QC 995 : . U61 F I no.58 M Technical Memorandum NWS CR-58 c . 2
GUIDELINES FOR FLASH FLOOD AND SMALL TRIBUTARY FLOOD PREDICTION
Lawrence A. Hughes, Scientific Services Division Lawrence L. Longsdorf, Hydrology Division
ATMOSPHERIC SCIENCES LIBRARY ■MjirLi—... DEC 4 1S75
N.O.A.A. Central Region Headquarters U. S. Dept, of Commerce October 1975
NATIONAL OCEANIC AND / National Weather noaa ATMOSPHERIC ADMINISTRATION / Service
fb 4337 ROM TECHNICAL KfHOfUMO* Rational Vaathar Sonrlco, Cantral Region Subsarl** _ ... , v , s.rriea Central Rerlon (CR) eub.eries provides an Informal medium for the documentation and quick dla.emlnatto* Ua Rational V.AtWS.rrte f#f, forMl The a.rlaa 1. used U, report on work In progre.a, to d.acrlb. of result* not arproprietSD r u mUu to a limited audience. Those Technical Memoranda will report on lnveatlga- technlcal procedure* and pr« e ^ 1#m1 bl„, ,f interest ualnly to regional pereonnel. and haneo will not be widely distribute, tlona devoted primarily to r»r* . , . „ the forwar ■erlee, ES54 Technical Memoranda, Central Region Technical Memoranda (CRTM)| papers 16 to 36 er* Papere 1 to H are In tne tore W„th.r Bureau Technical Memoranda (VBTN). Beginning with 37, the papere are now part In the for»*r earl**, '■ ->» . .... * of the .arias, NO»A Technical Memoranda W5. . .V . a m or COM number are available rroe the WatlonU Technical Information S.nrlca. °* S* D.partmant of Corm.ro., Fap«r« that h*** a or .. 221)1. MB; BMMMBMVl Order by accaaalo n mvnbar ah own la 518} Port Royal Road, Sprlnrr e , ,r^7IEblTr»Sn tha National Vaathar Service Central Region, SclentlMe Service. M^iIIi^! Roti 183S, toi r.. 120* street, Mn,a* city, Po. m.104. pr^ces vary for all paper copy $2. 25 microfiche TSSA Tachnlcal Kemorand* - i.At.tlon Probability forecaat Tarlflcatlon Surmar7 Noe. 1965-Mar. 1944. •‘VSD Staff, VDCKH - Nay 1964 cm 1 4*^1 ulr of S«>r Showers Over the Colorado Mountains. Vn. C. Sullivan and James 0. Severson - Juno 1946 CRH! 2 W Shower Dlatrlbutlor — °* I"'"on " JuM 1964 CR7N J CP.IN 4 Heavy Rain, In Colorado^ cm 5 The Plum Fire. w»* • .f vbcrh _ September 19SS Precipitation ProbebllK October 1944 cm 6 Effect of Diurnal V«.th< #££^1964 cm 7 Climatic Frequency of P. ,Tmb*r 1966 cm 8 CP.IX 9 Heavy Snow or GIaxInf.* 10 Detection of a Weak Front by WSR-57 Padar. ------0 - cm *111. Probability Foroca»t*. S5D Sta/f, WDCRII - January 1967 cm 11 «■ 5 forecasting In tha Central United SUtee (An,Interim Report). SSD Staff - January 1967 cm 12 Dturnal Surfaea Grostrophle Wind Variations Over tha Croat Plains. Vayne E. Sangster - Kerch 1967 CRH! 1) .*...... Probability of Sunmertlma Precipitation at Denver. Vo, C. Sullivan and James 0. Sevaraon - Kerch 1967 cm 14 t _ .._. Prvcloltatlon Rroboblllty Forecs-ts Uelng tha Central Region Verification Printout. Lawreneo 4. Hughs* - Nay 1967 cm 15 Small-Scale Circulations A,,oclsted With RaJlalional Cooling. Jack R. Cooley - June 1967 VBIK CR 16 rmhikltltv Verification Peoulta (6-Konlh and 18-Month). Uwrenc# 4. Hughes - June 1967 ___ WBT11 CR 17 On tha Use and Misuse of the Drier Verification Scoro.. Uwrenee 4. Hughes - Auguat 1967 (PB 175 771) VDTK CR 18 Probability Verification Results (74 Months). Lawrence 4. Rughet - February 1968 VDTH CR 19 Radar Depletion of the Topeka Tornado. Herman F.. I'ro.eer - April 1968 vtmi CR 20 Vinrl Vavaa on th. Croat Lakae. Lawrence A.JIughee - Hay 1968 VBTN CR 21 find Vavaa on th. Cr-at taka,. U.renee 