklahoma, in February 1986.
11. U.S. Basic Weather Radar Network
In 1946, the Weather Bureau established the U.S. Basic Weather Radar Network, although it was not called that at the time. Initially, the network consisted of early WSR-series systems, and a few air force, civil government, and cooperative radars. The network grew slowly in the 1950s and remained an amalgam of heterogeneous systems (Rockney and Jay 1953). The air force agreed to add CPS-9 radars to the network as they became available. In 1956, the Weather Bureau established the Radar Analysis and Development Unit (RADU) as a part of the forecast center in Kansas City, Missouri. The RADU was formed to resolve the problems associated with using a radar network consisting of such a diverse array of radars and reporting practices, to prepare and transmit a Teletype summary of the radar echo distribution over the United States, and eventually to prepare and transmit a graphical radar summary chart from all the radar weather data collected. Beginning in 1959, WSR-57s began to replace WSR-1s, -1As, -3s, and -4s and AWS stations on the network. A few joint-use sites were established at Air Defense Command (ADC) radar sites that filled gaps in network coverage (Foster 1957;Rockney 1960; Bigler 1961). Beginning in 1966, 22 FAA air traffic control radars were, in effect, added to the network. Data from these radars were collected by NWS meteorologists serving at four western U.S. air route traffic control centers (ARTCC) (see section 15). Many primary stations on the network had alternates to provide backup in case of outages; many AWS radars had alternate network responsibilities in the 1960s and 1970s.
12. Radar reporting code and radar summary chart
The RAREP code, used to transmit radar weather observations, originated in World War II in a form not unlike the later Manually Digitized Radar (MDR) code, and evolved to the SD or “azran” (azimuth–range) code in the 1960s and 1970s. These early “plain-language” codes were used by the RADU in Kansas City, Missouri, to prepare manual analyses of radar weather data in order to produce and transmit hourly radar summary bulletins on Teletype. In 1960, the RADU also began transmitting a manual radar summary chart, also called a radar composite chart, at three-hourly intervals (Bigler 1961). The utility of the radar summary charts was soon recognized (e.g., Wilson and Kessler 1963), and suggested improvements were gradually implemented. In the 1970s, NWS began making use of the MDR code, in which numerical echo intensity data were reported at each position in a grid system; the MDR data were more easily assimilated into computerized applications of radar weather data (Moore et al. 1974).
In 1976, in order to facilitate production of an automated radar summary (ARS) chart, a new digital radar code was implemented as a separate section of the RAREP, supplementing the human-readable azran data. Production of the ARS began one month later. In 1978, the ARS production system, including the new, numerical RAREP code, allowed NWS to cancel their MDR program. The frequency of the ARS increased to hourly, compared to the three-hourly products available when the chart was being produced manually. Timeliness of the chart was also improved.
The RAREP code, in whatever form, required extensive manual work by the observer. It made no sense to continue these manual methods with the advent of the sophisticated, partially automated NEXRAD or WSR-88D. NEXRAD’s automatically generated Radar Coded Message (RCM) was in part devised as a replacement for the RAREP. Editing of the RCM, originally contemplated as a way to disseminate key radar data, including severe storms information, was never implemented because of the manpower requirements involved in the editing process. An unedited form of the RCM is transmitted over WSR-88D communications channels and is used as the basis for preparation of the ARS graphic. That product is still disseminated to the Automation of Field Operations and Services system and transmitted on the facsimile circuits. Software running at the Aviation Weather Center in Kansas City, Missouri, prepares the ARS from RCM data, removing as much ground clutter and anomalous propagation as possible. Radar echo tops, which were included in the former RAREP code but are not found in the RCM, can be obtained by using the WSR-88D PUP or the products of the NIDS vendors (P. Hexter 1997, personal communication).
13. Wartime use of fixed weather radars
During the Vietnam War, CPS-9 radars were relocated to Tan Son Nhut AB, Republic of Vietnam (RVN), and Nakhon Phanom AB, Thailand. Three WSR-57M Storm Detection Radars, almost identical to the WSR-57 radars then in active use by the Weather Bureau, were purchased from Raytheon, retitled AN/FPS-41s, and installed at Udorn Royal Thai Air Force Base (RTAFB) and Ubon RTAFB, Thailand, and Pleiku AB, RVN. For the most part, these radars were used conventionally, for observing, analysis, and forecasting at individual weather stations, as they did not form a very good network. The FPS-41s at Udorn and Ubon and the CPS-9 at Nakhon Phanom were also used to collect half-hourly radar scope photography to assess the effectiveness of cloud seeding operations conducted from 1967 to 1972 by the First Weather Group along the Ho Chi Minh Trail in Laos (U.S. Congress 1974; Fuller 1990b). Later the FPS-41 at Pleiku was relocated to Nakhon Phanom, where it was also used to collect scope photography. These operations were intensive but odd by the standards of today’s computer-oriented weather operations. Photographs were collected using a repeater scope and photography equipment. The resulting film packs were then rushed to military photographic processing facilities at the bases, developed, and the transparencies flown nightly to Tan Son Nhut by courier aircraft. Then, in an all-night operation, weather forecasters in a classified facility projected the photographs, analyzed them by densitometer, and attempted to correlate changes in the appearance of photographed precipitating storms with the known time and location of cloud seeding activities, which had been sent to Tan Son Nhut by classified message. While records were kept and attempts were made, using radar and other data, to quantify the effects of cloud seeding on rainfall and enemy activities, the cloud seeding efforts were designed as a weather operation, not a statistical experiment capable of proving the effects of weather modification. The United States is now a signatory to United Nations conventions prohibiting environmental war, and operations of this sort are no longer conducted.
