VOL. 13, NO. 5 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY OCTOBER 1996A Mobile Mesonet for Finescale Meteorological Observations JERRY M. ST~ICASchool of Meteorology, University of Oklahoma, Norman, Oklahoma ERIK N. RASMUSSENNational Severe Storms Laboratory, Norman, Oklahoma SHERMAN E. FREDRICKSONNational Severe Storms Laboratory, Norman, Oklahoma(Manuscript received 29 March 1995, in final form 27 July 1995) ABSTRACT A mobile weather observing system (mobile mesonet) was designed to augment existing meteorologicalnetworks in the study of severe local storms and other mesoscale weather phenomena in conjunction with theVerification of the Origins of Rotation in Tornadoes Experiment (VORTEX). Fifteen mobile mesonet unitswere built, each consisting of meteorological instruments mounted on standard automobiles, for high temporaland spatial resolution observations. While the most accurate measurements are possible from stationary mobilemesonet vehicles, accurate observations also are possible from moving vehicles. The mobile mesonet instrumentsmeasure pressure (600-1100 mb), temperature ( - 33 - to 48-C), relative humidity (0%- 100% ), and winddirection and speed (00-360- and 0-60 m s-~ ). Onboard each vehicle, a Global Positioning System (GPS)receiver and a flux-gate compass obtain universal time, vehicle location (latitude, longitude, altitude), andvehicle heading and speed. A standard laptop computer stores data, computes derived variables, and providesreal-time data display. Instrument compatibility with the Oklahoma Mesonet allows for high-quality instrumentcalibration and maintenance. The purpose of this paper is to provide a technical overview of the mobile mesonet system. The rationale forchoice of instrumentation and justification for method of exposure are discussed. The performance of the mobilemesonet is demonstrated with two examples of data collected during VORTEX-1994 and comparisons with datafrom an Oklahoma Mesonet site.1. Introduction Studies of many small-scale weather phenomenasuch as severe local storms can benefit from very highspatial (order of hundreds of meters to tens of kilometers) and temporal (order of 10-100 s) resolutionmeteorological observations. For example, detailed information at the surface of winds and equivalent potential temperature can provide evidence to support thehypotheses that low-level baroclinic zones are a sourceof vorticity for low-level mesocyclone development insupercell thunderstorms (e.g., Klemp 1987; DaviesJones and Brooks 1993). Unfortunately, very few severe storms have passedover finescale meteorological networks during the past50 years providing limited insights into low-level stormstructures (e.g., Barnes 1978; Brown and Knupp 1980; Corresponding author address: Dr. Jerry M. Straka, School ofMeteorology, University of Oklahoma, Energy Center, 100 EastBoyd St, Rm 1310, Norman, OK 73019.c 1996 American Meteorological SocietyGoodman and Knupp 1993, etc.). Moreover, there havebeen even fewer instances of tornadoes and their parentcirculations being in close proximity to weather observation stations (e.g., Tepper and Eggert 1956; Ward1964; Davies-Jones and Kessler 1974, unpublishedNSSL reports). The task of obtaining comprehensive,very high resolution data with fixed position meteorological mesonets for studies of many small-scaleweather phenomena such as severe storms simply is notpractical; too few storms pass over these networks. In response to the scientific needs and the economiclimitations for obtaining finescale meteorological observations, to support the objectives of the Verificationof the Origins of Rotation in Tornadoes Experiment(VORTEX; Rasmussen et al. 1994), fifteen mobile mesonet units have been built and tested by the NationalSevere Storms Laboratory (NSSL), Center for Analysis of Storms (CAPS), and University of Oklahoma(OU). The mobile mesonet units are designed to makemeteorological observations of various phenomena associated with thunderstorms including rain and hailcores, front- and rear-flank downdrafts and outflows,921922 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUMI*. 13gust fronts, mesocyclones, flows near tornadoes, andstorm inflow regions. In addition, the mobile mesonetcan be used for studies of surface boundaries (e.g.,fronts, dry lines), shear lines, circulations, and othersmall-scale weather phenomena. The concept of developing a mobile/deployablesurface observing system is not entirely new. For example, Bedard and Ramzy (1983) and Bluestein(1983) built and deployed the Totable Tornado Observatory (TOTO). In addition, Fredrickson (1993)described a surface meteorological measurement system on NSSL's Mobile Cross-chain Loran Atmospheric Sounding System (M-CLASS) mobile atmospheric observatories (Rust 1989; Rust et al. 1990).Both of these systems are mobile/deployable meteorological measurement platforms designed to increasethe opportunities to make valuable measurements nearsevere weather phenomena during storm intercepts(Bedard and Ramzy 1983). Other deployable systems include the Stationary Automated Mesonetwork(SAM; Barnes 1981), Portable Automated Mesonet(e.g., PAM-II; Brock et al. 1986), and small instrument packages called "Turtles" for measurements ofpressure and temperature in and near mesocyclonesand tornados (Brock et al. 1987). Each of the 15 NSSL-CAPS-OU mobile mesonetunits described herein consists of a standard automobile, an instrument rack and mast, a suite of sensors,and a data collection, storage, and display system. Thesensors and data collection systems are derived, in part,from NSSL's M-CLASS mobile atmospheric observatories (Fredrickson 1993) and the Oklahoma Mesonet's systems (Crawford et al. 1992; Brock et al.1994). The design of the instrument mast provides ameans to obtain accurate wind, pressure, temperature,and humidity measurements from moving and stationary vehicles. Each instrument package also includes aflux-gate compass to provide vehicle heading information when stationary, and a Global Positioning System (GPS) receiver to provide vehicle heading and vehicle speed information when moving. Universal timeand position also are obtained from the GPS. A programmable data logger is used to collect and processdata directly from the sensors. A program on a laptopcomputer then periodically requests data from the datalogger and the GPS for storage and real-time display. The primary purpose of this paper is to provide atechnical overview of the mobile mesonet. The systemrequirements and instruments are described in section2, including specifics about instrument inaccuracy, resolution, and atmospheric coupling. The calibration procedures for the mobile mesonet are reviewed in section3. A brief discussion of how the mobile mesonet iscoordinated and managed in the field is given in section4. Examples of data collected using the mobile mesonetare described in section 5. Finally, a summary is presented in section 6.2. Mobile mesonet descriptiona. Instrument performance requirements The maximum local pressure perturbation associatedwith severe local storms that might be experienced bya mobile mesonet vehicle is on the order of -20 to 4-10mb in as few as about 10-30 s (e.g., Tepper and Eggert1956; Davies-Jones and Kessler 1974; Bedard and iramzy 1983; Bluestein 1983). (Much larger pressurefluctuations are possible in tornadoes.) Pressurechanges with larger mesoscale weather phenomena canbe on the order of 1-10 mb h-~. In addition, stationpressures (absolute surface pressures) in the centralUnited States can vary from less than 850 mb to morethan 1000 mb. Accuracy requirements are estimated tobe on the order of 0.01-0.1 mb. The resolution of thepressure measurement (and others as well) should beconsistent with the accuracy requirements. Wind speeds near thunderstorm mesocyclones, :rearflank downdrafts, gust fronts, inflows, etc., typically