{"Bibliographic":{"Title":"Pike-Buchanan flash floods of July 15, 1979","Authors":"","Publication date":"1979","Publisher":""},"Administrative":{"Date created":"08-17-2023","Language":"English","Rights":"CC 0","Size":"0000051442"},"Pages":["GB\nGB\n1399.4\n1394.4\n.K4\nRHN3\nN3\n1979\nNatural Disaster Survey Report\nPike-Buchanan\nFlash Floods of July 15, 1979\nCENTRAL\nLIBRARY\nMAR 12 1980\nN.O.A.A.\nU. S. Dept. of Commerce\nNational Weather Service, NOAA\nU.S. Department of Commerce\nOctober 1979\nSilver Spring, Maryland\n0788\n80","GB\n1399.4\nKY\nN3\n1979\nPREFACE\nSummary, Findings and Recommendations\n1.\n2.\nThe Flash Floods\n3.\nThe Flash Flood Warning System\n4.\nMeteorological Conditions\n5.\nHydrological Conditions\n6.\nObserving Systems\nRadar\nSatellite\n7.\nResponse\n8. Significance for Pilot Flash Flood Program","PREFACE\nThe Natural Disaster Survey Team named to investigate the Pike\nCounty, Ky. and Buchanan County, Va., flash floods of Sunday, July 15,\n1979, was comprised of:\nR.L. Carnahan, Team Leader and Chief, Weather & Flood Warnings\nCoordination Staff, NWS* Headquarters, Silver Spring, Md.\nA.F. Flanders, Assistant to Associate Director, NWS (Hydrology),\nNWS Headquarters, Silver Spring, Md.\nLars 0. Feese, Hydrologist in Charge, River Forecast Center,\nHarrisburg, Pa.\nRonald A. Willis, Principal Assistant, Weather Service Forecast\nOffice, Minneapolis, Minn.\nDr. Walter J. Saucier, Consultant, Professor of Meteorology,\nNorth Carolina State University, Raleigh, North Carolina\nThe team gathered in Louisville, Ky., July 24, and began the survey\nJuly 25, traveling to Charleston, W. Va., into the flooded areas of Pike\nCounty and Buchanan County and terminating July 27 in Tri-Cities, Tenn.\nPrimary areas reviewed were:\n1) internal NWS operations;\n2) community preparedness and individual public response; and\n3) flash flood observation system and local warning aspects of\nthe Appalachian pilot flash flood program.\nMedia response was not a part of the survey since no flash flood\nwatch or warning had been issued for the event by NWS.\nThe event was a series of classic flash floods occurring from intense\nthunderstorms over a small area on the order of a 15-mile diameter circle\nin the steep mountainside country of Appalachia in extreme eastern Kentucky\nin the headwaters of the Tug Fork of the Big Sandy River.\nThree deaths were reported as direct results of the flash flood and\ndamages may reach $13 million in Pike and Buchanan Counties.","ii\nAssistance was provided by numerous individuals in the conduct of\nthe survey but the help of the Appalachian Regional Commission (ARC)\noffice in Pikeville, Ky., , was invaluable. Mr. Keith Kelly, Director\nCentral Appalachian Development Project, ARC, made the arrangements that\npermitted the team to meet with local and State officials in the two\ncounties. Messrs. Benny Ward and Allan Reed of ARC accompanied the\nteam into the disaster area. Mr. Robert Morrison, Regional Coordinator,\nKentucky Disaster Emergency Services, was most helpful in guiding the\nteam to the areas that received flood damage. Mr. Stanley Schneider\nof NOAA NESS * performed a special analysis of satellite imagery.\nCover picture of Majestic, Ky . , on Poplar Creek courtesy Dave\nChaffins, Pike County, Ky , Information Officer. Photographs in\nFigures 2.3 and 2.4 were taken by A. F. Flanders during the survey.\n* U. S. Department of Commerce\nNOAA - National Oceanic & Atmospheric Administration\nNWS - National Weather Service\nNESS - National Environmental Satellite Service","1. Summary, Findings and Recommendations\nIndications from eyewitness accounts, local disaster services\nofficials, firsthand examination of the flooded areas and data sources\navailable to the NWS all lead to the conclusion that the storm activity\nthat produced the flash flooding was concentrated over a very small area.\nRainfall was intense and of short duration. The steep terrain contributed\nto the rapid runoff, causing a classic flash flood situation.\nThe NWS observing system is not sensitive enough to detect these\nsmall scale short-lived events. The scale of the event was so small\nthat the usually accepted practices in the field offices failed to reveal\nits significance. The problem was further aggravated by its occurring on\nthe midnight shift when staffing was at a minimum. The Louisville staff\nwas preoccupied by an even stronger system to the west of WSFO Louisville.\nNeither the satellite imagery nor the radar information available to the\nduty forecasters was of significant intensity to warrant special attention.\nAlthough the pilot flash flood program is in its beginning stages\nin both Pike and Buchanan Counties, no real-time surface reports of rain-\nfall are currently available for monitoring. Flash flood rainfall\nobservers were scattered through the area but none lived in the area of\nheaviest rainfall. When they did communicate among themselves and to\ntheir coordinators and local officials none of this information ever\nreached NWS channels until routine reporting time the following morning\nafter the flood was over. Loss of life was undoubtedly kept low by the\nfact that most residents were either awake or awakened by the thunder-\nstorms. As stated by one Flash Flood Coordinator \"There was no way on\nGod's Green Earth you could have warned those people. There was no\ngreat criticism of the NWS for not detecting the storm and issuing a\ntimely warning. This event demonstrates that local initiative and\nindividual response is vital to protect life and property from flash\nfloods in such rugged terrain.\nFinding 1\nNOAA Weather Radio false tone alerts have created a loss of confidence\nin the system. The low power output of the repeater at Pikeville\nhas also limited its usefulness.\nRecommendation 1\nExamine NOAA Weather Radio transmitters within range of Pike and\nBuchanan Counties to detect any possible source of false turn-on\ntone alerts at the transmitters. Advise state and local personnel\nof probable causes of malfunctions of receivers.\nFinding 2\nThere was a serious lack of real-time rainfall reports available\nto radar operators and offices with watch/warning responsibility.\nSome local volunteer observation sites currently have no backup\nto compensate for locations that are closed at night or on weekends.