HEAVY-FOG REGIONS in the CONTERMINOUS UNITED STATES ROBERT B

i 16 MONTHLYWEATHER REVIEW Vol. 97, No. 2

UDC 551.575.2:561.582.2(73) HEAVY-FOG REGIONS IN THE CONTERMINOUS UNITED STATES ROBERT b. PEACE, JR.

CornellAeronautical Laboratory, Inc., Buffalo, N.Y.2

ABSTRACT The heavy-fog statistics for 256 first-order weather stations were utilized to update analyses of the geographic distribution of fog within the conterminous United States. The survey shows that heavy fog (visibility one-fourth mile or less) occurs more than 20 days a year at approximately 50 percent of the first-order weather stations (of which 229 are air terminals) and that the mean heavy-fog frequency reported per station appears to be generally higher than it was 30 yr ago. The geographic distribution of heavy fog is shown in two ways-by a conventional isopleth analysis and by fog climatic regions with characteristic frequency distributions.

1. INTRODUCTION number of days with heavy fog as reported in the 1963 There is periodic need for knowledge of the frequency annual ~ummaries.~These statistics a,re based upon the distribution of heavy fog at air terminals in the United total length of record at each station and representwidely States. A survey of the literature in 1964 revealed some differing numbers of years of observationand often a old or simplified fog-frequency distributions(Ward, number of different locations in or near a given city. Some 1925, Stone, 1936, U.S. Department of Agriculture, stations have as few as 2 yr of record, while the frequencies 1941, and Haurwitz and Austin, 1944) and distributions of heavy fog for other stations arebased upon as many as of fog bytype or mechanism(Stone, 1936, Byers, 60 to 80 yr of observations. A few stations hadlong records 1959) but noreasonably current, detailed, quantitative for earlier years, but a substantial change in station loca- fog-frequency analy~is.~This lack motivateda brief tion reduced the record to 10 yr or less as of 1963. Earlier climatological study and the summary presented here. data were acquired for these stations and compared with The firstfog-frequency distribution produced was a the morecurrent, shorter period statistics. In most conventionalisopleth analysis such as those that have instances where such a doublerecord was available, been published in the past. This provided a basis of com- the long-period statistics indicate fewer days with heavy parison with older studies, but such analyses can be mis- fog than do those of the morerecent, shorter period. leading. Fogis largely a localized weatherphenomenon Table 1 shows the comparison between long- and short- and the natural station-to-station frequency variation in some areas requires subjective analysis. TABLE 1.-Comparisonbetween long- and short-termheavy-jog Study of the isopleth analysis reveals that the United frequency States can be divided into regions of reasonably common Mean Period zgl Period annual of I I of terrain and geographic properties that are strongly related Station heavy-fog record heavy-fogrecord to fog frequency. The conterminous United States hasbeen frequency (VI) frequency (yr) (dwslyr) divided into sevensuch regions and the heavy-fog frequencydistribution for each provides an essentially Tallahassee, Fla...... 55 2 41 22 objective insight into the frequency distribution of heavy Lake Charles, La...... 51 2 37 23 Laming, Mich ...... 24 9 13 45 fog at air terminals. Rochester, Minn...... 38 3 16 11 Saint Louis, Mo ...... 10 6 8 22 2. DATA BASE ANDPERIOD OF RECORD Lincoln, Nebr ...... 7 8 5 68 Atlantic City, N.J ...... 43 5 21 53 CONSIDERATION Williston, N. Dak ...... 9 2 7 45 Toledo, Ohio...... 20 8 8 a3 The source of data for this study was the U.S. Weather Fort Worth, Tex ...... 10 10 8 55 45 Bureau, Local Climatological Data summaries for 256 Alpena, Mich- ...... 18 4 20 Glasgow, Mont ...... 11 8 12 18 first-orderweather stations in the conterminous United Havre, Mont ...... 5 3 5 56 14 States. Thebasic fog statistics used were the mean annual Victoria, Tex...... 24 2 28

' This work was sponsored by the National Aeronautics and Space Administration under Contract No.NAS-156. Present affiliation, Atmospheric Water Resources Research, Fresno State College Foundation, Fresno, Calif. Since this work was performed a new fog-frequency distributionhas been published by Court (1966).