Many operational customers had emerged for the rainfall data, including limited interest by the 7th Air Force Intelligence and strong interest by army terrain analysts at the Combined Intelligence Center Vietnam. Beginning in 1970, the First Weather Group, which managed air force weather support in Southeast Asia, started collecting rainfall data from the radars quantitatively as well as by photograph. Hourly radar scope tracings were taken over a 10-n mi × 10-n mi grid system at participating radars, and the radar reflectivity factor was estimated manually using a power reference line comparison technique. Echoes were labeled according to their intensity, summed over 6- and 24-h intervals and the totals relayed to the analysis center by Teletype bulletin. By 1971, VIP-contoured scope images were traced and the VIP levels used to determine the radar reflectivity factor and estimated rainfall rate. The data were keypunched and analyzed by computer. Customers received this new capability so enthusiastically that the 1st Weather Group rated the effort as the single most successful in 1970 from a customer satisfaction point of view. Interest in the effects of precipitation on war remains strong today. As late as the Gulf War, the army terrain analysis team at U.S. Central Command’s Army Component Command Headquarters said they needed information about areas where there was a significant potential for flash flooding (Air Force 1993).
14. Tactical weather radars
Experience has shown that many military operations are radar permissive, in the sense that radars can be used without significant threat that they will become a target attracting enemy attack. In such situations, particularly when the wartime weather conditions are expected to include significant precipitation, a weather radar might very well be useful. This was certainly true in southeast Asia, where the demand for weather data that radars could provide far outstripped the operational weather personnel’s ability to provide useful products. For these practical reasons, air force weather personnel have always had either a true tactical weather radar, relatively lightweight and configured to facilitate transportability, or an equivalent like the APQ-13 in World War II.
In early 1968, the 7th Air Force, Air Component of the Military Assistance Command Vietnam, purchased from Bendix (now Allied Signal) 16 WTR-1s, similar to the commercial RDR-1 sold by Bendix mostly for use in aircraft. The WTR-1, an X-band radar, included a receiver/transmitter, antenna, radome, and operator’s console (Allied Signal 1978). Of the 16 WTR-1s, all but one were initially installed in Vietnam. The nomenclature center applied the designation AN/FPS-103 Tactical Weather Radar to all WTR-1 models. The FPS-103 was a nonquantitative radar with an echo intensity contouring capability where the contours did not represent known thresholds of the radar reflectivity factor. A small radar that could be readily moved and mounted on primitive towers or the tops of buildings, the FPS-103 met wartime needs for flexibility and ease of maintenance. After 1971, FPS-103s were shipped to other locations in the Pacific and the United States or turned over to the Republic of Vietnam Air Force. None are in the operational inventory today.
In 1974, the Air Force’s Electronic Systems Division (ESD) responded to a Tactical Air Command required operational capability for a tactical weather radar. ESD asked the Naval Air Systems Command’s Naval Avionics Center Indianapolis (NACI) to adapt a radar it was designing for the marine corps using the DOD Standard Electronic Module for use as an air force tactical weather radar. The marines’ version of this radar did not meet all the air force’s needs; among other shortcomings, the marines’ version could not measure the radar reflectivity factor. Two of the present authors (PLS and ACH) spent many hours informing the NACI team about characteristics of a weather radar that make it effective for quantitative measurements. The air force ordered six systems. The program encountered delays while NACI attempted to meet all the air force needs. Problems surfaced during acceptance testing, and a second series of tests was scheduled. Major R. Snell, HQ AWS, spent three months between the two acceptance tests helping the engineers at NACI correct deficiencies in the system. The corrections required a major system redesign. It was only through Snell’s efforts that the radar was finally able to pass the second acceptance test. No shelters for these radars had actually been accepted due to contractual problems with the manufacturer, so the radars were placed in bonded storage at NACI. Further delays were encountered in manufacture and integration of the radar’s shelter. In 1977, the system was designated as the AN/TPS-68 Tactical Weather Radar. In late 1982, the air force accepted delivery of the first six systems, with acceptance having been delayed a full three years, mostly due to the shelter problems.