areon the order of 10-30 m s-l, while gusts can exceed40-60 m s -~ (although the integrity of the mobile mesonet vehicle and instruments might be compromi, sedby such wind speeds and accompanying debris). Basedon previous wind measurements associated with thunderstorm phenomena (e.g., Bluestein 1983), speed anddirection need to be observed on timescales of secondLs.Moreover, accuracies of the wind measurements nee, dto be better than 1 m s -~ for speed and 5- for direction. Tornado, mesocyclone, and gust front proximitymeasurements described in Tepper and Eggert (1956),Bedard and Ramzy (1983), and Bluestein (1983) suggest that temperature gradients on the order ,of10-C km-~ and 10-C min-~, and relative humidity gradients of 30% km-~ routinely can be expected. To beuseful over a wide range of conditions, the mobile mesonet must measure temperatures over many tens ofdegrees Celsius and relative humidity from 0% to100%. These parameters need to be measured on timescales of tens of seconds and with accuracies of 0.5~'Cfor temperature and 5 % for humidity.b. Instrument platform Each mobile mesonet unit consists of a standard automobile (1993 Chevrolet Corsica for VORTEX-1994and 1993 Ford Tempo for VORTEX-1995), an instrument rack and mast, a suite of meteorological and positioning sensors, a data logger, a power system, and alaptop computer. A block diagram of the mobile mesonet system logic is shown in Fig. 1. The instrumentrack and mast are constructed from 3/4" aluminum robing and are mounted to the roof of the vehicle usingoff-the-shelf roof-rack-mounts (which have proven tobe adequate for long-term stability and security). Aphotograph of the sensors and instrument mastmounted on a vehicle is shown in Fig. 2a and a closeup of the sensors and instrument mast is shown in Fig.OCTOBER 1996 STRAKA ET AL. 923BLOCK DIAGRAM OF MOBILE MESONET SYSTEMSENSORSetcVEHICLE-RELATIVEWIND SPEED AND R M YOUNGWIND DIRECTIONTEMPERATURE (FAST) YSITEMPERATURE (SLOW) VAISALARELATIVE HUMID TY HMP35AC-LVEHICLE (HEADING) I KVH FLUXGATE VEHICLEVEHICLE SPEED I TRANSMISSIONPRESSURE I VAISALA PTB205AASPIRATION FAN I PANASONICBATTERY VOLTAGE PULSE, I ANALOG, I CAMPBELL SCIENTIFIC PC, WER j CR10 DATA LOGGER r q r ANALOG, PULSE, ~ RS232 J RS232 R~ RS232 SAMPLING ~ MUX [~ ~ / GAIN/OFFSET ~ 2-SE' COND'I ALGORITHMS'-:~--" "1~:~a~:~ ....... // GPS (each 1 ~ec) VEHIcLEBACKUP BATTERy ~. ~ e TIME - LATITUDE BATTERyI~ 12 ~ e LONGITUDE .... e ALTITUDE e VEHICLE SPEED e VEHICLE DIRECTION r SATELLITES . .~,_; ;.~-.~ 'FIG. 1. Block diagram of the mobile mesonet system. The main components are the sensors, data logger,RS-232 MUX, GPS receiver, 386 PC laptop computer, VHF radio, and power supply.I VHF RADIO] - I386 PC LAPTOP REAL-TIME DISPLAY e ARCHIVE OF ~W &SEC DATA DERtVED VARIABLES (e,g.,D~POINT, EQUIVALENT PQ~N~AL ~MPERATURE) - TRUE WIND SPEED/ DIRECTION, etc e ADDITIONAL VEHICLE DEPENDENT OFFSETS - RADIOTRANSMI~ER * ~_._:...; ............ : ~ 2 ~..~ LONG-TERM POWER / ARCHIVE AND SUPPLY [ O~ER DISPLAYS2b. A summary of the various sensors incorporated intothe mobile mesonet, including (when applicable) theirheights above ground, factory specified inaccuracies,estimated total inaccuracies, ranges, and resolutions, ispresented in Table 1. The estimated total inaccuraciesare based on factory, field (e.g., instrument-atmosphere coupling), and system [e.g., analog-to-digital(A/D) conversion with the data logger] inaccuraciesusing the root of the sum of the squares of each. Discussions of the sensors and problems associated withcoupling them to the atmosphere are described below.c. System descriptions 1 )POWER SYSTEM, DATA LOGGER, LAPTOP COMPUTER The mobile mesonet data collection system consistsof a power distribution system, meteorological sensors,data logger, GPS receiver, flux-gate compass, and laptop computer. Power for each mobile mesonet system is suppliedfrom a standard 12-V automobile storage battery separate from the vehicle's main battery. The system's battery is charged by the vehicle's 12-V charging system. A programmable Campbell Scientific Incorporated(CSI) model CR10 data logger is used to scan, sample,and manage the suite of sensors as well as store datalocally if needed. The CR10 is capable of measuringvoltages from 12 analog channels in the +2.5-V range.Two pulse input ports are available measuring fromswitch transitions to 256 kHz. Three channels are capable of providing sensors with a preprogrammed,precision excitation voltage of up to _+2.5 V. Optional software allows the data logger to interfacewith up to three serial output sensors. The CR10 allows the user to locally program the order that inputchannels are read and processed and the order thatoutput channels transmit. All channels can be sampled with independent gains and offsets applied.Channel sampling frequency is user controlled. Serial sensor timing requirements initially necessitateda scan of all sensors once per 2 s. The data from thesescans are used to generate 6-s averages. Communications between the data logger, GPS, and laptopcomputer (programming and data transfer) arethrough standard RS232 interfaces. Note that A/Dconversions by the CR10 limit, to a tolerable extent,the accuracy of some of sensors (e.g., wind). The GPS navigation receiver is a Motorola modelPVT-6 system designed with characteristics making itsuitable for in-vehicle satellite tracking. Details of theGPS system are discussed later but in summary thereceiver makes available universal time, longitude, latitude, altitude, vehicle vector, and tracked satellite information once each second. The laptop computer for the mobile mesonet systemdescribed herein can be any AT class laptop computeroperating under DOS. General requirements are oneserial port, power adaptors to allow the laptop to be runand charged from the vehicle's 12-V battery system,and a screen capable of easily being read in a widevariety of lighting situations. A 25-MHz Gateway 2000924 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 13 FIG. 2. (a) Photograph of the mobile mesonet instrument mast and instrument ports. Note the meterstick in figure. (b) Instrument ~nastlayout, including positions of the pressure port (1), aspirated weather shield for temperature and relative humidity (2), wind speed anddirection (3), Global Position System (4), and flux-gate compass (5). Note the meterstick in the figure.OCTOBER 1996 STRAKA ET AL. 925 TABLE 1. Mobile mesonet meteorological sensors. (NM means not meaningful.) Height Inaccuracy Inaccuracy fromMeasurement Sensor type Sensor model Output Sensor range (factory) (est. total) Resolution groundPressure Silicon Vaisala RS232 600- to 1100 mb <_+0.4 mb <_+0.6 mb 0.01 mb 1.0 m capacitive PTB202-ATemperature Resistance YSI 44203 Analog -30- to +50-C <-+0.15-C <_+0.3-C 0.01-C 2.25 m(Fast)Temperature Resistance Fenwal Analog -30- to +50-C <-+0.4-C <-+0.5-C 0.0I-C 2.25 m(Slow) UUT51J 1Relative Capacitance Vaisala Analog 0% to 100% <2%-3% <5% 0.03% 2.25 mhumidity HMP35AC-LWind speed Propeller-vane R.M. Young Pulse 0 to 60 m s-~ <2% reading <2%-4% 0.03 m s-t 3.0 m 5103 readingWind direction Propeller-vane R.M. Young Analog 0- to 355- <-+3- <-+30-6- 0.05- 3.0 m 5103Vehicle heading Flux-gate KVH AC-75 Analog sine (-1 to 1) <-+2- <-+2- <1- 3.0 m(stationary) compass cosine (-1 to 1)Vehicle heading GPS Motorola PVT-6 RS232 0 to 360- Est. avg. Est. avg. < 1- NM(moving) < -+ 2- <-+ 2-Vehicle speed GPS Motorola PN'T-6 RS232 0 to >50 m s-~ Est. avg. Est. avg. <1 m s-t NM <1 m s-~ <1 m s-~Vehicle location GPS Motorola PVT-6 RS232 0- to -+90-N/S <100 m <100 m Est. I m NM(latitude/ 0- to -+ 180-E/Wlongitude)Notebook 386sx, with standard 4-Mbyte RAM and80-Mbyte hard drive is being used for VORTEX. The laptop serves as the main driver and focal pointfor the mobile mesonet system by 1) communicatingwith the data logger for loading the initial program