013734-1-F THE UNIVERSITY OF MICHIGAN COLLEGE OF ENGINEERING DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING Radiation Laboratory DETECTION OF INCIPIENT TORNADOES BY A SPACEBORNE DOPPLER RADAR - A LITERATURE SEARCH Ralph E. Hiatt 31 December 1975 Final Report Purchase OrdArr S-54214A j^ Prepared for NASA Goddard Space Flight Center Greenbelt, Maryland 20771 13734-1-F = RL-2264 Ann Arbor, Michigan

D)ETCTION OF INCIPIE;NT TOINA nOES IBY A SPACE KHOINK I)OPI'L:R RAIARI - A TIIITIRATURK S:AICII Tntroduction This report is to summarize the results of a limited crfort literature search for information needed for the construction of a simplified model of a tornado vortex such as would be seen by a Doppler radar. Thc project was initiated as a result of our interest and that of NASA Goddard to develop the information needed to design a spaceborne radar system for the detection of incipient tornados. Ilencc, in our literature search, we have placed more emphasis on radar observables as scen from a satellite. It is intercsting to note that none of the several dozen papers reviewed had as its major emphasis the use of satellite borne radars for the early detection of tornadoes. Several papers, however, made refcrcnce to the subject. Many papers have been written on the use of ground based radars in the detcctlion and study of tornadocs. With respect to the space platflorm, two choices arc possible-a synchronous satellite or a low orbit vehicle. The synchronous satellite would have the advantage of long looks at chosen areas of interest and the much greater field of view. On the negative side, the power required would be much -greater even with the larger antcnnas which would be required, the resolution would be less and perhaps of greatest importance, the near-vertical viewing angle would make it difficult to properly observe some of the important tornado signatures. Tn addition, the decrease in radial motion in the storm system as viewed from the synchronous altitude would seriously limit the advantlages of the Ioppler detection system. A!)opplcr radar system on a low orbit satellite would have the obvious advantages with respect to power required, antenna sizc,rcsolution, I)opplcr returns arid the improved aspect of the viewing angle. Kvcn so, several dirriculties arc evident. 1Thcse include the following: 1

1. The transmitted and reflected signals nmut penetrate 5 to 10 km or more of turbulent cloud cover. It may bc possible to compensate for the resulting attenuation and noise in the data processing system. 2. If interest is restricted to the North American contincnt, viewing time cfficiency will be lcess by 75 percent or so while the satellite is over the oceans and other continents. 3. Once the satellite is in orbit, there is little or no opportunity for changing the area viewed (the footprint of the radar.) Thus it is frequently not possible to view a particular area as long or as frequently as desired. It is possible that a tornado would develop and becomne destructive while the satellite was outside of the U.S. viewing area. 4. It iwll be difficult to discriminate between the ground and weather targets of interest since there is little difrerence in their velocities relative to that of the satellite. This problem can be alleviated to some extent by using a narrow bcam antenna and by limiting the zone covered. With this limitation it should still be possible to observe the top of a tornado vortex and this may be an important clue for discriminating betwccn tornadoes and other storms. 5. The height of the satellite and the associated long ratnge introduces range ambiguities if the P1W (pulse repetition frequency) is high enough to provide the desired number of looks and resolution. Icspitc these somnwhat formidable problems, it is believed that the overall advantage is with the low orbit satellite. Methods of solving the above problcms have been outlined in the study made by those who prepared the Active Microwave Workshop Report (Matthews, 1975). lRadar engineers arc confident of their ability to design a satellite borne radar that could collect information that would bc cffcctivc in detecting the presence of tornadoes (R. Larson, 1975, personal communication). The results of the literature search arc discussed in the following sections. Scction II contaits a general description of tornadoes. Section TTT discusses the clectromagnetic obscrvablcs associated with a tornado as seen by both a passive and an active system. T1 Section IV somec of the pertinent I)opplcr radar systems arc discussed and conclusions arc presented in Section V. Since this is a report on a literature search, frequent use is made of quotations, sonmec of which arc direct quotations. 2