From the user’s perspective, the TPS-68 was like the FPS-77, having many of the same technical characteristics (such as a C-band wavelength), features, and controls. The TPS-68’s electronics were solid state, more modern than those of the FPS-77, and some limited use was made of digital electronics and light-emitting diode displays. The radar had a conventional, light-trace PPI with center blanking and a conventional RHI. It used an oscilloscope as an A-scope. The 6-ft antenna was stored in the shelter during transport and relocated to the roof of the shelter for operation. No operator’s manual or training was purchased with the TPS-68. In 1987, after a KC-135 crash destroyed the FPS-77 tower at Fairchild AFB, Washington, a TPS-68 was used there successfully for a year until a new tower could be installed (D. Michalewicz 1996, personal communication). The TPS-68 was also used successfully at Diego Garcia, an island in the Indian Ocean, in Operations Desert Shield and Desert Storm. Thunderstorms at Diego were interfering with takeoff of B-52s prepositioned there for employment in the Gulf War. The commander of the air force bomber unit at Diego credited skillful use of the TPS-68 by assigned weather personnel with much of the success of the first night’s bombing in Desert Storm, saying he would not have met his bombing schedule without using the radar to pick out usable “holes” in the thunderstorm activity in which to generate, launch, and recover the B-52s. The timing was intricate because the unit had to fuel the aircraft, load ordnance, launch, and recover B-52s returning from the first wave, all in the short time available between thunderstorms. Of the six systems procured, five remain in the inventory today; however, four are used to provide parts for the one operational system, installed at Taif, Saudi Arabia (R. Kandler 1996, personal communication).
The TPS-68 has been declared unsupportable from a maintenance point of view. The number of requirements for tactical weather radars now far exceeds the number of available systems, and the air force is considering buying a replacement tactical weather radar from commercial off-the-shelf sources. Acquisition of the replacement system is a lengthy process, so four Special Operations Forces Tactical Weather Radars (Ellason 400-P systems) have been purchased, and four Interim Tactical Weather Radars (Kavouras 2070-C equivalents) are under contract. The navy is also considering purchase of a limited number of tactical weather radars for navy and marine corps use; the requirements for the navy/marines tactical weather radar differ from those of the air force system.
15. Use of air defense and air traffic control radars for weather detection
The high cost of installing, operating, and maintaining radars in mountainous terrain and in remote locations caused the Weather Bureau to shy away from installations there and seek cooperative programs with other agencies. It had already been demonstrated that ADC radars could be effective at detecting precipitation (Ligda 1957; Bigler 1957). A pilot effort to employ such observations was set up at an ADC long-range radar site in Tennessee in 1960. In 1961, a similar effort was established at a site in West Virginia. In both cases, a staff of five was assigned for round-the-clock preparation of RAREPs and other data. These programs were phased out by 1968 but were cost-effective interim solutions to filling gaps in weather radar coverage in the eastern United States. Summer thunderstorms cause hundreds of wildfires every year in the central region of Alaska. From 1968 to 1975, NWS personnel were assigned to Alaskan ADC radar sites during the summer to report, using the standard RAREP code, the locations of weather echoes and associated increased likelihood of lightning to firefighting agencies. Thunderstorms were regularly observed from the Arctic Circle south to the Alaska range, with echo tops sometimes approaching 40 000 ft. As lightning detection networks grew, the need for this specialized service diminished.
Wilk et al. (1965) showed that the long-range, long-wavelength ARSR-1D radars used by the FAA for air traffic control were approximately equal to the WSR-57 in precipitation detection capability as long as the moving target indicator (MTI) and circular polarization (CP) features of these radars were turned off; however, operational problems caused by equipment designed and primarily used for aircraft detection caused some serious compromises (Benner and Smith 1970). Air traffic controllers regularly used several features of their equipment to suppress echoes that might mask aircraft targets. The first of these features, MTI, reduced ground clutter by eliminating stationary targets but also reduced the strength of weather echoes. The second feature, CP, partially canceled weather echoes. The third feature used to suppress masking echoes was STC, which reduced the signal strength of targets depending on their range from the radar; in operation, the effect of STC was to virtually eliminate all precipitation echoes within 30 n mi of the radar. To collect useful weather data, these features would have to be disabled, at least temporarily.
After a successful test in 1965, Weather Bureau personnel, beginning operationally in 1966, were assigned to FAA ARTCCs located near Salt Lake City, Utah; Los Angeles, California; Albuquerque, New Mexico; and Seattle, Washington. Data from four to seven long-range radars were available in these centers, for a total of 22 radars, providing virtually complete coverage of the western United States and offshore. Data from Weather Bureau radars located at Sacramento, California; Missoula, Montana; Catalina Island, California (for a few years); and Medford, Oregon (after 1971) were combined with data from the ARTCCs and disseminated in a very different way from that used east of the Rocky Mountains. With four to seven overlapping radars available in each center, outlines of individual echoes were drawn with the greatest detail possible on acetate overlays especially prepared for each radar’s surveillance area. The overlays were then composited to produce one chart per center. Annotations describing echo motion, pertinent data from pilot reports obtained by the controllers, and data from conventional hourly observations were added to the charts before transmission. The composite chart from each center was then transmitted by facsimile to the Salt Lake ARTCC, where all the western U.S. data were assembled into a single chart for distribution, once again by facsimile. Eventually the western U.S. radar data were added to the radar summary chart prepared for dissemination over the national weather facsimile systems. The data proved to be extremely useful to forecasters throughout the data-sparse western United States.