andreceiving data from the data logger; 2) communicatingwith the GPS receiver for data and for changing necessary setup parameters; 3) acting as controller for theserial multiplexer switch (needed as there is only oneRS232 interface on the laptop); and 4) acting as themain data handling device for data storage (hard drive)as well as algorithm application for real-time computation of derived information and data display.2) SENSORS AND INSTRUMENTS(i) Pressure A Vaisala PTB202-A silicon capacitive integratedcircuit pressure sensor with serial ASCII output is usedfor the mobile mesonet. The pressure range of this sensor is 600-1100 mb with a resolution of 0.01 mb anda response time of less than 1 s. The temperature operating range is -25- to +50-C. The factory specifiedtotal error is _+0.40 mb. A microprocessor in thePTB202-A makes all adjustments to the pressure including those for factory calibration and temperaturecorrection. The actual pressure sensor is placed aboutt m above the ground in the trunk of each vehicle. Considerable literature exists demonstrating the degradation of pressure measurements by dynamics effectsowing to the ambient wind and turbulence. In one test,with the mobile mesonet pressure sensor inlet exposedto the flow over the top of the vehicle, absolute pressureperturbations were found to be at least as large as 1.51.8 mb at speeds of up to 25 rn s-1. In another test withthe pressure sensor inlet placed in the trunk of the vehicle absolute pressure perturbations of at least 0.7-0.8mb were observed at speeds of up to 25 m s-~. Pressureerrors in both tests roughly were proportional to flowvelocity squared. A Qualimetrics (model M104598) quad-plate staticpressure port (e.g., Woo et al. 1989; Nishiyama andBedard 1991) is used to reduce the dynamic pressureassociated with wind and turbulence (Qualimetrics andBelfort quad-plate design ports similarly performed intest applications on the mobile mesonet). The quadplate pressure port design is advantageous over othersin that it has omnidirectional response and its performance is not severely effected in adverse weather conditions (precipitation, dust, sand, etc.). Moreover, thequad-plate design is relatively insensitive to being atsmall angles of attack (0- to _+25-) in laboratory tests.At wind speeds of 36.6 m s-~ (20.0 m s-~) and turbulence fluctuations of 0.8% (17%), pressure errors atport angles of attack between 0- and _+ 10- are less than0.04 mb (0.17 rob) to 0.2 mb (0.2 rob), respectively(Woo et al. 1989). For wind speeds less than 10.0m s -~ and turbulence fluctuations of 0.8% (17%) pressure errors that are less than 0.04 mb (0.1 mb) at anglesof attack between -10- and +25-. These results areimportant as roads are not always flat surfaces, and vehicles tilt backward/forward when accelerating/decelerating. To minimize adverse pressure perturbations associated with flow over a mobile mesonet vehicle the pres926 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 13 FIG. 3. The flow field over a standard automobile (Nissan) similar to the one used for the mobile mesonet (Society of Automotive Engineer's,INC. report HS J1566 JAN86 and Nissan). The vehicle shown is a one-fifth scale model. The flow was visualized using multiple sourcesmoke lines in a smoke tunnel.sure port is placed in a position that is about 0.62 m infront of the instrument mast, and about 1.00 m abovethe roof of the vehicle, which is about 2.5 m above theground. This location is ahead and above the significantflow and pressure perturbation produced by a standardautomobile as determined from examining factorywind tunnel tests (e.g., Fig. 3) and from mounting978.0977.5977.0976.5PRESSURE PORT TEST44- COURSE DATA (Test B: 80 pts) 976.0 0 5 10 15 20 25 30 FLOW SPEED (M/S) FIG. 4. Correlation between airspeed indicated by the R. M. Youngwind monitor and the pressure measured using the pressure port. Thetrue pressure is shown as a solid line for this sample test.pressure ports and wind monitors at various heightsabove an automobile's roof top. To test the quad-plate pressure port's effectivenessfor the mobile mesonet application, an instrumentedvehicle repeatedly was driven along two roads whereelevation changes were less than 1 m (i.e., pressurechanges were less than 0.1 mb) over distances of several hundreds of meters. Test course altitude deviationsof 0.5-1 m might account for up to 0.1 mb of the pressure fluctuations measured during the moving vehicletests. Data were collected for accelerating and nonaccelerating conditions. The primary tests were done whenthe wind was nearly calm and 5-min ambient pre. ssureaverages did not change over a period of about 3 h. Additional tests were conducted when winds of 5-8 m s-~were observed and ambient pressures were steady. Pressure port field tests showed that the differencesbetween pressure observed from a moving vehicle andthe mean stationary pressure were on the order of thesmall-scale pressure fluctuations (0.0-0.3 mb) observed when the vehicle was stationary and compar~tbleto those expected with the port. For moving vehicles,in both calm and windy conditions, it was found in theworst case that about 80% and 95% of all pressure observations were within about 0.1 and 0.2-0.3 mb, respectively, of the average stationary vehicle pressureobservations. The correlations between flow speed .andpressure were less than 0.005 for all tests perfo~~ned.An example of a test with the pressure port is shownOCTOBER 1996 STRAKA ET AL. 927in Fig. 4, where the correlation for test route pressuremeasurements is 0.002. From the tests described above it is estimated thatthe total (factory + field) error with the pressure portis less than about +0.6 mb during normal vehicleoperation. In perspective, this error would amount toa _+2 m s -~ error (about 10% relative error) in computing the cyclostrophic wind for a pressure gradientof 4 mb over a radius of 2 km (typical values for amesocyclone). Because velocity is proportional tothe square root of the pressure gradient for cyclostrophic balance (typical for thunderstorm mesocyclones), these pressure errors are not as large as thosewhen computing the geostrophic wind for whichspeed is proportional to the pressure gradient. Forexample, Brock et al. (1994) shows a geostrophicwind error of 13.3 m s -~ for a 0.4-mb pressure errorover 30 km. As pointed out by Brock et al. (1994),other errors including incorrect elevation in computing mean sea level pressure also might cause significant errors in calculations involving pressure gradients. (ii) Relative humidity Relative humidity (RH) is obtained using a Vaisala HMP35AC-L capacitive sorption sensor. Thissensor is packaged inside a cylindrical, liquidwater repelling, vapor permeable barrier (GORTEXmembrane) for protection from water and dust. Withthe barrier, the unaspirated RH sensor response timeis 15 s at 20-C and 90% RH (without the barrier theresponse time is less than a few seconds). The rangeof the HMP35AC-L is 0% to 100% RH and the resolution is 0.03% RH. The inaccuracy against fieldreferences, at 20-C, is +2% in the 0% to 95% RHrange and -+3% in the 95% to 100% RH range. Thetemperature operating range is -20- to +60-C. Asdiscussed in the next subsection, a thermistor is collocated with the RH sensor (beneath the GORTEXmembrane) so that a dewpoint can be computed. TheVaisala probe suites the mobile mesonet platformwell as it is rugged, well behaved, inexpensive, andhas low power requirements. An alternative is to incorporate a more expensive chilled-mirror dewpointsensor. As discussed below in (iii), an aspirated weathershield was developed to protect the RH and temperature sensors from rain, hail, and water spray from theroad. Without this shield serious errors were possiblein adverse conditions. Finally, it is worth noting that the HM?35AC-L RHsensor is particularly sensitive to radio frequency (RF)interference from VHF radios during transmission atgreater than 3-5 W (with antenna less than 1 m to therear of the sensor). After transmission the RH sensorreadings immediately return to normal. Radio frequency shielding of the instrument's cables and housing might help alleviate this problem in the future.