1911Bulletin of the American Meteorological Society
1. Introduction
As the twentieth century ended, considerable atten-tion had been focused on reducing the losses fromnatural disasters. To enhance disaster awareness, theUnited Nations declared the 1990s to be the Interna-tional Decade for Natural Disaster Reduction (Na-tional Research Council 1991). Mitigation efforts,such as Project Impact operated by the Federal Emer-gency Management Agency, have illustrated the im-portance of involving stakeholders in public safetyfrom federal agencies through the local communitylevel. Access to quality information and adequatetraining for local officials are required for disastermitigation programs to be successful. These resourcesalso address the other emergency management func-tions of preparedness, response, and recovery (Na-tional Research Council 1991, 1999).
Although the general weather information needs ofemergency managers and other public safety decisionmakers are well known, only limited progress has beenachieved to provide specific, tailored, and real-timeweather information to these users. Several panels(e.g., U.S. Weather Bureau 1964; National Research
Council 1980) concluded that the “link between me-teorological service and the users of weather informa-tion was markedly deficient.” Many examples havepointed to communication failures between the meteo-rological community and local decision makers thatdealt with weather hazards like floods [National Oce-anic and Atmospheric Administration (NOAA) 1977,1991], tornadoes (NOAA 1994, 1997), and severe heat(Changnon et al. 1996). In most cases, the NationalWeather Service (NWS) produced excellent forecasts;yet, local decision makers either did not receive or didnot act upon the NWS information. In other cases, theNWS could not make timely warning decisions be-cause they did not receive critical information fromlocal communities. In short, adequate data communi-cation and dissemination systems designed for localofficials—and that are affordable—have not beenavailable. Therefore, decisions affecting the safety andprotection of citizens have been made without the ben-efit of critical and local information.
Based upon our perception of a near-complete lackof real-time weather information available to localpublic safety agencies, the Oklahoma ClimatologicalSurvey (OCS) began an initiative in 1996 known asOK-FIRST. Demonstration funding was provided bya telecommunications grant from the U.S. Departmentof Commerce. The goal of OK-FIRST was to developa transportable, agency-driven information and sup-port system to help public safety agencies to harnessthe information age.
OK-FIRST: A MeteorologicalInformation System for Public Safety
Dale A. Morris, Kenneth C. Crawford, Kevin A. Kloesel, and J. Michael WolfinbargerOklahoma Climatological Survey, University of Oklahoma, Norman, Oklahoma
The Oklahoma Climatological Survey has supported local public safety agencies (fire departments, law enforce-ment agencies, and emergency management) through its OK-FIRST program since 1996. OK-FIRST provides real-timeuse of weather data to help public safety agencies fulfill their respective operational missions. Users are taught how tointerpret and apply weather data in their operations. The OK-FIRST system has been applied during a wide variety ofweather events, including severe weather, floods, wildfires, and hazardous materials incidents.
The lack of weather information at the communitylevel results from near-zero budgetary support for ru-ral emergency management agencies (most have onlyone funded or part-funded position and cannot affordcommercially available data feeds); an inadequate in-formation infrastructure (modern personal computersand access to the Internet); and local officials whooften have very limited knowledge of the availableweather information and even less experience in us-ing modern information. Thus, the OK-FIRST pro-gram was conceived to include critical elements toaddress these three deficiencies. First, OK-FIRSTwould provide access to localized, real-time weatherinformation that consisted of data from the OklahomaMesonet (Brock et al. 1995), from regional WeatherSurveillance Radar-1988 Doppler (WSR-88D) unitsthrough a unique Next Generation Weather Radar(NEXRAD) Information Dissemination Service(NIDS; Baer 1991) data-sharing arrangement, andfrom the NWS and other agencies. This access wasdesigned for fire, law enforcement, and emergencymanagement officials. Additional elements includedinstruction on the interpretation and use of the avail-able data and follow-up support on a continual basis.
As a result of OK-FIRST, Oklahoma’s publicsafety personnel have reshaped their work routineswhen dealing with weather-impacted situations.Rather than simply reacting to weather situations, lo-cal officials are proactive and predeploy resources.Local communities have benefited, as OK-FIRST par-ticipants have become local focal points for weatherinformation in a wide variety of situations. The emer-gency manager for the city of Moore, Oklahoma,stated, “the ultimate benefactor of [our response to arecent weather event] was, of course, our citizens andthose traveling through, as we had only a very smallamount of vehicle accidents during the entire event.”Another participant used information from OK-FIRSTthat “was critical to provide documentation of the cur-rent drought. A state [emergency] declaration forLatimer County resulted.” Senior citizens have benefitedbecause caregivers are informed about dangerous heatand wind chill indices. OK-FIRST information alsohas been applied forensically. In one case, data fromOK-FIRST disproved a murder suspect’s alibi, result-ing in an arrest and confession.