\n1","Recommendation 2\nIncrease recruiting activity in both Kentucky and Virginia in an\neffort to obtain more observers and improve observation site\ncoverage.\nFinding 3\nThe flash flood rainfall reporting network did not respond. Early\nreports of flooding and other indicators of a potentially serious\nsituation were not passed along to the NWS. Training programs of\nboth police and community flash flood observers, when implemented,\nhave been beneficial to improved response.\nRecommendation 3\nExpand training of flash flood cooperative observers to improve\nunderstanding of mutual roles of observers and NWS forecasters\nand to increase the number of real-time reports.\nFinding 4\nImproved data sources and communications in the 12-county area are\nnecessary for more assessment of extent and magnitude of flood events.\nRecommendation 4\nContinue and accelerate when possible installation of automatic\nreporting rain and stream gages in the area.\nFinding 5\nWarning responsibilities in the 12-county area should be reviewed\nto assure maximum coordination and optimum communications.\nRecommendation 5\nInitiate an examination of NWS warning responsibility assignments\nin the 12-county area with a view toward shortening communications\nlines and facilitating warnings operations.\n2. The Flash Floods\nThunderstorm rainfall in the early morning hours of Sunday,\nJuly 15, 1979, caused extensive flooding to numerous small creeks in\nthe headwaters of the Tug Fork of the Big Sandy River basin. A resident\nalong one of these small creeks said \"It started at 3 a.m. like pouring\nout of a bucket, but close to 5 a.m. it sounded like a transformer\nexploding and then all hell broke loose.\" That was the description of\nhow the waters came down Camp Creek, one of the typical creeks in the\nbasin. The event was over around 7 a.m. EDT.\n2","uchardson\nLawmansville\nMila\n115\nFigure 2.1\nCalf\nKEY TO MAP\nStambaugh\nCreek\nAdd\nDHNSON\nSitka\nWhitehouse\nFREEBURN\nWittensville\nHc\nA\nTweb\nInez\nKey\nTomahawk\nStaffordsville\nDr. DANIELS STOPOVER\nB\nDavisport\nKealka\nBeaut\nPHELPS AREA THOMPSON\nC\nPaintsville\nWilliamspott\nDebord\nMARTIN\nPHELPS AREA BOOTH\nD\nHaperhill\nMeally\nVan\nFISHTRAP LOUSVILLE REPORT FROM DEWEY\nLear\nDavella!\nE\nDeliver\nplease\nVanlear\nFEDS CREEK LOUISVILLE REPORT FROM DEWEY\nF\nOdds\nPoint\nMcClui\nCAMP CREEK UNCONFIRMED\nG\nAuxter\nTh\nWhitaker\nMR HASH ESTIMATE RIGHT FORK PAW PAW\nH\nMR HASH ESTIMATE LEFT FORK PAW PAW\nI\nMaree\nRonanza\nMR HALL BUCHANAN C.D. DIRECTOR\nJ\nPrestonsburg\nThomas\nLancer\nHeenon\nWilliamson\nMcCoinbs\nTurkey Cr\nEmma\nWatergap\nGoody\nWoods\nHag Knoh\nSloan\nDwale\nGulnare\nPiso\nVarney Road Fork\nAlien\nflex\nBanner\nTole\nSidney\nRisne\nIvel\nBelfry\nHardy\nMartin\nF\nLO\nCanada\nTram\nY\nMcCarr\nPramid\nHuddy\nMare Creek\nStone\nMaulan\nDana\nBelsylayne\nMeta\nPinson\nMcAndrews\nBroad Bottom\nPrinter\n\"Ransom\nHarour\nPETERS CREEK\nMorcoal\nFreeburn\nBoldinan\nEastern\nZebulon\n(Pinsonfork\nCoal\nHunter\nRun\nP\nMcVeigh\nGarrett\nMajestic\nGrethel\nDrift\nK\nOwsley\nLarkey\nWoodman\nMinnie\nKimpe\nEstill\nLittle\nMcDowell\nPikeville\nPhelps\nRaccoon\nKNOX CREEK\nDixie\n(605)\nWayland\n2032\nJambbree\nMcDowelk\n2010\nColeman\nKewarie\nFishtran\nArgo\nTeaberry\n2152\nShelbiana-\nFords\nPhyllis\nPrice\nGasveston\nDunlap\nBeaver\nKelsa\nLick Cr\nSutton\nPaw Pad\nRobinson\nCTY\nHat\nHylto\nCreek\nMillard\nIra\nKnob\nHurley\nLigon\nSteele\nTopmost\nBurton\nFedscree\nGreasy\nMouthcard\ntollybush\nMelvin\nCreek\nWheelwright\nWales\nWolfpit\nDrowbone\n(Kite\nVirgie\nWeeksbury\nBig\n(Blackey\nDraffin\nBelcher\nRockhouse\nHalo\nRock\nTHartley\noWolford\nJonancy\nRoseann\nHome\nHome\nHenry\nLookout\nBeaver\nCreek\nClay\nElkhorn\nMyra\nGap\nHarmon Junction\nHellier\nStacy\nSpeight\n2846\nBreaks\nDorton\nMaxie\nCr\nEtty\nLion/ui\nHarmon\nSlate\nshcamp\nDemocrat\nBeefhideHylton\nGrundy\n02\nTivis\nBartlick\n11050\nPatterson\nCHANNAN\nHemphiti\nShelhy\nMcRober\nGap\nBlowing\nJewell\nValley\nRock\nFlemide\nFlannagan\nVansanto\nPrate\nFork\nBearwallow\nHaysi\nOakwood\nWhitewood\nSeco\nVicey\nIsom\nTarpono\nKona\nKeen Mountain\nPound Gap\n(MavisdaleO\nClintwood\n©Birchleaf\nGeorges\n15R\nFork\nMarvin\nYewell\nAlmira\nClincho\nRidee\nMount Heron9\nPage\nColley\nPound\nFremont\nGrimsleyville\nShortt\nCoald\nCounts\nGap\nRowe\nHugh\nC\nE\nMurphy\nGlen\nRid\nS\nClell\nFlat\nDewey\nBurke\nN\nGap\nMcClure\nLynn\nSe\nDavenport\nAAly\nBuck Knob\noSpring\nQRed Ash\noDriN\nDuty\nNora\nCouncil\nRichlands\nBucu\nDuncan Gap\nRaven\nDyed\npWakenva\nS\nBig A Mtn\nPardee\nCarrier\nLIDTS\nSwands\nLeck\nGardner\nStephens/\n\"Trammer\nbereek\nPutnam\nQunbar\nWant\nwins\nWise\nCranes\nONoral\nNest\nGAGE #1 BUCHANAN CO.\nEsserville\nK\nToms Cr\nNortan\nGAGE #7 BUCHANAN CO.\nAndover\nTacoma\nL\nBanner\nWHITESBURG REGULAR REPORTING GAGE TO LOUISVILLE\nCoeburn\nM\nNORTH FORK REGULAR REPORTING GAGE TO LOUISVILLE\nN\nFLANNAGAN REGULAR REPORTING GAGE TO LOUSIVILLE\no\nGAGE #9 BUCHANAN CO.\nP\nGAGE #6 BUCHANAN CO.\n&\nGAGE #3 BUCHANAN CO.\nR\nGAGE #5 BUCHANAN CO.\nS\nGAGE #4 BUCHANAN CO.\nT\nU\nBUCU\n3","Adams\nHannah\nBiaine\nFigure 2.2\nCordell\nGallup\nMartha\nHifford\nRainfall Amounts July 15, 1979\nWilbur\nGeorges\n:ton\nCharley\nDavisville\nWinifred\nUlysses\nRichardson\nedbush\nFlatgap\nLowmansville,\nMilo\nCalf\nCreek\n115\nAdd\nStambaugh\nFuget\nHode\nSitka\nOHNSON\nWhitehouse\nInez\nTomahawk\nWittensville\n0\nJutof\n5\n10\n20nm\nKey\nBeauty\nDavisport\nWarfield\nBarnetts\nStaffordsville\nDebord\nLovely\nCreek\nThealka\nWilliamsport\nMARTIN\nPaintsville\nOil Sprs\nMeally\nDavella\nPilgrim\n159\nHagerhill\nW\nVan\nLear\nPreecé\nDenver\nVan Lear\nMcClure\n1408.\nThreeforks\nOdds\n050Z0\nE Point\nAuxier\nRiceville\nWolf\nLaura)\nWhiteker\nMoree\n2.00\nHatfield\nliville\nThomas\nS\nBonanza\nWilliamson\nPrestonsburg\nTurkey Cr'\nGoody\nHeenen\nLancer\nMcCombs\n1658\no\nEmma\nFlag Knob\nflex\nRoad Fork\nWoods\nPiso\nTole\nWatergap\nWarney\nGulnare\neville\nSloan\nDwale\n.00\nSidney\nBelfr\nHard\ndville\nAllen\nBanner\nMcCarr\nGoodiee\nCanada\n0715\nHuddy\nIvel\nDavid\nStone\n(029)\n2.00\nRisnel\n3.00\nGMartin\nTram\nPinso\nRansom\nF\nY\n1.00\nMare Creek\n195\nMeta\nMcAndrews\nFreeburn\nBetsy Layne\nPyramid\nDana\nMorcoal\nManton\nPinsonfork\nBroad Bottom\nLangley\nHapotd\nZebulon\nMcVeigh\nMajestic\nPrinter\nHippo\nCoal\nBaldman\nWoodman?