4. Hughes - Ha, 1968 . ieaaonal Aap-cte of rrobabUlty Forecastai 1. Summer. Uwrenc# 4. Hughes - June 1968 (TB TiJ 1 VBTN CR 22 • eaaonal Aerecle of ProbablUty forecaatai 2. Fall. Uwrenee 4. Hughes - September 1968 (PB 185 734) VBTN CR 23 The Importance of Areal Coven re In Precipitation Probability Pornenelln*. John T. Curran and Ldiwycne*Uwvanoo aA. - 5*pi 1964 WOTlt CR 24 Pa teo ml or* cal Condition, a" Relale-l to Mr Pollution Chicago, Illtnole, April 12-13, 1943. Charlee H. JS waa - October 1964 VDTH CR 25 “Kelt of Probability Forecaatai J. Wlntar. U-r.nc. A. Hu,he. - U.cemb.r 19M r» 105 2* VDTK cn 24 'aaeonal Aapect, of Probability Tor-casti: 4. Spring. Uwranca 4. Hughes - fabruary 1949 (PB 105 736J VBTN CR 27 Minimum Temperature Forecaellng During. Poealbla fro.t Parloda at Agricultural Vaathar Stations In Western Michigan. VDTTM CR 28 Marshall 4. Sod.rb.r* - Parch 1969 . ,______. ,OM An Alo ror Tornado Warnings. Harry V. Valdheusar and Uwrenee 4. Hughes - April 1969 VBTN CR 29 An Aid In Forecasting Significant Uka Snoeea. H. J. Rolhrock - Hovember 1969 VBTN CR. 30 van CR. 31 4 Forecast Aid for Boulder b'lnda. Wayne E. S.angatar - February 1970 ...... An Ob'eetlv# Method for Fstlmatlng the Probability of Sever# Thunderstorms. Clarence L. David - February 1970 VBTN CR 32 vm CR 33 Kentucky Alr-Soll Temperature Climatology. Clyde B. Ua - February 1770 Effective U«e of llon-Structural Kethoda In Water Kanavement. Varna Alexander - March 1970 __ VBTN CR 34 A Note on the Categorical Verification of Probability Forecasts. Lawrence A. Hughes and Wayne I. SangaUr - Auguat 1970 VBTN CR 35 A Comparison of Observed and Calculated Urban KLtlng Depths. Donald E. Vuereh - Auguat 1970 VBTN CR 34 IJOAA Technical Memoranda NWS
Forecasting Maximum and Minimum Surface Temperatures at Topeka, Kanaaa, U.lng Guidance fron tha PE Humartcal Pirodlctlo* HV5 CR 37 Pcriel (Ftsjs). Morrio S. Webb - November 1970 (COM-71-OOU8) Vnow'rorecajtlnr for Southcaatem Vlaconaln. Phrlnhart W. Hama - Hovwmbar 1970 {COK-71-00019) , 1.VS CR 3* A Svnootlc ClLmatology of Billiards on tha forth-Central Heine of tha United States. Robert. E. Black - reb 1971 (CON-71-00369) W5 CR 39 the Spring 1969 PJdveet Snowmelt Floods. Herron F. Hond,ehtln - February 1971 (COH-71 -00489) VWS CR 1.0 Th^Tmoeroturo Cyela^o^Uka Michigan 1.(Spring and Sumer). Uwranca 4. Hughes - April 1971 (C0H-71-0O545) KVS CR a MS Duet Devil Meteorology. Jack R. Cooley - May 1*971 (C0H-71-00628) , CR 42 Summer Shower RrobabUlty In Colorado a. Related to Allltuda. Alol. G. Topll - Mar 1971 (COM-Tl-OOTU) NWS CR 43 An Inveatiratlon of the Resultant Traneport Wind Wttnln the Urban Complax. Donald E. Wuarch - Juno 1971 (COM-71-00764) IMS CR 44 The Relationship of Soma Cirrus Formations to Severe Local Stoma. William E. Villi am. - July 1971
(Continued on back inside cover) VT0< h S’ NOAA Technical Memorandum NWS CR-58 <£, 'JL
GUIDELINES FOR FLASH FLOOD AND SMALL TRIBUTARY FLOOD PREDICTION
Lawrence A. Hughes, Scientific Services Division Lawrence L. Longsdorf, Hydrology Division
LIBRARY
N.Q.A.A. U S. Dept, of Commerce
Central Region Headquarters October 1975
UNITED STATES NATIONAL OCEANIC AND DEPARTMENT OF COMMERCE ATMOSPHERIC ADMINISTRATION F Rogers C. B. Morton, Secretary Robert M. White, Administrator d GUIDELINES FOR FLASH FLOODS AND SMALL TRIBUTARY FLOOD PREDICTION
Lawrence A. Hughes, Scientific Services Division Lawrence L. Longsdorf, Hydrology Division Central Region Headquarters