TT. Tornado Characteristics - (Gencral To add to the background information, it may bc helpful to have a brief nonclcctromagnctic description of a tornado. I)onn (1965) says, "Tornadocs arc by rar the most destructive manifostations of all nature. Nothing else is comparable to their fury." The power required to drive a tornado vortex has been cstimalted to excecd 10 kw (Vonnegut, 1960). A more complete description is quoted from l'cttcrssen (1969):... A tornado is a vortex of small horizontal extent and great intensity which extends downward from a thundercloud. It is usually visible as a funnel-shaped tuba cloud, with a broad base in the cumulonimbus and a narrow tubular cxtension down to the ground. The lower part oC the tornado cloud is ortea surrounded by an ugly column of dust and debris, which arc sucked up from the ground and thrown outward by the centrifugal force in the whirl. At the time of inception thc funnel is more or less vcrtical, but as the mother cloud moves on, the upper portion of the whirl becomes slanted and sometimes detached. On occasion several funnels may build down from the mother cloud, and some of these may not reach the ground. Thc diameter of a tornado may vary from a few meters to a few hundred meters, with an avcrabc of about 250 m. The winds arc strong also outside the funnel cloud, and the width of the path, as determined by the destructive crccts on the ground, may be two to four times as wide as the funnel cloud. The passage of a tornado is accompanied by a sudden pressure drop of the order of 25 mb, though sometimes much larger pressure changes have been observed. The decrease iii pressure from the rim to the center of the tornado represents a tremendous force, which few buildings can withstand. As a tornado goes ovcr a building, the outside pressure drops so suddenly that the pressure within cannot follow suit. Much of the damage caused by tornadoes is due to the prcssure diffrerntial through the walls and roof: buildings tend to "explode" rather than to be "blown over". 'Th circulation of the wind around a tornado is almost always in a counterclockwisc (cyclonic) direction. The wind speed near the core varies within wide limits. Since no ancmomenter has survived the passage of a severe tornado, no rcliable measuremenits of the wind speed have been madc.:stimatcs, however, indicate that the wind speed is of the order of 100 m/sec, or about 225 mph, and may be twice as high in extrcmc cases. From the formula given on page 303 it follows tltt, in extreme cases, the wind pressure may be as large as 800 lb/t. Only few structures can withstand such wind pressures. 3

The life of a tornado varics within widc limits. On the average, the length of thc path, as detcrminecd from thc destructive errects on the ground, is about 5 to 10 km (3 to 6 mi). On occasion paths as long as 300 km have bccn observed. The majority of the tornadocs in the United States occur in connection with squall lines or severe cold fronts. These tornadoes often occur in familics and movc it the direction of the wind in the warm air; they usually have long paths. Tornadoes arc observed also in connection with scattered thunderstorms of great intensity. Such tornadoes arc usually short-lived and have irregular paths. Although tornadoes may occur anywhere, except in the polar regions and over cold northern continents in winter, they arc most frequent to the cast of the lRocky Mountains, to the cast of the Andes, alnd in eastcrn India. This suggests that the influences of mountains on the air currents and thc distributions of temperaturc avnd moisture along the vertical arc important, but little is known about the mcchanisms that produce tornadocs. In the United States about 95 percent of all tornadoes move from a direction between south and northwest, with a strong prcfcrcncc (61 percent) for a southwesterly direction. Most of the tornadocs in the Unitcd States occur in connection with squall lines atnd cold fronts when there is a layer of warm and moist air near the ground overlain by a layer of dry air. Gray (1971), who has emphasized the importance of the statistical approach inl the study of tornadoes and tornado genesis states that the individual tornado will always be undorobscrvcd and oFtcn misreprcscnted. Based on his studies and those of his colleagues, he concludes that tornadoes, funnecl clouds anld many lesser intense mnicro-scale vortices form utlder conditions when strong cumulus updrafts penetrate lower tropospherin layers of large vcrtical wind shear. The combination of a strong buoyant updrart and vertical shear arc the important cnvironmelntal requirements. The grcat plains area of the United States leads the world in having these conditions a relatively high percentage of time and in the production of tornadoes. nased on the studics made by TIudlam (1963), tornado funnels appear to protudc from thn arch cloud in the forward right flank of severe storms and outside the precipitation region. III. Tornado electromagnctic Obscrvablcs Kmphasis in this study is on the use of active clectromagnetic systems, i.e., radars. llencc, in this section, rmost interest will be on the scattering and reflection of electromagnetic waves by tornadoes. However, one should also be 4