Unauthenticated | Downloaded 09/25/21 04:17 AM UTC February 1969 Robert L. Peace, Jr. 117 termmean heavy-fog frequency for the 14 stations for Figure 2 is the dense-fog frequency analysis published which such data were available. byStone (1936) for data through 1931 or 1932. (Dense Althoughthere are several possible explanations for fog was defined as a fog that restricted visibility to 1,000 the difference between more current andolder means-e.g., ft or less in contrast to the one-fourth-mile (1,320 ft) changes in observing procedure, climatic change, increased criteria presently used for heavy fog.) Relatively few of industrialization,changes airportin or observation the stat’ions in figure 1 had more than 30 yr of record in sites-the available data were not sufficient to ascertain 1963. So, Stone’sanalysis and figure 1 were essentially thetrue cause. However, the mere existence of these drawn for mutually exclusive data. A comparison of the differences, coupled withthe wide variationin record two patterns shows that both have similar basic features, length for the 256 stations used in the analysis, does sug- but the current analysis displays a 10- to 20-day higher gest that length of record may contribute to differences overall fog frequency everywhere except in the dry Great in recorded fog frequency between stations ina given area. Basin region. Such an increase is also evident in a com- Separation of this effect and that of local environment parison of the fog-frequency distributionpresented in differences is difficult or impossible where two stations in “Climateand Man” (U.S. Department of Agriculture, close proximityhave substantial differences bothin 1941) for data through 1938 with thedistribution pub- length of record and in reported fog frequency. For ex- lished by Court andGerston (1966) for data through 1960. ample, only 23 mi separate two Detroit, Mich., stations The Court and Gerston analysis in turn show essentially having 5 and 30 yr of record and mean annual heavy-fog the same frequency distribution as that in figure 1. This frequencies of 25 and 12 days, respectively. Toledo, Ohio’s generalincrease in fog frequency lends support to the 83-yr record (table 1) represents five city and two airport frequencyincreases found at most of theindividual locations as much as 9 mi apart. The 8-yr record began stations in table 1. It also supports the suspicion that at witha move of only 15 mi. The substantial change in least some of the station-to-station fog-frequency differ- heavy-fog frequency at Toledo could bedue either to ences are due to variable lengths of station records. different periods of record or to different locations, or both. Because of such uncertainty, the fog frequency used 4. REGIONAL ANALYSIS in the analysis for the stations in table 1 was that which most closely agreed with adjacent stations. Isopleth analysis is founded on the assumption that the analyzedvariable is continuously distributed in space. 3. ISOPLETHANALYSIS Although such analysisis often used to describe a smoothed distribution of a variable that isnot continuously dis- By nature,fog is largely alocalized weather phenomenon t,ributed, thegreater the unorganized variation of the because the causative factors of moisture and cooling are analyzed parameter the more subjective andless meaning- greatly influenced by local terrain and geography. Never- ful is such an analysis.When thenatural station-to- theless, large geographical areas are sufficiently uniform station variability of heavy-fog freqyency is compounded in characterto allow a continuous analysis of fog-frequency by variable lengths of record covering a period of basic distribution for the density of reporting stations in most frequencychange, an isoplethanalysis can be very parts of the United States. Figure1 is such a conventional misleading in certain regions of the United States. Because isoplethanalysis of annual heavy-fog frequency. It was of this, otmher meansof displa,ying the data were explored. drawnas objectively as possible to all 256 stations for As differenttypes of analyses were tried, it became which data mere available. The solid isopleths were evident that the United Statescould be divided into seven drawn to constant 10-day increments up to 100 days SO regions of reasonably common terrainand geographic that frequencygradients and singularities are evident. The most notable exceptions to a continuous distribution properties thatare strongly related to fog frequency. are the obvious discontinuities indicated by steep gradi- Figure 3 shows the boundaries of these regions and the ents along the Sierra Nevada Mountains and the central individual heavy-fog frequency for all 256 stations used Appalachian Mountains, the singularities at some coastal in the study. The boundaries are drawn where there is a stations such as New Orleans, La., Point Mugu, Calif., change in thebasic fog-frequency distributionas indicated Duluth, Minn., and Nantucket, Mass., and some moun- by eithera strong gradient of frequency or a distinct tain stations suchas Mt. Washington, N.H., and Stampede change in gradient. As would be expected, the boundaries Pass, Wash. between some frequency regimes coincide approximately Most of the area east of the Sierra Nevada and Cascade Jvith orographic features such as the Sierra Nevada and Mountains and west of t,he Tennessee Valley canbe CascadeMountains, the Continental Divide, and the analyzed in greater detail without appreciable degenera- tion of the pattern. On the other hand, along the Pacific foothills of the Appalachian Mountains. Others appear to coast andin the centralAppalachians fog is such a localized represent the limits of influence of large bodies of water. phenomenon that with the station density available even But inevery case the criterion for drawing aboundary was the 10-day frequencyisopleths cannot be located with the heavy-fog-frequencydistribution and only minor certainty. adjustments were made to agree with geographic features-