(iii) Temperature A Fenwal Electronics UUT51J1 thermistor is placednext to the RH sensor on the HMP35AC-L probe. Theworst case error of this thermistor is less than _+0.4-Cover a temperature range of -33- to +48-C, and canbe reduced by single-point calibration. The resolutionof the thermistor is 0.01 -C. The disadvantage of placinga thermistor next to the RH sensor is the influence ofthe surrounding GORTEX membrane. While highlypermeable to vapor fluctuations, this membrane is asignificant thermal barrier to the transmittance of outside temperature fluctuations. This transmittance alsois highly dependent on ventilation. Simple observationsand detailed tests performed by the authors and personnel at Oklahoma Mesonet (e.g., S. J. Richardson1994, personal communication) show that good ventilation can result in time constants of less than 1 min.Light wind conditions, however, can result in time constants of more than 5 min. Therefore, the temperaturefrom this probe is referred to as the "slow temperature.'' A YSI (Yellow Springs Inc.) 44205 thermistor alsois used to measure temperature. The YSI produces anoutput voltage that is related linearly with temperature.The maximum specified inaccuracy of this thermistoris +0.15-C over a range of -30- to +50-C. Similar tothe Fenwal, the resolution is 0.01-C. This sensor is notsurrounded by a liquid repelling barrier. However, itdoes have a thin coating of protective epoxy surrounding the thermistor allowing for an unaspirated responsetime of about 3 and a response time of 0.6 s whenaspirated by a 2 m s -~ flow. The temperature from thisprobe is called the "fast temperature." As the thermistor inside the HMP35AC-L (slowtemperature) can lag considerably behind the outsidetemperature (e.g., fast temperature), it follows that anoutside change in the vapor content will produce a"modified" value of RH upon reaching the volumebehind the GORTEX membrane. A solution is to derivea value of dewpoint, which is conserved for constantpressure and vapor content, from the HMP35AC-Ltemperature and RH. Then the YSI thermistor can beused for the actual "outside" air temperature to determine the approximate "outside" RH. There are adverse characteristic of standard algorithms typicallyused for this conversion. Fredrickson (1993) and Brocket al. (1994) show that derived dewpoint errors fromany temperature-RH sensor typically are less than-+ I-C for RH greater than 40%. These errors increaseexponentially, however, as RH decreases in value (Fig.5). Other errors can result from improperly matchedmeasurements owing to uncertainties in response lags. The temperature and RH probes are positioned about0.75 m above the vehicle (about 2.25 m above theground). Initially, an R. M. Young, model 41002 naturally aspirated, multiplate shield (e.g., Gill 1983 ) waschosen for protection from solar radiation and precip928 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 13 SENSITIVITY OF DEW POINT TO RH OFFSET ERRORS6 \\N. i , ! %u , ~.._:~ ~ i ~ i ~ ~r,- 2, ~ +2o/o ~ ; ; ; ; :o ..~~~ ~ ~ ~; ' +1% ~ ~~=~~ I ~~~~m 0 '~ ' ~ i~~ ~ .2 .! - ............. i ~- [~t -~% .$ 0 10 ~0 ~0 ~0 $0 $0 70 80 ~0 100 RESTIVE HUMIDI~ (%) F~. ~. Calculated dcwpo~nt ~or as a result of RH sensor offsetc~o~s at co~a~ p~e~s~m o~ ]000 mb ~d co~stant temperature of40-C. - In rainy conditions (including very light rain)when the vehicles were moving and/or the windsspeeds were high (>15 m s-1) Tgm was close to thewet-bulb temperature, while rshield was close to the actual temperature (via a benchmark thermometer). Atypical example when humidity was high (>90%) isshown in Fig. 7. For rainy conditions with low relativehumidity (RH < 75%), differences between Tsh~e~,~ andTgm have been noted to be greater than 5-C. - In cloudy or dark conditions (small or no solarradiation), with no precipitation, rshield and Tgin ess.entially were identical (differences were significantly lessthan the inaccuracy of the YSI thermistor). - In dry conditions, Tgul responded faster than '.Fsa~elawhen the mobile mesonet vehicles were moving atspeeds greater than 20 m s-~. This was a result of veryeffective aspiration of the vehicle-mounted Gill shield. - When the mobile mesonet vehicles were stationary, exposed to high sun angles (60--90-), and innearly calm conditions (wind speed less than 1 m .s-.1),T~in and rshield both probably read high by 0--3-C (e.g.,itation. While this shield has good radiation characteristics given sufficient airflow, instrument probes andshield surfaces easily were wetted from water spray inwindy and rainy conditions. In these circumstances,both RH and temperature measurements were degradedand stayed so for longer periods of time than were acceptable for the mobile mesonet requirements. Subsequently, an aspirated weather shield was designed andconstructed from schedule 40 (about 1/4" thick), 3"wide, white polyvinyl chloride (PVC) tubing (Fig. 6). The design of this shield is such that the flow mustturn three sharp corners before reaching the sensors.This helps to prevent rain and water spray from wetroads and other vehicles from reaching the sensors andhelps keep the inside of the shield dry (as determinedby in situ observations and by tests with an intensespray of water--for more than 30 min at a time--fromall directions relative to the shield). Moreover, sensorsalso are protected from solar radiation, dust, debris, andoccasional large hail (2-8-cm diameter). The shieldinterior naturally is aspirated when moving through theambient air. A Panasonic fan, which pulls air at a rateof more than 60 ft~ min-~, is used to aspirate the shieldinterior when there is a lack of ambient flow. Various temperature comparisons were made byplacing a YS! thermistor inside the aspirated weathershield discussed above (rshield) and another inside astandard Gill-type radiation shield (Tgi,) made byR. M. Young. (See Gill 1983 or Brock et al. 1994 forfurther discussion on the performance of Gill-type radiation shields). The two shields were mounted neareach other on the mobile mesonet instrument mast. Theresults include the following:51~VICW - C~ AWAY 0- 1t~ A~I~1~ V~A1~ 5HI~L;? ~ ............... ..... 12.7 m ~ ~n~~.~l.~m.~aal lal~. am~of ~n~kume.~,A~at,~ ~JInstrument; ~u~,2,O an7~2 cm FIG. 6. Side view cutaway schematic of the mobile mesonet aspirated weather shield. A cross section of the instrument tubing showing the relative positions of the HMPAC-35L and YSI probes also isshown. The weather shield is made from schedule 40 (about 1/4"thick) 3" wide, white PVC tubing. Note the direction of the airflow,the position of the instruments, and the aspiration fan.OCTOBER 1996 STRAKA ET AL. 929Gill 1983; Brock and Fredrickson 1993; Brock et al.1994). - During VORTEX-1995, nearly step-function temperature (T~h~ela) changes of 6--8-C in 12 s were observed while passing through sharp drylines and gustfronts at vehicle speeds of about 20 m s-~, suggestingadequate performance for our proposed efforts. (iv) Wind direction and speed The wind direction and speed sensor is a R. M.Young 5103 wind monitor propvane. The vane orientation of the wind monitor is determined by applying avoltage across a 10f~ (0.25% linearity) precision conductive potentiometer whose slide contact is fixed tothe direction vane. The output from the sensor is ananalog voltage that is proportional to the azimuth angle(0--355-). The vane threshold is about 1 m s-~, witha damping ratio of 0.25. The direction resolution is0.05- and the maximum inaccuracy is 3-. An error of_+0.5- was tolerated in alignment of the anemometer onan automobile. The