aware of the electromagnetic radiation generated by a tornado. This subject will bc discussed first. 'ornado obscrvables as viewed by passive clcutromagnetic systems Tt is rather obvious that the usual high level of lightning and thunder associated with a tornado would result in clectromagnctic radiation. Thc frequencics commonly associated with a spark discargc cxtelnd from near dc values to those approtximlatcly equal to the reciprocal of the rise time of the pulse discharge. Frequency componncts as high as many megahertzl arc known to be associated with tornadoes. According to Scoutcn ct al. (1972), the parent thunderstorm associated with the tornado and not the vortex is the likely source of the intense nlectrical activity. Taylor (1973) states that the clectrical charges within a thunderstorm cloud arc usually dispersed throughout the cloud and arc discharged to the earth by the discrete current surges of lightning strokes. The electromagnetic fields radiated during lightning strokes arc called atmospherics or sfcrics. Again, quoting Taylor, "The rcsponse of receivers tuned to frequencies less than about 30 klIz arc discrete and arc easily associated with a return strokc in a cloud-to-ground discharge. As the observing frequency increases, the number of responses increased to several thousand at frequencies above 10 Mllz. " The higher frequencies arc thought to be associated with the rapid occurrence of short-distancec dispersive processes within the parent storm cloud. In his 1973 paper Taylor reports on his investigation of electromagnetic radiation from tornado-producing severe storms occurring in Oklahoma in 1970. lie observed the rate of occurrences of atmospherics at frequencies from 10 Hz to 3 IVI ITz using short time constant circuits to prcserve the burst nature of the received impulse signals. TTc concluded that the parameter most indicative of tornadic activity was the number of bursts of high atmnospherics rates at frequencies above 1 M11J7. TTis fridings arc supported by the work of Hlughes anld Pybus (1970) and by Stanford ct al. (1971). Radar obscrvables nattan (1973) states that the identificatioln of a tornado by means of conventional radar is not easy. It has frequently been noted that tornadoes arc associated with 5

thunderstorm echoes having protruding fingers, V-shaped notches or doughnut shapes, i.e., small dry holes or notn-echoes within fairly irtensc echoes. Ilowever, these echo shapes arc rtequnitly seen whetl no tornadoes arc prcsenit and tornadoes frequently occur when the above echo shapes arc not seen. The presence of a hook-shaped echo in association with a severe thunderstorm is believed by several observers to be a rcliable indication of the presence of a tornado. BRiglcr (1956) described a scries of radar observations in association with a tornado that occurred near College Station, Texas. lie was able to observe a completely developed hook before the tornado touched the ground. Although the hook echo appears to be a good positive tornado indicator, many radar observers have followed the developmenti of tornadoes in which no such echoes were seen. Frcund (1966) ceamincd the radar echo signatures of 13 tornadic storms occurring near Norman, Oklahoma in 1964. lie found no definite radar signature common to all of the tornadoes. lie concludes that the hook echo might belong to a special class of tornadoes and states that the classic hook might be more characteristic of the severe storm literature than of real tornadic phenomena. I)onaldson ct al. (1975) conclude that no more than half of all tornadoes, under the best of conditions, indicate their existancc with an unmistakable hook echo. This is based on their own cxperience and information gained from other workers in the field. nattlar (1975) slates that most of the timne the shape of the echo is of little value in iidetifying tornadoes. In the same paper he states that long-lastin.g thunderstorms extleding to great altitudes and producing intense echoes arc likely to have associated violent weather such as hail storms or tornadocs. The intensity of a radar echo is thus another clue to the presence of a tornado. Donaldson (1961) reports that the echoes from New englanld thunderstorms which produce tornadoes arc more intense than the tornado-free storms. Jf the crrcctivc reflcctivity factor is greater than 10 mm /m at an altitude greater than 10 km as seen by a 3 cim radar, this is a good ilndication iti New England that the thundcrstorm will produce a tornado. (The effective reflectivity factor, z, is (6X 4M)/(7r k12 p ) ) where M is the liquid water contentl with particles of dianmctcr n1 and density p. o is the radar cross section of the particles and k|2 is a function of the index of refraction convcntionally taken as.93 for water.) 6

rThe maximnum height of the radar echo associated with a slorm is an important signature. For example, cyclonic storms occurring in the middle latitudes have maximum height echoes of 5 to 8 km. Thunderstorm echoes rcach heights of 10 km or so; hail storms have echoes up to 13 km in height while sevcrc storms with tornadocs commonly have echoes up to 20 krn in height (Katz, 1975). Atlas (1963) proposes a mtcthod or tornado dctection based on fluctuation analysis based on radar returns. The fluctuatioris arc due to two or morc scattcrcrs moving in an d out of phasc due to the wind velocity and the rotational mnotion. I igher fluctuation rates arc an indication of greater wind speeds. According to Atlas, fluctuation frequencies greater than 2 kiIz should provide a unique indication of a tornado, assuming a 10 cm radar system. Such a detection system would, unfortunately, require a pulse repetition frequency of 4000. The radar signature which, without doubt, is the most reliable mecans ofr identifying the presence of a tornado is that which shows the characteristic wind vorticity. I)onaldson (1971) makes the following statement: "Tornadocs and other damaging wintds in severe local storms provide destinctivc reatures for their remote identification by I)opplcr radar. Smith and l olmes (1961) dcmonstrated the capability of a cw I)opplcr radar for detection of a tornado, and Lhermittc (1964) considered techniques for presenttatior of unique pulse D)opplcr velocity patterns associated with tornado vortices." Velocities arc most easily measured with a I)opplcr radar; mean wind velocities arc determined from the I)opplcr returns from the scattercrs carried by the wind. I)oppler velocity information is, of course, limited to that component which is in the radial direction with respect to the radar system. TTcncc, special techniques arc required to obtain rotational information with a single I)oppler radar. I)ual Dopplcrs have been used for this purpose, but that techtnique is not adaptable to a space platform. I)onaldson (1970, 1971) and Dotialdson ct al. (1975) were successful in developing a very crffectivc method for recognizing a vortex signature. They found a scanning I)oppler radar has capability for mecasurement of the tangential shear of radar velocity. This is dcfined as the gradient of the radial compoinent of velocity in the direction normal to the radial vector. This type of shear is an indication of 7

