MONTHLY WEATHER REVIEW OCTOBER 1934 ment, which, afterward, was found to have an excellent vacuum and to read apparently as a t first; this method of removing air is rarely successful. No. 1019, the errors of which apparently have remained unchanged since it was purchased, appears to be the best of t,he four barometers and is considered the standarcl of the observatory. Director Patterson states that t,he Newman standard, imported in 1839 and compared wit8h the standard of the Royal Society at, Somerset House, originally had a cor- rection of -0.004 inc.h. Whether the British standard has been changed from that of the. Royal Society is not known, but, in any case, the corwction, now $0.003 inc.h, has not changed more than 0.007 inch in 93 years. The author acknowledges indebtedness to Mr. Arthur Rotch, of Boston, who kindly contributed the expenses of travel incidental to t'his work, and to the Instrument Division of the Weather Bureau, and to Director Patter- son of the Canadian Meteorologicd Office, for providing necessary facilities and assistance. At the time the coniparisons described herein were completed (early in 1933) no normal or absolute standarcl was in use in America. Since t8hen two iniportitnt, advances in precision barometry have occurred, of which one is the new primary standard of the National Physicd Laboratory, England, and the other the const,ruction o€ an instrumont at the United State.s Bureau of Standards. News of the latter will be welcomed by American mete- orologists who have hoped for the ma.intenance. of an absolute standard of pressure in this country. The new British standard is described in a p p e r , A New Primary Standard Barometer, by J. E. Sears and J. S. Clark, in the Proceedings of the Royal Society, A, vol. 139, 1933 of which the following summary is quoted: The paper cont,ains an account of a new primary standard barometer recently installed at the Nnt,ional Physical Laboratory t o serve as a basis of reference for all measurements relating t,o barometric pressures. The body of the instrument is constructed in stainless steel, wit.11 optically flat parallel glass windows through which the mercury surfaces are observed. These windows can be removed, if neces- sary, for cleaning, and the vacuum can be restored by means of suitable punips whenever the instrument is required for use. The average temperature of the mercury column is ascertained by means of a mercurial thermometer with a bulb 30 inches long, immersed in a hole bored in the stainless-steel body parallel to the barometric column. Two micrometer microscopes are fiT.ecl, one above the other, t o a massive vertical column which can be translated laterally so as t o view either the mercury surfaces, or the di\ ihions of a btandard invar scale set up at the side of the barometer hody. The height of each mercury surface is taken t o be the mean of two microscope readings, one of the direct image of a horizontal cross wire projected into the space above the mercury, and one of the reflection of this image in the mercury. The desiqn and general accuracy of workmanship are such t h a t individual readings should be correct to the order of 0.001 mm. In practice i t is found t h a t the mean residual error of a single observation is of the order of 0.005 min., this being probably attri- butable in the main t o minute fluctuations of barometric pressure which are continually taking place, even when atmospheric con- ditions are reasonably steady. The new instruments of the Bureau of Standards are described briefly in the following estract from a letter recently received from the Director of the Bureau: I n testing mercurial barometers me are using a Fuess barometer as the standard which had been modified so as t o have an all- glass cistern. Special methods of filling tlie tube have been developed in which the mercury ib distilled into the tube while under a high vacuum. We have found that tlie vacuum above the mercury column when so filled holds for a, number of years. This has been checked by intercomparison between 4 Fuess baro- meters, 3 of which are of the modified type. This intercomparison has usually taken place immediately after the refilling of one of the in3trurnents. In order t o eliminate possible errors due t o the low of vacuum above the mercury column s e have recently constructed a mer- curial barometer in which the vacuum space above the mercury column is connected to a mercury vapor pump and t o a McLeod gage. It is thus possible t o both control and measure the degree of vacuum. The scale and vernier from one