\nP\nE\nRun\nEasterh\nPhelps/OO\nHunter\nBosco\nKimpe\novsley\nGrethel\nGarrett\nJamboree\nDritt\nRaccoon\nPikeville\nArge\nColeman)\nLittle\nLackey\nMinnie\n(605)\nDixte\n2020\nMcDowell\nEstill\nBetty\nFishtrap\n52\nFishma\nKelsal\nWayland\nKewa\nMcDowell\nPhyllis\n3.70\nLake\nPaw Pay\nnlap\nShelbiana-\nDemar\nTeaberry\nLick\nIra\nFord\nHylto\nHurley\n.00\n6.00\nMausie\n5570201\nGalveston\nSutten\nKnob\nPrice\nSteele\nRaven\nMillard\nRobinson\nBeaver\noffShell\nCreek\nedscree\nHi\nHat\nMouthcard\n2\nBlackey\nLigon\nGrea\noWolford\nBig\nBurton\nCreek\nMarrowbone\nTopmost\nGarner\nRock\nor\nRoseann\nWolfpit\n050\nMelvin\nburn\nBelcher\nHome\nWales\nHon\nHollybush\nVirgie\nDraffin\nCreek\nan\nWheelwright\nRockhouse\nPippa\nWeeksbury\nStacy\nKite\nHarmon Junction\n1.60\nPasses\nHartley\nJonancy\nvis\nHalo\nLookout\nElkho\nHenry\nCity\n0\nSlate\nMallie\nClay\nBreaks\nMaxie\n412\nPuncheon\nHall\nHellier\nMyra\nHarmon\nBeaver\n.2846\nOmaha\nle Top\nGrunde\nGap\nDorton\nPatterson\nSpeight\nJewell\nAshcamp\nANNAN\nLionly\nEtty\nB\nValleyo\nBartlick\n86\nBea\nDeane\nBeefhideHylton\nTivis o\ntcarr\n78\nVansane\nDemocrat\nWhitewood\nPORK\nPrater\nShelby\nBlowing\nOakwood\nedfox\nFlannagan\\\n1.70\nColson\nHemphill\nGan\nRock\nKeen Mountain\nVicey\nMcRoberts\nHaysi\n1.32\n400!\nMavisdate\nJewell\nMarvin\nNeon\nFlemin\npound\nRidge\nTarpono\nIsom\npBirchleaf\nIsom\nage\nThornton\nO\nMount Heron'\nShortt\nSeco\nClintwood\nGrimsleyville\nJeremiah\nMillstone\nPound Gap Georges\nClincho\nGap\n133\n.1771.\n1.05\nColley\nKona\nForko\nGlen\nMayking)\nher\nRowe\nRidge\nBurke\nFCHER\nOAlmira\nCiell\nFrement\nCounts\nMurphy\nLynn\nas\nPound\nQRed Ash\noSpring\nS\nine\nFork\nWhitesburg\nC\nK\nE\nN\nDavenport\nRichlands\nRese\npDriN\n3\nHugh\nDongo:a\nMcClure\n2.25\nQAily\nOscaloosa\nCouncil\nDuty\nRaven\nDewey\nFlat\nDyeo\no\nBucu\nDay\nNora\nGap\nBig A Mtn\neven\nBuck\nKnob\nSwords\nKings\nFork - Eolia\nWakenva\nbereek\nGardner\nDuncan Gapo\nCarrie\nPutnam\nS, E\nartridge\nLeck\nTrammel\nHonaker\nWilder\nCoulwood\nPardee\nWest\n©Blackfc\nLIPP\nGlamorgan\nDante\nDante\nStephens\nSHerald\nCranes\n©Wise\nland\nClinchfield\noNest\nDunbar\n12454\nArtrip\nLyn\nRoda\nToms Cr\nS Clinchfield\nEsservilled\nStoneg\nBanner\nGuest\nTacoma\nDerby\nNorton\nCoeburn\nCLINCHEIN\nAndover\nLaurel\n4\nInman","5","","There were two confirmed deaths and a 4-year old child is missing\nand presumed dead from the flood.\nProperty damage is reported at $2.6 million in Buchanan County with\ndisaster survey reports indicating a possible final figure of $6 to 8\nmillion. In Pike County $2.4 million is reported with possible upper\nlimits of $6.5 million.\nRainfall locations (figure 2.1) and amounts (figure 2.2) show the\nconcentration of heavy rainfall in a small area with general estimates\nof 6 to 7 inches.\nDamage survey reports indicate more than 250 homes destroyed or\ndamaged in Pike County and about 200 homes in Buchanan County. Several\nhundred people were evacuated to public shelters. More than 250 private\nbridges of the kind in figure 2.3 (lower right) were destroyed. About\n50 miles of public road and a dozen bridges were destroyed (figure 2.4).\nEvents during the early morning hours, as reconstructed from\ninformation furnished by the Flash Flood Coordinators in Pike County\nand Buchanan County:\nBUCHANAN COUNTY, VA\nReport of Flash Flood Coordinator - John Hash - Grundy, VA\nCounty Civil Defense Director - Mr. Hall\nTimes (EDT)\nMr. Hash noticed thermal buildups northwest of area.\n3 a.m.\nMr. Hash was called by employee with information on\n5:30 a.m.\ncreeks washing out.\nMr. Hash called dispatcher at Grundy. They reported\n5:35 a.m.\n4.2 inches of rain.\nApprox. 6 a.m. Mr. Hash called CD Director in Hurley area. He had\n2.7 inches and still raining. Storm total amounted\nto 3.70 inches. CD Director is Mr. Hall.\nMr. Hash arrived at Grundy and confirmed 4.2 inches.\n7 a.m.\nHe proceeded to call gages to locate his trouble area.\nCalled northwest area at Indian Creek Gap, Boyd gage\nread .70. Total for Boyd, 1.05 inches.\nHe tried with no success to reach the #1 Rife-Slate\ngage--no response.\nBy this time he felt he knew where his trouble area\nwas and proceeded to take care of his problem areas.\n7","Gage Reports in Flash Flood Program\nPcpn\nGage\n1.60\"\n1\nRife-Slate NW of Grundy\n2\nNo observer\n1.78\"\nJaney SE of Grundy towards Richlands\n3\n1.05\"\n4\nBoyd Indian Creek Gap\n4.00\"\nMarvin SE East on 460 from Grundy\n5\n1.70\"\nShort near Deel on Route 460\n6\n4.20\"\n7\nGrundy\n3.70\"\n8\nPounding Mill Hall\n.86\"\nPrater Airport\n9\nNone of these gages drain into the Knox Creek or Peter\nCreek drainage.\nMr. Hash's estimates are: Right Fork of Paw Paw 7\";\nLeft Fork of Paw Paw 6\" by bucket measurement.\nPIKE COUNTY, KENTUCKY\nReport of Flash Flood Coordinator - Joyce Hale, Pikeville, Ky.\nPcpn\nPhelps area - Thompson, 2 miles out of Phelps area\n2.58\"\ntowards Pineville. This gage was not read immediately.\nPhelps area - Booth, South edge of Phelps - gage was\n3.50\"\nread immediately.\nSolid Waste Treatment Plant - Wolford, gage was cracked\nunknown\nand possibly had been emptied. When Benny Ward, ARC,\narrived there was 2.31 inches in gage. Probably not\nconnected to this particular storm. Gage is located\n4 miles from Phelps towards Pineville on right fork\nof Peter Creek.\nStopover - Dr. Daniels, 7 inches was reported, however\n6.50\"\n6.50\" was reported on Form E-16 sent to Louisville WSFO.\nIt is confirmed the gage is located on the side of a\ntelephone pole. Station is located one mile south of\nStopover on Turkey Creek.\nCamp Creek area - Thursday, July 16 still unable to con-\n13:00\"\nfirm this report. Benny Ward, ARC, was in the area and\nunconfirmed\nunable to find the gage. The lady with the baby who was\nawake at 5 a.m. when storm started and at 5:45 a.m.\nobserved water out of the creek. NOTE: Highway patrol\ninvestigated with follow-up by ARC determined initial\nreport of 13.00\" was in error, somehow confused with\nreport from stopover.\nEstimated in swimming pool along right fork Peter Creek\n7\" to 9\"\nbetween 3 - 6 a.m.