vorticity and its presclce is determined by scalnning the radar beamn in azimuth while its elevation is fixed. Ielcvation angles betwccn 0~ and 10~ arc sclectcd to observe quasi-horizo0ntal wind components at various heights within the storms. I)onaldson ct al. (1975) found that high shear values generally were first found at middle altitudes in a storm and then progressed downward. -1 They found that a shcar value of 0.02 sec proved to be a rcliable threshold crileriot for identifying severe storms. Storms were charactcrizcd as severe if they deposited hail with diameters of 3/4 inch or larger and/or inflicted wind damage by tornadoes or other imcans. These criteria enabled them to detect all but three in a series of 48 observed severe storms and the criteria were present in only six of more than 150 nonscvcrc storms which the group studied. IBrowt ct al. (1973), Kraus(1973 a, b) and Tnurgcss ct al. (1975) have made use of similar or related techniques using a single l)oppler radar to determine rotational characteristics of tornadoes and tornado cycloncs. I)opplcr radars have other advantages over conventional radars: they arc able to locate a tornado vortex with greater precision and, by suitable signal processing, stationary ground clutter canl be climinated making it more feasible to track storm echoes at short ranges and over hilly terrain. IV. I)opplcr Radar Systcms As indicated earlier, there appear to have been few, if any, investigations of the possible use of satellite borne radars for the detcction of incipient tornadoes. Iowevcr, rattan (1975) mentions the idea and states that specialized radars carried on satellites should play a part in the identification of violent weather in future years. lie believes that research on the use of satellites for the observation or severe local slorims and for use in communication of forecasts and warning deserves generous support. lie mentiois one cxaipple where a strong vortex was detected in the middle levels of a storm area about 23 minutes before an associated tornado touched the ground (with a ground-based system). I)ctcction well before touchdown is, of course, quite important as a means of saving lives and rminimnizing destruction. The most intensive and also the most recent study relevant to our objective was that performced by a working group assembled by NASA (Lyndoun B. Jolmson 8

Space Center). The group, approximatcly 70 in numbcr, mcl at thc Space Centcr in July 1974 to consider the utilization of active spaccbornc microwave systems in application progranms concerned with observations of (1) the land areas of the earth, (2) the oceans and (3) the atmosphere. Our present intcrcst is in the work of the 16-member panel on the atmospherc. A summary of this work appears in a paper by Isadore Katz (1975) who was one of the two co-chairmen. The complete report is contained in the recently released NASA report, number SP-376, edited by R. I. Matthews (1975). The charter of the atmospheric panel called for their investigation of the meteorological applications of a satellite borne radar. The following applications and objectives were considered feasible: map maximum echo heights in rain to provide and indication of storm intensity and rain rall production tropical storms, si7ec and location mapping precipitation and drop size spectra height of the melting level surracc wirlds phenomcna associated with the formation or sea ice cirrus cloud dtcction. Radar types considered for satellite borne mctcorological applications included a more or less convcntional pulsed radar, multiwavclcngth and dual polarization radars, a ow Doppler radar operating at the CO laser frequency (X = 10 ~mn) and a microwave pulsed Dopplcr radar. The last system appeared to have the most potential and it will be discussed in more detail later. Both synchronous and low orbit satellites were considered as possible platforms for these systems. Each had some advantages, but for several reasons the lower orbit is to be preferred. A satellite in a near polar orbit with an inclination angle of 58 and a height of 556 km was proposed. Such a satellite would have an orbit period of 1.6 hours. With the antenna bcamwidth and scanning arrangemecnt proposed, the complete carlh surfacc betwccn 1 68 latitude would be covered cvcry 12 hours. 9

