848 . . .I ;.. MONTHLY WEATHER REVIEW. JUNE, 1914 . .-
i .. , J
TBE ~E B S T O R M AND ITS PHENOmNA.
By W. J. HUIPEREYR, Professor of Meteorological Physics.
[DW, Weather Burem, Waahiugton, D. C., July 17,1814.1
Introduction..-A thunderst.orm, as its name implies, is a storm characterized by t.hunder and lighhing, just as a dust storm is one charsct?rizcd by.& great quantit,y of flyin dust. But the dust is never in any sense tho cause o 'i the storm that carries-it along, nor, so far as known, does either thunder or lzhtning have any influ- ence on the course-genesis, deqelopnient, or termina- tion-of even those storms of which. they forni, in some respects, the most impo.rt,?nt feat-ures. No mat.t.er how impressive nor how tprrifymg these phenomena may be, thoy never are anybhmg more t.han mere incidents t.o or pr0duct.s of the pecuhar storms the accompany, as
w i l l be made clear by what follows. fn short, t.hey are never +I any sense either storm-originating or storm- controkng factors. Origin of thunderstorm e?ectri&y.-A knowledge, or at least a good working hypothesis, of how t.he great amount of electricity incident to t;hunderstorms is generated, is absolutely essential to their lo cal explanation; t,hat is,
between their many phenomena. Fortunately such an hypothesis, or theory, rat.her, since it is abundantly sup- ported bv observat,ions and b laboratory experimentis,
18 available as a result of war% done on t.his subject m India bv Dr. G. C. Simpson (1) of the Indinn Moteoro-
lo 'cd Department. %r. Simpson's observations ust referred to, were
feet above sea level, and covered all of the monsoon sea- sons, B a t is, roughly, April 15 to Sept.ember 15, of 1908
and 1909. -He also obtained observations of the elec- trical conditions of the snow at Simla during the winter of 1908-9.
A tipping-bucket rain gage gave an automatic con- tinuous record of the rate and time of rainfall, while a Benndorf (2) self-registering electrometer marked the sign and potential of the char e acquired during each two-
tered the potential gradient near the earth, and a coherer of the t e used in radiotelegraphy registered the occur-
All obvious sources of error were esamined and care- fully guarded against. Hence it would seem that the conclusions drawn. from the thousands of observations e;iven in the memoir are fully justified; and especially so since several mdependent series of similar observations made a t different times, by different people, and at places widely se arated, have given confirmatory results
m every case. 8impson's records show that-
(1) The electricity brought down by the rain was sometimes positive and sometimes negative. (2) The total quantity of ositive electricity brought down by the rain was 3.2 times greater &an 'the total quantity of negative elec- tricity. (3) The period during which positively charged rain fell was 2.5
times longer than the period during which negatively charged rain fell. (4) Treating charged rain as equivalent to a vertical current of electricity, the current densities were generally smaller than 4X10-I5
amperes per square centimeter; but on a few occasions greater current densities, both positive and negative, were recorded.
(5) Ne tive currents occurred less frequently than positive cur- rents, anrthe greater the current density the greater the preponderance of the ositive currents. (6) ' h e charge carried by the rain was generally less than 6 electro- otat~c units per cubic centimeter of water, but larger charges mere occasionally recorded, and in one exceptional storm (May 13, 1908)
the negative charge exceeded 19 electrostatic unite per cubic centi- meter.
to a clear understanding of t f- e probable int,errelations
obtained at Simla, India, at an ' i e evat.ion of about 7,000
minute interval. A seconc P Benndorf electrometer regis-
rence o 9p each lightning discharge.
.
(7) As stated in paragraph (3) a b m , positive electrki wae recorded more frequentl than negative, but the exceee wao ge leae marked t,he higher the czarge on the rain.
(8) With all rates of rainfall positively ch rain occurred more
pmtive1.y ch-ed rain increased rapidly with increased rate of mh- all. With ramfall of less than about 1 millimeter in two minutes; poei- tively charged rain occurred twice aa often a~ negatively charged rain,
while with greater intensities it occurred 11 times as often.
(9) When the rain was falling at a less rate than about 0.6 millimeter in two minutes, the charge per cubic centimeter of water decressed 88
the intensity of the rain incread. (10) With rainfall of greater intensity than about 0.6 millimeter in two minutes the positive charge carried per cubic centimeter of water waa independent of the rate of rainfall, while the negabve charge car- ried decreased aa the rate of rainfall incretwed. (11) During periods of rainfall the potentaal gradient was more often negatwe than positive, but there were no clear indications of a relation- @hip between the sign of the charge on the rain and the e n of the potential gradient.
(12) The data do not suggxeaf that the negative electricity occurs more frequently during any particular period of a storm than during any other.
Concerning his observation on the electrXcation of snow . Dr. Sinipsnn sn.ys :
As far as can be judged from the few measurements made during the winter of 1908-9 it would appear that:
(1) More pwitive than negative electricity is brought d o n by snow in the proportion of about 3.6 to 1. (2) Positively chaged mow falls more often than negatively charged. (3) The vertical electric currents during snowstorms are on the aver- age larger than during rainfall.
(4) The charge per unit mass of precipitation is larger during snowfall than dunng rainfall.
While these observations were bein secured a number
frequently than negatively charged rain, and P e relative frequencyof
of well-devised es eriments were ma f e to deterniine the electrical effects o P each obvious process that takes place
some spray, positive and important resu Y ts were found,
in the thunderstorm. Freezing and thawing, air friction, and other things were tried, but none produced any electrification. Finally, on allowing drops of disti7lcd water to fall through
a vertical blast of air of sufficient stren th to produce
showing:
(1) That breaking of drops of water is accompanied by the production of both ositive and negative ions.
(2) T L t three tinies as many negative ions as positive ions are released.
Now, a strong upward current of air is one of the most conspic.uous features of the thunderstorm. It is always evident in the turbulent cauliflower heads of the cumulus cloud, the parent, presumably, of all thunderstorms. Besides, its inference is compelled by the occurrence of hail, a frequent thunderstorm phenomenon, whose for- mation requires the carrying of raindrops and the gow- ing hailstones repeatedly to cold and thcrefore hi h alti-
inferred that an updraft of at least S nieters per second must often occur wit,hin the body of the storm, since, as
experiment shows, it requires ap roximately this velocity to support the larger drops, anleven a greater velocity to support the avera .e hailstone.
air o P ordinary density whose upward velocity is greater t,hnn about 8 meters per second, or itself fall with greater velocity t.hrough still air; that in such a current, or with such a velocity, drops lar e enough, if kept intact, to force their way down, or, Lough the action of gravity, to attain a great.er velocity than S mete? per second with reference to the air, whether still or in mobon? are so blown to pieces that the increasea ratio of supportin
smaller drops. Clearly, then, the updrafts w i t h a
tudes. And from the existence of hail i t is 9 urther
Ex eriment also s E ows that rain can not f J l through
area to total mass ca11ses the rcsultin spray to be Carrie % aloft or left behind, together with, o 9 course, all oripnsl
JDXIC, 1 9 1 d t . MONTRfiY PPEATHER REVIEW. 340
cumulus cloud frequently must break u a t about the
cence, have grown beyond the critical size, an% thereby according to Simpson',s experiments, roduce electricai separation within the cloud itself. gbviously, under the turmoil of a thunderstorm, its chop y surges and puJses, such dro s may be forced througt the cycle of
division, of coalescence and disruption, from one to many times, with the forniation on each at every disru tion, agtrin according to expcrimtnt of a corresponcfingly increased electrical charge. The turnioil compels me- chanical contact between the drops, whereu on the charges break down the surface tension an8 insure coalescence. Hence, once started, the electricity of a
thunderstorm rapidly grows to a considerable maximum. After a time the larger drops reach, here and there, laces below which the updraft is small-the air can not !e rushing up everywhere-and then fall as poqitivel charged rain, because of the proce3ees just explainei The negative electrons in the meantime are carried up into the higher ortions of the cumulus, where they unite with the alouc? particles and thereby facilitate their coalescence into negatively charged dro s. Hence, the heavy rain of a thunderstorm shoul lf be Po.-itirely charged, as it almost always is, and the gentler portions ne atively char ed which very frequently is the cace.
the electricity in thunderstorms, a t eory that full,v accounts for the facts of observation and in turn IS
itself abundantly supported by laboratory tests and simulative experiments. If this theory is correct, and it seems well founded, it must follow that the one e-sential to the formation of the giant cumulus cloud, namely, the rapid uprush of moist air, is also the one essential to the generation of the electricity of thunderstorms. Hence the reason whv lightning seldom if ever occurs escept in connection with
a cumulus cloud is understandable and obvioun. It is sim ly becaute the only proce;s that can produce the one is &o the process that is necessary and sufficient for the production of the othor. TIM wiolpnt motions of cumulus cLnds.--From observa- tions, and from the graphic descriptions of the few bal- loonists who have experienced the trying ordeal of pasa- ing through the heart of a thunderstorm, it is known that there is violent vertical motion and much turbulence in the middle of a large cumulus cloud, a fact which SO far as it relates to the theory alone of the thunderstorm, it would be sufficient to accept without inquiring into its cause. However, to render the discu3sion more nearly complete, it perhap3 is worth while, since it is a mooted question, to inqulre what the. robable cause of the violent motions in large cuniulu+ c f oudv really i+-motions which, in the m nitude of their vertical components .and
ot d% er kind nor met with elsewhere by either manned, sounding, or pilot balloons. I t has been shown by von Bezold (3) that suclclen con- densation from a state of supersaturation, and also sud- den congelation of undercooled cloud droplets, would, as a result of the heat thus liberated, cause an equal1 sud-
motions analogous to those observed in large cumuli. However, as von Bezold hiiiisolf points out, it is not evi- dent how either the condensation or the freezing could suddenly take place throughout a cloud volume great enough to produce the observed effects. Besides, these eruptive turmoils, whatever their genesis, undoubtedly
same level innumerable drops which, t ?I rou h coales-
union (facilitate E; by any charges they may carry) and
guch in brie fg is Dr. Simpson's theoy of the origin of
vree of turmoi ?i , are never exhibited by clouds of any
den espansion of the atmosphere, and thereby tur E ulent
originate and run their course in re 'om already filled
able degree of su ersaturation can occur. Hence the
ously niust have some other cause; and this we shall find in the difference between the actual teiiiperature gradient of the surrounding atmosphere and the adiabatic temper- ature gradient of the saturated air within the cloud itself. Coiuider a warm summer afternoon, tern erature 30'
C.,.say, and assume the dew point to be 15" 8. Now,.the adiabatic decrease of temperature of nonsaturated am is about 1 O C. per 100 meters change in elevation, and there- fore, under the assumed conditions, vertical convection of the surface air causes condensation to be in at an eleve tioii of approximately 1.5 kilometers. B rom this level, however, YO long as the cloud particles are carried up with the rising air, the rate of temperature decrease, for at least a couple of kilometers, is much less-at fist about one-half the previous rate. After a considerable rise above the level of initial condensation, half a kilometer, the raindrops have so increased in size as to lag
sat be ind the upward current and even to drop out, while, at the same time, the aniouiit of moisture condensed per degree fall of temperature grows rapidly less. Hence, for both reasons-because the heat of the condensed water is no longer available to the air from which it was condensed, the drops having been left behind, and because but little latent heat is to be had from further condensa- tion, there being but little water vapor left-the rate of temperature dec.rease a ain approachee the adiabatic g r e
client of dry air, or 1'8 per 100 meters change of eleva- tion. Obviously, then, for some distance above the level at
which condensation begins to set free ita latent heat, the temperat.ure of t.he rising ma.ss of moist' air departs fnrt,her and farther from the temperature of the sur- rounding atmosphere a t the same level, and therefore its buoyancy for a t.ime as steadily increases. But, of course, as explained above, this increase of buoyancy does not contmue to a.ny great altitude. In talle lower at-mos here cont.inuous and progressive
that no great clifierence between the temperature of the rising column and that of the adjacent atmosphere is anywhere possible. Hence, under ordinax condit,ions, the uprush in this region is never violent. ut whenever the vertical movement of the air brings about a consider- able condellsation it follows, as a.bove explained, that there is likely to be an increase in its buoyancy, and hence a more or lees rapid upward movement of t.he central portion, like air up a heated chimney, and for the same
reason, together with because of viscosity, a rolling and turbulent motion of the sides, of the type so often seen in
column of air must ascend somewhat hevond i t s point o 9
towering cumulus clouds. Obviously, t.oo, the uprushin
equilibrium, and then, because slightly undercooled, correspondin ly drop back. Figure 1, tasnd upon approximately averago condi- t-ians, illustrat~es the oints just explained. The elevation is in kilometers a n f the temperature in degrees centi- grade. AB is t.he approximate temperature gradient for non- saturat,ed air, about 1°C. per 100 meters change in elevation. GCKDPF is t.he supposed temperature gra- dient before convection hegins, or a decrease,m accordance with observations, of 6' C., approximately, per kilometer increase of elevat.ion, except near the surface where the t.emperature decrease, before convection Las begun, ordinarily is less rapid.