of the Fuess barom- eters is used on this instrument which by comparison with the standard meter bar can be relied on t o about 0.01 mm. We esti- mate the over-all accuracy which will be secured with this new arrangement t o be f 0.05 mm. of mercury. This accuracy is better than that of any portable barometer which is likely t o be submitted for test. METEOROLOGICAL CONDITIONS PRECEDING THUNDERSTORMS ON THE NATIONAL FORESTS 1. WESTERN AND CENTRAL OREGON By W. R . STEVENS [Weather Bureau, Wnshington, D. C ., November 19341 The grea.t n1enac.e of thunderstorms to forested areas in all the western fire-wen.t,her districts of the Weather Bureau has been emphssize,cl many times by ad1 offizials connectecl with the fire-weathe,r work n.nd by forest- protection agencies. In t,he region uncler consideration in this paper (western and c,entrad Oregon), 1ight)ning causes more than one-half of all the forest fires. It is not apparent to the student of the daily weather map that lightning should be suc.h a great fire-causing factor in tlus region, because few thunderstorms are observed a t first-order Wea.ther Bureau stations in t,his area. However, a t higher elevations in the national forests of western and central Oregon, thundershornis are frequent, and occasionally so widespread, with so many cloud to ground flashes, that forest-protection a.gencies are not able to cope with the situation unless they are warned a t least a few hours in advance. It is the purpose of this paper to discuss thunderstorms, particularly in relation to the national forests of western and central Oregon, and to present an analysis of the meteorological conditions that ordinarily precede their occurrence in that region. CAUSES OF THUNDERSTORMS The thunderstorni is the result of vigorous vertical convection of humid air. The thunder and lightning which attend the storm play no part in its mechanism. When vertical convection of air occurs, the air is said to be unstable. Instability may be brought about by strong surface heating; by overrunning of one layer of air by another a t a considerably lower temperature; by underrunning and uplift of a saturated layer of air by ti denser layer; and by forced ascent of humid air masses up mountain slopes. There are two classes of thunclerstorms, (1) the heat thunderstorm, and (2) the cyclonic thunderstorm. This classification is based upon the cause of the instability which produces the storms. There are other classifica- tions of thunderstorms, but this one suits best for the present discussion. Conditions are favorable for the genesis of heat thun- derstorms when the pressure is nearly uniform and slightly below normal over a wide area. When this situation prevails, the winds are light and the surface air becomes OCTOBER 1934 MONTHLY WEATHER REVIEW 367 strongly heated, resulting in vigorous vertical convection currents and cumulo-nimbus clouds, provided the decrease of temperature with altitude (lapse rate) exceeds the dry adiabatic rate of 1’ C. per 100 meters, and sufficient water vapor is present to produce raindrops in the rising air. Genesis of heat thunderstorms is favored by drafts up the sides of mountain ranges, mountain peaks, and valleys. Storms of this type are very likely to form after 2 or 3 days of unusuully warm weather, when the lower air has become so heated that convection extends to high nl titudes. Cyclonic thunderstorms may occur in the southeast, quadrant of a cyclone, in which case the high lapse rate necessary for rapid convection results from the different directions of the lower and upper air currents. The sur- face air in the southeast quadrant flows from warmer regions, while the currents aloft, which flow more nearly from the west, are often sficiently colder to induce the convection necessary to the production of thunderstorms. Cyclonic thunderstorms also occur along the “cold front” of a cyclone. Warm tropical winds are associated with the eastern portion of a cyclone, while cold polar winds prevail over the western portion. The boundary between these two systems of winds is usually well marked, and ir known as a “cold front.” The cold air advances in the foriii of n wedge. Fiiction a t the surface of the earth retards the advance of the cold air, while the cold air aloft udvmces unirnpeded. This results in a wedge with its poirit some distance above the ground. Beneath the overhaiiging front of cold air, warm air is eii- trapped, which results in strong convectioii, either through the overrunning cold air or out in front of it, ancl the squally winds that are always associated