The parameters for a pulsed )oppler radar system were devcloped by the workshop members. It was emphasizcd that the specifications were listed only as an cxample of a radar system having a coverage roughly equal to that of existing mctcorological satellite systems. The system specified was considered to be a [airly conventional i)oppler radar. The selected specifications were not proposed as the optimum but the panel believed that they could serve as a takc-off point for further discussion and development. It should be noted also that the system was designed with general ncmtcorological applications as the objective and was by no means designed exclu.sively for the detectioni of incipient tornadoes. It is the opinion of this writer, however, that the design has mainy or most of the characteristics and capabilities that would be required for tornado detection. Since it may be dirrficult to obtain the eecded funds for a satellite borne radar to be used exclusively for studying tornadoes, it may be prudent to have as one's objective a system with two or more meteorological applications in mind. The system parameters arc set forth in T'able 4-VI of the Active Microwave Workshop Report and that table is reproduced here: T'ntativc Specification for a Multibcam Inoppler System Radar wavelength, cm l'cak power (per beam), kw Average power, w -4 PIlF (stability bettcrn than 10 ), kilz Pulse width, psc Antenna bcamwidth, deg Bcam footprint on the ground, km Teamn nadir angle, dcg (razing angle, deg Conical scan, beam scan time, sec Satellite displacement during complete scan, km Angular displaccmcnt of the antenna beam during echo round trip, deg Satellitc orbit 5.6 5 100 2.5 10 0.3 5 by 15 60 20 40 280 0.13 polar inclincd (cont.) 10

Satellite altitude, km,v500 Satcllitc orbit time, hr 1.6 Satellitc groundspccd, kmn/scc 7 Fforward-groundspccd Doppler shift, kllz ';v200 6-dml )oppler smearing (max), mn/scc 35 Nonambiguous velocity interval, mn/scc 70 An additional specification, not indicated in the table, is a requirement for multiple beams; a 5 to 10 beam system was suggested. With a single beam, the scan rate needed to cover the desired ground swath does not allow sufficient time on each target area for the radiated pulses to return to the antcnna. A parabolic reflector antenna is envisaged; it would be suspended below the satellite with its bcam at a nadir angle of 60. A spiral scan on the carth's surfacc is obtained by spinning the satellite and dish about a vertical axis as the satellite moves forward in its orbit. A swath width (perpendicular to the orbit direction) of 2000 km is obtained. The bases for the choice of wavelength, beam sizec, scan rate, pulse repetition rate, etc., arc discussed in the Workshop Report and that discussion will not be repeated here except for a few comments. One of the most important considerations was to choose parameters that would provide I)oppler infornmation on the atmosphere with a minimum of contamination by ground or sea return. The narrow bcam (0. 3~) and proper range gating help to eliminate this problem and mnakes it possible to effectively probe the horizontal and vertical distribution of atmospheric targets. Another problem inherent with satellite borne l)oppler radar systems is due to the high ground speed of the system ( 7 km sec ). This causes an undesirably large spread in the )oppler due to the variation int range of the mneteorological targets with respect to their position in the bcam. The spreading or smecaring of the D)opplcr makes it impossible to obtain useful metcrological data fromn the spectrum width. iKy using already available signal processing techniques it is possible, however, to obtain values for the nmcat n)opplcr. With the mean I)oppler data one can monitor mecan wind velocities within a precipitation system. Since the target area is viewed from two 11

directions as the satellite approaches and as it recedes, two Doppler velocity components can be measured. Another critical part of the system specification is the maximum usable PRY'. The workshop panel reports that if the antenna pattern is free of sidelobcs and with a grazing angle of 20, the maximum time between successive pulses is 400 to 500 Aseconds. A PRY of 2000 llz places the forward I)opplcr in the 100th ambiguity region and in the opinion of the panel this would still allow discrimination -1 between atmospheric and ground clutter speed with 35 m sec or less Dopplcr smearing. Katz (1975) makes a 9 point lisl of the cxpected capabilities of a spaccbornc radar system in a low altitude orbit. All nine points should be considered by those who would decide on the cost cfcctivencss of such a system. ecre, however, we list only the two capabilities which would tend to increase our ability to detect incipicnt tornadoes. Comments in parentheses have bccn added by this writer. 1. Ability to map the nmaxinmum echo heights in rain to provide an indication of storm intensity atnd rainfall production. (As slated earlier, storm heights above 13 km or so almost always indicate the prcsence of a tornado.) 2. Measure horizontal motion within a storm systcm. (This capability is to a large extent made possible by the forward and backward look provided by the proposed system. Tnformation on radial and horizontal wind motion should indicate the presence of a vortex and hence the existence of a tornado. The possibility of using the??plarneshear indicator" technique developed by D)onaldson and his colleagucs (Armstrong and I)onaldson, 1969; I)onaldson, 1970) should also be invcstigated. With this scheme it is possible to locate quickly, at least with a ground-based n)opplcr radar, regions in a thulrderstorm ccho where circular motions cxis. ) The microwave workshop panel concluded that a I)opplcr radar operating at a wavelength of 10.6 P/m (CO2 laser frequcncy) had good potential Cor meteorological applications. It orfers an attractive alternative to a 6 cm radar system in its ability to detcct and resolve speed and location of cloud particles. Their report states that such a systlem should be able to survey areas of the atmosphere larger than is 12