\n8","3. The Flash Flood Warning System\nThe systems designed to protect our citizens against flash floods\n--which, like tornadoes and severe thunderstorms, are short-fused and\ngenerally localized hazards -- rely on elements of the National Oceanic\nand Atmospheric Administration (NOAA), National Weather Service (NWS),\nthe awareness and readiness of local officials and the public, and the\ncooperation and availability of mass media channels. The system has\ntwo basic facets. One facet consists of the National Weather Service\n(NWS) efforts to produce and disseminate watches and warnings to which\nthe public and local officials can respond. The other facet depends\nupon developing local flash flood warning systems by which local\ncommunities can detect and respond to immediate threats without any\nexternal forecasts or warnings.\nWhenever possible, NWS attempts to issue timely watches and\nwarnings of flash floods. In those situations where flooding appears\nlikely, in terms of expected precipitation and the ability of the ground\nto absorb it or for the stream to handle it, flash flood \"watches\" are\nissued in advance. These watches serve to alert local officials and the\npublic, as well as NWS personnel in smaller warning offices, that flash\nflooding is possible. But NWS capability to predict heavy or extreme\nrainfall amounts over small areas is limited. This \"watch\" phase has\ntherefore not been very successful for summertime flash floods. Only\nslightly more successful, but still not adequate, is NWS capability to\nperform the close monitoring needed to determine when existing rainfall\naccumulations and stream conditions pose an immediate flash flood threat\nto an area. This monitoring relies on telemetered river and rain gages,\ndirect measuring networks (which are not usually dense enough to cover\nvery small drainage basins), estimates from remote sensors (radar and\nsatellite), and reports from volunteer observers.\nThe local flash flood warning system is dependent on more direct\nlocal initiative and cooperation. In this approach, NWS hydrologists,\nmeteorologists, and warnings coordination specialists prepare local\nforecast procedures based on rainfall and stream data, and, in some\ninstances, on automatic local flash flood alarms, to give timely warnings\nto a threatened area. This approach requires a large degree of local\ncooperation and interest; it is often necessary for NWS personnel to\nspend considerable time working with communities to generate that interest.\nIt should be noted that whether or not a local warning system exists\nin a particular community, the NWS is still responsible for disseminating\nappropriate statements and warnings. However, if the community and the\nlocal population are unprepared to receive or respond to messages from\nNWS, no amount of investment in equipment or new techniques will avert\nfuture tragedies. Where active community programs exist, the feedback\nto NWS to allow for downstream warning has been of high caliber and\nsuccessful.\n9","Several NOAA elements have responsibility to provide input to the\nflash flood warning program in Pike County, Ky., and Buchanan County,\nVa. The responsibilities of these NWS and National Environmental\nSatellite Service (NESS) units include the following:\n1. Weather Service Forecast Office (WSFO) Louisville, Ky., has\nweather forecast and flash flood watch responsibility for Kentucky and\ncounty flash flood warning responsibility that includes Pike County.\nWSFO Louisville is also responsible for hydrologic services in all of\nKentucky. Louisville has a local warning radar about 170 nm from Pike\nCounty.\n2. Weather Service Forecast Office (WSFO) Washington, D. C. has\nweather forecast responsibility for Virginia, Maryland, and Delaware\nand flash flood watch responsibility that includes Buchanan County.\n3. Weather Service Office (WSO) Tri-Cities, Tenn., has local\nweather and county warning responsibility that includes Buchanan County.\nTri-Cities operates a network radar about 75 nm from the flooded area.\n4. The National Meteorological Center (NMC), located at Camp\nSprings, Maryland, provides guidance consisting of large-scale analyses\nand prognoses, large-scale quantitative precipitation forecasts, and\nprobabilistic forecasts of thunderstorms and severe weather.\n5. River Forecast Center (RFC) Cincinnati provides guidance of\nrainfall needed to produce flash flooding. It also issues Headwater\nStatements which specify the amount of rain required to produce\nhalf-flood and twice-flood discharge at selected locations.\n6. NESS Satellite Field Service Station (SFSS), Kansas City, Mo.,\ntransmits satellite imagery (visible and infrared) and regularly\nscheduled (usually every six hours) satellite interpretation messages.\nIn addition, this unit initiates telephone calls to WSFOs when the SFSS\nmeteorologists consider observed cloud elements, patterns, or motion\nto be significant or threatening.\nOverall management of the NWS units mentioned here lies in the\nNWS Eastern, Central, and Southern Regions and NWS Headquarters.\nManagement of the SFSS resides in NESS Headquarters.\n4. Meteorological Conditions\nAll regularly scheduled facsimile analyses and guidance charts\noriginating from the National Meteorological Center (NMC) were\navailable to forecasters at WSFOs Louisville and Washington, D.C. prior\nto the flood event.\n10","The surface analysis for Saturday evening, July 14, showed a cold\nfront extending from Ontario, Canada, southwestward through central\nMichigan, central Illinois, and into eastern Missouri. Prognostic\ncharts indicated this front would move slowly southward during the next\nhours, but remain north of Kentucky. Thunderstorms were indicated\n12\nover the area. The front did move slowly southeastward through Sunday\nmorning, July 15, and by 8 a.m. EDT was approaching the northern tip\nof\nKentucky. A small high pressure ridge extended into extreme southeastern\nKentucky from a high pressure center in the Gulf of Mexico, and remained\nnearly stationary from Saturday evening through Sunday morning.\nAnalyses\nand prog charts at the 850 mb and 700 mb levels during the same period\nreflected the persistence of the Gulf of Mexico high pressure center and\nridge at these levels. A weakening north-south short wave trough was\nevident at both levels moving over the top of the ridge and was located\nin western Kentucky by 1200 GMT Sunday. Wind flow was from the northwest\nat both levels with an average speed of about 15 knots.