When the joint-use program began in the western United States, switching polarization between linear polarization (LP) and CP was a problem because controllers did not want to work around the clutter, and changing polarization had to be done at each of the individual radar sites. As the controllers and radar meteorologists gained experience in working together, the radars were operated more and more in the LP mode. The controllers learned that most weather echoes did not cause them serious control problems. Frequently it was desirable to display the precipitation echoes, since aircraft approaching or finding themselves in precipitation could request and receive a heading out of the weather. This example of effective interagency cooperation continued into the 1990s, until NEXRAD radars were installed in the west.
16. Advances in operational radar data processing and digital applications
The treatment of weather echoes in a radar system can be divided conveniently into two parts, signal processing and data processing. The demarcation between the two is not always distinct, either conceptually or in terms of the implementation. In broad terms signal processing refers to the manipulation of echoes from multiple transmitted pulses and multiple spatial locations (usually range bins) to derive basic quantities such as the radar reflectivity factor or mean Doppler velocity for each range bin. This signal processing can be carried out in a stand-alone, special-purpose unit such as the VIP or its digital equivalent, or the processing can be carried out in a general-purpose computer. Data processing, which follows signal processing in the data stream, refers to any further processing of these “base data” (to use a NEXRAD term) needed to perform functions such as determining rainfall rates, applying algorithms to infer storm characteristics, preparing products, and displaying data or products.
Weather radar equations are the underpinning for quantitative use of radar echo intensity data and are especially important in automated processing of radar echoes to obtain meteorological quantities such as the radar reflectivity factor and rainfall rate. Early investigations refer to underestimation by several decibels of the average backscattered power by the equation of the time (Marshall et al. 1955). Various investigators gradually whittled the discrepancy down. Probert-Jones (1962) found that part of the remaining difference could be attributed to an overly simplified model of the radar beam as having a uniform distribution of power between the half-power points. Probert-Jones’s rederivation of the radar equation, using a Gaussian beam model, explained most of the remaining discrepancy. Some systematic error sources may remain, such as the finite bandwidth error of Smith and Nathanson (1972), undetermined losses within the radar system, and propagation losses through the atmosphere. The matter is not fully resolved, and further improvements in the weather radar equations are still a possibility. Except for large errors caused by violations of the underlying assumptions made in deriving the weather radar equations, the errors are small enough that they are difficult to detect relative to the measurement uncertainty of the radar systems. The larger errors resulting from violations of the assumptions (e.g., the beam not always being filled uniformly with homogeneous precipitation) have serious limiting consequences on the usefulness of quantitative radar weather data beyond a range of about 200 km. Today, almost every radar meteorologist has at his or her disposal data from a high-resolution, multispectral meteorological satellite system, the Geostationary Operational Environmental Satellite-Next Generation series of satellites. It is possible to fuse the information from radar and satellite sources in such a way as to extend the usefulness of quantitative radar weather data out to ranges at which the beam totally overshoots the weather. The spacing of the present-day weather radar network also helps limit the consequences of violations of the assumptions of the weather radar equation at any one radar, as well as to help solve the cone-of-silence problem, in which data are not available directly above the radar antenna due to antenna motion limits or volume scan restrictions.
The discovery that raindrop-size distributions often followed an analytically convenient negative exponential form with an associated power–law relationship between the radar reflectivity factor Z and the rainfall rate R (Marshall and Palmer 1948) was initially greeted with some skepticism, triggering additional research. In the 1950s and 1960s, the U.S. Army and others commissioned considerable research on the relationship between Z and R and the associated drop-size distributions. The large body of such Z–R relations (see, e.g., Battan 1973) proved fundamentally useful to operational radar meteorologists in measuring precipitation using radar, although the range of the variation can be perplexing. An unknown and perhaps substantial part of the variability in the reported relationships can be ascribed to sampling limitations (Smith et al. 1993).
Off-line postprocessing and analysis of radar data, digitized either manually or electronically (e.g., Kessler and Russo 1963; Wilk et al. 1967; Wilk and Kessler 1970), provided strong indications that computer processing of the data could facilitate operational use of the radar information. Research applications of small, digital computers, attached directly to weather radars for data processing purposes, began in the 1960s. At first, special-purpose computers designed for the application were used (e.g., Atlas et al. 1963). The more versatile, general-purpose minicomputers that appeared on the market in the mid-1960s soon proved superior for this application (e.g., Smith and Boardman 1968). By the early 1970s, it was apparent to most that a weather radar without attached computer data processing and applications is, almost by definition, an underutilized system. Many uses of weather radar, especially hydrometeorological applications, severe weather detection, and storm movement forecasting, are greatly facilitated by an attached processor and applications software. By the 1970s, all research radars had been connected to computer systems. It was apparent that, with some investment in computer hardware and software, operational radars could provide much greater benefit to customers.
NWS conducted D/RADEX (McGrew 1972) from 1971 to 1976 to investigate radar digitizing hardware and applications software designed to exploit radar’s potential usefulness in meteorology and hydrology. Five network WSR-57s were equipped with the D/RADEX facilities, including minicomputers, storage devices, and displays. D/RADEX gave rise to the vertically integrated liquid water (VIL) technique, a useful thunderstorm intensity and flash-flood indicator (Greene and Clark 1972). VIL experienced increased use with the wider availability of computer data processing systems. D/RADEX also showed the operational feasibility of processing in near-real time the large amounts of data that are generated by using antenna tilt sequences to create a three-dimensional volume scan of the atmosphere. D/RADEX produced, from the volume scan data, radar echo tops maps, real-time accumulated precipitation, and a flash-flood monitor. The system also provided an implementation of Lemon’s weak echo region technique, an early severe storm identification algorithm (Saffle 1976), and data retrieval by off-site users using digital modems instead of remote scopes and slow-scan television-based remoting systems. Each D/RADEX system sent accumulated precipitation totals to the relevant River Forecast Center (RFC) every 3 h.