with cloud particles in the presence o ff which no appreci-
rapid uprush and t 1 e violent turbulence in question obvi-
convect.ion builds up t. Fl e adiabat.ic gradient so gradually
350 MONTHLY WEATHJIR REVIEW. JUNE, 1914
As convection sets in, the temperature decrease near the surface soon apprnximates the adiabatic gradient for dry air, and this condition extends gradually to altitudes, till in the rtssumed case, condc?nsat.ion gins at the level b, or where the temperature is 15OC. Tlere the temperature decrease, under the assumed conditions suddenly chanoes from 10°C. er kilometer increase 01
increases with increase of altitude and consequent de- crease of temperature. At some level as L, t.hc tempcra- ture difference batween the rising and the n.djacent air is
a maximum. A t D the t-em lerature of the rising air
presumably carries it. on to some such level as 11. Within
the rising column, then, the temperature gradient is
yter
elevat.im to ra%er less than ha 9 that amount, bn t slowly
is the same as that Of the air L i jacent,, but its momentuni
FIG. l.-Tempereture gradlenta within (CLD) and without (CKD) cumulus clouds.
approximately given by the curve BCLDHE, while the tern erature gradient of the surrounding alr is given by &e curve ACXDEF. The cause, therefore, of the violent. uprush and tur- bulent condition within large cvmulas clouds is, pre- sumably, the difference between the temperature of the inner or warmer portions of the cloud i t d f and that. of
the surrounding atmosphere a t the same level, as indi- cated by their respective temperature gradients CLD and CRD. Clearly, too, while some air must flow into the condensation column all along its length, t.he greatest pressure difference, and, therefore, the greatest inflow, obviously is a t ita base. After the rain has set in, how- ever, thls basal inflow is from immediately in front of the storm, and necessarily so, as will be explained later. Cbnvechml instability.-Rapid vertical convection of humid air, as we have seen, is essential to the production of the cumulus cloud and, therefore, to the generation of the thunderstorm. Hence it is essential to consider the conditions under which the vertical temperature gradient
necessary to this convection can be established. These are :
1. Strong surface heating, especially in regions of light winds; a frequent occurrence. The condition that the winds be light is not essential, or, perhaps, even favorable to the genesis of all thunder- storms-only the local or heat vanety, and favorable to these simply because heavy winds tend to prevent the formation of isolated rising columns of air, the progenitors of this particular type of storm.
2. The overrunning of one layer of air by another at a temperature sufficiently lower to induce convection, well- nigh the sole cause of ocean thunderstorms and also of frequent occurrence on land.
3. The underrunnin and consequent uplift of a satu-
to a greater or less extent and, presuma ly, therefore at least an occasional one of sufficient magnitude to produce a thunderstorm. Here the underrunning air lifts both the saturated layer and the su erincumbent unsaturated layer, and thereby forces ea% to cool adiabatically. But as both layers are lifted e unlly while, because of the latent heat
than the dr , it follows that a sufficient mechanical lift of
nonsaturated layer above a su eradiabatic temperature
clouds, and, perhaps, a thunderstorm. Periodic reciarrmce of thitnderstorms.-While thunder- storms ma occur at any hour of any day they neverthe- less have t s nee distinct periods of maximum occurrence: a, dail , b, yearly, and e, irregularly cyclic. Each maxi- mum iepends upon the simple facts that the more humid the air and the more ra id the local vertical convections
convection of humid air that roduces them.
here over land areas is most ronounced when the sur- face is most heated; that is, luring afternoons. Hence the hours of maximum frequency of inland or continental thunderstorms are, in most places, 3 to 4 p. m. Daily ocean period.-Because of evaporation and of the high specific heat of water the surface temperature of the ocean increases but little during the day, and because of convection it decreases but slightly at niglit. Indeed, the diurnal temperature range of the ocean surface usually is but a small fraction of one degree C., while that of the atmos here at from 500 to 1,000 meters elevation is sev-
over the ocean that are favorable to rapid vertical con- vection are most frequent during the early morning hours, and, therefore, the m&sinium of ocean thunderstorms usually occurs between midnight and 4 a. m. I'early 1m.d eriod.-Just as inland thunderstorms are
and for the sanie reason, they are, in general, most fre- quent over the land during the hottest months of the year, or, rather, during those months when theamount of surface heating and, therefore, the vertical tempera- ture gradient is a masimum. Hence, in middle latitudes, where there are no late spring snowa to hold back the temperatures, the month of niasinium frequency often is June. In higher latitudes, where the strong surface heating is more or less delayed, the maximum occurs in July or even August. Yearly ocean pekd.--Over the oceans, on the other hand, temperature gradients favorable to tbc genesis of
B
rated layer of air by a c K enser layer; a fre uent occurrence
of condensation, 3 t le saturated layer cools much slower
a saturated T ayer of air would establish between it and the
gradient and thereby produce f oca1 convection, cumulus
the more fre uent and a 7 so the more intense the thunder- storms, for 4n t e obvious reason that it is rapid vertical
Daily hnd period.-Vertica P convection of the atmos-
era1 fo P d as great (4). Hence those temperature gradients
most frequent 8 uring the hottest hours of the day, so too,
JUNE, 1914. MON'I'HLY WEATHER REVIEW. 351
thunderstom, and, therefore, the !toms themselves, occur most .frequently dunng the Tnter and least fre- quently dunng the summer. This IS because the tem- erature of the air- a t some distance above the surface., geing largely what it was when i t left the windward conti- nent, greatly changes from season to season while that of the watcr, and, of course, the air in contact with it, cha es
average decrease of temperature with increbe of eleva- but little through the year. 'ihat is, over the oceans "% t e
the same relation to the annual average windward tem- pcrature that the total annual precipitation over the en- tire world does to the annual average world temperature. In each case the amount of evaporation or amount of water vapor taken into the atmosphere, and, therefore, the amount of subsequent recipitation, clearly niust in-
test and complete support of this deduction is furnished by figure 3, in which the full line represents the anioothed
crease dnd decrease with t R e tcmpernture. Au excellent
Fro. %-Relatbn of European rainfall to eastern U. 8. temperature.
tion obviously is least and, therefore, thundemtorms few- est in summer! and greatest, with such storm.. most numerous, in winter. flyclic land period.-Since thunderstormq are accom- panicd by rain and since over land thev are most iiuiiier- ous during summer, it would appear that they niust oc-
cur most frequently either in warm or in wet yenrv and lewt frequently in cold or in dry yoarv. Further, if it should happen, as it actually does, that, for the earth as
annual European precipitation (5), and the dotted line smoothed annual average eastern American temperatures. Beyond a reasonable doubt, therefore, for the world as a whole, warm years are wet and cold ones are d . Hence, as above stated, it is practically certain that x e maxima of thunderstorms occur during years that are
wet, or warm, if we prefer, for the two are synchronous, and the minima during years that are dry, or cold. A partial and, so far as it goes, a confirmatory statistical
Fro. S.-Relation of snnusl number of thunderstom days to total annual pmipitatbn- Of destructive thur
a whole, warm years are also wet years and dry yenrs cold years, it would appear logical1 certain that, for the
must belong to the ears that are wet and warm and the
A complcte statistical examination o these statements is not possible, owing to the fact that nieteorological data are available for onl portions of the earth's surface and
sive data do exist. The annual rainfall, for instance, to the leeward of a large body of water obviously must bear
entire world, the maxlmum num g er of thundemtornlr
9
minimum to thoise t Tl a t are cold and dr .
not for the whole o 9 it. Nevertheless, well-nigh conclu-
-Holland. The uppermast, m v y curveshows the vsrfstlon in the smoothed numbm ide~tormn in Germany.
test of this conclusion is given by figure 3. The lower group of curves is based on m exhaustive study by Dr. von Gulik (6) of thunderstorms and lightning injuries in Holland. The continuous zigz line 'ves the actual annual number of thunderstorm T %l ays an the continuous curved line the lame numbers smoothed. The broken lines give,respectively, the actual and the smoothed values of the annual average preci itation. The up er curve
structive thundorstorms (7) (number of thunderstorm
days not readily available) in Germany.
represents the variations in t R e smoothed num er of de-
JUNE; 1914 352 MONTFIIJY WEATHER REVIEW.
Theoriginal dataonwhichthislastcurveishasedindicate
a continuous and ra id 'increase of thunderstormn:destruc-
to the extent that the country has become more densely populated and more thick1 studded with destructible
the curve and on1 the variation factor retained.
uencp for all Holland closely parallelv the curve of %understorm injury in all German . Hence, it seenis
pretty much the same way over both-countries, and, presumably, also over many other portions Of Europe; that IS, roughly as the rainiall varies, or, conadenng the world as R whole, roughly as the tem erature varies.
the annual number ot thunderstorms and the total annual precipitation, kindly assembled by the Climatologicn.1 Divi- sion of the United States Weather Bureau, P. C. Day in chargo,.is shown by figure 4, in which the upper line gives, in rmlbmeters, the smoothed aveiage annual preci ita-
tiveness. Presuma g ly, however, this feature is real only
property. At any rate, this e 9 ement has been omitted from
It will be note i that the curve of thunderstorm fre-
safe to infer that the frequency of ti understornLq vnries
Additional statistical evidence of t R e relation between
tions of 127 stations widely scattered over the who i e of
Fro. 4.-E&tlon of snnual number of thunderstom, T , to total snnual preclpltation, P, United States.
the United States, and the lower line the smoothed aver- age annual number of thunderstorms a t these same sta- tlona. It was thou' ht a t first that this relation might differ greatly for goso gortions of the United States whose climates are radica y dissimilar, and tor this reason the stations east of the one hundredth mendmi provision- ally were classed separate!y from those west of it; but
the resulk for the two sechons, being substantially alike, show that for this purpose their division is entirely unnec- essary. As will be seen from the figure the statistics of only the past 10 years have been used. This is because the annual number of such storms reported ra idly decreases from
thunderstorms re orted er station during the past 10
(practically the same stations) from 1880 to 1890. The transition from tho smaller to the lar er number was due
regulations that chan ed the official deaition of a thun-
without rain." This, however, does not account for the fact that from 1890 to 1904 the average annual number of thunderstorms reported per station increased, a t a nearly constant rate, almost 100 per cent. Either the
1904 back to about 1S90. Indeed, t F ie annual number of
years is almost K R ouble t e annual number per station
in great measure, doubtless, to an a 91 teration in station
derstorm rrom " thun % er with rain to "thunder with or
storms did so increase, which seems im robable, or else
to this particular phenomenon. At any rate, so contin- uous and so great an increase in the average numbor of thundeistorms can h a d l be accepted without abundant
storm records provisionally have been rejected. Obviously a niuch closer relation between the number of thunderstorms and total reci itation would hold for
subgrouping of the data has been made, though, -
purpose of this portion of the study was to arrive at some definite idea in regard to the cyclic change of thunderstorni frequency, to see with what other mete- orolo ical phenomena this change is associated, and, if
there was, on the average, an increase o I attention given
confirmation, and for t i!l is reason the earlier thunder-
some months and seasons t f P an or others, but no such
sumably, it would give interesting results. The w gre ole
possi % le, to determine its cause.
frequency of thunderstorms, and many other phenomena must perforce undergo exactly the same irregular cyclic variation. As already stated the statistical evidence bearing on these conclusions neither is nor can be com lete, but the
exanlined so confirniatory that but little doubt can exist of their general accuracy. Cyclic oct-an period.-The record of thunderstorms over the oceans is not sufficiently full to justify any con- clusions in regard to their cyclic changes. Possibly, as in the yeiirly and the daily periods, the ocean cyclic period may be just the reverse of that of the land, but this is not certain. Grographic distribu tion.-The geographic distribution of the thunderstorni may safely be inforred from the fact tshat it is caused b the strong vertical convection of
assume, and the assumption is supported by observs- tion, that the thunderstorm niust be rare beyond either polar circle, especially over Greenland and over the Antarctic continent, rare over great desert regions wherever situated, and, on the other hand, increasmgly abundant with increase of tenipecature and humidity, and, therefore, in general, most. abundant in the more rainy portions of the equatorial regions. The east coast of South America from Pernambuco to Bahia is said to be an exception. Pressure and tcm erature dktribu.tion.-In illustrating
disposition of isobars and isotherms, or the distribution of atmospheric pressure and temperature, typical weather maps OF the United States,' figures 5-1 9, have been used, not because the thunderstorms of this country are dif- ferent in any essential particular From those of other countries, but chiefly as a matter of convenience in making the clrttwings. To facilitate their stud each of
secutive maps. The first shows the 12-hour antecedent
deductions are so obvious and the sta,tistictt P data already
huinid air. From t K e nature of its formation one would
the occurrence of t 7l understorms with reference to the
the several types discussed is illustrated with t 3: Fee con-
1 The author wishes lo acknowledge the courteous cooperatbn of the Forecsst D i d slon U 8. \\eather Bureau, In selecting mnps typical 01 thunderstorm condltiozu In thr trnited statas.