This manuscript documents the components of theOK-FIRST program, provides an assessment from anindependent evaluation, and notes improvementsmade by rural communities in how they responded toweather-impacted situations. Section 2 details the early
history of OK-FIRST. Section 3 documents the datadissemination system of OK-FIRST, including itsWeb-based decision-support system. Section 4 high-lights the user-support components of OK-FIRST byfocusing on the instructional workshops. In section 5,the reader will find details from an independent evalu-ation of OK-FIRST and about the impact OK-FIRSThas had in local communities across Oklahoma. Theimportance of OK-FIRST to the citizens of Oklahomais summarized in section 6.
2. History
The availability of real-time weather hazard infor-mation historically has been nonexistent to publicsafety officials, especially those outside of major metro-politan areas. However, the Oklahoma Mesonet (Brocket al. 1995), developed as a multipurpose, mesoscale,observing network, addressed this deficiency by provid-ing real-time data to a host of users across Oklahoma.Potential user groups targeted from the Mesonet’s in-ception included agricultural interests, K–12 educa-tion, the meteorological community, and public safety.As the Mesonet was implemented, OCS initiated anoutreach program for math and science teachers in1992 known as EARTHSTORM (McPherson andCrawford 1996). This K–12 project initially providedcomputers and Mesonet data to 50 teachers acrossOklahoma using a dial-up bulletin board system andcustom software. Completion of a 4-week educationalinstitute was required so that teachers would developthe necessary computer skills and knowledge of howto use Mesonet data in the classroom.
As the WSR-88D network became operational inthe 1990s, the NWS implemented the NIDS conceptusing four private vendors. As a result of market forcesand costs incurred by the NIDS vendors, many localpublic safety agencies were unable to purchase WSR-88D data due to budgetary constraints. Even so, localemergency managers often assumed the role of storm-spotter coordinator [see the description of the inte-grated warning system for severe weather provided byDoswell et al. (1999)], and became frustrated know-ing that safe and efficient spotter coordination couldbe facilitated through access to real-time weather ra-dar data, if they could only afford it. Thus, radar datahad to be the cornerstone of a weather informationsystem for public safety agencies in Oklahoma.Because of the previous success in rural data dissemi-nation using the Mesonet and EARTHSTORM, op-
1913Bulletin of the American Meteorological Society
portunities surfaced for OCS to redistribute NIDS datafrom 15 radars around Oklahoma at reduced subscrip-tion rates (Crawford et al. 1999). With a capability toprovide radar data, Mesonet data, and additional prod-ucts from the NWS and other agencies, OK-FIRSTwas built upon the foundation of experience gainedfrom the Mesonet and EARTHSTORM.
3. Data dissemination
OK-FIRST disseminates weather data to users viaa comprehensive Web site. This site was designed anddeveloped based upon information needs expressed bylocal officials in Oklahoma and from results of a na-tional survey of emergency managers by NOAA’sForecast Systems Laboratory (e.g., Kelsch 1996).These findings indicated that an information systemfor emergency managers should include radar data,NWS watches and warnings, other observations suchas satellite data, local observations of rainfall, winds,and other surface parameters. Because NWS watchesand warnings have been widely available to publicsafety users via a variety of sources, OK-FIRST di-rected most of its resources into the delivery of radardata and Mesonet data.
a. Data sharing strategyOK-FIRST required an infrastructure to efficiently
deliver real-time weather information to a host of ru-ral users across Oklahoma. The strategy to shareMesonet and NIDS data was to deliver “raw,” or un-processed, data files via the Web so each user coulddisplay the data using custom software and popularWeb browsers (Netscape Communicator or InternetExplorer). These browser “plug-ins” (Wolfinbargeret al. 1998a,b) reduced server overhead because theweather data files did not have to be preconverted to adisplayable format (e.g., GIF images) at a central fa-cility before distribution.
Several benefits accrued from this strategy. First, thecentral server could process requests from many moreusers at once. This strategy is very important when oneconsiders that, during a typical day with precipitation,∼90 000 distinct radar products are received at OCSfrom the 15 radars around Oklahoma. While the vol-ume of Mesonet data products remains small, Mesonetdata are provided in a multitude of display formats.Providing complete control of the graphical display onthe end-user machine allows the data-disseminationsystem to be more scalable (i.e., more combinations
of Mesonet data displays are possible and additionalWSR-88D radars or products can be distributedthrough the server without substantially increasing theprocessing requirements at the central site). In addition,software residing on the user’s computer permits a highlevel of interactivity with the data (e.g., animation,querying of data values, zooming, and scrolling) andmaintains intuitive data access via the Web. Moreover,the plug-in software and accompanying Web site soft-ware gives each user the flexibility to have a custom-ized set of display preferences for geographic overlays,default radars, and Mesonet parameters.