\nAt the 500 mb level, wind flow was from the north-northwest at 15\nto 25 knots circulating around a weak high pressure center over western\nMissouri. The short wave trough was hardly evident at this level. The\nBarotropic, Baroclinic, and Limited Fine Mesh (LFM) model vorticity\ncharts available on Saturday evening valid for Sunday morning indicated\nlittle vorticity advection of any kind over the flood area. Numerical\nvalues of vorticity were low. Cooling of about two degrees Celsius took\nplace at the 500 mb level during Saturday night over the flood area, but\nthe ridge continued to dominate.\nAt the 300 mb level Saturday evening, a weak ridge was situated\nnorth to south from central Kentucky southward to the Gulf of Mexico.\nThis ridge drifted eastward but remained west of the flood area through\nSunday morning. Winds were generally less than 10 knots from the northwest.\nAtmospheric moisture is analyzed and forecast on several different\ncharts available from NMC. The LFM 700 mb relative humidity analysis\navailable Saturday evening placed most of the eastern quarter of the\nU.S. in an area of 70 to 90 percent relative humidity, including eastern\nKentucky. By Sunday morning this moisture area was forecast to shift\neastward, but with the westernmost edge cutting the eastern tip of\nKentucky. The average relative humidity from the surface to 500 mb was\nabove 60 percent over the eastern third of the U.S. on Saturday evening\nwith the flood area experiencing about 70 percent from late Saturday\nthrough Sunday morning. Precipitable water was moderately high and\nchanged little through the same period with an average value of about\n1.70 inches. The lifted index was calculated to be minus 4 both\nSaturday and Sunday afternoon with a value of minus 1 Sunday morning.\nThe K index averaged about 31. These indices were quite common over\nmuch of the southeastern third of the U.S. and indicate no extremes in\natmospheric instability.\nThe LFM precipitation prog chart available early Saturday evening\nindicated no precipitation over the flood area Sunday morning. However,\nprecipitation was indicated for much of central Kentucky and western\n11","West Virginia. The man-made Quantitative Precipitation Forecast Chart\nfrom NMC indicated precipitation up to 0.25 inch over the flood area\nduring the period. The Excessive Precipitation Forecast Chart did not\nindicate excessive rainfall over or near the flood area. The Hydrologic\nFlash Flood Guidance received from the Cincinnati River Forecast Center\nindicated a precipitation amount of 2.9 inches was needed in a three-\nhour time period to cause flooding in the area.\nNeither WSFO Louisville nor WSFO Washington, D.C. issued a Flash\nFlood Watch prior to the flood event, although the evening forecasts\ndid indicate the possibility of thunderstorms over the area. An analysis\nof the information available to the forecasters prior to the flood event\nindicated that a Flash Flood Watch would probably not have to be issued.\nSurface and upper air analyses and prog charts contained little\ninformation to indicate that heavy rains should be concentrated over\nthe flood area. Considerable moisture had been advected into the south\ncentral portions of the U.S. prior to the flood by a weakening low\npressure center which was the remains of Hurricane Bob. However, equally\nmoist air was evident over much of the southeastern part of the U.S. as\nwell, and no one particular area could be singled out as suspect for\nheavy rains.\nThe flood producing thunderstorms did not form near the surface\nfront, but rather formed just northwest of the flood area itself, which\nwas well south of the front. Since the air was relatively unstable\nprior to the thunderstorm activity, the effect of the 500 mb cooling\ncoupled with the presence of the weak 850 mb and 700 mb trough moving\ninto the area resulted inadditional destabilization in the general area.\nThis destabilization along with available moisture very likely accounted\nfor the heavier precipitation that occurred over the flood area. However,\nthese features looked very insignificant on the analyses charts available\nSaturday evening, July 14. There was no indication on the prog charts\nto cause the forecaster to suspect significant precipitation in the flood\narea. This was backed up by the fact that neither the NMC Quantitative\nPrecipitation Forecast Chart nor the Excessive Precipitation Forecast\nChart indicated significant precipitation for the flood area.\nSatellite photographs and radar observations provide shorter time\nrange tools which may be utilized for issuance of Flash Flood Watches\nand especially Flash Flood Warnings. Satellite photographs proved of\nlimited value in this situation, leaving radar observations as the probable\nbest tool. A post-storm analysis is described in Section 6.\nGridded infrared satellite photographs (resolution 5 nautical miles)\nwere available at half-hour intervals at WSFO Louisville and Charleston\nprior to and through the precipitation period. Satellite photos begin-\nning at 0600 GMT (2 a.m. EDT) Sunday, July 15, indicated two large areas\nof showers and thunderstorms approaching northwestern and northern\nKentucky. Neither of these areas crossed over the flood area, but did\noccupy the attention of the forecasters through Saturday evening and\n12","into Sunday morning. At this same time a tiny dot appeared over\neastern Kentucky, which was the activity that would ultimately grow\ninto the thunderstorms that caused the heavy rains between 0700 GMT\nand 1000 GMT (3 a.m. EDT and 6 a.m. EDT). The satellite grid was not\naligned correctly on the 0630 GMT and 0700 GMT photos, thus placing\nthe developing storm area over eastern Tennessee on these two photos.\nNeither the Great Lakes nor the Gulf of Mexico was easily identifiable\nto help the forecaster correctly hand grid the photos. By 0730 GMT\nthe grid was once again correct and remained so for the remainder of\nSunday morning.\nThe cloud area over eastern Kentucky at 0600 GMT grew and inten-\nsified slowly through 0930 GMT after which time dissipation began.\nThe second gray level was in predominance through 0830 GMT reaching\napproximately 60 miles by 100 miles in extent. A small black level\narea was evident at 0900 GMT and increased slightly by 0930 GMT before\ndissipating by 1000 GMT. The whole area was moving slowly toward the\nsoutheast over extreme eastern Kentucky and western Virginia.\nRadar information was available from the WSR-57 network radar at\nthe Tri-Cities airport near Bristol, Tennessee (TRI) and the WSR-74\nlocal warning radar at Charleston, West Virginia (CRW). The Tri-City\nWeather Service Office is located about 60 miles south of the flood\narea and the Charleston Forecast Office about 60 miles to the northeast.