JUNE, 191.1. MONTHLY WEATKER REVIEW. 363
364 MONTRLY W E A m R REVIEW. JUNB, 1914
J m , 1014. MONTHLY WEATHER REVIEW. 355
356 MONTHLY WEATHER REVIEW. JUNE, 1914
conditions, the second the particular pressuretern era- ture distribution in question, and the third the E l o u r subsequent conditions.
In these figures the isobars, in corrected inches of
, and the isotherms, in Fahrenheit degrees, are marke by full and by dotted lines, res ect,ively. The legend “LOW” is written over a region rom which, for some distance in every horizontal direction, the pressure increases. Similarly, the legend “HIGH” apphes to a
region from which, in every direction, the pressure de- creases. The arrows, as is custoniaxy on such maps, fly with the wind, while the state of weather is indicated by the usual U. S. Weather Bureau synibols. Obviously, the key to the geographic distribution of thunderstorms, vertical convection of humid air, is also
the key to their location with reference to the existing distribution of baronietric pressure. From this stand- point the places of their most frequent ocmrrence are: a. Regions of high temperature and widely estended nearly uniform pressure. (See f i ~. 5 , 6, and 7.) The conditions are still more avor&ble when the air is humid and the ressure, perhaps because of blie
above normaf
When the ressure is approsimately uniform the winds
air to become strongly heated ancl thereby finally to establish thunderstorni convections. Such storms, always favored by mountmain regions, and part.icularly by steep mountain peaks and stroiigly heated valleys, are, of course, most frequent of summer afternoons and are espe- cially liable to occur a t the end of two or three days of unusually warm weather. They develop here and there sporadically, hence the name ‘LZocaZ ” thunderstorm; last,
as tt rule, only an hour or two, and travel neither rapidly nor far. Those that form over mountain peaks oft,en do
not travel at all. The necessary initial convection is essentially, if not wholly, due to surface heating ancl therefore they frequently are referred to as “heat” thunderstorms. They are well-ni h the only type of
thunderstorm in the tropics, an$ perha s, the most
zones.
b. The southeast quadrant (Southern Hemisphere, northeast), or, less fre uently, the southwest (Southern Hemis here, northwes8, of a regularly formed low, or typica r cyclonic storni. (See figs. 8, 8, a.nd lo.)
In this case the temperature gradient essential to a rapid vertical convection is not produced chiefly by local surface heating, as it is during the genesis of “lieat” thunderstorms, but, in great measure, result,s from the more or less crossed directions of the under and over currents of air. The surface air of the quadrant in question normally flows from lower and warmer latitudes, while with increasin altitude the winds conie more ancl
crossing of the air currents, then, the lower from warier sectionsand the u perfroin regions not so much warme?
temperature gradient, or rate of temperature decrease with increase of altitude, and therefore may frequent.ly be, and doubtless often finally is, the determining cause of a rapid vertical convection and the formation of a
thunderstorm.
. This particular type of thunderstorm, commonly known as the ‘ yclonic” thunderstorm, is almost wholly c.on- fined to the temperate and higher zones, for the simple reason that the well-dehed cyclone, essential to its creation, seldom occurs in tropical or equatorial regions.
P mercu7
humidity, sli htly be P ow normal or, at most, but litt,le
are light anc P every opportunity is given for the surface
common type in the warmer portions of t fl e temperate
more nearly from t fl e west, or even northwest. This
possibly even colc Y er-progressively increases the vertica.1
c. The barometric valley between tho branches of a distorted or V-shaped cyclonic isobar. (See figs. 11, 12,
and 13.)
This region is also favorable to the formation of sec- oiiclary lows, which, though often of small area, some- tiiiies are intense even unto tornadic violence. Just how specific examples of this type of pressure
distribution originated may not always be clear, but however established, each necessarily leads to o posin surface winds am1 also to more or less oppositely Jirectei upper currents along adjacent paths; and each wind syst,em, the lower and the upper, tends to produce an independent effect. Thus the op osing or conflicting surface winds cause such an irregu f ar mixing of the air and such over and underrunningof currents as is likely to establish, here and there, a convection or thunder- storm radient. Hence the frequency of t,hunderstorms along t B ie valleys of low-pressure basins. On the other hand, the oppositely directed adjacent, not, conflictin , upper c.urrents by cat.ching masses of air, espccial B y rising masses, between them tend mechanically t.0 pro- duce, in t,he niitldle atmosphere, violent vortices of limited estent. The more violent of these vertica.1 atmospheric whirls, usually acconipanied by thunder and rain and often estending down to the surface of the earth, where they become dest.ructive, are known 8s tornadoes. Hence thunderstornis generated in the barometric region under discussion, the region in which t,ornadoes most frequently originate and develop, might properly be called “toomadzc ” thunderstorms. At11 os iheric conflict.s and turwoil, of the nature just
prot,rusion, or valley, of t,he low-pressure basin, and there- fore often do orcur? even simultmeously, here and there. along it,s entire length, and together form t.lie well-known
“ line sqnnll.” Besides, 8s t>lie whole cyclonic conclitioii moves forward in pmd from west t.0 east,. ~~aintaining. in a nieasiire, for many hours its identity of form arid nature, it follows t,lint its ~d l e y of low pressure, and tliore- fore it,s line of thunderst.orms, must also travel with it in the sanie general direction and with approxiniately the sanie velocity. A line or row of thunderstornis-a “line squall”--m observations show, alwa s moves across its own axis, not necessaril at right ang P es, but nevert.heless across and
mason for this is not the asia1 c irection of the low-pressure valley which, indeed, though usually running south, may have any orientation from t,he parent basin, but rather the fact that the valley it.self, together wit,h its accompanying t.hunderstorni conditions, travels acros3 and not along its own direct,ion. In t.his conncction i t is also worth noting that the tem- perature distribution in t-he wake of a thunderstorm ren- ders the ~cciirr~nco of an ininierlinte si:ccessor improbable, as will be esplainecl later. Hence while a considerable number of thunderstorii-.s may and often do travel abreast they can never follow each other closely in file. d. The region covtred by a low- ressure trough between adjacent high-pressure areas. (&qs. 14, 15, and 16.)
Along the adjacent borders of two neighboring anti- c clones-that is, alon the barometric trough between
directly opposed t.0 those from the other. Hence, because of the overru~:niii~, as explained under c, and the resulting tenipcrature gradients, this also is a region of frequent t,hunderstorn.is. Here, too, a nuniber of niorc or less inde- endent st.ornis way exist siniultaneously along the same !ne, and advance abreast for great distances across the country.
describcc, 1 obviously niay occur at any place along the
f
not pard T el t.0 it., nor oven ap rosimntdy so. The chief
t l tin-t.he surface win a s from one side are more or less
JUNE, 1914. MONTHLY WEATHER REVIEW.
358 MON'I'HLY WEATHER REVIEW. J m , 1914
$van, 1914. MONTHLY WEAlTlXR REVIEW.
360 MONTHLY WEATHER REVIEW. JUNE, 1914
There does not appear to be any independent. or dis- tiiic tive name for the thundeistorm enerated mcler this
some justification, be called the “ anticyclonic ”*thunder- storm, or even t,lie “trough” storin.
e. The boundary between warin and cold waves. (See figs. 17, 18, and 19.)
Along such a boundn.ry the direction of flow of thc
on the maps, t.0 that of the colder ones. Therefore i t must frequently happen that at irregular int.ervals along such a boundary t.he upper air, coming froin the cold arc:\.,
overruns a section of surface air beloiiging t o t.lie \va.rili
region; but, of course, only where tlic upper ai;. is 4 1 potentially warnier than the lowcr-if potciitially colder it would underrim. Xow, whercv er this overriiiining on t.he ptirt of t.he cold air does ocoiir t.he vertical t.e:~:]~~rnture gradient obviously is abrupt,ly and greatly incrrr.iscd. niid wherever, in the course of its fi.irt.litlr i?;o~eii:e~;t., tlic n c w
gradient esceeds the ticlinbat.ic rate of t.miq)crtitiire chnge, as analogous t.0 c.~~.sc? b, i t 0ft.r.n IiitiSt., ulitlr:. tdic given conditions, verticd coiivcction with r:.i.iiiq tIiuiicl(:r, and iightning is apt to occur. Ecncc, :M st;?td, t.iic boundary betm-oen warni and cold WBXW i.5 wot.lier p1ac.e
favorable to the thunde-wtorni, wliich, uiidcr tiic;;? con-
ditions, >ossibly niiglit bc cded thc ‘ibori7~1”! storill. ‘I he a b ore five distinct t:,-pe:s of. wratlii~r cunclitiona, together with their innuinrri~bl c‘ mriztions and C O J ~:~ &ii:-
tions, protinbly iiicludc dl t.1 tc t. nrc distinctly fiivornhlr to tho production of tl:uiiclcist,or!rIs. E:!.ch t.iiic1.G t,o
establish a.11 ndiabiitic or (?veil :iuy,eriidiabti tic. tecqwrtit,ii;:> gradimt up to the cloud !evc-l--tl:e O I ~P thins ~:~R C Y I tial t,o the production of 6 stroiig ycrticn! coiivcction, tlir lira-
genitor, as we 1in.c-e :wen, of thc t!liiiid:rat,i)rl;~. Thmderstonn uvXs.--Sliortly, sag 20 niiiiu t,cs or
so, before the rain of R thuntlei.storm reaches 8. even LI- cality the wind a t that place, which generdy is light itiitl
from the south or southwest across the pith of the storm, begins to die down to an approsimate c:ilni nnd to change its direction. ’Alien this cliaiigc is complete, it blows for a few minutes, rather gently, dircctlv t.:m-ard the nearest portion of the storm front, nitti ihaIlj-, as t,hc. rain is almost at. hand, again, but this time ahrupt,lT ant1
in rather violent, ust.s, away from trlie storm aitc! 111 t.ho
usually differs appreciably from that of t.lie original sur-
face wiiicl. Generally this violent gusty miid l:ist,s
through only the earlier portion of the strmi, :111il t,licn is
graclually but rather quickly succeeded by a contparn- tively gentle wiiid that, though following the storm a t first, frequently, after an hour or so, blows in tlie saiiit: general hrection RS the original surfiice wind. The cause of the thunderstorm winds needs to be care-
f idly considered if one would iiiiderstnnd nt all clearly the mechanism of tlie storm itself. As already explained, this typo of storin oms it.s origin to that vertical con-rection w1iic.h rcsults from a inore or less superndiabatsic temperature graclien t. I t is t.liis grn-
dient, no matter how established, uc-liet,lier by siniplc sur- face heating or by the oveI aiid uncler running of uii-
equally heated layers of air, that permits, or rather forccs:
the pIoduction of the cuniulcs cloud in which md by the motions of which the electricity that chm-actcrixes the storm in question is geneiated. Nevertheless, as everyone knows: the passap of a
cumulus cloud overhead, however large, so long as no
rain is falling from it, does not greatly affect tlie direction and mamitude of the surface wind-does not bring. on
t,ype of pressure distribution. Per f aps it midit, with
warm humid layers of air is more or less opposite, iis 21 low11
same direction t a a t i t i3 traveling, n directii.ni tliut
any of the familiar gush and other thunderstorm phe- nomena. Hence we must infer that somehow or other the rain is an important factor both in starting and in maintaining the winds in question, for they do not exist before tlie rain begins nor continue after it has ceased. On the other hand, i t cannot be assumed that the rain is the whole cause of these winds, for they do not accom- pany other and ordinary showers, however heavy the
downpour may be. ‘J’lie wtud course of evcnts, illustrated by figure 20,
ti9cn froiii the rccords obtained at Washington, D. C., dur-
90
80
70
9.9
9.8
60
so
43
3.5
20
IO
0
2J
1.5
1.0
0.5
0
”
FIG. ?cl.--rJurse of meteorological elements on a thunderstorm day, at IVashhgton, D. C. (July 30,1913).