From 1976 to 1980, NWS pursued acquisition of a new, automatic RADAP, which would replace the aging D/RADEX equipment and expand the former D/RADEX system to include all network radars and some local-warning radars. In mid-1980, RADAP procurement was halted due to higher-priority requirements. At that time, NWS decided to go ahead with a scaled-back RADAP II program, an interim system that would replace the D/RADEX equipment and add an additional five sites to the automated network. RADAP II was acquired in 1983, and the system was expected to meet NWS needs until deployment of NEXRAD (Shreeve 1980; Greene et al. 1983). Eventually the number of RADAP II sites grew to 12, including one in Panama. The VIL technique and echo tops map, developed in D/RADEX, made it into RADAP II and eventually into NEXRAD. The availability of RADAP II through the mid-1980s provided NWS with a computer-based radar test bed in which non-Doppler techniques such as VIL could be used by operational weather forecasters. The benefits in this approach were twofold. First, the techniques themselves could be evaluated in a practical, operational setting. Second, forecasters could be exposed to and trained in the use of digital radar algorithms of the sort they would eventually experience in NEXRAD. The work of Elvander (1977, 1980), Winston and Ruthi (1986), and others to define objectively severe weather echo signatures based on digital radar data was implemented as the severe-weather probability (SWP) algorithm of D/RADEX, a technique that later made its way into RADAP II. In the mid-1980s, in unpublished work, Elvander used a generalized operator approach to develop the version of the SWP algorithm that is actually implemented in the WSR-88D today; coefficients have so far been derived only for Oklahoma thunderstorms (R. Elvander 1997, personal communication). The use of reflectivity-only algorithms such as VIL and SWP led to measurable improvements in storm warning statistics during the mid-1980s at RADAP II sites and eventually elsewhere, as forecasters learned to apply them to quantify storm assessments made using non-Doppler radar. Later the availability of the same techniques on the WSR-88D helped give forecasters confidence in making the transition from non-Doppler to Doppler radar.
The advent of computer-driven color monitor technology in the 1970s made it easier for meteorologists to recognize characteristic weather echo features. In the summer of 1982, a test of the usefulness of color graphics displays to hydrometeorological forecasters was conducted at the Pittsburgh Weather Service Forecast Office. An Interactive Color Radar Display system was developed to display computer-generated radar products to forecasters in color. An expanded test of full operational use of automated color displays for radar observations and severe weather warnings was conducted at Oklahoma City, Oklahoma, in spring/summer 1983. The results were encouraging, and today color displays are an important component of the WSR-88D and other weather radar systems that rely on interpretation by a meteorologist.
17. Interpretation of radar weather data
Interpreting radar weather echoes and making judgments about the character of the weather phenomena associated with those echoes is the operational radar meteorologist’s daily work. The earliest such techniques attempted to link the shapes and configurations of radar echoes to the underlying weather phenomena. The classic example is the hook-shaped echo, often indicative of tornadoes. As mentioned earlier, the first hook echo was photographed in 1953 by D. Staggs of the Illinois State Water Survey (Stout and Huff 1953). Other characteristic echo shapes supposedly associated with severe weather were quickly reported. The horizontal configuration of echoes in lines and line echo wave patterns was said to be indicative of the potential for severe weather. Characteristic echo motion was also thought to be an indicator. The vertical extent of radar echoes, the presence of weak echo regions and bounded weak echo regions (echo-free vaults), and particularly the extent to which these echoes penetrated the tropopause were other indicators. Donaldson (1965) not only reviewed these and other indicators, but also collected data on the likelihood of experiencing severe weather, given their presence. The spiral rainbands associated with tropical storms are another recognizable echo characteristic and can be useful in identifying the location of the storm’s eye or center of circulation.
Research radars almost from the start had the ability to measure the radar reflectivity factor Z of weather echoes. In the late 1950s, the AFCRL conducted in-house studies and sponsored research at the Illinois State Water Survey and Texas A&M University in hail identification techniques for the CPS-9. Based on the results of this research (largely not reported in the journals but rather in technical reports), investigators such as Donaldson (1961) reported an association between profiles of Z with height in the storm and the probability of hail, severe thunderstorms, and tornadoes. Two-predictor techniques proved to be powerful, such as echo tops combined with the Z profile, and early algorithms were reported along with associated probabilities of detection and false alarm ratios (Boyd and Musil 1970). Operational radar meteorologists soon adopted these techniques also. In the mid-1970s, Lemon’s techniques (Lemon 1977, 1980), which exploit the organization of model severe storms, began receiving greater acceptance by operational meteorologists, as they could be used with a non-Doppler radar.