\nHourly overlays were available from TRI with observations taken about\n30 minutes past each hour. There is no requirement to prepare overlays\nat CRW, but several were available during the early hours of the morning\non Sunday, July 15. Neither radar is monitored continuously as station\nduties require the radar meteorologist to perform other assignments in\naddition to radar.\nPrecipitation appeared to move into the flood area first about\n0600 GMT with echoes moving over. the area until about 1100 GMT at\nwhich time all activity had moved to the east. Overlays indicated\nthat as many as four individual thunderstorm cells may have moved over\nthe flood area during this time period. Cell movement was from the\nnorthwest at an average of 17 knots. Video Integrator and Processor\n(VIP) levels of the strongest cells as observed at TRI were level 3\nwith maximum tops between 31,000 and 39,000 feet MSL. Charleston radar\nindicated an occasional VIP level 4 and similar maximum tops except for\na top of 45,000 MSL at 0900Z corresponding to a VIP level 4 over the\nflood area at that time.\nThis corresponds closely to the time of maximum intensity indicated\nby satellite photos. Level 3 VIP corresponds to an estimated precipita-\ntion amount category of 1.1 to 2.2 inches per hour, while the VIP level\n4 category is 2.2 to 4.5 inches per hour. Some attenuation may have\ntaken place due to the ground clutter in the hill and mountain area\nnorth of TRI. Also, the VIP 4 levels at CRW could have occurred at a\ntime when the TRI radar was not being monitored.\n13","Manually Digitized Radar (MDR) values are transmitted by the TRI\noffice and were available to the forecasters at neighboring NWS offices.\nHowever, with no value greater than 3 entered in the MDR grid square\nduring any one hour, flood producing rain was not suspected. The TRI\noffice does not issue narrative radar statements at night due to lack\nof personnel, and personnel on duty early Sunday morning did not think\nthe precipitation was of such significance as to warrant an information\ncall to WSFO Louisville, which has Flash Flood Warning responsibility\nfor Pike County, Kentucky. (The TRI office has Flash Flood Warning\nresponsibility for Buchanan County, Virginia.) The Charleston WSFO did\nmake two calls to the Huntington WSO during the early morning hours to\ndiscuss the precipitation event.\nProbably the most significant information to be gleaned from the\nhourly radar overlays from TRI is the likelihood that several cells\nmoved over or very close to the flood area in a period of three to four\nhours. If the strongest thunderstorms moved through late in this period,\nwhich seems to be the case, the heavy precipitation would have fallen on\nsaturated ground and would have fed into already rising creeks and\nstreams. Real-time detection of this pattern would require constant\nmonitoring of the radar scope, which can be done effectively only with\nradar data processing equipment, such as is planned for installation\nin the 1982-85 time frame. Monitoring of the radar echoes can be\nincreased by calling in extra help on overtime, if conditions warrant.\nAt Bristol during the night of the flood ,normal staffing was\nmaintained. No scheduled radar observations were missed. Weather\nconditions did not warrant calling in extra staff to assist in\ncontinuous monitoring of the radar scope. Even if extra help had\nbeen available, it is doubtful that the flooding situation would have\nbeen indicated by the radar using manual techniques.\nIn summary, analyses and prog charts were of little help in aiding\nthe forecasters to predict the occurrence of such locally heavy rain in\nadvance. Satellite photos did indicate a slowly expanding area of\nprecipitation over the flood area. However, the size and intensity\nof the area looked rather insignificant, especially compared to the\nother areas of precipitation in the same general area for which the\nforecasters were responsible. Radar could have been of some value in\ndetecting heavier precipitation over the flood area, but the identifica-\ntion of any train of cells that may have moved over the area would have\nrequired the services of a full-time radar observer, which were not\navailable at either TRI or at CRW the night of the flood. Weather\nstations utilizing radar data from TRI should be made aware that\nreported VIP levels may be too low in the mountain areas surrounding\nthat station to the north. Telephone coordination between radar\nstations, and between radar and nonradar stations, should be stressed.\n5. Hydrologic Conditions\nIn the headwaters of the Tug Fork, flooding was confined to two\nprincipal creeks, Peters Creek and Knox Creek. The Tug Fork forms the\nstate boundary between Kentucky and West Virginia. These creeks are\n14","fed by numerous small creeks such as Camp Creek, Hurricane Creek and\nPoplar Creek, most with drainage areas of less than 10 sq miles. They\nall sustained serious flooding.\nAcross the divide in the Levisa Fork Basin, flooding occurred on\nsmaller creeks, principally Slate Creek to the east of Grundy and the\nsmall creeks that drain into Slate Creek. Grundy has a flash flood\nalarm which was not activated as Levisa Fork rose less than half\nbankfull.\nHydrological guidance available to the forecast offices with Flash\nFlood Warning responsibility includes: Daily Flash Flood Guidance,\nLimited Area Fine Mesh Prog (LFM), Quantitative Precipitation Forecast\n(QPF), and Excessive Precipitation Forecast Chart.\nThe LFM, QPF, and Excessive Precipitation Chart showed no indica-\ntion of excessive precipitation over the flooded area. The Flash Flood\nGuidance indicated that it would take 3.2 inches of rain in three hours\nto cause flash flooding in Zone 8, which includes Pike County, Kentucky;\n2.6 inches were required in Virginia Zone 10, which includes Buchanan,\nVirginia.\nOn the Tug Fork at Williamson, West Virginia, the river crested\nat 20.5 ft., flood stage is 27 ft.\nThere are no formal river forecast points in the flood area. All\nthese creeks are small ungaged watersheds with warning services limited\nto community flash flood warning programs.\n6. Observing Systems\nInterpretation and use of available radar and satellite informa-\ntion by the duty forecasters was discussed in the earlier sections.