possible with a microwave radar. Further, the 10 pmn wavelength radar would have range and velocity information in a single pulse, several microseconds long, sufficient to provide mean velocity information at a much higher rate than call be had with a microwave system. As with the microwave system, use could be made of the rcturnl as scen in both the forward arid reverse direction so that vector wind direction can be determined. The possibility of using a microwave radar imaging system with a synthetic aperture as an alternate to the radar system with the very large antenna as specified in the above table should be considered. Highly sophisticated imaging systems have been developed by the personnel of the:nvironmcntal Research Institute of Michigan (EKIIM) and others. Tlarson (1975) makes the following statcment: "Tnased on the present slate of the art in imaging (or synthetic aperture) systems it would appear fcasible to design a spaceborinc velocity detection anld mappinrg system (for tornado study) having roughly the following capability: 1. velocity resolution 1 mph 2. range resolution < 10 m 3. azimuth rcsolution (in imaging mode) < 1 mile 4. rcacting time: near real timc. These capabilities would be heavily influenced by the data processing required for the I)opplcr signature detcrmination and imagc-forming roles of the system. The cxtent to which these capabilities could actually be achieved would require a detailed investigation." V. Conclusions In this literature search, we have found no reports of an investigation seeking to determine the feasibility of a satellite borne radar tornlado detection system as the prime objective. As indicated in the body of the report, we found numcrous reports on the use of ground-based radars for the detection of incipicnt storms. Tn those few studies which have been made on satellite borne meteorological radar systems, very little attention was given to tornado detection. We have cxamincd, however, several reports that arc quite relevant to our objective. 13

Scveral reports have been rcviewed which describc the clcctrornagnetic obscrvables associated with a tornado. Our emphasis has been on those which would bc the most reliable signaturcs for an active radar system. Of these the best arc the rotational characteristics of the winds, the echo height, the echo intensity and possibly the "hook" shape. The most complete source of information on a satellite borne meteorological radar system which we were able to find was the recent report of the active microwave workshop panel (Matthews, 1975). Information from this report which appears to be most relevant to tornado detcction is reported in some detail. nRased on the advances made in satellite load capabilities and in advanced radar systcms, we believe that a properly designed spaccborne I)opplcr system would be a valuable tool for use in the detection of incipient tornadoes. We have no estimate of the cost ffrrctivnncss of such a system. We believe this could be cnhanccd by designing a systemlcl with two or three or possibly more meteorological applications. We believe also that in the design of a satellite borne microwave I)opplcr radar system, considcration should also be given to the inclusion of a 10 Pum cw I)opplcr systcm such as was mcntioned by Matthews (1975). To enhance the tornado detection system, the satellite system might also include passive sfcric detection systems (Taylor, 1973). One objection made to the use of spaccborine radars for tornado detection was the undesirable length of tilrme between successive searches of specific areas (I)Dnn is, 1963). If the system would prove successful ard cost effective, the time between looks can be reduced by a factor of 2, 3 or 4 by adding additional satellites in corresponding numbers. We iquit again the recent statemcnt by Battant (1975), "Rcscarch on the use of satellites for observation of severe local storms and for use in communicating forecasts and warning deserves generous support." We quote also I)r. David Atlas (1975), 1'rcsidcat of the American Mctcorlogical Society, in a presentation made to the Hlouse Subcommnnittec on the Envionmn-ctl and Atmosphere at the AMS Confercence on Severe Tlocal Storms, 23 October 1975, Norman, 14

Oklahoma: "Amnong all the observation needs, however, I am sure we arc in unanimous accord with Professor Talttan (Itallan, 1975) who cmphasizcd the n.cd for and potlcntial of remote sensing tccuhniqucs-cspecially for I)oppler radar-for the unambiguous detection and pinpoirtilig of tornadoes. Ilerc is a tool whose time has come. With the exception of a modest amount of applied research to determine its possible limitations, such as false alarm and miss-rates, I believe it is ready for deploymnctt.... The phenomcnal advances which have been made in remote sensing of clouds and tenperaturcs from satellites over the last 15 years indicates one of the most promlising directions to take. " 15