This technical note, augmenting Manual Chapter E-13, was devised to pro vide all offices with guidelines for determining the threat and extent of flash floods and other small tributary floods. Basically these involve knowledge of when, how much and how fast rain came down and over how much of a particular river basin, and some method for converting this to river stage or flood potential. This note discusses these two points. Offices with a radar or a WBRR recorder play a key role in such efforts as they have knowledge of the extent and location of heavy rain that other offices, at times, do not have.
We will start with anticipating and locating heavy rain areas. Then we will estimate the extent of the flood, with increased precision and objectiv ity, depending on how much specific information is available for the pur pose, be it nothing but experience or the maximum consisting of flood in dices and advisory tables.
HEAVY RAIN THREAT AREAS
Practically all heavy rains in the Central Region occur with the convective type weather usually associated with thunderstorms, but they may occur without thunder. When thinking about the day ahead, one should consider that there is a significant possibility of heavy rain in the SELS AC area for thunderstorms. Further guidelines are available in the various QPFs on facsimile, particularly the maximum-precipitation panels. But regard less of previous guidance, one should always be alert to heavy rai" possi bilities whenever thunderstorms are imminent or occurring, especially if the moisture content of the air at and below 700 mb is higher than average for the area and season and/or a lifting mechanism such as warm advection or positive vorticity advection at 500 mb is present.
Once rain has started, radar data, especially when used in conjunction with satellite data, is the best initial means of being alerted to the extent and intensity of possible heavy rainfall. The manually digitized radar (MDR ) data is especially helpful. When the MDR data indicate sizable amounts have fallen in an MDR box (the sum of MDR digits is 16 or more in two hours or 20 or more for four hours) in your area of responsibility, the radar officj closest to the rain area should be consulted for information on location and movement of echoes. In these suspected areas of heavy rain, calls should be made to telemetered rain gages and to rainfall observers, the latter pre ferably by RDOs or normal collection centers. Observer calls during nor mal sleeping hours should be reserved for serious emergencies.
ESTIMATING THE FLOOD
Two types of flood indices to be discussed in detail later are routinely made available for practically all parts of the Central Region. Where such an index is not available for any reason, one would have to resort to experience and the table such as that below to judge the flash flood problem.
Rainfall Rates & Durations for Flash Flooding
R ate Duration at a point Extent of Problem (In/hr)
0. 5 - 1. 0 1 to 6 hour s Minor problems possible in some urban areas
1. 0 - 2. 0 1/4 hr to 1/2 hr Minor problems in some urban areas 1 hr Local flooding
2. 0 - 5.0 1/2 hr or more Flooding
Over 5. 0 1/4 hr or more Flooding
The table is most appropriate for the hilly areas for which flood indices are not provided, but it can be used anywhere as a rough guide. Obviously in flatter areas the lower limits would be less likely to produce the effects indicated.