b. Incorporation of user feedbackOne objective of OK-FIRST was to create an
“agency-driven” information system. In other words,the system was to be initially designed and subse-quently refined based upon continual input from us-ers to ensure that OK-FIRST would be functional tomeet their needs. This goal was achieved through thefrequent incorporation of user feedback into subse-quent software releases and accompanying Web pages.For example, the first release of the plug-in softwaresimply displayed (Fig. 1a) the NIDS product knownas composite reflectivity [CREF; see Klazura and Imy(1993) for details]. A separate text file of the storm at-tributes was also available (Fig. 1b), although these at-tributes are embedded within the CREF product. Thus,users manually had to match a given storm attributewith its reflectivity signature to determine if the stormattribute information applied to the user’s location. Inother words, the manual process of matching storm V8in Fig. 1b with its location east of the radar site inFig. 1a was difficult for some users. Thus, several us-ers inquired if an easier method could be developedto use the storm table. A later build of the plug-in soft-ware integrated the display of the storm attributes withthe CREF radar image via a “hot” cursor-tracking fea-ture. A graphical overlay of the attributes provided de-tailed information as the cursor was moved over aspecific storm centroid (Figs. 1c,d).
Numerous other improvements have been made.These include better rendering of the image data andgeographic overlays. The software also was extendedto produce time series graphs of Mesonet data on-the-fly, cursor-tracking, interactive legends, and easiermethods to format and present Mesonet data.
c. OK-FIRST Web site contentAs the plug-in display software evolved, so did the
Web pages that supported the plug-ins and other data
1914 Vol. 82, No. 9, September 2001
sources. The OK-FIRST Web site (Fig. 2) organizeddata into four categories. First, the NIDS content areaprovided intuitive access to the entire collection ofNIDS products from the 15 WSR-88D units aroundOklahoma. This unique approach is a major benefit toOK-FIRST: namely, that analysis and use of radar dataoccurs using a network approach. For example, a pub-lic safety official whose area of responsibility is lo-cated midway between Oklahoma City and Tulsa,Oklahoma, can view their area using two or more ra-dar units (e.g., Fig. 3). The advantage is twofold. Thepublic safety official can corroborate image informa-tion between multiple radars, and the user can miti-gate those rare but real-life situations when a givenWSR-88D unit might require maintenance at the mostinopportune moment. For example, during the tornadooutbreak of 3 May 1999, the WSR-88D unit near theTulsa area required maintenance. Using OK-FIRST,users in northeast Oklahoma could monitor their ar-eas using the radar units located in Oklahoma City;Fort Smith, Arkansas; Springfield, Missouri; andWichita, Kansas.
The second content area focuses on data from theOklahoma Mesonet, which is presented in the form ofmaps and time series graphs. Maps are dynamic sta-tion plots, line contour plots, and/or color-filled con-tour plots of measured or calculated parameters (e.g.,temperature, humidity, wind chill index, rainfall, orwinds). These maps are important for users to discernsignificant weather patterns for the development ofsevere weather or to alert firefighters to impendingwind shifts. Time series graphs present daily summa-ries of Mesonet data for given locations. This type ofinformation has been used in forensic investigations(i.e., airplane crashes and homicides). Other static plotsof Mesonet data, like long-term rainfall accumulationsand daily temperature extremes, are also available.
FIG. 1. Evolution of plug-in software capabilities (displaysshown use the composite reflectivity product from the WSR-88Dnear Oklahoma City at 2151 LT on 4 Oct 1998; range rings are30 n mi): (a) display using version 3.0 of plug-in software; (b) cor-responding storm attribute table; (c) display using version 6.0 ofsoftware (arrow represents cursor on storm V8, information aboutstorm V8 appears in legend); (d) storm attributes displayed inde-pendently from the radar image plus a 1-h storm projection (circlesrepresent hail algorithms, squares represent the WSR-88Dmesocyclone detection algorithm, triangles represent tornado de-tection algorithm).
1915Bulletin of the American Meteorological Society
4. Instructional workshops, usercertification, and follow-up support
Before gaining access to the OK-FIRST Web site,each participant is required to complete a weeklongworkshop. The workshop establishes a baseline ofskills in computer literacy and the application of mod-ern weather data in the decision-making process.Workshop participants are selected competitively; anarrative is required about the local community andthe applicant’s reasons to participate. Geographic di-versity is also a factor in selecting agencies. A prefer-ence is given to rural participants (65% of thecommunities represented have populations of less than10 000 and 49% have fewer than 5000 residents).
The Web site supports the workshops through ac-cess to extensive project information and instructionalmaterials. These materials include reference materialson basic meteorology and climatology, severe weather,
Access to NWS text products and graphical prod-ucts is achieved through links to various serversaround the United States [including a primary NWSserver in Washington, D.C., the Storm Prediction Cen-ter (SPC) in Norman, Oklahoma, and the Arkansas-Red Basin River Forecast Center (ABRFC) in Tulsa,Oklahoma]. These servers provide watches, warnings,forecasts, and discussions of current and forecastweather. Among OK-FIRST participants, the Convec-tive Outlook Graphic produced by the SPC is a high-demand product.