HEFE.R NC ES 1. Armstrong, G.M. and R.J. I)onaldson (1969),"Plane Shear indicator for real-time I)oppler radar identification of hazardous storm winds", J. Appl. Meteor., 8, pp. 376-383. 2. Atlas, 1). (1963), "Radar analysis of severe storms", Mctoorological Monographs, September 1963, pp. 177-223. 3. Atlas, I). (1975), Overview: The prediction, detection, and warning of severe storms. AMS ConFercnce on Severe LTocal Storms, 23 October 1975, Norman, Oklahoma. 4. Iatltan, T,.J. (1973) Radar Observation of the Atmospherc. The University of Chicago Plress. 5. IBattan, L.J. (1975) I)ctection of Severe local Storms. Presented at the October 1975 Conference on Severe local Storms, Norman Oklahoma. 6. Isigler, S. G. (1956) "A note on the succcssrul identification and tracking of a tornado by radar", Wcathcrwisc, pp. 198-201. 7. IBrown, R.A., I). W. Burgess and K.C. Crawford (1973)" Twin Tornado Cyclones within a Severe Thunderstorm: Single I)oppler Kadar Observations", Weather, April 1963, pp. 63-69. 8. IBurgess, 1). W., T.H. Lecmon and It.A. Brown (1975) Evolution of a tornado signature and parent circulation as revealed by single I)oppler radar. Itadar Meteorology Conference, Vol. 16, April 1975, published by the American Meteorological Society, pp. 99-106. 9. I)ennis, A.S. (1963) Fundamental limitations on precipitation observations from satellites. Proceedings of the Tenth Weather Radar Confcrence or the American Metcorological Society, Iloston 1963, pp. 348-354. 10. I)onaldson, lt.J., Jr. (1970) "Vortex signaturc recognition by a I)oppler radar," J. ofAmer. Metcor., pp. 661-670. 11. I)onaldson, l.J., Jr. (1971) I)oppler radar identification of damaging convective storms by plan shear indicator. Seventh Conference of Severe T,ocal Storms, October 1971, pp. 71-89. 16

12. I)onaldson, K.J., Jr.,. M. D)ycr and M.J. Kraus (1975) Operational 3enefits of Meteorological I)oppler Radar. eHport AFCRTL-T.H-75-0103 21 Fcbruary 1975, Bedford, Massachusetts. 13. I)onn, W. L. (1965) Meteorology, Third Edition. McGraw-HIill Iook Co., New York, N.Y. 14. Frcund, 11. -. (1966) Kadar:E ho Signature of Tornadoes. Twclfth Radar Meteorological Confcrence, Norman,Oklahoma, pp. 362-365. 15. Gray, W.M. (1971) lacacarch Mcthodology, Observations, and Tdcas on Tornado (encsis. Sevecrth Conference on Severe local Storms, October 1971, Kansas City, Missouri, pp. 292-298. 16. llughcs, W.T. and E:.J. l'ybus (1970) Severe Storm Sfcrics-Stroke rate. iFourteenth Conference on ladar Meteorology, American Meteorological Socicty, November 17-20. 17. Katz, I. (1975) Active Microwave Sensing of the Atmosphere From Satellites. l'reprints of the Sixteenth Radar Meteorology Conference, hlouston, Texas, April 1975, pp 1-7. 18. Kraus, M.J. (1973a) Calculating Air Flow from Single D)opplcr Kadar Velocity Components. P'reprints of the;ighth Conference on Severe Local Storms, pp. 44-47. 19. Kraus, M.J. (1973b) "iDopplcr radar observations of the Brooklinc, Massanhusctts tornado of 9 August 1972", 1ull. Amer. Metcor. Soc., 54, pp. 519-524. 20. Larson, H. W. (1975) Private correspondence (included in 20 January 1975 proposal submitted by the Univcrsity of Michigan to NASA Gloddard Space Flight Center). 21. Tlhermittc, K.M. and I). Atlas (1961) Plrecipitation Motion by Pulse D)opplcr Radar. Proceedings of the Ninth Wcather Hadar Confcrencc, pp. 218-223. 22. Tlhermitte, t. M. (1964) ")opplcr radars as severe storm sensors", TBull. Amer. Meteor. Soc., 45, pp. 587-596. 23. T.udlam, F. I. (1963) "Scvcre local storms: a review", Metcorological Monographs, 5, No. 27, pp. 1-27. 24. Matthew, R.., E:d. (1975) Active Microwave Workshop Report. NASA Kcport S P-376. U.S. Government l'rinting OFfic, Washington, 1).C. 20402 Stock No. 033-000-00624-1, $5.60. 17