Flood Indices
Flood indices show the amount of rain falling in a 3-hour period that will lead to flooding on the smaller streams. They are put out on RAWARC by the RFCs for their areas of responsibility, and are of two styles. One style is put out daily, and is called the "Average 3-hour Flash Flood Rain- fall Guidance by State Forecast Zones". It is grouped by states, and a value is given for every state forecast zone, except those in Colorado and Wyoming not in the Missouri Basin. Each RFC handles the zones in its area of responsibility, and if a zone straddles more than one RFC area of responsibility, the zone is assigned to a single RFC by their mutual agree ment.
-2- The Zone Index provides generalized guidance for multi-basin sized areas in practically all of the region. It is especially useful for determining the need for a flash flood watch before the rain has started, or before an esti mate of rainfall amount and coverage' can be made. It is also used in pre paring warnings where no other guidance is available. In such a case, a warning should be issued when it is quite likely that the amount of rainfall will exceed the Zone Index value. Where zones are large, there is a greater possibility that the zonal index is not representative for particular basins. Some basins in the zone may have greater runoff than average either because the basin has steeper slopes than average, has more impervious soil, or has more moist ground (usually the most recent rainfall). If runoff is larger than average, the amount of rain to produce flooding should be less than that given by the Zone Index, and the threat of flash flood correspondingly higher.
The other style of index is for a sizable number of representative river basins and is called "Flash Flood/Headwater Basin Crest Stage Guidance". We will call it the Basin Index for simplicity. The Basin Index is a better index for the basin to which it applies than the Zone Index, as it considers the specifics of that basin as to slope, soil moisture and permeability, and size, etc. At times the Basin Index for a nearby and similar basin can be used to improve on the Zone Index for a basin for which there is no Basin Index.
The values of the Basin Index are grouped by RDOs, are put out twice a week (Monday and Thursday afternoons), and are updated as necessary (after signi ficant rains). This index provides specific guidance for particular basins (which are generally quite a bit smaller than forecast zones), and also allows quantitative prediction of peak stages for specific cities when used in con junction with the Flood Advisory Tables discussed below (for each basin for which there is a Basin Index there is a Flood Advisory Table). Thus the Basin Index is primarily for flood warning purposes, when fairly good in formation is available on precipitation amount and areal coverage. However, the Basin Index is available for less than half the basins in the region. The location of the basin to which a particular index value applies is made easier by a listing in the RAW ARC message of the state forecast zone(s) appropriate to each basin.
Offices having flood index values for their areas of responsibility should plot, or at least know, these values. To facilitate such plotting, the Kansas City RFC has provided each office having flash flood responsibility with a map of rivers and basins and a plastic overlay for the map which outlines the state forecast zones and the basins for which indices are supplied.
If either index is under two inches, almost any thunderstorm has the potential to produce flooding (at least on very small streams). However, if the index is over four inches, it would take much more favorable thunderstorm con ditions to yield this amount, such as unusually high atmospheric moisture
-3- conditions and/or slowly-moving thunderstorms. These indices and know ledge of the meteorological situation will be the extent of the data needed for putting out flood watches and warnings. However, if the rainfall is heavy enough to exceed that given by the Basin Index, as soon as one can get a fair ly good estimate of the rainfall amount and areal coverage from rainfall ob servations or from radar estimates, one can get more specific about the time and magnitude of the flood for those basins for which a Basin Index is routinely available. This is because for each such basin there is also a Flood Advisory Table, described later, from which such information is obtained. To use the table one needs information on the average basin rainfall.
Determining Basin-Average Rainfall
Basin-aver age rainfall can be determined from rainfall reports or estimated from radar where one is available. No matter which is used, it is usually best to get the rainfall for about the number of hours indicated in the Flood Advisory Table. Usually this will be 3 hours or 6 hours. If the rainfall has been particularly heavy, one should use the table as soon as the basin-aver age rainfall exceeds the Basin Index value, so at these times the basin-aver age rainfall may be obtained for fewer hours than indicated in the table. The forecast can be updated later if rainfall continues to about the duration speci fied by the table. The advantage is earlier warnings when critical values are reached earlier than postulated by the table.