Unauthenticated | Downloaded 09/25/21 04:17 AM UTC 118 MONTHLYWEATHER REVIEW Vol. 91, No. 2

Unauthenticated | Downloaded 09/25/21 04:17 AM UTC February 1969 Robert L. Peace, Jr. 119

329-930 0 - 69 - 4

Unauthenticated | Downloaded 09/25/21 04:17 AM UTC 120 MONTHLYWEATHER REVIEW Vol. 97, No. 2

Unauthenticated | Downloaded 09/25/21 04:17 AM UTC February 1969 Robert L. Peace, Jr. 121

a 70,...... ~ ,,:j/i/,,,' ......

...... :,I::: :I' ,I ;j;jj i ...... l...... i...... G R EAT PL b I N Sl

0 p 0 IO 20 30 40 50 60 70 80 93 100 101t 0 4 z ANNUALNUMBER HEAVYWITHDAYS FOG LL OF FIGURE5.-The histograms of fogfrequency for two of the fog climatic regions delineated in figure 3.

No station in this region averages over 10 days of heavy fog per year, while the majority experience 5 days or less per year. CASCADETO THE ROCKY MOUNTAINS

0 IO 20 30 40 50 60 IO 20 30 40 50 60 This area is characterized by mountains separated by broad,flat valleys. Moisture,though not abundant, is ANNUALNUMBER OF DAYS WITHHEAVY FOG adequatelysupplied from modest precipitation. As a FIGURE4.-The histograms of fogfrequency for four of the fog result, all stations in this region experience an average of climatic regions delineated in figure 3. at least 11 days of heavy fog per year,but only two experience more than 30 days per year (fig. 4b). Data were grouped according to the areas in figure 3 GREATPLAINS AND MISSISSIPPIVALLEY and histograms showing the percentage of stations report- The uniform terrain and limited moisture characteristics ing heavy fog of given frequency were plotted for each of this extensive region are responsible for the narrow, area. These histograms and one covering the ent'ire con- moderate, heavy-fog-frequency distribution displayed in terminous United States arepresented in figures 4 through figure 4c. The principal water sourceis the Gulf of Mexico ; 7. To the extent that the reporting stationsare representa- no other large materbodies directly influence fog formation tive of their regions, the figures may be interpreted as the in this area. Orographic effects associated with the foot- fraction of the total area of a region that is characterized hills of the Rocky Mountains tend to broaden the fre- by the given fog frequencies. quenc,y distribution in that area,and five of the seven A brief description of the fog characteristics of each stations reporting over 20 days of heavy fog per year are region is given below. located along the eastern slopes of the mountains. SIERRA NEVADA TO THE ROCKY MOUNTAINS GREATLAKES AREA This arid region is cut off from moisture by mountain The influence of the Great Lakesextends south andwest barriers both east and west. The lack of moisture causes a of the lakes to an ill-defined border with the Great Plains narrow, uniformly low fog-frequency distribution (fig. 4a). and Mississippi Valley; to the east the Appalachians form

Unauthenticated | Downloaded 09/25/21 04:17 AM UTC 122 MONTHLYWEATHER REVIEW Vol. 97, No. 2