Doppler-based techniques for identifying severe storms had been in development by research groups and the Weather Bureau since the mid-1950s. Some such techniques are truly research-grade, in the sense that applying them requires multiple Doppler radars. Others, such as the tangential shear of the radial velocity and single-Doppler vortex recognition, are inherently more suitable for operational application in an affordable network. Donaldson et al. (1975) applied measures such as probability of detection, false alarm ratio, and a new metric called the critical success index to a large collection of cases and attempted to show whether Doppler radar is superior to conventional radar in detecting severe storms. A similar assessment of the utility of Doppler radar in storm warning was reported by Lemon et al. (1977). The results of both these studies, showing that the Doppler techniques were superior to non-Doppler methods for tornado and severe thunderstorm detection, were available to the participating agencies in time to be used in deciding the characteristics of the nation’s next-generation weather radar. In fact, these and related studies were probably the major factor, before the Joint Doppler Operational Project (see Part II of this paper), in establishing the operational utility of single-Doppler radar observations.
18. Education, training, and professional development activities
With deployment of the WSR-57, the Weather Bureau’s initial planning was that radar meteorologists would enhance their careers by working one or two years as a WSR-57 radar meteorologist as they progressed up their career ladder. As the network expanded, a shortage of interested meteorologists developed, and some of the best meteorological technicians were assigned instead. All radar staff personnel successfully completed a four-week undergraduate-level training course at the University of Miami, receiving five university course credits for their efforts. The lecture notes of the Miami course are in Hiser (1970). The training at Miami, which began in 1959, continued into the 1970s, when it was moved to what was then called the NWS Technical Training Center (NWSTTC) in Kansas City, Missouri. The electronics technicians received eight weeks of training at the NWSTTC and were responsible for complete maintenance of the WSR-57.
It was intended that additional development of the radar meteorologists’ skills would be facilitated by use of radar scope photography. Many of the WSR-1 and -3 radars and all the WSR-57s were equipped with 16- or 35-mm cameras and Polaroid cameras for the purpose. The scientific curiosity of the radar staffs was encouraged. Copies of the 16- and 35-mm photographs could be obtained for use in poststorm analysis. Emphasis was placed on accurate diagnosis of echoes and the reasons for them, not merely the routine reporting of their presence.
These early activities in the enthusiastically developing field of radar meteorology eventually triggered an effort by the Weather Bureau, air force, and navy to assemble and publish the most useful technical training material and operational procedures in the form of a standard and “official” weather radar manual. The first of these was called the Weather Bureau–Air Force–Navy (WBAN) Weather Radar Manual, first published in the 1960s. In the 1970s the WBAN Weather Radar Manual was replaced by FMH-7.
Because operational weather personnel do not always have time to follow radar meteorological research, AWS collected relevant results in technical reports, agency recurring publications, and letters. The material was provided to weather stations equipped with radar and to schools responsible for presenting technical training to weather and equipment maintenance personnel. These included distribution of Donaldson’s (1965) review, publication of a tornado case study based on meteorological analysis and data from the FPS-77 radar (Finley et al. 1973), a digest of the signatures of severe weather as seen on the radars of the day (Whiton 1971; Whiton and Hamilton 1976), and a report of Lemon’s (1977, 1980) techniques.
The NWSTC (the “technical” was eventually dropped from the name) in Kansas City, Missouri, offered three specialized courses. The foundation course, radar meteorology, provided four weeks of instruction to meteorologists assigned to offices having a weather radar. NWSTC also presented a three-week version of the foundation course. Finally, NWSTC offered a users’ course, three weeks in length, for meteorologists desiring to use radar data remotely (Covey 1984). The NWSTC also offered an intensive “short” course on radar maintenance for NWS electronics technicians.
The air force technical training facility formerly at Chanute AFB, Illinois, and now at Keesler AFB, Mississippi, offered short courses on radar meteorology and the operational use of weather radars. The air force’s short course in radar meteorology had, since the 1960s, been about one training week in length. From about 1970 to about 1975, the air force merged its radar meteorology course with its satellite interpretation course, but the number of training days devoted to radar was not affected. Shrinking air force budgets and a drawdown frustrated attempts made in the mid-1970s to increase radar content and course length in the short course. The air force also taught radar weather observation and scope interpretation as a part of its weather technician course. In addition, the air force taught radar maintenance as one of the largest units in its weather equipment maintenance technician course.
Only to a limited extent can these courses and study materials provide the breadth and depth of understanding of the whole field of radar meteorology that we might desire for ourselves as students of the masters who established radar meteorology so successfully in the 1940s. The field has now grown, formed alliances with the cloud physics, mesoscale meteorology and severe storms communities, and developed subspecialties to such a degree as to make comprehensive understanding of it a lifetime’s work.
19. Civil applications of weather radar
At a typical WSR-57 site in the early 1960s, the radar meteorologist would take and transmit a RAREP every hour via Teletype to the RADU. This entailed such activities as identifying echo areas or lines, measuring maximum echo intensities and representative echo top heights, recognizing important “signature” features, using the reflection plotter to determine cell and storm motions, and encoding the information into the RAREP format. The RADU would then assemble the RAREPs from participating radar sites and prepare composite radar summary products. Radar operations were closely attuned to severe weather diagnosis and forecasting. Radar often provided key information on which to base weather warnings. When severe weather was observed, the radar meteorologist took and transmitted special radar weather observations. When a hurricane or typhoon came within detection distance of a weather radar, the radar meteorologist was expected to take and transmit special observations of the location, eye size, movement, and other attributes needed by tropical cyclone forecasters.