in:: tl:c p t s s g e of the notable thunder squall of July 30,
In!:.;, s(v?tis to be about as follows:
I;’ii:.t. approximately adiabatic temperature gradi- vnt i.: ~.stiibli+:lird over a wide area, rough1 up to the base Iewl of tlw cui-!iulus clouds. But whi T e the uprising II~:?J:c]w~ of tdic Psisting convection currents, due to su er- adi.Lbntio gr:;clicnti, iiuiy be localized aiid here and t K ere mthcr r:q~id, tile return or downflow, though real1 the
cnuw of t h uy1r;l.f t, is widcsp-ead mid correspon 2 ingly gcntlc. Tlic condition cssantial to a local and ra id flo~~.iiflo~~~-t,lint i4, a local dccidcd cooling at a high a Y ti- tticlc-does not exist, and therefore the counterpart to tlic 1ipwa.rc1 currents is nowhere conspicuous. - Second. Aftw n time! RS a result of strong convection in a cunmlus cloud, rain is formed a t a considerable alti- tude where, of course, the air is quite cold, in fact so cold CI --
JUNE^ lQl4. MONTHLY WEATHER REVIEW.
162 MON!t'HLY W'EA'I23ER REVIEW. J m , 1014
Jmm, 1914. MONTHLY WEATHER REVIEW.
J m m , 1914
E . It is a common thing in semiarid regions to see a
shower, even a thundershower, have the base of a
heaT clou and yet, fail ut,terly, becacse of evaporation, t o reach the surface of tlie earth. Hence it appears quite certain that. in t.he aver e tliiinderstorm a considerable ortion of t,lie rain t.liat eaves tlie cloud is eva orated
E efore it, reaches tlie ground, and thwefore that t e tem- perat.ure decrease of the at.mosp1iere is largely owing to t.liis fact.. But if so, why? t,lien, one might properly ask, does not. an eqi: ally great temperat.ure drop accompany all heavy rains ?
The answer is obvious, becaim, as a rule, the tempera- tiire is Iiiglier and the relative humidity lower during a thunderst.orm than at, tlie time of any other ordinary rain. The chief, perhaps t.he soh, reaSon for this difference in relative humidity is the difference in the two cases, be- t,ween the movements of the air. In the thunderstorm t,he descending air, wliich can be no more than saturated at, top, dynamically warms EO rapidly and is so continu- ously renewed that. evaporation into it can not keep pace with its vapor capacity. During other rains, however, where t,liere is no atmospheric descent., and t,lierefore no dynamical heat,ing, ap roximate saturat.ion mcst soon ob-
but. little cooling. We will now return to the numerical values and com- pute a probable magnitude of cooling due to eva oration.
let one-fourth of the rain that, started, or half a cent.het.er, be evapciratod. This would consiime 303 heat, units from an aircolumn 3,000 meters highwhose heat capacit.yis that of only 50 cubic centimeters of water. Hence, as a result of eva oration alone, the temperat.ure of the air cclumn
\\vo1ild B e lowered on die average by about Go C. Evapc- rat,ion, therefore, appears to be both necessar and suffi-
storm. Since t.he molecular weight of water is 1s while t,be average mcolecditr weight of air is approsimately 39, it follows that, tlie amount, of evaporation above assumed woiild decrease the density of t.he atniospliere by, roughly, one part in a thousand. On the other hand, a decrease in temperature of Go C., that would be proclnc.ecl by the evaprat ion assumed, wolild increase t.lie density by about one part. in flfty. Hence the resdtant of t.1iek.e two cppos- ing effects is substantially that of the sec.oncl alone; that is, a distinct. increase in the density. Doubtless, as already stated, the evaporation of thun- derstorni rain, and t,herefore the drop in t.emperature and the consequent, gtiin in density, all increase wit,h decrease of elevation. In some nieusure, however, this effect is counteracted hy the increasing rate of dynamicltl heat- ing in the lower layers result.ing froni the correspondingly increased rate of ressure gain to change in elevation.
vary, it seeiiis quite certain that t.he cold rain of a thun- derstorm and its evaporation together must establish a local downrush of cold air-an observed important and characteristic phenomenon, redly the immediate cause of the vigorous circulation, whose rational esplanation has been atteinptcd in the ast few paragraphs.
tains in great nieasure its origind velocity and, there- fore, on reaching the earth rushes forward in the direc- tion of the storni movement, underrunning and buoying up the adjacent warm air. And this condition, largely due, as ex lained, to condensation and evaporation, once establishec! necessarily is self-perpetuating, so long tu the general temperature gradient, humidity, and wind
R P
t.ion; hence but 1itt.le ! urt.her evaporation and, of course,
As before, let, a hentimeter rain leave tlie c P oud, but
cient to produce all or nearly all the cooling o P a tliunder-
But no matter ; ow nor to wh3t estent the details may
As the colunin or shcet o P cold air flows down it main-
that hail is often roduced. Now this cold rain, or rain
from the level of its forniation all the way to tlie earth, artly as a result of its initial low tcmperaturc and partly !,cause of the evaporation that takes place during its fall. Heace this continuously chilled column of air be- cause, somewhat, of the frictional dra.g of the rain, but maid because of the increase, due to this chilling, of its
centrated and vigorous, or swiftly flowing, return branch of the vertical circulation. In fact, it (or gravity acting through it) becomes the sustJning cause of the storm's circulation. At the same time, bccause of the downward blow and because of surface friction, a,.. will be explained later, the btirometric pressure is abruptly increased. It will be worth while to consider some of thcse stnte- ments a little more closely, and to test them with possible numerical values. Omitting, as we may, the effeck of radiation, there seem to be hut three possible WRYS by which the cooliiig of a thunderstorm may be obtained: a. B the descent of originally otcntially cold air. b. By chilling the air
will be considered separately.
a. Obviously no portion of the upper air could main- tain its position if potentially even slightly colder thnn that near the surface. If a t all potentially colder it would fall until it itself became the surface air, as indecd is the case in all vertical circulation. Hence the great decrease in temperature that comes with a thundef- storm is not the result of the descent of a layer of air
originall pot-entially cold for, as explained, an upper
cooling could not exist. Again, any descending air muft
come from either below tho under surface of tho cloud or from above this level. If from below, i t must, brcause of adiabatic heating, reach the earth at substantially the original surface temperaturc?. If from above i t would, as IS obvious from figure 1, reach the earth ercn warmer than the originnl surface temperature. Hence, looked at in any WRY, case a ohviowly is inndniissible. b. Let the under surface of the thunderstorm cloud he
1,500 meters abore the earth, and the column of air cooled by the cold rain and its evaporation 2,000 n:etrrs high. Let the surface temperature be 30' C., and the temperature gradient before. the storm bcginn adidbntic up to the under-cloud level, and let there be a P-centi- meter rainfall. Now, at the temperature ssumed, a column of air 2,000 meters high wliose cross section is 1 square centimeter weighs, roi:ghly, 210 grams, and its heat capacity, tllarc- fore, is approximately that of 50 grama of wa?er. At tl:e top of this column tlie temperat,we can be, at most, only about 20' C. lower than at. t.he bottom, and if the rain leaves t.he top at t.liis temperature but reaches the eartli
7' C. cclder than the surface air before tlie storm ([em- peratures that seem at least to be of t.lio correct order) it will have been warmed 13' C. during its fall and tlie air cslumn cooled on the average about 0.5' C. But., as a
matter of fact, the air usually is cooled b from 5' C. to
sarily is reduced to some extent, by mere heat conduction to tlie cold rain much the greater portion of the cooling clearly must have some other origin. Further, since a is inadmissible and b only a minor contributing factor, it follows that by exclusion only evaporation, is left, to account for much the greater portion of tlie cooling.
Let us see, then, if evaporation really is adequate to meet these damds. .
and hail, as it fa1 P s, a.nd as long as it falls, chills the air
own B ensity, immediately and necessarily becomes a con-
with the col a rain. e. l3y evaporation. Each of these
layer su d ciently cold to give, uftcr its descent, the actual
10' C. Hence, while tlie t.emperat.ure o 9 t,he air ncces-
JUNE, 1914. MONTHLY WEATHER REVIEW.
366 MONTHLY W E A m R REVIEW. .JUNE, 1814
hmr, 1014. MONTHLY WEA!IXBR REVIEW.
868 MONTHLY WEA!I'HER REVIEW. J m , 1914
direction am favorable. It must be remembered, how- ever, that thunderstorm convection, rising -air just in front and descending air with the rain, does not occur in
a closed circuit, for the air that goes up does not return nor does the air that comes down immediately go up again, there simply is an interchange between the sur- face air in front of the storm and the upper air initsrear. The travel of the storm, by kee ing u with the under-
temperature contrast essential to .this open-circuit con- vechon as does continuous heatmg on one side and cooling on the other maintain the temperature contrast essentd to a closed circuit convection. The movements of the warm air in front of the rain, the lull, the idow, and the updraft resemble somewhat those of a horizontal cylinder restin on the earth where
the air is quiet and rolling forwar% with the speed of
the storm. Similarly, the cold air in its descent and for- ward rush, together with the updraft of warm air, also
running cold current, just as e 8 8 ectua y maintains the
foglike condensation which, of course,. renders my d s tached vortex at this position uite visible.
top is against the storm, ma be regarded as a thiid
complete than eithcr of the others. Schematic illustrutions.-The above conceptions of the
mechanism of a thunderstorm can, perhaps, be made a
little clearer with thc aid of illustrations. Figure 21, a
to fall, and the stream lines of
This squall cloud, in whic.h t B e direction of motion on
horizontal thunderstorm cylin d er much smaller but more
I
in the makin cloud from whic
Fro. !2l.--prlncipal a t movements in the development ora cumulus cloud.
resembles a horizontal cylinder, but one sliding on the earth and turning in the o posite direction from that of the forward r o b or alrwarm cylinder. I n neither
ex lained, the air that goes up remains aloft while the
to the lower levels. The concbtion of flow persists, as
do cataracts and crest clouds, but here, too, as in their case, the material involved is ever renewed.