wind monitor speed is obtained by measuringthe alternating current voltage sine wave produced byrotation of a six pole magnet mounted on the oppositeend of the propeller shaft. The frequency of the sinewave is measured by a pulse channel on the data logger,which converts the signal to speed in meters per second. The sensor threshold speed is about 1 m s-~ witha distance constant of 2.7 m. The specified peak measurable wind speed is 60 m s-~ (- 133 mph), althoughit has a maximum survivable gust on the order of 80m s-~. The resolution of the speed sensor is 0.03m s-~, and the inaccuracy is 2% of the reading. An estimate of the true ambient wind direction andspeed is derived by using either the vehicle headingand speed from GPS when the vehicle is moving or theheading from the flux-gate compass when the vehicleis stationary, the vane's orientation relative to the vehicle, and simple vector calculations. The average GPSerrors are estimated to be less than 2- and less than 1m s -~ for heading and speed. Originally the R. M. Young wind monitor wasplaced on top of the instrument mast. At a height of 3m, the wind monitor was above the significant flow( < 1 m s -~) and pressure perturbation field producedby the vehicle as previously described. Various testsusing three wind monitors mounted on a special rackdemonstrated that wind observations at 3 m aboveground were at a sufficient height above standard automobiles for accurate wind measurements. These testswere done with the 1 ) wind monitor at various heightsabove the vehicle (up to 4 m above ground), 2) windapproaching the vehicle from various directions, and3) vehicle moving at various speeds (0-30 m s-l).Comparisons also were made with nearby fixed-location anemometers (e.g., Oklahoma Mesonet facility inNorman, Oklahoma; see section 5). Further evaluation242322212O2230 2240 2250 ~ME(CD~I ..... F+++++++++++ RAIN INTENSrTY FIG. 7. Comparison of the fast temperature measured in the aspirated weather shield (Tshx-~a) and the fast temperature measured in aGill-type shield (Tgi,) using identical YSI thermistors. The Gill temperature essentially is the wet-bulb temperature. The humidity overthe time interval varies from about 90% to 97%. The start and stoptime for rain and rain intensity are indicated on the plot ("-" is lightrain or drizzle, "=" is moderate rain, and "+" is heavy rain). Thevehicle was moving at speeds of 10-25 m s-~ during this test.(after VORTEX 1994) showed that placing the windmonitor on a boom 0.62 m ahead of its original location(just forward of the front of the vehicle roof) allowedfor slightly more accurate measurements (errors lessthan 2% wind speed reading compared to a wind monitor at about 4 m above ground) at heights of about 23 m above the ground and at vehicle speeds of up to25 m s-l. This position has been adopted for all futureuses and is the one indicated in Fig. 2. Tests continueto determine and correct any biases in the wind speedsand directions. In summary, tests show that wind measurement inaccuracies are similar to wind sensor inaccuracies usingthe flux-gate compass heading when the vehicle is notmoving with the following limitations: * There are no nearby obstructions (trees, vehicles,bridges, buildings, etc.) to the flow. - The vehicle is on a flat road with longitudinal orlatitudinal axis inclinations of less than 10--15-. Furthermore, measurements of wind direction andspeed from moving vehicles are accurate to within_+30-6- and 2%-4%, respectively, using vehicle direction and speed from GPS. Additional limitations formoving vehicle calculations include the following: e The vehicle is moving at a speed of more than 3 -1ms ~ The vehicle direction and speed are varying byless than _+2- s-~ and less than 1 m s-2 over a periodof 6 s.930 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 13 - Wind speeds are greater than 2 m s-~. (v) GPS and fiux-gate compass A Motorola PVT-6 GPS, which tracks six simultaneous satellites, and has a serial RS-232C interface format, is used to obtain UTC time, latitude, longitude,altitude, vehicle velocity, and vehicle heading. Theseare available every 1 s. Due to U.S. Department ofDefense controlled signal degradation, termed selectiveavailability (SA), position accuracy is limited to 100m in the horizontal and 150 m in the vertical (90%confidence). In actual use these variations appear asrandom position and vehicle vectors errors. The vastmajority of these errors remain within a range so as tonot affect most mobile mesonet calculations. The DODsignal degradation can be removed, to a large extent,by using differential GPS (position errors can be reduced to less than 10 m). Computation of mean sealevel (MSL) pressure is not recommended using rawGPS altitude; a 9 m in altitude error is equivalent toabout a 1-mb pressure error. It might be possible toobtain accurate mean sea level pressures from GPS latitudes and longitudes and interpolated United StatesAir Force altitudes, or possibly from differential GPSaltitudes. A KVH-AC75 flux-gate compass obtains vehicle heading information to determine wind direction when vehicles are stationary (the GPS is not accurate enough atslow vehicle speeds). Field tests show that the KVH compass output varies smoothly for general road conditionsand when vehicle acceleration is negligible. Sine and cosine analog voltage corresponding to north and east components of the vehicle heading are output by the compass.The time constant of the compass is 500 ms, and theinaccuracy is 2- and 4- rms at tilt angles of + 10- and+30- from the horizontal, respectively, for an undistortedfield. The compass is located high enough above the ve~hicle to reduce most of the local field distortion. Inaccuracies introduced by this and other sources are removedthrough calibration (discussed in section 3). In addition,each flux gate has adjustments available to compensatefor most local distortions. Radio frequency produced by normal VHF communications can cause the output voltage of the KVHflux-gate compass to exceed levels tolerated by theCR10 data logger. This problem can lead to contamination of all other analog channels. Precision voltagedividers maintaining the ratio between the sine and cosine output while limiting the magnitude of the outputhelps prevent contamination of other analog channelsthough it does not solve problems of temporary distortion of (duration of communication) KVH measurements. Radio frequency shielding and bypassing wouldprobably solve these problems altogether.3. Calibration and data quality assurance Prior to use in the field, the mobile mesonet instruments are checked for being within factory specifiedtolerance using the Oklahoma Mesonet calibration facilities. Sensors are spot checked before and after eachday in the field to detect for instrument drift and otherproblems. Thorough reviews of the Oklahoma Mesonetcalibration procedures and facilities are presented inBrock et al. 1994, Richardson and Brock 1993, Brockand Fredrickson 1993, and Richardson 1992, 1995. Abrief summary of the relevant procedures follows: - Vaisala pressure sensors are calibrated using acontrolled pressure-temperature calibration facili*ty. Aparoscientific 1015A barometer is used for a reference.Temperature is varied slowly from -20- to +50-Cwhile the pressure is cycled over its range. Referencecomparisons are made after the sensor readings stabilize at specific pressure set points conditions. - HMP35AC-L and YSI thermistors are calibratedover their range using two reference thermometers submerged in a stirred antifreeze bath, with temperaturecontrolled by a heater and a freezer over a range from-20- to 50-C. - HMP35AC-L sorption humidity sensors are calibrated over a range of 5%-100% RH at constant temperature using a General Eastern humidity generatorand chilled-mirror dewpoint sensor for a reference. - R. M. Young wind