Local radar summaries were disseminated to the news media at some stations. With the fielding of radar remoting equipment in the late 1960s, remote radar weather data became more popular than such summaries.
The mission and capabilities of the RFCs did not include flash-flood forecasting; therefore, the local radar meteorologist issued flash-flood warnings, subjectively assessing the intensity, movement, and duration of rainfall events as seen on radar. This was a tricky business, as a flash flood could arise from a combination of several circumstances, including high rainfall rates, slow storm movement, or long duration. Peculiarities of soil type, slope, and basin disposition of the terrain relative to the rainfall patterns; existing soil moisture; expected absorption and runoff; and a number of other conditions compounded the difficulties. Today, the WSR-88D precipitation processing subsystem can be useful in flash-flood forecasting despite the system’s relative immaturity (see, e.g., Hunter 1996).
The RFCs in the 1960s found it difficult to apply radar data, as their computer models were more suited to the use of river stage reports and 24-h precipitation totals as reported by the conventional rain gauge network. Attempts were made by some RFCs to apply MDR data and digital information from D/RADEX and RADAP II when they became available in the 1970s and 1980s, but success was spotty, and practices were not uniform across the RFC system. As was the case with flash-flood warnings, it was not until the WSR-88D was fielded that these problems were addressed. Today the WSR-88D network provides digital precipitation data to the RFCs.
In the 1950s and early 1960s, radar meteorologists were sometimes required to give pilot weather briefings. Most pilots were very aware of the weather radar system, as it was widely advertised in the aviation literature and in ground school instruction. These pilots often asked the meteorologist for a summary of the weather as observed by radar along their route of flight. Later, airline meteorology offices and dispatch facilities had access to information provided by the radar remoting systems, as did the FAA’s Flight Service Stations, and the need for radar meteorologists to brief pilots decreased.
Radar scope photographs were taken routinely at all the WSR-57 sites and at other network stations. These scope photographs were provided to NCDC for use in research and studies. Copies of the film were maintained at each radar site so radar meteorologists could conduct and publish professional investigations of significant weather events and understand local weather processes characteristic of their location.
20. Military applications of radar weather data
Typically-military weather forecasters use radar weather data to improve the quality of severe weather warnings, making them more useful to military decision-makers for resource protection purposes. In addition, radar weather data are used to improve observations and forecasts vital to the conduct of military operations affected by weather. Two examples tell these stories.
At Kelly AFB, Texas, site of the vast San Antonio Air Logistics Center, inspections, modifications and repairs are made to the air force’s C-5 transport and B-52 bomber fleet. These aircraft are too large to be housed in any of Kelly’s hangars, and much of the work has to be done outside. To permit access by repair crews, the aircraft are lightened and jacked while repairs are made. These aircraft offer a lot of tail surface and so are always somewhat susceptible to being turned by wind. In the lightened and jacked configuration, they are even more vulnerable to these effects. If any of the aircraft are turned while jacked, they fall off the jacks and the jacks penetrate the airframe, causing great damage to multimillion dollar aircraft. Forecasters at Kelly are expected to provide 1-h warning of damaging winds from thunderstorms and other sources, as it takes that time to “de-jack” the aircraft. These forecasters have to use radar not only to help forecast the arrival and severity of thunderstorms but also to make inferences about wind from radar data to prepare the precise warnings needed. Traditionally, Kelly forecasters used wind inferences from echo motion and fine line data obtained from non-Doppler radars. Now they can use wind data obtained directly from Doppler weather radars to make their warnings.
Resource protection is not the only military application of weather radar. Some military operations are particularly vulnerable to weather, and forecasts can actually improve the success ratio of these operations. An example is airforce undergraduate pilot training, where inexperienced student pilots first earn their wings. At each stage in the pilot training curriculum, the students have different qualifications and, with them, different weather limits. At some stages of their training, student pilots can fly only under visual flight rules; increasing cloudiness and weather in the training areas is cause for bringing these students back to home base. Bringing students back from a training area requires time because the approach pattern at home base is usually far busier than that at the busiest commercial airports. Moreover, the direction of arrival of advancing weather can have a devastating effect. If the base becomes “weathered-in” in an unforecast way before the flying areas are affected, the result can be a dangerous “divert” of marginally qualified pilots to locations without the proper fuel and maintenance facilities. From there, the aircraft and pilots have to be recovered later, with tremendous impact to the flying training schedule. Something that would have almost no effect at a less busy air base can have tremendous effect at a busy pilot training base. An example is an unforecasted change in wind direction that requires the supervisor of flying to “turn around the pattern.” With so many aircraft taking off, landing, on approach to the runway, in the pattern, entering it, and departing from it, such changes can require time, valuable time during which the pattern cannot be used to recover aircraft. Most student jet training areas are adequately covered by a single weather radar, so radar can be used effectively to improve flying safety while helping to maintain the flying schedule.