In operations, it is best to get an estimate of the basin-aver age rainfall from radar first, when possible, as it can give earlier estimates through a bit of extrapolation, and it can be used for areas where no rainfall reports are avail able. Rainfall observations are used to check the radar estimate, or in loca tions or at times when radar estimates are not available.
1. From Radar Estimates
It is not easy to get a really good estimate of basin-average rainfall fi om radars in heavy rain situations because of the many simplifications that are usually made, as in most cases, each of these tends to overestimate the rainfall. For example, the peak rainfall rate usually covers only a small portion of the basin. The area of the radar cell is greater than the rain area at the ground because of beam spreading and other reasons; the radar rate derived from the radar intensity is usually greater than the rate at the ground because of evaporation of falling drops and suspension by updrafts of the smaller raindrops. It is more difficult to get a good estimate for a large basin than a small one because the longer time for echoes to cross it and because of the problem of the percent of the basin receiving rainfall.
The following simplified example illustrates the factors to be considered.
-4- It is a method of making an estimate for a small basin and time period. Such an estimate can be useful in anticipating future rainfall where, if one waits for rainfall observations, one may not provide adequate warning lead time.
r adar e cho
First, extrapolate the echo direction of motion and development to determine the portion of the basin covered by the echo as it moves across it. In this case it is about 55% of the basin (the cross-hatched area). Second, from the echo speed, determine the time it would take for the center of the echo (or any other point) to cross the basin. We will use three-fourths of an hour. Third, determine the average rainfall rate for this echo. The rate near point X (or at any point on a circle through point X), which is a bit over half-way from the center of the echo to the echo edge, would generally be appropriate. This assumes several things about the rainfall intensities in the echo, but it is a reasonable effort to get away from the peak or near peak values which cover such a small area. In this case let us take this average rate as 2. 0 inches per hour (the maximum radar rate may be 5.0 inches per hour or more). Fourth, obtain the basin-aver age rainfall as the product of the areal cover- age (0. 55), the time to cross the basin (0. 75 hours), and the average echo rainfall intensity (2. 0 inches per hour) i. e. 0. 55 x 0. 75 x 2. 0 gives 0. 8 inches as the basin average. It is easy to overestimate the basin average by a considerable amount unless the area of the basin covered and a less- than-peak rainfall rate is taken. For large basins and larger time periods, extrapolation of radar echo areas and intensities will probably n% be success ful, as changes in the echoes cannot be adequately anticipated for such time periods. For these basins, work with the past echo conditions, or mostly past conditions, but the task is more difficult with multiple echoes and chang ing echo areas and intensities.
Digitized radar is of little help for specific basins, except as an alerting tool. With the current manually-digitized radar (MDR) code, it has been noted on the small sample of cases studied that the largest amount of rainfall in inches in an MDR box is about 10% of the sum of the MDR digits if rainfall occurs for six hours or more. Of course the location of that rainfall is not certain
-5- with so large a box, but checking with the radar office should narrow it down.! If the proposed radar code change takes place, the MDR boxes will be one- fourth the current size, more precise locations will then be possible, and when the electronically digitized radar becomes widely available, even more precision will be possible in both amount and location.
2. From Observed R ainfall
Use of observed rainfall is an easier method than using radar, but if the reports are few or far from uniformly distributed, or, in a large basin in which multiple radar cells were involved, the rainfall pattern may be quite uncertain. Nevertheless, the method is the same for all. First, plot all rainfall amounts in and near the basin. Then draw lines of equal rainfall amounts across the basin. Determine the average basin rainfall by eye based upon these isolines.
Example: Given Basin A below
Total rainfall amounts at observation points:
(1) = 3.00" (2) = 2.00" (3) = 3.50"
(4) - 1. 50" (5) = 1.80" (6) = 0.50"
Average is about 1. 8 inches.
D ! ! 3"
-6- FLOOD ADVISORY TABLE
Flood Advisory Tables have been prepared by River Forecast Centers for hundreds of locations, and each RDO and selected WSOs has a copy of any available for its area of responsibility. These use the basin-aver age rain fall amount for (approximately) the period specified on the table to give the expected highest water level at a forecast point (river gage). The table also indicates the average amount of time from the end (or middle, as indicated on the table) of the heavy rainfall to the peak water level. The index allows for initial river stage and ground moisture by entering it with the Basin Index value for the day. On each Table there are instructions and an example of its use, so further discussion is not needed here; however, the table is not used unless the basin-aver age rainfall exceeds that given by the Basin Index, except in one or two cases shown on particular tables.