\nThis section concerns a post-analysis to ascertain what additional\ninformation might be derived from these sources in an attempt to better\nunderstand the weather system that caused the flash flooding and perhaps\nprovide insight to earlier detection.\n6.1 Radar Film Analysis\nExamination of the 16 mm film from the Tri-Cities WSR-57 radar\nadded little additional information. Frames were available every five\n(5) minutes for several hours prior to and through the period of flood-\nproducing rains. A real disadvantage was the fact that VIP levels\nwere not available on the film. The precipitation areas were visible\nwith minor white shade variation in the strongest cells, but these were\ndifficult to pick out from the general precipitation area. The film\ndid indicate that several cells passed over or very near to the flood\narea. Four cells appeared to move over the area at approximately 0750\n15","GMT, 0820 GMT, 0840 GMT and 0918 GMT. The latter cell seemed to linger\nover the area until nearly 1000 GMT. This corresponds closely with the\ntime the strongest cells were observed on satellite as well as the time\na VIP level four (4) with tops to 45,000 feet was observed on the\nCharleston radar. VIP level three (3) with tops from 35,000 to 39,000\nfeet was observed at the TRI radar station according to information\ntaken from radar overlays at 0835 GMT and 0935 GMT. The presence of\nan extraordinarily strong cell or group of cells could not be\ndetermined from the film. VIP was inoperative on the remote scope.\nAntenna elevation angles were set generally at about one-half\ndegree, which is normal for routine radar search operation. Further\nelevation of the antenna was not necessary to observe the echoes in\nthe flood area.\n6.2 Satellite Imagery Analysis\nNOAA's operational east coast satellite during this event was\nSMS-2. Images from NOAA geostationary satellites are made available\nto National Weather Service WSFOs from a series of Satellite Field\nService Stations (SFSS) The NESS SFSS at Kansas City provides image\nsupport through the GOES Central Data Distribution System to 23 WSFOs\nin the central third of the United States, including WSFO Louisville.\nThe SFSS at Washington, D.C. similarly serves 14 WSFOs including\nWSFO Washington, D.C. and WSFO Charleston.\nBoth visible and infrared (IR) imagery are transmitted during\ndaylight hours; only the infrared images are available at night.\nThe meteorologists on duty at each NESS SFSS prepare Satellite\nInterpretation Messages (SIMS) every six hours for transmission over\nthe RAWARC teletype circuit. The messages are routinely released at\nabout 0600Z and 1200Z. The 0500Z SIM from the Washington, D.C. SFSS\nwas released about an hour before convective cells began building over\nthe Virginia-West Virginia-Kentucky border area. It stated, \"High\nclouds over western West Virginia southward to western North Carolina\nis moisture from thunderstorms which dissipated Saturday AM. \" The\n1200Z SIM from Washington, D.C. was released 3 hours after the most\nintense phase of the storm and stated \"Loops show vorticity max in\nsouthwestern Virginia near Tri-Cities moving slowly southeastward.\nMid/hi level overcast, scattered thunderstorms near and ahead of\nvorticity max are drifting southeastward into southwest Virginia,\nwestern North Carolina and northern South Carolina.\" A check of telephone\nlogs at the Kansas City and Washington, D.C. SFSSs revealed no telephone\ncalls either outgoing or incoming on July 15 from Louisville, Charleston\nor Washington, D.C.\nThe SFSSs can also request Quantitative Precipitation Estimates\n(QPE) from the NESS Synoptic Analysis Branch (SAB) in Camp Springs,\n16","Maryland, for relay to the WSFOs. The technique gives half-hourly or\nhourly rainfall estimates for county areas using specially enhanced GOES\nthermal IR imagery together with high resolution visible images, if\navailable. The factors considered include cloud top temperature, rate\nof anvil growth, position of the cumulonimbus under the anvil, duration\nof the storm as well as the effect of merging cells, merging convective\ncloud lines and overshooting tops.\nDue to the small size of the cloud cells over the impacted area and\ntheir relatively warm tops, no real-time precipitation estimates were\ngenerated. Instead, a check of SAB logs showed that on July 15, QPE\nwere done for Kansas (twice), Iowa, Colorado and New Mexico. Convective\nsystems over these states showed up on the imagery as larger and deeper\nthan those over the Virginia-West ginia-Kentucky border area.\nRod Scofield of NESS prepared retrospective QPE from the satellite\nimagery on file and arrived at the following precip totals for July 15,\n1979, in the affected area:\nPike County\nBuchanan County\n0500-1030Z\n5.42 in\n3.85 in\nSatellite imagery did show a small convective cell beginning to\nbuild over Pike County, Virginia, at 0530Z. This activity developed\nas follows:\n0600Z - Cell over Pike County has doubled in size over past half-hour\nbut is still quite small. Cloud top temperatures are in the\n-41° to -52° range. Grid is 1° lat. too far WSW (250°).\n0630Z - Convective system has increased in sizeb y about a third in\npast half-hour and now covers part of Buchanan County as well\nas Pike. Grid 1° lat. too far NNW (340°). Convective system\nin Ohio warming.\n0700Z - Convective system has increased slightly in size. There has\nbeen a merger of cells over Pike and Buchanan Counties. Grid\nabout 1° lat. too far NW (320°)\n0730Z - (Figure 6.1) Convective system has more than doubled in size\nin past half-hour as shown by IR and now covers Lawrence,\nJohnson, Martin and Floyd Counties in Kentucky as well as Pike.\nTops still in the -41° to -52°C range. Grid OK. Notice large,\nintense convective systems over western Ohio and in the\nNebraska-Kansas-Iowa-Missouri region.\n0800Z - Convective system increasing in size. New cells forming in\nWest Virginia over Mingo and Logan Counties. Grid OK.\n17","0730 15JL79 22E-2MB 01041 18061 DB5\nFigure 6.1\n18","Figure 6.2\n19","0830Z - Cells over Mingo and Logan Counties have merged in the past\nhalf-hour. Overall system growing. Grid OK.\n0900Z - System still growing. Cell over Mingo County has merged with\nthe main larger system in Kentucky and now covers parts of\n10 counties along the Virginia-West Virginia-Kentucky border.