A fourth section provides access to “other” datasources that include derived products from the Okla-homa Mesonet. These value-added Mesonet productsinclude the Oklahoma Rainfall Update, used duringrainy episodes and for real-time drought monitoring,and the Oklahoma Fire Danger Model (Carlson andEngle 1998). This model is based upon the normal-ized difference vegetation index derived from Ad-vanced Very High Radiation Radiometer satellite data,land use information, and Mesonet weather conditions.Updated many times per day, this model provides de-tailed maps of overall fire danger, fire intensity, rateof fire spread, and the Keetch–Byram Drought Index(KBDI). The other category also links to other usefulweather-related Web sites.
These data sources are also organized according tosix specific weather hazards (Fig. 4). For example, asection on severe weather includes links to outlooksfrom the SPC, while the flooding section includesoutlooks and guidance products from the Hydrometeo-rological Prediction Center (HPC), the ABRFC inTulsa, and the Climate Prediction Center. Thus, theuser can easily determine the current weather threat.
A decision-support component of the Web siteblends instructional material (section 4) with severalcategories of current data sources. The products on thedecision-support pages are organized by their per-ceived utility before or during an event (Fig. 5;Table 1). Products of use during a specific event arefurther stratified by specific subthreats (e.g., radarproducts for hail detection are base and composite re-flectivity and vertically integrated liquid). These prod-ucts are displayed dynamically based upon the user’spreferences. Annotated examples of specific signaturesare included to assist with data interpretation of drylines, cold fronts, hail cores, tornadoes, floods, highfire danger (Fig. 6), and other hazards. In other words,the OK-FIRST site integrates current weather datawith online training materials appropriate for specificweather hazards.
FIG. 2. Illustration of the “current data” section of the OK-FIRST Web site. Content areas are accessed using the four but-tons at the top left of the screen (Radar, Mesonet, NWS, andOther). The display is a plug-in image of vertically integrated liq-uid from the WSR-88D at Frederick, OK, at 1755 LT on 3 May1999. In the radar section, users can choose any of the NIDS prod-ucts (buttons on the lower-right panel under the radar image) fromany of 15 radars (map on the lower-left panel). Bottom panel il-lustrates how users customize their geographic overlays.
1916 Vol. 82, No. 9, September 2001
floods, fire weather, and a comprehensive set of datafrom 10 case studies.
During the workshops, “hands-on” computer labo-ratory exercises are integrated with lectures (Fig. 7).Typically, lectures introduce concepts while labora-tory exercises reinforce the lecture material by usingactual radar signatures, for instance. During these ex-ercises, project staff assist participants when they en-counter difficulty. Project staff also are availableduring the open computer lab during most eveningsof the workshop. This one-on-one instruction withmany of the participants produces additional confi-dence using the OK-FIRST system. The laboratoryexercises are based upon data from actual weatherevents. Four cases deal with severe weather (a classictornado outbreak on 26 Apr 1991; a major hail andflood event in Dallas–Fort Worth, Texas, on 5 May1995; a major hail and wind event in northern Okla-homa on 17 Aug 1994; and a tornado and hail eventon 2 Apr 1994). Additional case studies focus onfloods, on winter weather, on wildfires using the Okla-homa Fire Danger Model, and on a significantheatburst and wind event (22 May 1996). All labora-tory exercises and other instructional materials (in-cluding a cross-linked glossary) reside on the Web site
so participants can refresh their memory when theyreturn to their local jurisdictions.
Throughout OK-FIRST, training workshops be-yond the initial institutes were deemed to be essential.Two-day refresher workshops have been conductedperiodically since the first weeklong workshop. Therefresher courses discussed updates to the software andthe Web sites in addition to providing new materialson data interpretation. Refresher workshops provideduser feedback, which was incorporated into future soft-ware releases, Web page updates, and training regi-men. In addition, intensive 2-day workshops for“assistants,” who work and volunteer in trained OK-FIRST agencies, covered basic data interpretations.
Because of the technical nature of the data, thetraining, and the frequent updates to the Web site andassociated software, a user certification policy wasdeveloped based upon near-unanimous feedback fromparticipants. The policy required users to be certifiedthrough completion of the weeklong course, and tomaintain their certification through the refresher work-shops. User accounts are terminated if a user does notparticipate in a refresher course in an 18-month period.Since the policy was implemented, 91% of the usershave maintained their certification. Of those who have
FIG. 3. Illustration of the “two-screen mode” available in OK-FIRST, which allows users to display products from any of the con-tent areas. Both images are from 4 Oct 1998: on the left, base reflectivity from Oklahoma City and, on the right, the correspondingimage from Tulsa.
1917Bulletin of the American Meteorological Society
left the program, most are due to attrition issues notrelated to OK-FIRST (e.g., retirement, job reassignment).
Follow-up support for participants is an importantcomponent of the program. This support includes help-ing individuals to interpret imagery between workshops,to ensure that the most recent software updates have beendownloaded and installed, and to solicit user input priorto the next generation of improvements to the OK-FIRSTsystem. In addition, user success stories and criticismshave been collected and data requests have been filledto assist participants in documenting specific events.