The squall cloud.-Between the u rising sheet of warm
zontal vortices are sure to be formed in which the two currents are more 01: less mixed. The lower of these vortices can only be znferred as a necessary consequence of the opposite directions of flow of the adjacent sheets of warm and cold air, for there is nothing to render them visible. Neither can any vortices that may exist within the cloud be seen. Near the front lower ed e of the cumulo-nimbus system, however, and immezately in front of the sheet of rain, or rain and hail, the rising air has so nearly reached its dew point that the somewhat lower temperature, produced by the admixture of the dewending cold air, ui~ ouffieient to produce in it e iight
case, however, is t 5l e analogy complete, for, as above
GO P d air that comes down is kept by its greater density
air and the adjacent descending s % eet of cold air hori-
concentrated or local down current, only an impcrceptible counter settling of the air round about, because as - viously explained, the air cataract requires locai cooEg to subpotential temperatures, and this in turn requires local rain. Figure 33 scheniatically represents a well-developed thunderstorm in rogress. The falling rain, often mixed
temperature oTaclient was a ready closely adiabatic it follows that &e actual temperatures will be subpotential from the sui.fac.e of the earth to within the cloud, or throughout and a little beyond the nonsaturated or evaporating levels. As soon, then, as this column or sheet of air is suflicientl cooled i t flows down and for-
thundeistorm are established substantially as shown. Referrin to the ure: The warm ascendin air is in
(in dry weather) at D', the s uall cloud at S- the storm
primary rain, due to initial convection at R ; and the secondary rain at R'. This latter phenomenon, the recondary rain, b 8 thbg of frequent occurrence and
K
with hail, cools t P le air throu h which i t falls, and as the
ward and all the atmosp 1 eric movements peculiar to the
the region 8; the co3descending air a t D; the f ust cloud
collar a t C; the thunder hea 2 s at T; the haif a t H; the
J m , 1914. MONTHLY WEATHER REVIEW. 368
often is due, M indicated in the figure, to the coalescence and quiet settling of drops from an abandoned portion of the cumulus in which and below which winds-and con- mction are no lon er active. Mmmato-cumu% rarely, false cirri frequently, and cap-cbuds occasionally, accom any thunderstorms,. but as tbp are not essential to i t g e y therefore are omitted
Thundmstomn pressurF.-Before the onset of a thun- derstorm there usually if not always is a distinct fall in the barometer. At times this fall is extended over several hours, but whether the period be loiig or short the rate of fall usually is greatest at the near approach of the storm. Just as the storm breaks, however, the pressure rises very rapidly, almost abruptly, usual1 from 1 to 2
storm asses again becomes rather steady but at a some- what figher pressure than prevailed before the storm began. The cause of these pressure changes is, doubtless, rather complex. The decrease in the absolute humidity
from & e above schematic illustration.
millimeters, fluctuates irregularly, and h ally a,, the
in a thunderstorm is not at all aecurafel known, but while
14 meters er second seems to be excessive; in fact, its
it is not, then, since the ressure of a wmd varies as the
Hence it wou d seem that there probably is at least one other pressure factor, and, indeed, such a factor obviously esists in the check to the horizontal flow caused by ver- tical convection. To make this point clear: Assume two la ers of air, an upper and a lower, flowing parallel to eats other. Let their respedve masses per unit length in the direction of their horizontal movement be M md m, and their velocities V and v. Now, if, through convection, say, the whole or my portion of the lower layer is carried aloft, obviously it must be replaced below by an equal amount of the upper layer. Let the whole of the lower layer be carried up. This layer, to produce the rain that was above assumed, 2
at times probably very considerable, t E e above value of
average va P ue may not be even half so @?at. If in reality
s uare of its velocity, it P ollows that less than one-fourth
o 9 the actus1 ressure increase can be caused in this way.
Y
thunderheads,
the atmospheric pressure, and, presumably, each con- tributes ita share. Both thcse effects, however, are comparative1 permanent, and while they may be mainly responsible &r the increase of pressure that peis::its after the storm has gone by, t h y probabl are not the
produced pressure maximuiu. Here a t lc& two factors, one obvious, tho other inconspicuous, are involved. These are: a. T h e ra id downrush of air, and b. the intrr-
tion. The downrush of air clearly produces a vertically di- rected pressure on the surface of the earth, in the s.me manner that a horizontal flow produces a horizontally directed pressure n ninst the side of a house. But a
pressure increase frequm tly reached in a thunderstorm, would mean about 2.72 grams per square centimeter, or 27.2 kilograms per square meter, and require a wind velocity of rou hly 50 kilometers er hour or 14 meters
chief factors in the production of the iiiitia 9 aid quickly
fercnce to horizonta P flow caused by thc vertical circula-
pressure equal to t a a t given by 3 nim. of mercury, a
per second. #ow, the velocity o I the downrush of air
and the decrease in temperature both tend to increasp--centimeters, will have to be a t least 1 kilometer deep, and if it should merely change laces with the upper am, or if the different lavers shou P d minde and assume a
coninion velocity, V’, ihcre obviously gould be no change
iil t!ic total linear momentum, nor in the flow. In sym- bols we would have the equation
iKV+mv=(M+m) V’.
uo MONTHLY WEATHER REVIEW.
JUNE, 1914. MONTHLY WEATHER REVIEW.
372 MON!I'HLY WEATHER REMEW. JUNE, 1914
J-, 1914 MONTHLY WEATHER REVIEW. 373
one&fth its former value. This would reduce the total flow by about 1 part in 400, and thereby increase the barometric reading by nearly 2 millimeters. It would seem, then, that the friction of the thunder- storm gust on the surface of the earth, through the con- sequent decrease in the total linear momentum of the atmosphere and, therefore, its total flow, must be an important contnbuting cause of the rapid and marked increase of the barometric pressure that accompanies the onset of a hcavy thunderstorm. To sum u : The chief factors
tude: a. Decrease of horizontal friction. b. Vertical wind pressure, due to descending air. c. Lower temperature. d . Decrease in absolute
storm the temperature commonly is high, but it begins rapidly to fall with the first outward gust and soon drops often as much as 5’C. to 10’ C. because, as already es- lained, this gust is a portion of the descending air cooled gy the cold rain and by its eva oration. As the storm
though it seldom regains its original value. Thunderstm. hu.m.idihj.-As previously explained, heavy rain, at least up in the clouds, and therefore much hu- midity, and a temperature contrast sufficient to pro- duce rapid vertical convection, are essential to the genesis of a thunderstorm. Hence during the early forenoon of a
thunderstorm day both the absolute and the relative humidity are likely to be high. Just before the storm, however, when the temperature has greatly increased, though the absolute humidity still is high, the relative humidity is likely to be rather low. On’ the other hand, during and immediately after the storm, because chiefly of the decrease in temperature, the absolute humidity IS
corn arativel low and tlie relative humidity hi h. “fZainqu8~”--It has frequently been note8 that the rainfall is greatest after heavy claps of thunder, a fact that appears to have given much comfort and great en- couragement to those who maintain the efficacy of mere noise to induce precipitation-to jostle cloud particles togother into raindrops. The correct explanation, how- ever, of this phenomenon seems obvious: The violent turmoil and spasmodic movements within a large cumulus or thunderstorm cloud cause similar irregularities in t-lie condensation and resulting number of raindrops a t any given level. These in turn, as broken by the air currents, give local excess of electrification and of electric dischar e or li htning flash. We have, then, startin toward t8e the same level, mass, sound, and light. The ight travels with the greatest velocity, about 300,000 kilometers per second, and therefore the lightning flash is seen before the thunder is heard; its velocity bein , roughly, only 330 meters per second. But the rain fa& much slower still and there- fore reaches the ground after the thunder is heard. In reality it is the excessive condensation or rain formation up in the cumulus cloud that causes the vivid lightning and the heavy thunder. According only to the order in which their several velocities cause them to reach tshe surface of the earth it mi ht appear, and has often been
heavy thunder, next in order, was the cause of the esces- sive rain, which most certainly it is not. Thundptorm vdoeit .-The velocity of the thunder-
increase of t 1 e barometric storm appear to be, possibly
humidiY. Thun erstomn temperutures.-Before the onset of the
passes the temperature genera Y ly recovers somewhat,
eart E at the same time and from racticaly f
so int.er reted, that the lig 5 tning, first perceived, was the cause o P the thunder, which, indeed, i t is, and that the
storm is simply the ve P ocity of the atmosphere in which
the bulk of the cumulus cloud happens to be located. Hence as the wind at this level is faster by ni ht than by day and faster over the ocean than over lan%, it follows that exactly the same relations hold for the thunderstorm, that it travels faster over water than over land and faster by night than by day. The nctual velocity of the thunderstorm, of course, varies greatly, but its aver e velocity in Europe is 30 to 50 kilometers per hour; in #e United States, 50 to 65 kilometers per hour. Hail.-Hail, consisting of lumps of roughly concentric layers of compact snow and solid ice, is a conspicuous and well-known phenomenon that occurs during the early portion of most severe thunderstorms. But in what ortion of tlie cloud it is formed and b what process the
being obvious, become clear only when the mechanism of the storm itself is understood. As before, let the surface temperature be 30’ C. and the absolute huniidit 40 per cent, or tlie dew point 15’ C.
fore, cloud formation will begin wlien the surface air has risen to an elevation of 1.5 kilometers. Immediately above this level the latent heat of condensation reduces the rate of temperature decrease with elevation to about half its former value, nor does this rate rapidly increase with further gain of height. Hence, usually, for the abwre assuniptions correspond in general to avera e thunderstorm conditions, it is only beyond the 4-ki B 0-
meter level that freezing ternpcratures are reached. It is therefore only in the upper portions of cumulus clouds,
and undercooled fog or cloud droplets, that Barticla ail can the portions that clearly must consist of snow
either originate or greatly grow. But whnt, then, is the process by which the nucleus of the hailstone is formed and its layer upon layer of snow and ice built up? Obrioudy such drops of rain as the strong upclraft within the cloud may blow into the region of freezing teniperatures will quickly congeal and also gather coatings of snow and frost. After a time each inci ient hailstone mill get into a weaker updraft, for this
edoe of the ascending co umn. In either case it will then fa17 back into the region of liquid dro .:, where it will
%e frozen by tge low temperature of the kernel. But again it meets an upward gust, or falls back where the ascendin drnft is stronger, and again the cyclic journe
time-there may be several-the journey 1‘: completed a
new layer of ice a.11~1 z fresh Inyer of snow are added. In general the size of the hailstone3 will be roughly propor- tional to the strength of the convection current, but smce their weights vary approxinintmely (they are not homo- geneous) as the cube of their diameters while the sup- porting force of the upward air current varies, also a11 rosimat8ely, as only the square of their diameters it
evident, from the fact that a strong convection current is essential to the forination of hail, that it can occur only where this convection exists; that is, in the front portion of a heavy to violent thunderstorm. Some meteorologists hold that the roll scud between the awending warm and descending cold air is the seat of hail formation, but this is a mistaken assumption. Centrifugal force would throw a solid object, like a hail- stone, out of this roll probably before a single turn had been completed. Besides, and this objection is, Jerhaps, more obviouvl fatal than the one just given, t .h e tem- perature of d e roll scud, because of its position, the
fayers of ice and snow are built up are 9 acts that, far from
Under these conc T itions saturation will obtain, and, there-
P
is a H ways irregular and ufl’y, or else will tumble to the
ather a coatina of water, a portion of w E‘ ich will at once
K from rea f m of ra.in to region of snow is begun; and eac
fol r ows that n limiting size i;j quickly reached. It is also
374 MONTHLY WEATHER REVIEW. JUNE, 1914
lowest of the whole storm cloud, clearly must be many de ees,above the' freezing point. Indeed, as-the above c aF culation shorn, temperatures low enough for the for- mation of hail cmi not often obtain at levels much less than three times that of the scud, and therefore it clearly is in the higher levels of the cumulus and not in the low scud that hail must have its genesis and make its growth. &Mning.-About the middle of the eighteenth cen- tury Franklin and others clearly demonstrated that the lightning .of a thunderstorm and the discharge of an ordinary electric machine are identical in nature, and
simultaneous but locally disconnected streaks. Fre- quently the dischaye continues flickeringly (on rare occa- sions evm steady, like a white-hot wire) during a percep- tible time-occasionally a full second. But all these phenomena are best studied by means of the camera, and have been so studied by several ersons, among whom Walter, of Hamburg, and Larsen, of &cago,
are two of the most persistent and successful. Stationary cameras, revolving cameras. stereoscopic cameras, cam- eras with revolving lates, and cameras with s ectro-
jointly, and the results have abundantly justified the time and the labor devoted to the work. Figure 34, eopied by permission from one of Walter's un ublished negatives, shows the ordinary tracery of a
camera. It is on y a perinanent record of t.he ap marance of the lighning to the unaided eye. Figure 35, owever, also copied by Walter's kind perinission from one of his unpublished photo raphs, is a record of the same dis-
that the more nearly vertical discharge occurred but once or was single; that this discharge was quickly followed by a second along the same path to about onefourth of the way to the sarth where it branched off on a new course; that the second discharge was followed in turn at short but irregiilar intervals by R whole series of sequent dis- charges; that most of the dischar es appeared as narrow
discharges appeared, not to t.he eye, but on the plate of the rotating camera, as a broad band or ribbon. On close inspection it will be obvious that the plaidlike ribbon effect is due, the warp to irre ularities in the more or less
and therefore brighter ortions of the strea -. Another point particular1 wort F, y of attention is the fact that
following ones remain entire from end to end and am nowhere subdivided. Figure 36, taken. froni a hotogra h obtained by Mr.