By the mid-1970s, U.S. weather services had fielded a variety of conventional, non-Doppler weather radars and were considering the technology with which to replace them. All NWS radars except the most primitive were equipped with hardware-based integrators for signal processing; a few had attached data processors. AWS radars were, almost without exception, totally manual. Research in radar meteorology, especially for severe weather identification, had turned toward the use of Doppler techniques. Operational radar meteorologists found recent research increasingly difficult to apply.
Leaders of operational civil and military weather radar programs recognized an opportunity for change but were trapped by budgetary realities. AWS was caught in the air force’s protracted, post-Vietnam drawdown. NWS had a somewhat more positive outlook on the possibility of fielding a high-quality replacement radar. Neither AWS nor NWS corporately favored the alternative of a jointly acquired, next-generation weather radar.
In the second part of this history, we show how a few courageous leaders of operational weather radar programs in both the services, backed by years of difficult, intensive research in radar meteorology and inspired by leading senior scientists in radar meteorology who were proactive in wanting to apply research results, were able to show the nation the way to field a fully capable, next-generation weather radar system.
The authors thank R. Donaldson Jr., consultant to Hughes STX Corporation, for his thorough review of an early version of this paper. We also appreciate the assistance provided by Allied Signal, H. Benner, C. Bjerkaas, E. Dash, R. Elvander, P. Hexter, R. Kandler, R. Miller, the National Climatic Data Center, V. Rockney, and R. Saffle for providing valuable information. The contribution of one of the authors (PLS) was supported in part by National Science Foundation Grant ATM-9221528.
List of Acronyms and Abbreviations
ADC—Air Defense Command
AFB—Air Force base
AFCRL—Air Force Cambridge Research Laboratories
AFGL—Air Force Geophysics Laboratory
AGC—Automatic gain control
A&M—Agricultural and mechanical
ARS—Automated radar summary
ARSR—Air route surveillance radar
ARTCC—Air route traffic control center
A-scope—Type-A indicator, display of signal amplitude vs range or time
ASDE—Airport surface detection equipment
ASR—Airport surveillance radar
AWS—Air Weather Service
CEICON—Calibrated echo intensity control
DOD—Department of Defense
DOT—Department of Transportation
D/RADEX—Digital radar experiment
DSP—Digital signal processing or processor
DVIP—Digital video integrator and processor
ESD—Electronic Systems Division
FAA—Federal Aviation Administration
FMH—Federal Meteorological Handbook
IEEE—Institute of Electrical and Electronics Engineers
MDS—Minimum discernible signal
MDR—Manually digitized radar
MIT—Massachusetts Institute of Technology
MTI—Moving target indicator
NACI—Naval Avionics Center, Indianapolis
NCDC—National Climatic Data Center
NEXRAD—Next-Generation Weather Radar
NIDS—NEXRAD imagery dissemination system
NSSL—National Severe Storms Laboratory
NWS—National Weather Service
NWSTC—NWS Training Center
NWSTTC—NWS Technical Training Center
ORR—Operational radar replacement
PAR—Precision approach radar
PPI—Plan position indicator
PUP—Principal user processor
Rad Lab—Radiation Laboratory
RADAP—Radar data processor
RADU—Radar Analysis and Development Unit
RAF—Royal Air Force
RATTS—Radar to Telephone Transmission System
RCM—Radar coded message
RFC—River forecast center
RTAFB—Royal Thai Air Force Base
RVN—Republic of Vietnam
Sh—Pulse Short pulse
STC—Sensitivity time control
SWP—Severe weather probability
TDWR—Terminal Doppler Weather Radar
VIL—Vertically integrated liquid
VIP—Video integrator and processor
WBAN—Weather Bureau–Air Force–Navy
WBRR—Weather Bureau Radar Remote
WSR—Weather surveillance radar
* Current affiliation: Science Applications International Corporation, O’Fallon, Illinois.
Current affiliation: Institute of Atmospheric Sciences, South Dakota School of Mines and Technology, Rapid City, South Dakota.
Current affiliation: Amherst Systems, Inc., Warner Robins, Georgia.
Corresponding author address: Dr. Roger C. Whiton, SAIC, 619 W. Hwy 50, O’Fallon, IL 62269.
Terms such as “today” and “currently” refer to 13 March 1997.
Standardized nomenclature of military (and optionally civil) electronics systems grew out of the experience of World War II. The nomenclature is positional. The two letters before the slash, often dropped, indicate army, air force, and navy. The first letter after the slash indicates the class of installation or platform [A—airborne, C—air-transportable (no longer used), F—ground fixed, T—ground transportable]. The second letter indicates the type of equipment (M—meteorological, nonradar; P—radar). The third letter indicates the purpose of the system (N—navigational aids; Q—special or combination of several purposes; R—receiving such as passive detection systems; S—detecting or establishing azimuth and range, i.e., search;T—transmitting). This is not an exhaustive list. The Federal Aviation Administration (FAA) uses the military nomenclature for radars it shares with the military but has developed its own naming system (ARSR—air route surveillance radar, ASDE—airport surface detection equipment, ASR—airport surveillance radar, and PAR—precision approach radar).