with the passage of a cold front. That thunderstorms do not occur a t times when the pressure distribution appears favorable for their inception is due to certain factors that are not apparent from sur- face observations. For instance, the lapse rate may not be high enough to permit strong convection currents, or there may not be sufficient water vapor present to cause condensation within the limits of the vertical curren ts that do form. I n this study, moisture conditions a t the surface, as in- dicated by the vapor pressure, are used in the discussion of thunderstorin forecasting. It is true that tempera- ture and moisture conditions aloft are as important, or more so, as those a t the surface in producing thunder- storms, but observations of these are not available for the region under consideration. PRECIPITATION I N THUNDERSTORMS Rain does not fall con huously during a thunderstorm, (in fact, none may fall), but generally in very heavy showers. Everyone is familiar with the fact that rain after almost ceasing may begin again with great violence after a lightning flash. Condensation occurs in the ascending current as soon as the temperature of the rising air is reduced to the dew- point. The raindrops do not fall immediately, but are carried upward. Srnall raindrops fall very slowly through still air, and can be carried upward by a relatively slow ascending current. Lenard has shown that raindrops cannot fall through air of normal density whose upward velocity is greater than 8 meters per second, nor fall themselves with greater velocity through still air. When raindrops form larger than 5 or 6 millimeters in diameter, 1 Lenard. P., Met. Zeit., 21; 248, 1904. they are unstable, a.nd imnrediat>ely break up into smaller drops. The ascending current. in a thunderstorin is not steady, so that the raindrops intermittently rise a.nd fa.11, alter- nately breaking up into srnaller drops and coalescing into larger ones. The drops whic.h get to the edge of the ascending current, or rea.ch the top of the current and spread out, horizontally, fall to the ground, and produce the heavy rain during t8he early part of the storm. The occasional occurrence of hail in thunderstorms is definite proof that axending currents frequently are very violent, and estend t,o high altitudes. The rain- drops a.re carried upwn.rd into the region of freezing temperature, where they congeal and obtain a coating of snow. After a time, during a lull in their diameters, while the sup- porting force of the current vtwies approsiniately as the square of bhe dimiet,e,rs of the hailst,ones, n limiting size is quickly reached. states t t h t . esperiment~ shows that the vertical velocity iiecessrtry to sustain a hailstone 1 inch in diameter is at least 59 miles per hour, rind 116 miles per hour if the st,one is 3 inches in diameter. Formation of raindrops is essential t,o the occurrence of lightning. However, it is a common occurrence in semiarid regions, and less frequently in other sections, to see lightning but no rain reaching the ground. The reason for this phenomenon is that t.he lower air is so dry that the drops are evaporated before t.hey reach the earth’s surface. Huniphreys INSTABILITY I N NATIONAL FORESTS OF WESTERN AND CENTRAL OREGON The convect)ive processes which produce the majority of thunderstornis in the national f0rest.s of western and central Oregon are induced by strong surface heating in c.onnection wit’h a trough of low pres.sure which extends in a genernl north-south direckion from Briton.” A summary of Mr. Morris’ paper immediately follows this art)icle on page 370. TYPES OF THUNDERSTORM DAYS Three types of thunderstorm days have been defined by Mr. Morris. The “10cal’~ st8orm clay is one on which one or a few storms occur that affect only a small area 3 Humphrey& W. J., Physic3 of the Air, second edition, p. 346. 368 MONTHLY WEATHER REVIEW OCTOBER 1934 of the region. The (‘general1’ type either has many small storms which affect two-thirds of the area, or hfts one or more storms which make a c.ont,inuous trac.k at’ least two-t,hirds the length of the region. The “int,er- mediate” storm day is one on whkh the st,orms are more widespread than on the (‘local” day, but less extensive t8han on the “general” day. I n this paper, the terms ‘(int8crmediate” and (‘gen- eral ” are replaced by “ scat,tered ” and “ wide-spread ”, respectively, because they correspond better with fore- cast terminology. A “local” storm day caiise,s only one, forest fire, on the Rverage. Fires started on t,his type of day are, as R rule, quickly controlled