Smithsonian Institution, shows another series of sequent discharges siniilar to those of figure 25, exce t that in t,his case there was no ribbon discharge. T l e time of the whole discharge, as calculated by Mi. Larsen, was 0.315 second. Here, too, side branches occur with tlie fint but. only tlie first discharge. This, however, is not an invariable rule for occasionall , as illustrated by
the side branches ersiilt through two or three of the
case each tributary &en re eated follows, as does the nisin stream, its own oiigin ap channel. Tile phenonienon of sequent discharges, all along the same path, and the clisn psarance of the side branches with or quick1 7 after the t 1st discharge both seem reason- :hly clear. $he first discharge, however produced, ob- viously takes place against very great resistance, and therefore under conditions the most favorable for the occurrence of side branches or ramifications. But the clischarge itself leaves the air along its path temporaril highly ionized-puts a temporary line conductor wit here and there a poorer conductin branch, in the atnios- pliere. This conductor is not on P y temporary (half the
icms are reunitedin about 0.15 second, the air being dust )
tlie atmospheric violence it itself creates. Because partly, perhaps, of just such interruptions, and because also of the volunie distribution of the electricity which prevents
n sudden and coiiiplete discharge, the actual discharge is tlivitbd into a nuiiiber of artials that occur sequent1 .
if they esist, are only here and there and but little more than sufficient to interru t the flow. Hence the next discharge, if it occurs uicE1 must follow the wnducting and, therefore, origin3 discgarge path. Besides, in the
graphic attachments lave all been used, separate f y and
lig g tning dischar e when photographed with a stationary
charge obtained wit % a rotating caiiiera. It will he noted
h f
intensely luminous streaks, and t 7l at one of the sequent
continuous discharge, and t a e woof t.0 rou lily end-on
while the first dy ischarge has several side branches the
Larsen, of Chicago, and kin a ly loane 3 for use here by the
a
figure 27, copied from apublished p K otograph by Walter,
first successive disc B aroes, but not through all. In such
K
but also so estreniely fragile as to be liable to rupture z y
Ol>viously, the breaks in t f le conducting (ionized) pat K ,
M. W. R., June, 1914.
CIC
To face page 374.
II
--
I.
FIG. =.--The growth of an electric spark discharge (Walter).!
i a
R
II
FIG. 24.-Streak lightning stationary camera; companion to Fii. 25 (Walter).
FIG. 26.-Streak lightning (sequent discharges), rotating camera (Larsen).
Y
FIG. 27.--Streak lightning (sequent discharges), rotating camera (Walter).
FIG. %.-Streak lightning (sequent dlSChm es) rotating camera; companion to Fig. 24 (Wafterj.
Idghtnine.
I
FIG. 28.-Spectrum of lightning (Fox).
JUNE, 1914. MONTHLY WEATHER REVIEW. 875
subse uent discharges the original side branches will lie
or, what comes to the same thing, because of the ni0re
abundant ionization and consequent higher conIluc tivity of the path of heaviest discliargo. This leav.es tlie enesis of the initial clischrtrge, ofben if
probably is, at present, tlie least uaderstooil of d l the many thunderstorm phenomena. Judging froill the voltages required to produce laboratory spa.rke, roughly 30,000 volts per centimeter, it is not obvious how such
tremendous voltage tliffereiices can be establisliccl l ~e -
tween clouds or between a cloud and the enrth as would
semi to be necessary to produce a discharge kilometers in length, as often occuis. Of course the potential of individual drops may grow in either of two wt~ys: a. By coalescence of equally charged smaller drops into lnrger
ones. In this case, since capacity is directly roportioned to the radius, the potentials of the intlivicluaf clrops must be proportional to the squares of their radii. 6 . By evaporation of ec@l~ charged drops. Here the potcn- tials of the individua drops obviously is inversely NO-
portional to their radii. In each case the tendency o h the separate drops to discharge is increased, I.mt tlie potcntinl of the cloud as a whole remains uncha.ngcc1. At, prescnt, therefore, one can do hut little niom thnn specula.i,e on
the subject of the primary liglitning discl~:i~i-ge, h i t (wen
that much may be worth w1111c4 since it helps one to Tu-
meniber the facts. As already esplainerl the elec.tricn1 sopnixtkni witliln 1%
thunderstorm cloud is such as to p1a.c.e a liemily ch:~rgc.d positive layer (lower >ortion of the cloutl) hetwcen tlie
layer (upper portion of the cloud). Hencn tlie tlisclini~gt~a. or lightning, from the interme1li:ite or positively chargntl layer niay be either to the negative p(.trtion diw-e, in
soiiie cases even to an entirely tliflereat clc)uil, or fo f91:e
earth below. Further, through the sustaining inflti(mct? and turbulence of the uprushing air there imist, 1-ie foi~ni~tl at times and pln.ces pr~tctically continuous sheetnu a.ii~i streams of water, of coiirse henril charge#l nntl a t high potential, a.iic1 also layers ani1 strea%s of liiglity ictiiizetl an;
that is, electrically speakin , heavily cliai ge(l ccriiductinp sheets and rods, whether o B coalesced drops ri? of ionixetI air. are over and over, so long tis the 4h:irii: liists, I;:I(:-
nieiitaiily placsd here and there within Idle positive17 charged iiiass of the storm cloud. Let us see, then, what mi lit be espected as the result, of this peculiar disposition o B charges and conductors, t.he
result, namely, of the esistence of a heavily suifnce- charged vertical concluct.or in a strongly volunie-charged horizontal layer or region above ant1 below- w1iic.h tlicre are steep otential gradients to negatively charged parallel su I? aces. The conductor will be a t the same potential throughout,, and therefore the mnsima of potent.inl grnt1ient.s nornial
to it will be at its ends, where, if these gratlicnt,~ tire steep enough, and the longer the conduct,or the steeper t,he gradients, brush discharges will t,ake place. Assume, then, that a brush discharge does take place and t.11n.t t,here is a supply of electricity flowing int,o the concluct,or t.o make good the loss. The brush and tlie line of it.s most vigorous ionization necessarily will be directed along the potential gradient or toward the surface of 0pposit.e charge. But this very ionizat,ion aut,omatically increases the length of the conthxtor, for a pat.h of highly ionized air is a conductor, and as the Iengt,li of the conrluct,or grows so, too, does the steepness of the potential gra.dicnt
at its forward or terminal end, and as the steepness of
quick i! y abandoned because of their greater mc;istmco,
not usually the on f y one, to be esplained, and incleetl this
earth and a much liig t er, also heildy charged, iiegatiw
this gradient increases the more vigorous the discharge, alwa s assuming an abundant electrical sup ly. Hence,
cloud has a good chance, by making it.s own conductor as it goes, of geometrically growin into a li htnin flash of
is siiiall the lightmiiig is feeble and soon dissipated. Whether the discharge actually does burrow its way through the atmosphere in sonie such manner as that indi- cated probably would be difficult, though not necessarily inipossible, of observation. Indeed, a roughly analogous phenonienon (10) can be produced on a photographic plnt,e by bringing in coiit,ac.t wit,h the film, some dlstance apart, two c.onducting points attached to the o posite poles of an influence machine. Brush discharges i! evelo about each oint, but the glow at the negative r%
. As it oes it continually toward the positive builds for itself, out o the silver of t e eniulsion, a con- ducting path. RocX.et lightning.-Many persons have observed what at least seemed to be a progressive growth in the length of a
streak of lightning. In sonie cases (11) this growth or progression has a peared so slow as actually to suggost
At first one might well feel disposed to re ard the phe-
definitely described and too frequently observed to justify such suniniary dismissal. Naturally, in the course of thousands of lightnin discharges, many degrees of ioni-
potential eraclient are encountered. Ordinarily the growth of %e discharge, doubtless, IS rn a eometric ratio
possible for t,he conditions to be such that the dischar e can barely more than sustain itself, in which case t % e
niovenien t of the flash terminal may, possibly, be rela- t.ivrly slow. Bull Iig7dniv1 .-Curious luminous balls or masses, of
account., have tinie ancl again been reported among the phenoniena observed during a thunderstmom. Most of theni appear to last only a second or two and to have bren seen a.t close ran e, some even passing thou h a
st.one (13), or like a meteor, &om the storm cloud, and along the approsiniate path of both previous and subse-
c uent liglit,ning flashes. Others appear to start from a
c\oud and then quickly return, and so on through an endless vwiety of places ancl conditions.
are en tirely spurious, being either &sed or wanderin Erush discharges or else nothing other than optical il f usions, tluc in most cases probably to persistence of vision. But here, too, 8s in t.2ie cttse of rocket,light,ning, t,he amount and escelleiice cA observational evidence forbid the aseunipt~ion t.hat. nll such phenemena are merely sub- ject,ive. Possibly in some instances, especially t-hose
111 w-hich i t is seen to fall frmi the clouds, ball lightning
I P R .~ be only est.renie cases of rocket lightning, cases in whwh the clischarge for a t.inie just sustains iteelf. A clusrly sinijlar ic!en has been developed in detail by Toe der (141. It may either disappear wholly and noise- Icssky, as oft.eli reported, or it. could perhaps suddenly gain ii! stren t.li and instantly diea.p ear, as sometimes
those that are not. niere optical illusions are stalled tfunder- bolts, certainly may sound very strange. But that
an e Y ectric spark once started within a t E. understorm
large dimensions. Of course w fl en the e P 5 ectrica supply
detaches itse 9 f and slowly nieandeis across the p ate
?Pit a
the flight of a roc r -et, hence the name.
~ionienon in question as illusory, but it E as been too
zat,ion, availability o f electrical charge, and slopes of
alii1 the progress of it,s end exceedingly swi 7 t, but it seems
which C. De 59 ans (12) probably has given the fullest
house, but they have a 4 so ap eared to fall, as wou ! d a
I)oubt.less many reported cases of ball lightnin
observed, wit 7 i a ebnrp abrupt clap o F thunder. Tu say that all genuine c.s.ses of ball Ijght-nin
376 MONTHLY WEATHER RETrlEW. JUNE, 1914
indeed is just what they are ac.cording t.o t.he ahove apeculat.ioii, a specula.t.ion tmhat. reco izes no difference in kind between streak, rocket, and?22ll Ijghining,. only differences in the amounts of ionization, quant.it.ies of
availa.ble electrici ty and steepness of pot.entia1 grdients. SA.eet Zigh.tning.-When a dist.ant. t~liuadercloud is observed at- night one is uite certnin to see in it. bcauti-
often flicker and glow 111 exactly the snnie inmner as
does streak lightning for well-nigh a whole second. In the daytime and in full sunlight the phenomenon when seen at all ap ears like a sudden sheen t,liat tmivels :u:cl spreails here an s there over the surface of the cloucl. Certninly in most cases, so far 8s definitely known in all cases this is only reflect.ion from t.lie body of t.lic cloud of streak lightning in ot.her and invisible portions. C'onceivnbly a brush or coronal discha.rge may take >lnce from the
expect this to be either a faint cont.inuuus glow c)r else
a momentary flash coincident wit.11 a discharge from the lower port-ion of the cloud to earth or to eunie other cloud. But, as already stated, only reflection is deli- nitely known to be the cause of sheet lightning. C'oroiiid effectss seem occasionally possiblet but that they are ever the cause of the henumenon in queshn has ncver
It has often been asserted, too, that tlierc? is ti railicnl difference between the spectra of storl.eali n1:d sheet lightning, but even this does not tippear ever t.cr h a ~e been phot.ogra.phically >roved. Beaded ~~g~.t~~g.-$iscontiiiuous 'or beaded streaks of
lightnin have been reported from tinie to tinie. Indeed
impression of seeing, this phenomenon, but with one or
two doubtful esceptions lie felt prwticnliy ccrtaiii that it was only an o tical illusion. In nddition to visual
graphs showing streaks of light broken i1it.o inore or less
evenly spaced dashes have been obtained and rcporbetl aa photographs of beaded lightning. Without excep- tion, however, these seein certainly to be nothing other than the photo ra lis of alternating current elect,ric lights, taken wi t f f i t ie ctiineru in motion. The objectire reality, therefore, of beaded ligiitning does not, seein at :ill well established, at least, not sufficiently well to justify an serious effort to esplain it.
fretum Ziqhtniii. .-This is comnionly referred to as the
discharges that take plrtce here and i.lisre from objects
on the surface of the earth coinciclentl-~ wit,li liglit4iiing flashes, and as a result of the suddenly chitnged elect.ricn1 strain. These discharges are always siiinll in coinparison with the main lightning flash, but st, tinies they are suiii- cient to induce explosions, to start fires, and e v ~n t,o take life. Dark Zightwin.g.-When t~ hotogpplic. plate is esposecl
that one or more of the streak images, on development, exhibits the " Clayden effectJJ-tliats is, a ~petm coiiipletrly
viously, then, on prints froni such a negative the reversed streaks must a pear as dark lines, and for that reiisoii rodiiced thein have been
called "dark liglitning." '&ere is, of course, no such thing as dark lightning, but tlie pliotogttpliic. phenome- non that ave rise. to the nanie IS real, interesting, and
ful illuminations, looking Y ike great eheet,s of flnme, that.
upper surface of a tohunderat.orni cloud, L u t one w-ould
clearly been establis P led and appeals verS doubtful.
the au z or liniaelf has several tinics seen, or hud t.he
observations of t r ie kind just described iiiiuiy Iilioto-
return shock, an (9 is only those relatively sinal! electrical
to a succession of lightning rp ashe!, it occctsionnlly lispjmis
reversed-while the others show no suc 1 i tendency. 011-
the lightning i ashes that
reproducib 9 e at m11 in the laboratory(l5).