speeds are checked with a multispeed motor drive. Propeller shaft bearing torques arechecked for being within tolerance. - R. M. Young vane direction potentiometers arechecked using a factory provided calibration set. Thewind vanes are aligned with the center line of the vehicle using two plumb bobs (one for the front and onefor the back of the vane). The alignment measurementsare checked three times for repeatability (better than0.5-). [After the field phase of VORTEX-1994.. thewind vanes were rechecked and found to be within0.1--1.0- (typically 0.5-) of their initial alignment.] - Flux-gate compasses were calibrated by comparing more than 10 000 pairs of flux-gate to GPS headings per vehicle for noncardinal directions. This wassimpler and approximately as accurate as calibrationusing a surveyor's transit and compass rose. Post-field-experiment data-quality assurance is asimportant as any other component of a field experiment. For the mobile mesonet, data are flagged whenthey are known to be in error (e.g., wind speed anddirection when vehicle was accelerating), are suspicious, or need to have corrective measures applied tothem. Suspicious data are determined through use ofdata bounds, standard deviation thresholds, various :filters, and instrument time constants (e.g., to identifyunphysical changes owing to RF).4. Communication and field management The mobile mesonet was first deployed during VORTEX-1994 in a highly coordinated fashion to put instrumented vehicles in positions where they might: obOCTOBER 1996STRAKA ET AL.Cursortlat: 35.647 Ion: loo. 183 5.6kmSand7.SkmWofAIlison (4.14.7mi)931 T.a.~m~e,Ente~ Position E,qr apolat~ ]Track Histories ,I Features ]Team Summ~ Teels RulerCameraWhere is? ZoomDegreesI r-- ~ Age - "~ " f ~-~ ~ ~~ 4 ...... ',I .: :::: I ...... ., ' ~.:!.::~..:i ....::! ' ' __. ' I Field/-~~.q~ .ql 3'53 ,~m:.:..J~7 ~ [.NSSL4 2~224 .[ 06~ ~ [. Coordinator .............................. .: .................... ..................... .......................................... ....... .......................................... ......................... :: .......... FIG. 8. A portion of the computer display from the mobile field coordinator's computer. The map shows intercept team icons (seen asvarious polygons and circles, some with weather information--in color on a live display), weather features (note the box with TOR), roads,places of interest, topographical features, railroads, transmission lines, etc. Team status is indicated with icons at the bottom. Cursor locationand nearest town location appear in the window at the top. The distance between semibold section lines is 1.6 km.tain the most useful measurements of weather phenomena associated with severe thunderstorms. A centralvehicle in the field (the field coordinator) acts as thehub for all communication relating to coordination.This vehicle is equipped with cellular telephones, two45-W VHF radios (ICOM), aircraft radios, and a satellite messaging system (QUALCOM INC. OmniTRACS communication system). The VHF range (typically 8-16 miles) is greatly enhanced (to as much as25-100 miles, depending on altitude) when a repeateronboard an aircraft (i.e., the NOAA NP3W, which wasused during VORTEX) is part of the field operations. In the field, the coordinator can track the locationsof the mobile mesonet vehicles by periodically requesting latitude and longitude from each vehicle via 4550-W VHF radio or cellular phone. These are input intoa portable UNIX-based field coordination computerthat displays high-resolution maps derived from USGSDigital Line Graph data (Fig. 8). With a well-practicedcoordination crew (Erik Rasmussen and Kathy Kanak),the location of 15 vehicles can be obtained and manually input within 90-120 s. Each mobile mesonet vehicle system also uses tone encoding software on a BlueEarth microprocessor to automatically transmit information, such as pertinent weather conditions and GPSposition, over VHF radio to the coordination computer.All mobile mesonet vehicle positions are displayed asicons on the coordination computer maps. The "age"of the position is indicated using a color code. Mapdata includes USAF terrain information, all roads,trails, railroads, towns, landmark, and man-made features. The display system can zoom in and out on mapswith square domains of 0.1--4.0- of latitude and longitude on a side. The software permits real-time triangulation of field reports of phenomena such as wallclouds and tornadoes, tracks and extrapolates positions932 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUlvIE 13of logged weather features, and generates reports usefulfor nowcasting and crew safety.5. Mobile mesonet examplesa. Comparison with Oklahoma Mesonet A comparison of meteorological observations including pressure, temperature, relative humidity, andwind direction and speed between a mobile mesonetunit and the Norman, Oklahoma, Mesonet airport siteshow good agreement. Observations are from 9 September 1994 from 2004 to 2034 CDT (0-30 min inFig. 9). The wind comparison is mostly qualitative asthe Oklahoma Mesonet wind is at 10 m, and the mobileis from 3 m. Nevertheless, the wind directions (Fig.9a) show the same general fluctuations, and 5-min averages agree to within a few degrees for most timesexcept between 20 and 25 min when they differ by 8-10-. This is to be expected considering the height differential between the two anemometers. Similarly,wind speed fluctuation patterns are in agreement, withthe 10-m wind of the Oklahoma Mesonet reading 0.126 1.4 m s-] (typically about 0.6 m s-1) higher thanthe mobile mesonet unit 3 m wind (Fig. 9b). The pressure traces averaged over 5 min agree within the 0.4mb specified inaccuracy of the instruments (Fig. 9c).Temperature trends agree, although the Oklahoma Mesonet is about i-C warmer than the mobile mesonet unit(Fig. 9d). This is probably because the 1) strong insolation at this time and the mobile mesonet unit's temperatures are fan aspirated while the Oklahoma Mesonet' s are not, and 2) inaccuracy of the instruments isas much as 0.15--0.4-C. Finally, 5-min averages ofrelative humidities (not shown) agree to within 1%2% RH.b. Kaw Lake, Oklahoma, tornadic thunderstorm ati~d mesocyclone On 6 May 1994, a supercell moved southwardthrough north-central Oklahoma, producing at least onebrief tornado. North of Pawnee, Oklahoma, two parallel swaths of F1 damage (on the Fujita damage intensity scale) were produced. These appeared to be dueto the mesocyclone intensifying to the surface (thisstorm will be the subject of detailed analysis that willbe described in a later paper). Two vehicles collecteddata in the path of the mesocyclone, and in 'the vicinityof one of the damage swaths. The time series of pressure and equivalent potential temperature, temperatureand humidity, and pressure and wind are shown in Figs.10a-c. Wind barbs for the two vehicles are indicatedon Figs. 10a and 10b for convenience. Several details of these time series are especiallynoteworthy. For the period 1935 to almost 1945 CST,the vehicles were less than 100 m apart, with the cardesignated "P4" due west of "FC". Then P4 departedwestbound, and moved to a position 2200 m due westof FC. The pressure traces (Fig. 10a) from both vehicles show a pressure fall larger than 6 mb in about 4min (1940:30-1944:30 UTC). A "pressure spike:" ofalmost 3-mb amplitude occurred over a period of about30 s just prior to 1945, with the minimum at FC 18 sbefore it occurred at P4. This pressure signature islarger than any ever recorded with TOTO 'Bedard and36O315'270'225'180'135'45WIND DIRECTION a~MMOM5 10 15 2'0 25 30 lIME (MIN) 980.0 ~ 979.5 !~ 979.0 ~ffi~. 978.5' 978.0 - 977.5 ~ 97'/.I ~ 9?8.$ - 976.0 - 9?5.