\nMost intense part of system is directly over Pike County where\ntops have dropped into the -52° to -58°C range.\n0930Z - (Figure 6.2) System still intensifying as shown by blown-up\nIR. Dark gray patch (coldest cloud tops) now covers parts of\nBuchanan and Mingo Counties as well as Pike.\n1000Z - System still expanding but coldest tops have been lost\nindicating slackening in rainfall intensity.\n1030Z - System no longer growing.\n1100Z - System weakening over the Virginia-West Virginia-Kentucky\nborder.\nIn summary the small cell that eventually increased in intensity\nand size to produce the flash flood was noticeable in the post analysis\nat 0530Z (0130 EDT). It continued to develop with cell mergers evident\nat 0700Z (0300 EDT). From accounts of residents in the area, some of\nthe heaviest rainfall began about this time lasting up to two hours\nbefore ending. This timing coincides with the apparent maximum intensity\nobserved by satellite between 0900Z and 0930Z (0500 to 0530 EDT).\n7. Response\nThese creeks are all small ungaged watersheds with no formal river\nand flood forecast service provided. Local community flash flood\nwarning systems are being implemented in the 12-county Appalachian\nflash flood pilot project and are the only services available to the\nresidents. Such systems have been started in Pike County and Buchanan\nCounty.\nThe NWS operates a NOAA Weather Radio at Hazard, Ky., with a low-\npowered repeater at Pikeville. The flash flood coordinators, police\nand others have receivers which are their primary means for alert when\nthe NWS issues a Flash Flood Watch or Warning. However, complaints\nwere noted from a number of people that their receivers are subject to\nfalse \"turn-on\" alerts. Some no longer monitor the radio because of\nthis problem. The NOAA Weather Radio is vital as most radio stations\ngo off the air at night, some as early as 9 p.m.\n20","False \"turn-on\" alerts can generally be attributed to the fact\nthat many of the earlier and more inexpensive radio receivers pick up\nextraneous tone pulses because their response selectivity is poor.\nThese receivers usually do not have suitable tone decoders to screen\nout tone frequencies other than those being transmitted by the NWS at\n1050 Hz. Some receivers have been known to be triggered by signals\nranging from 700-1800 cycles.\nSince the radio receivers are for the most part privately owned,\nany modification to improve the tone signal discriminating capability\nhas to be done by the owners. It is usually less expensive to replace\nthe older receivers by those presently being marketed than to make the\nnecessary modifications.\nAlthough the alert tone can inadvertently be switched on by the\noperator at the NWR Broadcasting Console, this is not a very common\noccurrence.\nThere is no general warning system in the area such as fire alarm\nor civil defense siren to warn residents. The hills are confining and\nroads winding and no warning system except one designed for each of\nthe small creek basins, known locally as \"hollows\" - can be truly\neffective.\nThe local flash flood rainfall reporting network did not function.\nVery few of the observers lived in the disaster area and those who\ndid, failed to call in their reports. Some called their reports in\nat 7 a.m. One raingage is at a location that isn't open on weekends.\nSome gages are questionably located--such as on telephone poles.\nEven though there was no NWS watch or warning issued prior to the\nevent people responded fairly well within their areas. Neighbors\ntelephoned neighbors as long as the telephones were in operation\nadvising each other the creeks were rising, overflowing or that a\nwashout had occurred. This information was passed to the civil defense\ndirectors. None of the information apparently left the area -- at\nleast the NWS never was advised until routine data collection time around\n7 a.m. -- well after the rain had ended.\n8. Significance for Pilot Flash Flood Program\nSince the event, NOAA representatives have met with local and state\nofficials to consider means for improving the effectiveness of the local\nwarning systems as a part of the pilot flash flood program. of course,\nmuch of the instrumentation is yet to be placed. Many of the observers\nhave yet to be recruited.\nBeyond these rather easily surmountable problems, there remains\nthe fact that those observers already recruited and equipped did not\nreport until after the storm was past. Apparently the observers did\nnot feel a report was necessary. Additional training is planned for\n21","all observers and coordinators, and it is anticipated that once the\nsystem is complete, participation will be increased.\nEven though, in this case, no information was relayed to NWS\nuntil after the fact, it is notable that there was little loss of\nlife. This may be fortuitous, or it may mean that the neighbor-to-\nneighbor warnings were effective in saving lives. It is clear that\nthe lead time for possible warnings is indeed very short for events\nof this kind in Appalachia. The terrain may demand more flash flood\nalarms and automated response devices. Communications are very\ndifficult in the area and more work needs to be done on NOAA Weather\nRadio coverage.\nThe appropriateness of the issuance of a delayed warning has\nbeen questioned. It was justified in the minds of the forecasters\non the chance that the threat was not yet ended. It may be that a\ndelayed warning causes confusion. Failure to issue a warning while\na threat still exists is probably a worse error. On balance, it\nwould seem that a warning should be issued so long as the forecaster\nrecognizes some continuing threat.\nThe widely dispersed warning responsibility for the counties\naffected by this event leads one to questions whether a better\narrangement might not be found. In order to maximize the probability\nof success for the pilot program in the area, it would seem desirable\nto review the assignment of warning responsibilities to assure maximum\ncoordination and optimum communications Ideally, the communication\nlines to the responsible office (s) should be short and highly dependable\nReassignment of warning responsibilities could be achieved within a\nrelatively short period of time and might have a beneficial effect on\nthe success of the pilot project in the months immediately ahead.\n22"]}