Tern cmturc.-What the temperature along the path of a ligttning discharge is no one knows, but it obviously is high, since it frequent.ly sets fire to buildings, trees,and many other objects struck. In an ordinary electrical conductor the ainount of heat generated in a ven time by ai1 dectric current is proportional to the pro 9 uct ORT, in which C is the strength of the current, R the ohmic resistitlice, and T the time in question during which C and R are su>posed to remain constant. I n a spark discharge of t / it! nature ol lightning some of the energy protlucrs cfiects, such as decomposit!ion and ionization, othcr th:m mere local heating, but as esperiment shows, n great deal of helkt is generated, according, so far as
we know? to the same laws that obtain for ordinary con- ductors. Hence est,ra heavy discharges, like estra large currcnts, roduce excessive heating, and therefore are far more lial>!e than are light ones to set on fire any objects
T-isibJity.-Just how a. lightning discharge renders the :it.mosphere through which it passes lunlinous is not clefinitcly known. It must and does niake the nir path very hot, its me hitvt. seen, but no one has yet succeeded, by m y niiiount of ordiiiiwy heating, in rendering either
osj-g:n or nit.rogcn hniinous. Hence i t seem well nigh cerbaiii that the light of lightning flushes owes its origin
to winething other than high teliiperature, probably to
i11t,crliiil atomic disturbances incluced by the surlftly inoving elcctrons of tlie discliwge. A'pc ctl.iim.-Li~litiiiiig flashes are of two colors, white nnd pink or rose. The rose-colored flashes, when esam- iiicd in the spectroscope, show severd lines due to hydro- gen which. crf COUPSI?. is furnished by the decomposition of sollie of the wat.or dona the lightning path. The white flashes, on the other hm8, show no hydrogen lines or at iiiost but. fniiit oncs. As one might suspect, the spectruin of ii lightning h s h and that of an ordinary electric spark in uir i~re 1jxcticnll.v identical. This is well shown by i'igurr. ?S, copied from an article on the spectrum of lightning hy Yos (16). in w1iic.h the up er or wavy por-
portion to :I hhorat,org sp;~rk in air.
it is oi'tcii asscrted that tile spectrum of strcalc light- ;ling coiisists ~vliolly of briglit lilies and that shect light-
:iiiig @y s onIy nitzogen 11ands; and from this i t is argtieu tmtU the latt:>r is not a mere reflnction of tlie first. This msa.tlon is not supportcd by figure 37, tlie brightrst portions of which, the portions that mould the 1oug:at IN s,wi as rrflcctioii grow stcadily fcrbli+r, coincide with strong nitrogen bnuds. In this c.onnection, how-
ever, i t should be rcnicmbercd that accurate wave- lengt!i mpasurementa. and therefore positive identifica-
to tiic smdl disl>cimm or separation of the lines on al 7
tioii of the liiics of li&ting spectra, is not possible, owin
such ncgativrs so far reported. Dumtion.-Tiio duration of the lightning discharge is excmdingly variable, ranging froni t 1.90W2 second for a singlr flash too, in rare caszscs, even L full secoiid or inore
for a niultiplr flash consisting of n primary and a series of scyumt ilaslies. 011 rare occasions a dischare of long duration app~ms to th.e eye to be steady like a glowing solid. Possibly tlic? best nieasurements of the shorter intrrva.Lq w ~r c iiiaclr by De Blois (17) with the aid of a high-frec1iicmc.y oscillograph. He found t-he durations of 35 single peaks, averaging 0.00065 second, to range from
O .(X J O ~ secoiid to t~.tNJ16 se,cond. Flashes that last as lon
are ahnost certainly multiple, consisting of a succession
th2Lt t h y n1uy hit.
tioii is due to the light,ning niid the r ower or straight
ns a few tenths or even a few hundredths of a secoii 2
JUNE, 1914. MONTHLY WEATHER REVIEW. 377
of apparentl individual discharges occurring at unequal intervals. Jccasionally a practical1 continugus ais- charge of va ing intensity, but a i the time strong enough to ro 7 uce luminosity, will last a few 1iundrc.dtlis of a seconP It must be remembered that the duration of eren a single discharge and the length of time to coiiiplcte the circuit,'or ionize a ath, from cloud to earth, say, are entirely different gings. The latter seems usua.lly (rocket and ball lightning may furnish exceptions) to be of exceedingly short duration, while the former depends upon the supply of electricity and the oliniic resistance directl
for some reason or other, or for no reason, made the statement that
trary, it became a favorite dogma of the testbook, aased on unquestioned from author to author and Eanded down inviolate from edition to edition. Therc often are a number of successive discharges in a fraction of B second, as photographs taken with a revolving camera show, but they are not only dong tlie samv This is obvious
1 rom the fact that when the side branches persist, as in figure 26, through two or more partial or srquent clis- charges, they are always tiirnecl in the same direction. It is also proved by the direct evidence of the oscillo-
and upon the potential difference invcisely. Disc K arges darect, not alternating.-Years ago soxiie one
high fre uency, forthwi%, despite
ath but also in the same direction.
graph (18):
In the case of each seDarate discharge also the direc- tion seems constant. It may vary in stl.cmgth, or pulsate. but, apparently, it does not alternate. Tlirre are several reasons for concluding that li$i tning discliargcs are direct and not alternating, of which the following cover
a wide range and probably arc the best: a. Lightning operates telegraph instriimcmts. If the discharge were alternating it would not do so. 6. At times it reverses tlie polarity of dynaiiios. This re uires a direct and not a high-frequency alternat-
c. The oscillo a h (19) shows each surge or pulsation,
d. The relative values of the ohmic resistance, the self-induction, and the capacity, in the case of a lightning discharge, appear usually, if not always, to be such as
to forbid the possibility of oscillations. It has been shown that whenever the product of the capacity by the s uare of the resistance IS great.er than four times the selkduction, or, in symbols, that when- ever
oscillations are impossible. Undoubtedly all t-hese terms vary greatly in the case of lightning discharges, but R, presumably, is always sufficieii tly large to maintain t,he above inequality and therefore absolutely to prevent oscillations. Possibly a calculation giving roughly the numerical order of the tei-rus involved would be he1 ful. For this
w i g a radius of 3 kilometers, and whose height above the round is 1 kilometer, and let there be a ducharge from &e center of the cloud base straight to t,he earth: Find n robable value for the self-induction and capacity, and Korn these the liniitin value of the resistance to prevent
ing disc 1 arge.
as well as the w 73 o e flash, to be unidirectional.
ORa > 4L
pu ose assunie a cloud whose unclersur F ace is circular
oscillations, or the v 3 ue of R in the equation
OR2= U.
To find L wo have the fact that the coefficient of self- induction is numerically equal to twice t-he mer y in the
ther fact that per unit volunie this energ is numerically eqiial to H/Sn, in which H is the m netic force. Let a
density in it to be uniform. Let 6 he t.he equivalent radius of trhe cylinder, concentric with the lightning path, along which the return or clisplacement current flows. In this case the energy W of the niagnet,ic field per centimeter length of the discharge is given by the equation
magnet.jc ficld per unit current in the circuit, an d the fur-
he the radius of t,he lightning p t h an 3 assume the current
Let 6=2 kilometers and a=5 centimeters. Then TILlog,4 x lo(+ 3 = 11, approsimately. Hence the ener of the niagnetjc field per unit current for the mho e length, 1 kilometer. of the flash is represented by the equation
W106= 11 x 106,
hence the self-induction = 82 x IO5 =23 X lo-' henry. To find C' we shall assume a uniforni field between the cloud and the earth. As a nirttter of fact this field is not uiiiforn;, and the calculated value of 0, based u on the above assuniption, is somewhat less than its actu3 value. Assuming, t.hen, a uniforni field we have
gs
Q= a = 7;9x10'o~~85x10s=35x10-8 farad, about. 4xd 41L~105
Hence, substituting in the equation
C B = 4L,
R= 190 ohms per kilometer, approximately.
R'eitlicr a, the radius of the lightning path, nor 6, the eqiiivihit radius of the return current is accurately known, but from the obviously large amount of suddenly espanded air necessary to vocluce the atmos heric dis-
tiiiieber would be the niinimuni value for a. Also, from the size of thi.inder clouds, it a pears that 10 kilometers
On substituting these cxtrenie values in the above equations, we get
we got
turhaiiees i n d e n t to thunc \ er it would seem t 1 a t 1 cen-
would be the masimuni value P or 6.
as the alt.ituc1e of the cloud i t follows t E at, other things
the value of R er unit 5 ength. If the nltitucc P is kept constant and the size of the
R = 200 ohms per kilometer, roughly.
From the fact that C' varies inverse1 and L directly
remaining equd, the hei ht of the cloud has no effect on
cloud varied C! will increase directly as the area, and L will increase directly as the natural logarithm of the equivalent radius of the cylinder of return current. Assuming the area of the cloud base to be 1 square kilo- meter, which certainly is f a r less than the o r h a r y size,
and computing as above we find
R = S50 ohms per kilometer, roughly.
Again, assuming the base area to be 1,000 s uare kilo- meters, an area'far in escess of that of the 1 me of an ordinary thunderstorm cloud, we find
R = 35 ohms per kilometer, roughly.
It would seem, therefore, that a resistance along the
light.ning ath of the order of 200 ohms per kilometer,
absolutely to prevent electrical oscillations between or 0.008 o E ni per centimeter, would suffice, in most cases,
378 MONTHLY WEATHER REVIEW. JUNE, 1914
cloud and earth. I n reality the total resistance includes, in addition to that upon which the above c+lculations are based, the resistance in parallel of the numerous feeders or branches within the cloud itaelf. In other words, the assumption that the resistance of the con- denser plates is negligible may not be strictly true in the case of a cloud. Nor is this the only uncertctinty, for no one knows what the resistance along the path of even the main discharge actually is. though, judging from the resistance of an oscillatory eiectric spark (ZO),
it, presumably, is much reater than the cdculated
have seen, must be unidirectional and not alternating. Length of strea&.-The total length of a streak of light- varies greatly. Indeed the brush discharge so
to distinguish between them, nor, therefore, to set L
minimum limit to the length of n lightning ath. When
path is seldom more than 2 to 3 kilometem, but, in the case of low-lying clouds, may be much less, and especially
so when the envelo a mountain peak.
to cloud the path generally is far more tortuous and its total length much greater, amounting a t times to 10, 15,
and even 20 kilometers. Discharge, w h e to w7wreP-As already ex lained,
one art to another of the same cloud, or from cloud to clou~. But since the great amount of electrical separa-
tion, without which the lightning could not occur, takes lace within the rain cloud, it fouows that this is also
%kely to be the seat of the stee est potential gradients.
frequently between the lower and the upper portions of the same cloud, and this is fully supported by observa- tions. The next in frequency, especially in mountain- ous regions, is the discharge from cloud (lower portion) to earth, and the least frequent of all, ordinarily, those that take place between one and mother entirely inde- pendent or disconnected clouds. Ezpbsives efleds.-The excessive and abrupt heating caused by the lightning Qurrent explosively and great1 expands the column of am through which it passes. even explosively vaporizps such volatile objects as it may hit that have not sufficient conductivity to carry it off. Hence, chimnep are shattered, shingles torn off, trees stripped of them bark or utterly slivered and clemolislied, kite and other wire fused or volatilized, holes melted through stee le beUs and other large pieces of metal, and a
thousand o er seemin freaks and vagaries wrought. Man of the effects o lightning appear at fimt difficult to exp 9 ain, but, exce t the physiological and, probably, some of the chemicJ all de end upon the sudden and intense heatin along its p a d . t%emicuZ &cts.-h is well known, oxides of nitngen and oven what might be termed the oxide of oxygen, or ozone, are produced along the path of an electiic spark in the laboratory. Therefore one might expect an abundant formation, during a thunderstorm, of these
same compounds. And this is esactly what does occur, as observation abundantly shows. It seems probable, too, that some aninionia niust also be formed in this way,
the hydrogen being supplied by the decomposition of dnindrops and water vapor. In the resence of water or water vapor these several compoun s undergo important chan es or combinations.