5. Program evaluation and impact
A formal program evaluation was included duringthe first three sets of weeklong and refresher work-
shops. An independent evaluator determined theeffectiveness of the instruction and the impact OK-FIRST had on participants’ work habits. The instruc-tion was evaluated through the use of pre- andpost-tests, while focus groups provided anecdotaldocumentation of the program’s impact. Anecdotes(not part of the formal evaluation) also were collectedinformally by project staff. The project evaluator
FIG. 4. OK-FIRST Web page used during flood threats. Userscan choose from any of the six “threats” by clicking on the drop-down box in the upper right. The product shown is the Day 1Quantitative Precipitation Forecast from the HydrometeorologicalPrediction Center. The image expands to full screen when clicked.
FIG. 5. Example of a decision-support page for use during se-vere weather. Products are stratified by perceived utility beforeand during an event. They appear in a new browser window whenchosen. The products “during an event” are further classified bya specific “subthreat” (hail, tornadoes, high winds). Interpretationguides for many products are linked for side-by-side comparisonwith the most current edition of the chosen product.
1918 Vol. 82, No. 9, September 2001
(James et al. 2000) concludedthat “OK-FIRST staff were ableto enhance significantly theknowledge and skills of theproject participants in a veryshort period of time. This iseven more impressive given thevery technical nature of thematerial and the fact the partici-pants did not come from techni-cal backgrounds.” The formalevaluation provided a quantita-tive analysis of the effectivenessof the instruction during theworkshops. For example, theevaluator provided a question-by-question breakdown plussummarized results by contentarea. Evaluation results were re-leased periodically to projectstaff so midcourse improve-ments in instructional strategiescould be made. The evaluatoralso collected information fromparticipants at a refresher work-shop that involved all threeclasses. This investigated thesupport provided to the users byOK-FIRST staff. The evaluationdetermined that 98% of the par-ticipants were very satisfiedwith the OK-FIRST Web site.The participants also had over-whelmingly positive responsesabout training and support (95%said that the training had helpedthem a great deal and 97% re-ported that they were very orsomewhat satisfied with ongo-ing project support).
In addition to the formalevaluation, project staff alsoroutinely examine Web site us-age statistics. These numbersconsist of the total number ofdownloaded data files classifiedby radar data and Mesonet data.Between 1998 and 2000, the an-nual number of Mesonet filesdownloaded by OK-FIRST par-ticipants are 62 628, 126 731,
Severe weather (before the event)
SPC day 1 and day 2 outlooks Example
Surface temperature and dewpoint contours Thunderstorm initiation
Equivalent potential temperature Thunderstorm initiation andpropagation
Base reflectivity Dry lines; outflow boundaries;intersections of boundaries
1919Bulletin of the American Meteorological Society
function, type of Internet access,and the weather patterns. Forexample, several participantshave been able to acquire fasterInternet access over the pastseveral years, allowing them toaccess the OK-FIRST Web sitemore frequently. For this andother reasons, product usagetends to increase from year toyear. This increase also resultsfrom adding participants eachyear and to increasing appli-cation of the weather data toon-the-job challenges by theparticipants.
The Web site usage statisticscan also be stratified accordingto each user to reveal product us-age patterns or an apparent levelof interest in using OK-FIRST.For example, all but 24 of theregistered users had more than5000 accesses during the year2000. Of the users who accessedthe site more infrequently, 10were granted access in Decem-ber during a weeklong data inter-pretation workshop. Most ofthe remaining 14 users had notattended a recent refresher work-shop, underscoring the impor-tance of continuing educationfor a support program like OK-FIRST.
Negative criticism of OK-FIRST has been very minor.The most vocal critics havebeen agencies that were deniedaccess to the program. Althoughthese denials resulted from adesign criterion to enforce geo-graphic diversity and to elimi-nate overcrowded workshops, afew agencies were disturbedwhen they were rejected and anearby agency was accepted.
Many of the agencies that were originally denied ac-cess were subsequently approved to participate in laterworkshops as the geographic diversity requirementwas slightly relaxed.
Winter weather (before the event)
SPC mesoscale discussion Heavy snow example
Observed thickness and 850-mb Snow and other winter precipitationtemperature
Winter weather (during the event)
Base reflectivity Snow in clear-air mode;distinguishing precipitation type
Upper-air soundings Freezing rain example; snow example
Regional temperature and wind chill Wind chill chart
Flooding (before the event)
HPC quantitative precipitation forecasts Example
HPC quantitative precipitation discussion Contractions used in discussions
Mesonet temperature and heat index Heat index chart
TABLE 1. Continued.