$ - 9~$.0 0PRESSURE MM~ OMWIND SPEED1' ~ OCS 0 5 10 15 20 25 30 ~ME(MIN)~' 28' 27' 26' 25'< ~4' 23' 22' 21' 20TEMPERATUREd$ 10 15 20 25 30 5 1'0 15 2'0 2'5 30 TIME (MIN) TIME (MIN) F~a. 9. A comparison of meteorological measurements (1-min averages) from the Oklahoma Mesonet (OM) and a collocated ~nobilemesonet (MM) unit: (a) wind direction (MM at 3 m and OM at 10 m); (b) wind speed [as in (a)]; (c) pressure; and (d) temperature.OCTOBER 1996 STRAKA ET AL. 933 976 i i [ i , t , i tt~ .~ i ] ~ i * , .......... .... 'i i074 ' ......... F' t,t~ , i ~ [ I ~ i i :-~.~'.1 i i i i i ~ I ~'~ ' ~ ~ i i i . [ [ i i * ] [~ 972 , , ~ , , , , [ i. i [ [ [E , ~ ~ ~ ~ ~~ ~ ~ J). ~ ~ ~~ [ ~ :~.'~ [~ [ ~70 [ r ~[ [~ ~ %x ~.k t ~t ~ ~ ,~ t [ ' ~8 ' '~ i i ~ t ~ i ~ ~ ~ ~ ~ ',I ',~ 7 I~ ',', rl' ',' ', ',',', ] t t t [ , ~ , [ t i ~ I 966 ~ t = ~ I ~ ~ ~ I ~ [ ] ~ I I ~ t t [ 328 1935 1940 1945 19~0 1955 time (CO~ ~xx, trrx rfff f f~' J~ 24~2 ~338 ~ 22 O336~ -3~ ~332 ~ 20330 .~9O8O 18 [ ] [ [ [ [ [ [ ~0 1935 1940 1945 1950 1955 time (CDT) ,-,-,',' - ,. ',.. ' f ~r~"" I 976 974 972Q' 970 968 Ilttltv'.+~. trill* Itllll Illll[I Itllltt Itllllt ,Ill ,~ .... I~ tllllll966 [~][~ 1935 1940302s ~2o~.t5 ~10 FIG. 10. (a) Traces of observed and calculated fields between 1936 and 1955 CDT from two mobile mesonet vehiclesthat collected data in the path of a mesocyclone, and in thevicinity of one of the damage swaths. Observations fromprobe 4 (P4) are indicated with dashed lines, and field coordinator (FC) with solid lines. From 1936 to 1945, P4 was 100m west of FC. From 1945 to 1955, P4 was 2200 m west ofFC. Pressure (thin lines) and equivalent potential temperature(bold lines). One wind barb equals 5 m s-~ and one flag equals25 m s-~. Pressure is not shown for P4 after 1945 as it movedto a location to the west at a different altitude. Note that thepressure spike at FC's location occurred 18 s before than itdid at P4's. Also, equivalent potential temperature increasedto a maximum less than 60 s after the pressure spike. (b) Asin (a). Temperature (thin lines) and relative humidity (thicklines). One wind barb equals 5 m s-~ and one flag equals 25m s-~. Notice the dramatic temperature rise and relative humidity drop immediately after the pressure minimums at theFC and P4 locations (indicated on plots). (c) As in (a). Pressure (thin lines) and wind speeds (thick lines). The peak windsoccurred shortly after the pressure spike. There also appearsto be wind maximums, with the strongest wind recorded atP4's location at about 1948 CDT.Ramzy 1983; Bluestein 1983). It is possible that someof this pressure drop is a result of turbulence, but measurements with the pressure ports for turbulent flow of20-36 m s-~ should be in error by at most 0.2 mb.Furthermore, two separate, nearby vehicles, measurednearly the same pressure drop. Interestingly, equivalentpotential temperature increased as the pressure minimum approached, rising nearly 8 K in about 3 rain. Themaximum occurred about 1 min after the pressure minimum and was near the "ambient" values in thestorm's warm inflow. Later, the equivalent potentialtemperature decreased to values more representative ofthe storm's cold pool. Other details, which never have been observablein mesocyclones before (notice that these would notbe recorded at temporal resolutions less than about30 s), can be seen in the temperature and humiditytraces (Fig. 10b). The temperature fell for severalminutes prior to the approach of the pressure minimum, then rose for about 2 min. Just as the minimumpassed, the temperature trace represents a stepchange of larger than 2-C. The temperature remainedelevated for about 2 min, and then fell back towardcold-pool values. The relative humidity rose until thepressure minimum, reflecting first cool, moist outflow, and then nearly saturated inflow air. Just afterthe pressure minimum passed, the humidity fell veryrapidly by more than 10%. It then rose to over 90%as conditions returned to those representing the coldpool. The wind data (Fig. 10c) reveals strong peaks (>30m s-t) within a few seconds of the pressure minimum.A second period of gustiness can be seen in the tracesfrom both vehicles at around 1948 CST, perhaps inassociation with the passage of the edge of the storm'scold pool. The traces are very similar prior to 1945,934JOURNAL OFATMOSPHERIC ANDOCEANIC TECHNOLOGYVOLUME 1343,72527/Research 1/M MData/26Mav94/maoFilesS/OO35.mrn1.6 km FIG. 11. Mobile mesonet data -+2 min from 1935 CDT 26 May 1994. Observations each 1 min are plotted for moving vehicles, and 1935CDT observations for stationary vehicles. Distance scale is indicated with arrow at bottom. The station model has temperature (-F) in theupper left, dewpoint (-F) in the lower left, equivalent potential temperature minus 300 K in the upper right, and vehicle I.D. in the lowerright. A half wind barb equals 5 kt and full wind barb equals 10 kt. The heavy solid line indicates the primary wind shift; a north-southelongated updraft base was near this line as noted from visual observations. The heavy dashed line is a secondary wind shift. Only weakoccasional cloud-base rotation was observed with this storm.despite the fact that the vehicles were about 200 mapart. The peak average (6 s) wind speeds measuredat 3-m height were around 35 m s-l, and peak 2-swinds approached 38 m s-~. Widespread F0-F1 winddamage occurred near these vehicles, including aboutone dozen large power poles being broken near groundlevel. Although much analysis work remains, the detailsin these data seem to indicate that the mesocyclonewas encircled with by three identifiable air masses:OCTOBER 1996 S T R A K A E T A L. 935inflow, cold pool, and a unsaturated downdraft thatpresumably was dynamically forced. Details such asthese could not be obtained without the ability toposition sensors with adequate temporal responseand resolution.c. Lubbock, Texas, hailstorm As a final illustration data are shown for a nearlystationary supercell hailstorm that occurred near Lubbock, Texas, on 26 May 1994 (Fig. 11). This stormproduced 10-18 cm of rain, and hail 0.5-3.5 cm indiameter that accumulated to depths of 15 cm. The datashown are 1-min observations between 0033 and 0037UTC for moving vehicles, and single observations at0035 for stationary vehicles. Preliminary analyses ofimportant storm-scale frontal boundaries have beenadded. This storm had a mesocyclone in middle levels,whereas the lower-level circulation did not develop.The mobile mesonet revealed at least one unexpectedfeature: a region of strong low-level divergence locatedjust west through southwest of the main updraft andmesocyclone. There are suggestions in the data thatdownburst activity was occurring in this divergence region. Present in this mapping of mobile mesonet dataare remarkable details of the low-level structure of thissupercell storm.6. Summary A 15-unit mobile mesonet for finescale meteorological measurements has been developed and tested. Themobile mesonet was designed to augment existing meteorological networks in the study of severe localstorms and other mesoscale weather phenomena in conjunction with VORTEX field program. Special carewas taken to contend with difficult atmosphere-instrument coupling problems. The system does not seem tobe vehicle-type sensitive when considering standard sedan automobiles (three different types of sedans havebeen used with comparable performance from each).Each mobile mesonet measures pressure, temperature,relative humidity, and wind. A Global Positioning System (GPS) receiver and a flux-gate compass onboardeach vehicle obtains universal time, position (latitude,longitude, altitude), vehicle heading, and vehiclespeed. A standard laptop computer stores data, computes derived variables, and provides real-time datadisplay. Instrument compatibility with the OklahomaMesonet allows for high-quality instrument calibrationand maintenance. The total equipment cost for eachunit including a laptop computer is on the order of$7,500. The cost for labor, tools, and calibration facilities is