The nitrogen peroxide (most stab e of the oxides of
limiting value; and if so, t 7l en lightning flashes, as we
merges into the spark and the spark into an thunderbolt that it is not possible sharply
the discharge is from cloud to earth the -P ength of the
On the ot 31.1 er han , when the discharge is from cloud
lightning discharges may be from cloud to earti, Y from
Hence it would appear that fig E tning must occur most
P tR
f B
nitrogen) combines with water to produce both nitric and nitrous acids; the ozone with water gives hydrogen peroxide and sets free oxygen; and the ammonia in the iiiain nierelp dissolves, but probabl also to some extent foriiis caustic ammonia and hy (9 rogen. Symbolically the reactions seein to be as follows:
2N0, + H20 = mT03 + HNO,.
2NH3 + 2fi20 = 2h H30H + H2. 0, + H,O = HqOz + 0,.
The ammonia aiid also both the acids through the pro- diiotion of soluble salts are valuable fertilizers. Hence, wherever the thunderst*orm is frequent and severe, especially, therefore, within the tropics,, the chemical actions of the lightning may materially add, as ha. recently been shown (211, to the fertility of the soil and the growth of crops. Danger.-It is im ossible to say much of value about
indoors than out &iring a thunderstorm, 8s ecially if the house has R well-grpunded metallic roof. 9 f outdoors it is far better to be in a valley than on the ridge of a hill, and it. is always dangerous to take shelter under a two-the taller the tree, other things being equal, the greater the danger. Some varieties of troos appear to he more frequently struck, in proportion to their nunibers and ex)osure, than others, but no tree is immune. It seenis hat,, in general, the trees most likely to be struck are those that have either an estensive root system, like thc locust, or deep tap roots, like the ine, and this for
and therefore offer, on the whole, tho least electrical resistnnce. Finally, if one has to be outdoors and exposed to the danger of. a violent thunderstorm i t is advisable, so far
as danger from the li-htning is concerned, to get soakin met,, because wet c810tfbs are iiiuch better conductors, an dry m e a poorer, than the hunian body. In extreme cases
it niight.even be advisable to lie flat on the wet ground. As just implied, the contour of the land is an important factor in deteriiiining the relative dan er from lightning
c.h~rge, the only kind that is dangerous, varies inversely as the distance between them. Hence thunderstorms are more dangerous. in iiiountuinous regions, at least in the higher portions, than over a level countiy. For this same reason also, distance of cloud to earth, there exists on high peaks a level 3r belt of maximum danger, the level, approsimateljr of the base of the average cumulus cloud. The tops of tht; highest peaks are seldoni struck, simply because the storm generally fornis and runs its course at a lower level. Clewly. 'too, for any given sect,ion the lower the cloud the grnaler :.ho dan er. Hence a high degree of humidity is favorahlo t.o a finngerous storm, partsly because the c.louda will forin at a low level and partly because the pre- cipitation will be abundant.. Hence, t.00, a wmter thun- derstorm, because of it,s generally lower clouds, is likely to be mor0 dnngerous than an equally heavy summer one. The c m u.n.ogmph..--various mst,rument,s, based upon t,he principles of 'wireless " receivers, have been devlsed for recording the occurrence of 1ight.ning discharges. Of courso t.he sensit,iveness of the inst,runient, the dist,ance and tho mngnit,ude of the discharge all are factors that affect t,he record, but by keeping the sensitiveness con- stant,, or nearly so, it is possible with an inst,rument of
this kind to estimat.e the appr0simat.e distance, pro ess,
and t.o sonie estent even the direction of the st,orm. %ev- ertheless t,here does not appear to be much demand for
the danger Ironi lie R tning. Generally it is safer to be
the very obvious reason that they are t 1 e best grounded
8
because, obviously, the chance of a c s oud-to-earth dis-
JUNE, 1914. MONTHLY WEATmR REVIEW. 379
this information, and therefore at present the cerauno- graph is but s arin ly used.
satisfact,ory idea in regard to the cause of thunder, and it is not a rare t,hing even yet to hear such a childish e s p l s nat.ion as that it is t.he noise caused by the bumping or
Th,wnder.- $7 or a ong while no one had even a remo t.ely
cles of which he is t.he center, while ot.her portions are directed more or less rndinlly from him. This would account for, and doubt.less in a iiiensure is t.ho coi.rec,t. 8s )In-
accompany t.hunder. When, as often happens, sev- erd discharges follow each ot.her in rapid succession t,liere is every op ortunity for all sorts of irregular mut.ual inter-
sound impulses they send out. Under favorable conditions t.he echo of thunder from clouds, hills, and ot,lier reflecting objects cert.ainly is effect,ive in accentuating and prolonging the noise and rumble. But. t,he import.ance of this factor gen- erally is g r e d y overestimated, for ordinarily the rumble is sub$t.ant.iaUy t.he same whet,her over t,he ocean, on t.he prairies, or among the mountains. Distance heud-The distance to which thunder can
be heard seldom esceeds 35 kilometers, while ordinarily, perhaps, it is not heard more than half so far. To most ersons, familiar with tlie great distances to which the h n of large caniion is still perceptible, the relatively smafi distances to which thunder is audible is quite a surprise. It shoiild be remembered, however, that both the origin of the sound and often the air itself as a sound conductor are radically different in the two cases. The firing of cannon or any other surface disturbance is heard farthest when the air is still and when, through temperature inversion or otherwise, it is so stratxed as in a measure to conserve the sound eiier y between horizontal planes.
the atmosphere irregular in respect to either its tem- perature or moisture distribution, or both, for these conditions favor the production of internal sound reflections and the dissipation of energy. Now the former or -favorable conditions occasional1 obtain during the production of ordinary .noises, including the firing of cannon, but never obtam during a thunderstorm. In
nat.ion of, soinn of t,he loud booming effects or crashes t I mt
ference an t; reinforcement of t.he compression waves or
Surcession of discharges.
Reflect.ion.
Conversely, sound is hear f to the least distance when
66828-lkI.2
fact, the thunderstorm is especially likely itself to establish the second set of the above conditions, or those least favorable to the f a r carrying of sound. Then, too, when a cannon, say, is fied the noise all starts from the same place, the energy is concentrated, while in the case of thunder it IS stretched out over the entire length of the lightning path. In the h t case
the energy is confined to a single shell; in the second it is diffused throii h an estensive volume. It is these
tlie energy that cause the cannon to be heard much far- ther than tlie heaviest thtinder, even though the latter almost certainly produces much the greater total atmos- pheric disturbance. N o m 1 atmospheric electrici.ty.-The only reason for mentioning norninl atmospheric elmtricity rn connection with thunderstorms is to emphaslze the fact that, con- trary to what man suppose to be tlie case, there is but
these two phenomena. Thus while the difference in electrical potential between the surface of the earth and a point at constant elevation is roughly the same at all parts of the world, tlie nniiiber and intensity of thun- derstorms vary greatly from place to place. Further, while the potmtinl gradient at any given place is in winter the number of thunderstorms is most requent in siiiiinier, and while the gradient, in the lower layer of the atmosphere, at ninny places, usually is .greatest from S to 10 o’clock, both morning and evening, and least at 3 to 3 o’clock p. ni. and 3 to 4 o’clock a. m., no closely analooous relations hold for tlie thunderstorm. Probably tKe most interesting conclusion in regard to normal atmospheric electricity so far drawn from obser- vation and esperiment is this: That the earth every- where, land nncl water and from pole to pole, is a nega- tively charged s >here of practical1 constant surface
that it is constantly carrying away a current that amounts on the whole to about 1,000 amperes. Where the sup ly of negative electricity comes from
charged in spite of this steady great loss, or how the out- going current is compensated, no one knows. Rain does not help matters for, as we have seen, that is prevailingly ositive, whereas we need, to com ensate the loss, to grin back negative electricity a n l a great deal of it.
means of the lightning, for, in the great majority of cases,
this, too, is positive from cloud to earth. And so the
puzzle remains. As Simpson (22) puts it:
A flow of negative electricity takes place from the mfme of the whole globe into the atmosphere above it, and this neceasitates a return current of more than 1,000 amperes; et not the Eli htest indication of any mch current has 80 far been founi, and no eatisf&9.ory explanation for ita absence has been given.
Much more, of course, might be said on this subject, for i t is a big one on which many have labored, but er- haps the above is sufficient for the purpose of this %rial section, namely, to show that, contrar to opinions often
thundemtorm and normal atmospheric electricity; that, according to our best evidence, they are distinct and inde- pendent phenomena.
REFERENCES.
differences in t E e concentration and the conservation of
little relation, in t T ie sense of cause and effect, between
ptest
density, surrom d ed by nn ntmosp 9 iere so conducting
that keeps the su R ace of the earth on the whole negatively
Neit a er, so f a r as known, is compensation supplied by
held, there is no obvious and close re 9 ation between the
Memoirs, Indian met’l. dept., Simk, 1910, 20, pt. 8. Physikal. Ztachr., Lei zi , 1906, 7: 98.
Braak, Beitr. z. Phyaik d. fr. Atmoaph., Le~pzig, 1914,8: 141, Sitzber., R. preuse. Aid. d. Wiss., Berlin, 1892, 8: 279-309.
380 MONTHLY WEATHER BEVIEW. JUNE, 1914
(5) Hellnuann. Die Niederschliige in den Norddeutschen Stromge-
6 Meteor01 ische Zeitsctrift, Braunschweig, 1908, 25,. Jhrg., 10s. Ztschr. f. d. geeamte Vemcberungswlss., Berlin,
(8) Humphrey, W. J. Bull., Mt. Wea erobs'y., Washington, 1913, 33: 567.
(9) Annalen der Physik u. Chemie, Leipzig, 1899, 68: 776.
(14) Annalen d: Physik, Lei zig, 1900, 22: 623: (15) Wood. Science, New A r k (N. S.), 1899, 10: 717. (16) Astruphysical jour., Chicago, 1903, 18: 294. Ill) Pluceed'mgs, Anier. instit. elec. eng., New York, 1914, 83: 563. 18) De Blois. Proceedings, Am. instit. elec. eng., New York, 1914,
(19) De Blois. loc. cit.
(20) Flendng. Tbe principles of electric wave telegraphy and Leduc. Comptes rendus, Paris, 1899, 129: 37. telephony. 2d ed. 1910. S . p. 228-557. Euerett. Nat.ure, London, 1903, 68: 599. (21) Capus, Guillnume. Annales de &ographie, Psris, 1914, 88: 109.
Vwlb. Comptes rendus, Paris, 1901, 132: 1537.
bieten. Berlin, 1906, v. 1, 336-337, and elsewhere.
1904, 4, pt. 4. (A& Dim-Berlin.
6: 1.
Gleflm, %to
lQMd
Ciel et terre, Bruxelles, 1910, 31: 499. (22) Nature, London, 1912, 90: 411.