25. Pcltcrsscn S. (1969) Introduction to Meteorology. Third Kdition. Mc(rawIlill 1Book Company, New York, N.Y. 26. Scouten, I).C., I). T. Stephcnson, and W.(G. Biggs (1972) "A sfcric rate azimuth-profile of the 1955 l1ackwell, Oklahoma tornado", J. Atmos. Sci., 29 (5), pp. 929-936. 27. Smith, K. T,. and!). W. IIolmcs (1961) "Use of D)opplcr radar in meteorological obscrvations", Mon. Weather Rcv., 89, pp. 1-7. 28. Stanford, J.T1., M.A. Tind and G.S. Taklc (1971) "Klectromagnctic noise studies of severe conllnclivc storms in Iowa: the 1970 storm season", J. Atmos. Sci., 28(3), pp. 436-448. 29. Taylor, W. T. (1973) "E;lectromagnotic radiation from severe local storms in Oklahoma during April 29-30, 1970", J. of (ieophys. Kcs., 78 (36), pp. 8761-8777. 30. Vonncgut, 1i. (1960) ":lcctrical theory of tornadoes", J. Gcophys, lcs., 65 (1), pp. 203-212. 18

GI I IAT. F R NCFS 1. liggs, W.(c. and P.J. Waite (1970) "Can T. V. rcally detect tornadocs?", Wcathcrwisc, June 1970, pp. 120-124. 2. Irown, K.A., W.C. Bumgarner, K.C. Crawford and 1). Sirmans (1971) "Preliminary D)opplcer velocity mcasurcne rits in a developing radar hook echo", Ilull. Amcr. Metcor. Soc., 52, pp. 1186-1188. 3. Chalmers, J.A. (1954) Atmospheric Electricity. Reports on progress in Physics, Vol. 17, pp. 101-134. 4. Fctl, K. W. arid S. Isrand (1975) "Tropical cyclone movement forecasts based on observations from satellites", J. of Appl. Meteor., 14, pp. 452-465. 5. Fujita, T. 1. (1970) "The Tlubbock tornadoes: a study of suction spots", Wcathcrwisc, August 1970, pp. 161-173. 6. Fujita, T.T'. (1971) Proposed Mechanism of Suction Spots Accompanied by Tornadocs. Seventh Confrcence on Severe T.ocal Storms, October 1971, Kansas City, Missouri, pp. 208-213. 7. Fujita, T.T. (1973) "Tornadoes around the world", Wcathcrwisc, April 1973, pp. 56-62. 8. Hlouse, ). C. (1963) "Forecasting tornadoes and severe thunderstorms", Meteorological Monographs, Scptmcber 1963, pp. 141-156. 9. Kropfli, ii.A. and T..J. Miller (1975) Thunderstorm Flow Pattcrns in the Three )imcnsions from )ual-l)opplcr ladar. Sixtccnth ladar Meteorological Conference, April 22-24, 1975, louston, Texas, pp. 121-127. 10. Thermittc, K. (1974) "lcal-timc monitoring of convcctivc storm processes by I)ual- )opplcr radar", Atmospheric Technology, Winter 1974-1975, pp. 26-33. 11. Miller, T,.J. and I.G.C Strauch (1974) "A I)ual-l)opplcr radar method for the determination of wind velocities within precipitating weather systems", Remote Scnsing of the environment, 3, pp. 219-235. 12. Pierce, E. T. (1968) Sfcries. Proceedings of the Scientific Meetings of the Panel on Rcmote Atmospheric Probing, April 18-20, May 16-17, 1968. 19

13. l'urdom, J. r'. W. (1971) Satcllite Tmagery and Scvcre Weather Warnings. Scventh Conference on Scvcre Tlocal Storms, Kansas City, Missouri, October 1971, pp. 120-127. 14. Scg-nan, R. (1970) "tSSA D)opplcr radar systcm", Wcathcrwisc, April 1970, pp. 70-73, 83 and 103. 15. Scrafin, F.J. (1974) "Tnformation cxtraction from meteorological radars", Atmospheric Technology, Winter 1974-1975, pp. 46-54. 16. Smith, P.T.., Jr. (1973) "Wcather radar: achievcmcrts, promiscs and problems", Atmospheric Technology, 2, June 1973, pp. 57-65. 17. Strauch, K.(c., A.S. I'risch, and W. I. Sweezy(1975) I)opplcrlladar Mcasurcments of Turbulence, Shear and D)issipation Hates in a Convective Storm. Hadar Mctcorology Confcrcncc, Vol. 16, April, 1975, published by the American Meteorological Society, pp. 83-88. 18. Taylor, W.L. (1972) Atmospherics and severe storms from remote sensing of the tropospherc. Edited by V.. D)crr, Chapter 17, U.S. Govcrnment Printing Office, Washington, I).C. pp. 17-1 to 17-17. 19. zrnic,!).S. and lt.J. noviak (1974) Vclocity Spectra of Vortices Scanned with a l'ulsed i)opp]cr Radar. UIISJ Spring Meeting, Atlanta, (;eorgia, June 10-13 (Abstract p. 24). (Completc report obtained from I).S. zrnic.) 20