Product Interpretation guides
and 366 732. Corresponding totals for radar data are936 448, 1 404 988, and 1 959 144. Some care mustbe taken when interpreting these figures, becausemany factors influence them, including participant job
1920 Vol. 82, No. 9, September 2001
a. Overall impactThe overall attitude and response by the program
participants can be summarized by the following:
• “I’m proactive now rather than just reactive.”• “This is the most fantastic tool emergency manag-
ers have ever had.”• “Each community can have access to data specifi-
cally aimed at their community or region. This al-lows the emergency manager to check his ownspecific area while the NWS and TV stations areconcentrating their coverage on other areas of thestate.”
• “I don’t know what I’d do without it. I’d probablyresign. It’s that important.” (H. Gunter, The DailyOklahoman, 11 May 1999).
Specific benefits with respect to type of weatherevent include the following.
b. Severe weatherUser comments about the use of OK-FIRST dur-
ing severe weather highlighted more efficient use ofresources:
• “I can look at OK-FIRST and figure out where Ineed to send spotters rather than just scatteringthem out across the county and hoping that theyare in the right places.”
• “The number of times storm spotters are activatedhas been drastically reduced, and when they are ac-tivated, it is for a shorter duration. Also, fewer spot-ters are needed.”
Comments collected after specific events highlightspecific actions that informed personnel could make.
• “One spotter was assigned to a location west ofMoore. As the storms moved in, our spotter coor-dinator decided—due to her OK-FIRST display—to move the spotter a couple of miles south. It wasthis spotter who gave us first knowledge of thelarge wall cloud that eventually spawned the tor-nado that destroyed/severely damaged a dozenhomes and apartment buildings. The spotter’scall—along with a warning from NWS Norman—caused us to activate our warning system, and weprovided our residents about 10 minutes of warn-ing. There were NO injuries or fatalities from the
storm. The spotter later toldus (numerous times) that hadthe EOC not moved him, hewould not have been in theproper location to see thewall cloud! This scenario isEXACTLY what OK-FIRSTwas designed to do! Itcertainly worked here!!!”(G. Kitch, following a tor-nado in Moore on 4 Oct1998).
• “Following the devastatingstorms in Oklahoma City,Moore, and Del City, numer-ous Logan County policeand fire crews were eager togo there to assist. Only byreal-time OK-FIRST datawere we able to persuade de-partment chiefs that our ownthreat was not over; thatcrews and equipment shouldbe retained in our area untilseveral additional supercellscapable of producing torna-does passed. Thus, emer-
FIG. 6. Example of a data interpretation guide for the Keetch–Byram Drought Index, oneof a series of images produced by the Oklahoma Fire Danger Model.
1921Bulletin of the American Meteorological Society
gency response was still strong in Logan Countywhen we were struck about 90 minutes later. Whenpolice and rescue crews arrived at the first LoganCounty damage site near the city of Crescent, oneof the first tasks was to open the highway suffi-ciently to get an ambulance through from Crescentto the hospital in Guthrie. All efforts were to getthat ambulance moving with a critically injuredtornado victim. About the time they succeeded, asecond tornado approached in the dark. The am-bulance and the tornado moved on intersectingpaths. Emergency management, aware of bothevents, was able to stop the ambulance until thetornado had passed just in front of it. The town ofMulhall, devastated by the initial tornado after itpassed Crescent, was warned primarily by two lawenforcement units sounding their vehicle sirens inthe town. The units had been dispatched there bythe Sheriff’s office based upon OK-FIRST data.Both units continued warning residents until theywere each hit by debris: one by power lines downacross his car, the second by a large tree on top ofhis unit. Both officers were uninjured—and sowere all but one Mulhall town resident!” (J. Lewis,following the 3–4 May 1999 tornado outbreak).
• “I promise you we’d have had deaths in LincolnCounty without it. I firmly believe that” (B. Spring-field, following the 3–4 May 1999 tornado out-break). Lincoln County residents were informed byfrequent updates broadcast via scanner. As result,these rural citizens took shelter. Mr. Springfieldalso consulted with other local officials to evacu-ate the Tanger Outlet Mall, which was later de-stroyed. No physical human casualties resulted,although the town of Stroud lost its three majoremployers and over half of its tax base in a matterof minutes.
Several other rural emergency managers preventedeven greater tragedies from occurring on 3 May bydirecting crews from their areas around interveningtornadic supercells as they responded to needs in theOklahoma City area.
c. Fire weatherOK-FIRST participants have alerted firemen to ap-
proaching wind shifts. Participants at fire departmentshave their adjusted staffing and operational procedures.
• “I used OK-FIRST to forecast a significant windshift during a large grass fire. From Oklahoma
Mesonet data, I initially detected a wind shift aheadof an advancing cold front. NIDS data from theOklahoma City radar, in clear-air mode, [also]clearly showed the wind shift line. I then projecteda precise arrival time in coordination with the NWSforecaster in Norman, and passed the informationto the Incident Commander (IC), giving him 45minutes lead-time. With the advance warning, theIC refueled, rewatered, and repositioned brushpumpers near structures at the southeast edge of thefire. When the sudden wind shift occurred, equip-ment was in position to protect two rural housesand several outbuildings. Had the wind shift oc-curred without warning, the response time to thenew head of the fire would have been 5–10 min-utes after the IC recognized the wind shift. The ex-posures on the southeast flank would certainly havebeen seriously threatened” (J. Lewis, following awildfire event in Dec 1997).