.30

-25

-20

0 . 15 W

0 1 IO 20 30 90 50 60 70 80 90 100 101+

ANNUAL NUMBER OF DAYS WITH HEAVY FOG ANNUAL NUMBER OF DAYSWITH HEAVY FOG FIGURE6.-The histograms for the Atlantic and Gulf Coast regions Figure 7.-The combined histogram for the conterminous United delineated in figure 3, and the combined histogram for the two States. regions. spread is due to the inclusion of mountain stations such amore distinct border. Heavy fog isunderstandably asStampede Pass, Wash. (altitude 3,958 ft, frequency higher near the Great Lakes than on the Great Plains, 251 days/yr). At such stations the statistics represent low and the spread of fog frequencies is also greater inthis mea cloud base at or below station elevation more often than (fig. 4d). The highest fog frequencies are near the water, fog in the more conventional sense. However, such surface but fog frequency does not vary as much with distance obscurations fall within the USWB and WMO definition from the water as it does among stations at comparable of fog, so theyhave been included in the analyses and distances from water (fig. 3). frequency distributions. WEST COAST APPALACHIAN MOUNTAINS The area west of the Sierra Nevada and Cascade Moun- The Appalachian Mountains region resembles the West tainsis under the influence of the Pacific Ocean and Coast region inmany ways. Both representareas of the orographic lift provided bythe mountains. As a irregularterrain and substantial moisture together with consequence, heavy-fog frequency is generally high while orographic lift. As a result, the fog-frequcncy distributions the irregular terrain and varying proximity of reporting in the two regions arevery similar, both showing high stations to the ocean cause a wide spread in frequencies average incidence andlarge frequency spread (fig. 5b). (fig. 5a). The greatest frequencies occur along the coast But the AppalachianMountains region exhibits a more of southern California and at the stations located high in distinct peak (from 25 to 30 days) not found on the West the mountains, as would be expected. But much of the Coast. The annual number of days with heavy fog in the

Unauthenticated | Downloaded 09/25/21 04:17 AM UTC February 1969 Robert L. Peace, Jr. 123 Appalachians varies from a low of 22 days to a high of distribution of fog frequencies onlyto theextent that 307 days.(The latter actually represents primarily low reporting stations are representatively distributed within cloud occurrence atMt. Washington.) In general, the a region. This is not a restriction to interpretation of the greatest fog frequency is found to the west, of the highest histograms in terms of the frequency distribution of heavy terrainand is due to orographic lift of predominantly fog at air terminals, since most of the observations mere westerly winds. made at air terminals. ATLANTIC AND GULF COASTS When all regions are combined into EL single histogram of fog frequency (fig. 7), the significance of fog as an air- These two coastal regions were at first treated separ- terminalhazard becomes apparent. Approximately 50 ately, but because of thestrong similarity of their fog- percent of t’he reporting stations (229 of the 256 represent frequency distributions they have been combined into a air terminals) experience over 20 days peryear with single region (fig. 6). Both coasts display a moderately periods of one-fourth-mile visibility or less. Most of the broad frequency distribution around a relatively highmodal extremely high frequency reports come from stations that frequency. However, despite the wide variationin the are not air terminals, but the majority of the 244 stations proximity of stations to large bodies of water,the fre- reporting up to 70 days with hea.vy fogs are air terminals. quency spread is substantially less than that of either the Appalachian Mountain region or the West Coast areas- REFERENCES where irregular terrain creates more widely diverse, local- Byers, H. It., GeneralMeteorology, McGraw-Hill Book Co.,Inc., ized fog-producing conditions. The one extreme local Kew York, 1959, 540 pp. (see p. 491). influence inthis area is found inthe Nantucket-Block Court, B., andGerston, R. D. “FogFrequency in theUnited Island area where the heavy-fog frequency is about t’wice States,” TheGeographical Review, Vol. 56, No. 4, Oct. 1966, pp. 543-550. that of the other stations along these coasts. I-iaurwitz, B., andAustin, J. M., Climatology, McGraw-Hill Book Co., Inc., New Pork, 1944, 410 pp. (see Plates 12 and 13). 5. CONCLUSIONS Stone, R. G.,“Fog in the United States and Adjacent Regions,’’ The regional histograms provide insight into both the The Geographical Review, Vol. 26, No. I, Jan. 1936, pp. 111-134. U.S. Department of Agriculture, “Climate and Man,” Yearbook of variations of heavy-fog frequency from one point of the Agriculture, 1941, Washington, D.C., 1941, p. 737. .Country to anotherand the variability of frequencies Ward. R.D., The Climates of the United States, Ginn and Co., Boston, within each region. However, they represent the spatial 1925, pp. 335-338.

[Received January 26: 1968; reuised July 18, 19681

CORRECTIONNOTICE Vol. 96, No. 1, Jan. 196s: 1). 26, 2d paragraph, 3d line from bottom, F is to be read instead of E, and G instead of F; 1). 28, 3d line after equation (4) 8, instead of up,and 10th line aft’er equution (4), release instead of reaseel; p. 30, fig. 11 caption, [G][e]instead of [u][VI.

Unauthenticated | Downloaded 09/25/21 04:17 AM UTC