SUMMARY
1. Heavy rains that produce flash floods come mainly from thunderstorms, but thunder won't necessarily be reported, so watch the SELS AC, the QPF facsimile guidance, and later the radar observations, including manually digitized radar data.
2. If thunderstorms are a possibility, look up the latest RFC flood indices (from RAWARC). When the index is under two inches, almost any thunderstorm can produce at least localized flooding.
3. If the expected 3-hour rainfall equals or exceeds the zone or basin flood index value for an area, issue a Flash Flood Watch (WSFOs), as re quired in accordance with Manual Chapter E-13. Remember that a watch on the average indicates a considerably lower threat of the flood than a warning, so it should not be expected to verify as often.
4. When thunderstorm rainfall is about to occur over a basin, especially a small basin, or is already occurring, estimate the basin-average rain fall. This can be done using radar, especially for small basins, or, later, using rainfall observations.
5. If the basin-aver age rainfall exceeds the Basin Index, issue a Flash Flood Warning. If no Basin Index is available, use the Zone Index.
6. When a basin-average rainfall value exceeds the Basin Index value, RDOs and selected WSOs should use the appropriate Flood Advisory Table in conjunction with the Basin Index to forecast and issue the time and magni tude of the maximum stage at the particular point on the stream. If a Flood Advisory Table is not available for a desired location, Tables for
-7- adjacent areas can be used as a guide. The closer and the more similar the basin to that of the table used, the better. Statements on the degree of flooding for such situations must be more generalized than when a specific Flood Advisory Table is appropriate.
ACKNOWLEDGEMENT
The authors wish to express their appreciation to Mr. Herman Mondschein, HIC of the RFC at Kansas City, for his editorial and technical assistance, and Mr. Guy Gray, NSSFC, for his ideas on using radar to determine basin- average rainfall. Special thanks is given to the field offices which contributed to the intensity vs. duration table on page 2.
A REQUEST:
For those basins for which a Basin Index is not available, we would appreci ate receiving information on how local experience suggests that the Zone Index be modified for each of these basins, or, if no zone index exists, how the table on page 2 is modified for specific basins. We would also appreci ate knowledge of specific cases where the Flood Advisory Table for a basin did not provide reasonably good crest values.
-8- (continued from front inside cover)
NWS CR 48 Manual of Great Lakes Ice Forecasting. C. Robert Snider - December 1971 (COM-72-10143) NWS CR 49 A Preliminary Transport Wind and Mixing Height Climatology St. Louis, Missouri Donald E. Wuerch, Albert J. Courtois, Carl Ewald, Gary Ernst June 1972 (COM-72-10859) NWS CR 50 An Objective Forecast Technique for Colorado Downslope Winds Wayne E. Sangster December 1972 (COM-73-10280) NWS CR 51 Effect on Temperature and Precipitation of Observation Site Change at Columbia, Missouri March 1973 (COM-73-10734)
NWS CR 52 Cold Air Funnels. Jack R. Cooley and Marshall E. Soderberg September 1973 (COM-73-11753) NWS CR 53 The Frequency of Potentially Unfavorable Temperature Conditions in St. Louis, Missouri October 1973 Warren M. Wisne]
NWR CR 54 Objective Probabilities of Severe Thunderstorms Using Predictors from FOUS and Observed Surface Data (COM 74-11258) Clarence L. David May 1974
NWS CR 55 Detecting and Predicting Severe Thunderstorms Using Radar and Sferics John V. Graff and Duane C. O'Malley June 1974 (COM-74-11335) NWS CR 56 The Prediction of Daily Drying Rates Jerry D. Hill Nov. .1974 (COM - 74-11806) NWS CR 57 Summer Radar Echo Distribution Around Limon, Colorado Thomas D. Karr and Ronald L. Wooten Nov. 1974 ( COM-75-10076)