• “Our fire dispatchers use the KBDI to page mul-tiple rural fire departments when the values get toohigh. This way, we can direct more efforts in get-ting the fires knocked down when they are stillrelatively small, instead of having them get out ofcontrol” (H. Trottier, following a major statewidefire emergency during Sep 2000).
d. Flooding• “I used the OK-FIRST system (NIDS and Mesonet)
to estimate that 6½ inches of rain fell in southernPittsburg County. I was able to use that informa-tion to call the county commissioners and warnthem that a particular bridge might wash away twohours before the bridge was destroyed. The peoplehere in McAlester and Pittsburg County are reallysold on the system” (B. Young, Oct. 1997, onlydays after returning from a weeklong workshop).
FIG. 7. Workshop of 19 emergency managers during Aug 1999.K. Kloesel is shown delivering a live weather briefing where par-ticipants are asked to follow using their computers.
1922 Vol. 82, No. 9, September 2001
• “We had an ambulance that needed to transport anindividual to a different hospital during a flood. Iwas able to tell them the latest information aboutwhere roads were covered and where they could getthrough.”
• “On Christmas Day the river was within 6 inchesof flood stage. Instead of having to run down to theriver every couple of hours or pulling people awayfrom their families, I was able to use OK-FIRSTand see that the crest had already passed and it wasgoing back down.”
e. Other situationsOK-FIRST has been used during outdoor athletic
events, concerts, parades, and festivals to protect thepublic and provide guidance to event organizers. Localusers have disseminated heat index and wind chillvalues to the elderly and to outside workers. Forensicuse includes the investigation following homicides,arsons, and aircraft accidents. Other communities haveused OK-FIRST for more efficient scheduling of pub-lic works projects and directing evacuations duringhazardous-materials incidents:
• “We had a chlorine gas leak at the water treatmentplant. We used current data to assist respondingagencies and for evacuation of residents of thearea.”
• “This winter we had a storm moving in. We wereable to watch it over the course of a few hours andcould tell that the snow wasn’t going to hit us. Thecity could send the snow crews home and saved alot of overtime pay.”
• “I’ve used it on several occasions to assist countycommissioners on their road work.”
6. Summary
OK-FIRST has convincingly demonstrated thatlocal public safety officials, when properly trained andprovided with accurate and real-time weather informa-tion, can make decisions beneficial to the protectionof life and property and to save precious resources. Inaddition, OK-FIRST has made the transition from ademonstration project to an annually state-funded pro-gram. To date, over 140 public safety officials havebeen trained. Workshops continue to be conducted asparticipants are recruited from communities not cur-rently served. The workshops also help to continuelocal involvement following the attrition of partici-
pants. Through the process of developing and sustain-ing the program, OK-FIRST staff have gained consid-erable experience and knowledge regarding therequirements for decision-support systems that ben-efit public safety and in the training methods that aremost effective for nonmeteorologists. As access tofederal radar data becomes less restrictive, the experi-ence gained in Oklahoma should benefit other privateand public sector groups who desire to build similarsystems. Some interest has been expressed by a fewneighboring states in building programs like OK-FIRST, although none have formally begun the pro-cess. While the legislative mandate of OCS limits itsdirect involvement in building systems like OK-FIRSToutside the borders of Oklahoma, the intellectual prop-erty created by OCS is available via licensing and con-sulting arrangements. However, OCS has also beeninvolved in replicating OK-FIRST to solve problemsof other user groups (e.g., rural electric cooperatives).
Acknowledgments. OK-FIRST began using a grant from theU.S. Department of Commerce and its Telecommunications andInformation Infrastructure Assistance Program (now known asTelecommunications Opportunities Program) of the NationalTelecommunications and Information Administration. The Na-tional Weather Service provided additional funding to help testthe decision-support system. Annual support is provided by theOklahoma legislature through the Oklahoma Department of Pub-lic Safety. Additional lesiglative support was achieved throughthe efforts of Oklahoma State Senator Cal Hobson. UnisysWeather Information Services provided the opportunity to redis-tribute NIDS data for reduced subscription fees. Assistance inthe training workshops has been provided by staff from the NWSForecast Offices in Norman and Tulsa, the ABRFC and theWSR-88D Operational Support Facility. In addition, the successof the program is largely due to the efforts of the participantsthemselves.
Numerous other individuals contributed to the success OK-FIRST. The following present and former OCS staff provided as-sistance in a variety of ways: Andrea Melvin, Andrew Reader,Christopher Duvall, Renee McPherson, Mark Shafer, HowardJohnson, Tim Hughes, Derek Arndt, Justin Greenfield, AmyCameron, Cerry Leffler, Mike Wolfinbarger, Brad Stanley, RobbYoung, Stdrovia Blackburn, Jared Bostic, Jessica Thomale,Sridhar Kulasekharan, Billy McPherson, the Mesonet field tech-nicians and student operations staff. The evaluation was conductedby Tom James and Paula Long from the Institute for Public Af-fairs at the University of Oklahoma. The support of OklahomaGovernor Frank Keating is also acknowledged.