not known. The performance of the mobile mesonet was demonstrated through comparisons with data from anOklahoma Mesonet site and with examples of data collected during VORTEX. These showed that valuablefield measurements can be made from both stationaryand moving mobile mesonet units. An improvement being made to the existing systeminvolves refining the moving wind observation techniques and calculations. Only one GPS position is usedevery 6 s to compute the wind speed and direction. Thisrequires that the vehicle does not accelerate (at least onaverage) during this period for viable measurements.Wind calculations might be improved by incorporatingmore GPS information, which is available every 1 swith the receiver being used. Tests continue toward thiscause using higher-resolution GPS information in various manners) Software limitations and data acquisition timing are limiting factors. Another problem thatneeds additional attention is determining representativedewpoint temperatures and coupling thermodynamicsmeasurements with different time responses. Finally, afuture addition to the system might be a precipitationoccurrence sensor. Care, though, would be needed todelineate precipitation from water spray from vehiclesand the road. Acknowledgments. We gratefully acknowledge PaulGriffin and Dennis Nealson at the National SevereStorms Laboratory for help designing, constructing,and maintaining the mobile mesonet. Thanks also aredue to Keith "Casey" Crosbie, Ken Eack, Frank Gallagher, Jane Hornbrook, Kathy Kanak, Scott Richardson, Yvette Richardson, and Tom Sheperd for graciously helping with instrument calibration, systemstests, and last minute construction details; SoniaLasher-Trapp for providing data analysis assistance;and Dr. Fred Brock and Dr. Ken Crawford at the Schoolof Meteorology of the University of Oklahoma andOklahoma Climate Survey for use of the OklahomaMesonet facilities, and invaluable expert consultation.Scott Richardson helped with the preliminary design ofthe aspirated weather shield. In addition, Doug Forsythe; Jim Moore; and Drs. William Beasley, KelvinDroegemeier, Robert A. Maddox and Steve Nelsonprovided much encouragement and support. We alsothank Ms. Kathy Kanak, Dr. Robert Davies-Jones, andDr. Robert Maddox for reviewing and editing early versions of this paper and helping in many ways duringthe project. Without all of these people and many others, this project would not have been possible andbrought to completion. Finally, we are very appreciative of insightful reviews from Jim Moore, Dr. RogerWakimoto, and anonymous reviewers. The mobile mesonet was made possible through funding from the National Oceanic and Atmospheric Administration, National Severe Storms Laboratory, University ofOklahoma Graduate College and School of Meteorology, and National Science Foundation under Grant ~ The mobile mesonet now (as of 1995) uses samples of about 1 sfor typical operations.936 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 13ATM91-20009 to the Center for Analysis and Prediction of Storms at the University of Oklahoma.REFERENCESBarnes, S. L., 1978: Oklahoma thunderstorms on 29-30 April 1970. Part 1: Morphology of a tornadic storm. Mon. Wea. Rev., 106, 673 -684., 1981: SESAME 1979 User's Guide. NOAA, 223-226.Bedard, A. J., Jr., and C. Ramzy, 1983: Surface meteorological ob servations in severn thunderstorms. Part I: Design details of TOTO. J. Climate Appl. Meteor., 22, 911-918.Bluestein, H. B., 1983: Surface meteorological observations in severe thunderstorms. Part II: Field experiments with TOTO. J. Cli mate Appl. Meteor., 22, 919-930.Brock, F. V., and S. Fredrickson, 1993: Oklahoma mesonet data qual ity assurance. Extended Abstracts, Eighth Symp. on Meteoro logical Observations and Instrumentation, Anaheim, CA, Amer. Meteor. Soc., 311-316.--, G. H. Saum, and S. R. Semmer, 1986: Portable Automated Mesonet II. J. Atmos. Oceanic Technol., 3, 573-582.--, G. Lesins, and R. Walko, 1987: Measurements of pressure and air temperature near severe thunderstorms: An inexpensive and portable instrument. Extended Abstracts, Sixth Syrup. on Mete orological Observations and Instrumentation, New Orleans, LA, Amer. Meteor. Soc., 320-323.--, K. C. Crawford, R. L. Elliot, G. W. Cuperus, S. J. Stadler, H. L. Johnson, and M. D. Eilts, 1995: The Oklahoma Mesonet: A technical review. J. Atmos. Oceanic TechnoL, 12, 5-19.Brown, J. M., and K. R. Knupp, 1980: The Iowa cyclonic-anticy clonic pair and its parent thunderstorm. Mon. Wea. Rev., 108, 1626-1645.Crawford, K. C., F. V. Brock, R. L. Elliot, G. W. Cuperus, S. J. Stadler, H. L. Johnson, and C. A. Doswell III, 1992: The Oklahoma Mesonetwork--A 21st century project. Preprints, Eighth Int. Conf Interactive Information and Processing Sys tems for Meteorology, Oceanography, and Hydrology, Atlanta, GA, Amer. Meteor. Soc., 27-33.Davies-Jones, R. P., and E. Kessler, 1974: Tornadoes, Weather and Climate Modification. W. H. Hess, Ed., Wiley and Sons, 552 595. , and H. E. Brooks, 1993: Mesocyclogenesis from a theoretical prospective. The Tornado: Its Structure, Dynamics, Prediction, and Hazards. Geophys. Monogr., No. 79, Amer. Geophys. Un ion, 105-114.Fredrickson, S. E., 1993: National Severe Storms Laboratory mobile atmospheric observatories; Surface meteorological measure ments. Extended Abstracts, Eighth Symp. on Meteorological Observations and Instrumentation, Anaheim, CA, Amer. teor. Soc., 219-224.Gill, G. C., 1983: Comparison testing of selected naturally ventilated solar radiation shields. Report submitted to the NOAA Data Buoy Office, Bay St. Louis, MO.Goodman, S. J., and K. R. Knupp, 1993: Tornadogenesis via squall line and supercell interaction: The November 1, 1989, Hunts ville, Alabama, Tornado. The Tornado: Its Structure, Dynamics, Prediction, and Hazards. Geophys. Monogr., No. 79, A~ner. Geophys. Union, 183-199.Klemp, J. B., 1987: Dynamics of tornadic thunderstorms. Annu. Rev. Fluid Mech., 19, 369-402.Nishiyama, R. T., and A. J. Bedard Jr., 1991: A "quad-disk" s'Latic pressure probe for measurements in adverse atmospheres with a comparitive review of static pressure probe designs. Rev. Sci. lnstrum., 62, 2193-2204.Rasmussen, E. A., J. M. Straka, R. P. Davies-Jones, C. E. Doswell III, F. Carr, M. Eilts, and D. R. MacGorman, 1994: Verification of the Origins of Rotation in Tornadoes Experiment: VORTEX. Bull. Amer. Meteor. Soc., 75, 1-12.Richardson, S. J., 1992: Automated temperature and relative fimnid ity calibrations for the Oklahoma Mesonetwork. M.S. thesis, School of Meteorology, University of Oklahoma, 65 pp. , 1995: Automated temperature and relative humidity calibra tions for the Oklahoma mesonetwork. J. Atmos. Oceanic. Tech nol., 12, 951-959. , and F. V. Brock, 1993: Temperature and relative humidity probe calibration and intercomparison for the Oklahoma sonet. Extended Abstracts, Eighth Symp. on Meteorological Ob servations and Instrumentation, Anaheim, CA, Amer. Meteor. Soc., 281-286.Rust, W. D., 1989: Utilization of a mobile laboratory for stom elec tricity measurements. J. Geophys. Res., 94, 13 305-13 3 l 1. , R. P. Davies-Jones, D. W. Burgess, R. A. Maddox, L. C. Show ell, T. C. Marshall, and D. K. Lauritsen, 1990: Testing of a mobile version of a Cross-chain LORAN Atmospheric Sound ing System (M-CLASS). Bull. Amen Meteo~r. Soc., 71, 173 180.Tepper, M., and W. E. Eggert, 1956: Tornado proximity 'traces. Bull. Amer. Meteor. Soc., 37, 152-159.Ward, N. B., 1964: The Newton Kansas tornado cyclone of May 2,4, 1962. Preprints, llth Weather Radar Conf, Boulder, CO, A~ner. Meteor. Soc., 410-415.Woo, H. G. C., J. E. Cermak, and J. A. Peterka, 1989: Calibration of atmospheric pressure sensor quad-disk probe. CSU Project Re port 2-9 7960. Fluid Mechanics and Wind Engineering Program and Fluid Dynamics and Diffusion Laboratory. Dept. of Civil Engineering. Colorado State University, Fort Collins, CO.