References
Baer, V. E., 1991: The transition from the present radar dissemi-nation system to the NEXRAD Information DisseminationService (NIDS). Bull. Amer. Meteor. Soc., 72, 29–33.
1923Bulletin of the American Meteorological Society
Brock, F. V., K. C. Crawford, R. L. Elliott, G. W. Cuperus, S. J.Stadler, H. L. Johnson, and M. D. Eilts, 1995: The OklahomaMesonet: A technical overview. J. Atmos. Oceanic Technol.,12, 5–19.
Carlson, J. D., and D. M. Engle, 1998: Recent developments inthe Oklahoma Fire Danger Model, a mesoscale fire dangerrating system for Oklahoma. Preprints, Second Symp. on Fireand Forest Meteorology, Phoenix, AZ, Amer. Meteor. Soc.,42–47.
Changnon, S. A., K. E. Kunkel, and B. C. Reinke, 1996: Impactsand responses to the 1995 heat wave: A call to action. Bull.Amer. Meteor. Soc., 77, 1497–1506.
Crawford, K. C., D. A. Morris, and D. E. Lewis, 1999: OK-FIRST:A NIDS success story. Preprints, 15th Int. Conf. on Interac-tive Information and Processing Systems for Meteorology,Ogeanography, and Hydrology, Dallas, TX, Amer. Meteor.Soc., 44–47.
Doswell, C. A., III, A. R. Moller, and H. E. Brooks, 1999: Stormspotting and public awareness since the first tornado forecastsof 1948. Wea. Forecasting, 14, 544–557.
James, T. E., P. O. Long, and M. A. Shafer, 2000: An indepen-dent evaluation of the OK-FIRST decision support system.Preprints, Second Symp. on Environmental Applications, LongBeach, CA, Amer. Meteor. Soc., 40–47.
Kelsch, M., 1996: Severe weather information for emergencymanagement decision making. Preprints, 18th Conf. on SevereLocal Storms, San Francisco, CA, Amer. Meteor. Soc., 597–601.
Klazura, G. E., and D. A. Imy, 1993: A description of the initialset of analysis products available from the NEXRAD WSR-88D system. Bull. Amer. Meteor. Soc., 74, 1293–1311.
McPherson, R. A., and K. C. Crawford, 1996: The EARTHSTORMProject: Encouraging the use of real-time data from the Okla-homa Mesonet in K–12 classrooms. Bull. Amer. Meteor. Soc.,77, 749–761.
National Oceanographic and Atmospheric Administration, 1977:Johnstown Pennsylvania flash flood of July 19–20, 1977.Natural Diaster Survey Rep. 77-1, 60 pp.
——, 1991: Shadyside, Ohio, flash floods, June 14, 1990. Natu-ral Disaster Survey Report, 127 pp. [NTIS PB91-204966INZ].
——, 1994: Southeastern United States Palm Sunday tornado out-break of March 27, 1994. Natural Disaster Survey Report, 75pp. [Available from the National Weather Service, 1325 East–West Highway, Silver Spring, MD 20910-3283.]
——, 1997: The central Texas tornadoes of May 27, 1997. ServiceAssessment, 99 pp. [Available from the National Weather Ser-vice, 1325 East–West Highway, Silver Spring, MD 20910-3283.]
National Research Council, 1980: Technological and ScientificOpportunities for Improved Weather and Hydrological Ser-vices in the Coming Decade. National Academy Press, 87 pp.
——, 1991: A Safer Future: Reducing the Impacts of Natural Di-sasters. National Academy Press, 76 pp.
——, 1999: Reducing Disaster Losses through Better Informa-tion. National Academy Press, 72 pp.
U.S. Weather Bureau, 1964: The national research effort on im-proved weather description and prediction for social and eco-nomic purposes. Prepared for the Federal Council for Scienceand Technology, Interdepartmental Committee on Atmo-spheric Sciences, 84 pp.
Wolfinbarger, J. M., R. A. Young, and T. B. Stanley, 1998a: In-teractive software for viewing NEXRAD level 3 data on theWorld Wide Web. Preprints, 14th Int. Conf. on Interactive In-formation and Processing Systems for Meteorology, Ocean-ography, and Hydrology, Phoenix, AZ, Amer. Meteor. Soc.,208–212.
——, ——, and ——, 1998b: Delivering real-time interactive datafrom the Oklahoma Mesonet via the World Wide Web. Pre-prints, 14th Int. Conf. on Interactive Information and Process-ing Systems for Meteorology, Oceanography, and Hydrology,Phoenix, AZ, Amer. Meteor. Soc., 213–217.