and suppressed wit,liout, anv extra preparat,ions being niade. For this reason, when it is obvious from t,he weat,her map or relnt,ionships given in this paper that if any try officials t,liat meteorological conditions are not menacing, but that a few storms are likely to develop. Modifying terms such as “scattered” and “wide-spread” would indicate that st>orms of a more dangerous type, and to a degree defined by the respective ternis, are, likely to develop. SEASON OF LIGHTNING FIRE HAZARD The sea.son of lightning fire hazard begins in w-estern and centntra.1 Oregon in June and ends in September, n s a rule. The following table shows the number of thunde,rst,orm days on the. national forests of this area from June bo September, inclusive, classified with respect to type, for the 7 years studied. 1 Local 1 Spattered 1 Wide-spread I Total June ..-.-......--.-.-.-...---.-....---- 38 Julv ________....____....----.--....--. Au~ust September _.___..._.__......_..--..... 50 lili .- The season included in this study is the months of J d y and August, as the above figures indicate that the thun- derstorm situation is most acute during this period. RELATIONSHIPS A number of relationships between various meteoro- logical element,s and the occurrence of thunderstorms will be presented in t>lGs paper and in succee.ding articles trea t’- ing of other nntional forest,s. It, is emphasized thnt t,hese re,lationships are intended as adjuncts to t,he daily went,her maps. It is not presumed that bett,e,r forecasts can he, made, from t,hese relationships alone than t,hose tha.t, nn esperience.d forecnster c,an make who is fa.milia,r w-i t,h meteorologic.al conditions that ordinarily produce. t,hun- derstorms in t,he region. However, it is believed that they are valuable aids in thunderstorm forecasting, because they indicate when conditions are relatively safe, that is, if thunderstorms develop a t all the probability is great t,hat there will be only a sufficient number to pro- duce a “locnl” storm day; a.nd when conditions are likely to be dangerous. It must be borne in mind that only A 7-year period is included in this study, and therefore the conclusions reac.hed necessarily are provisional. Use of barometric pressure reduc.ed to sea level ns one of the factors is obvious, as it has been known for long time thnt there is a relation between pressure distribution and thunderst,orm activity. Pressure difTerenc.es between various stations are used as an indieatmion of torms were more frequent and more dmgerous. On all ‘(s~nttered ” and ‘I wide-spread ” storm days thnt occurred with pressure a.bove 30.05 inches a t Seattle, the morning pre,ssure a t Roseburg was below 30.10 inc.lies during July, and below 30.12 inches in August. No “ wide-spread ” storm days occurred during July when the Seattle pressure was below 30.05 inches and t,he Roseburg pressure higher than 29.98 inches. Three “ scatt.ered ” storm days occurred under t’hese circum- stances out of a total of 44 days. Here again we find a very sharp line of demarcation. Whenever the pressure in the morning is below 30.05 inches at Seattle in July and lower than 30 inches a t Roseburg, the day must, be regarded as potentially dangerous, especially if other nieteorological conditions indicate that thunderst,ornis are probable (fig. 1). When the Seattle pressure is below 30.05 inches and the Rosebur pressure higher than 30.02 inches in “scattered ” or “wide-spread” storm day occurred under these conditions out of a total of 33 days. Another important factor in thunderstorm activity which is apparent in August, but shows up to only a small extent in July is the 24-hour pressure change at Portland. Any day in August that the pressure is below 30.05 inches at Seattle, below 30.04 inches at Roseburg, and the pressure at Portland has fallen 0.04 inch or more in 24 hours must be regarded as dangerous (fig. 3). During the July months, it was found that when the pressure is below 30.05 inches at Seattle, the vapor pres- sure ahove 0.35 inch a t Roseburg, and in addition the pres- sure a t Roseburg is not more than 0.08 inch higher than at Baker, a dangerous situation exjsts. Under these circum- stances many thunderstorm days of the “sctittered ” and (‘wide-spread” types occur. However, any day that the Seattle pressure is below 30.05 inches and does not meet the other requirements is relatively safe (fig. 4). Out of n totnl of 50 such cases, there wns only 1 L‘~vide-spread’l and 1 “scnttered” storm day. When the pressure is above 30.05 inches at Seattle, the vapor pressure at Rose- burg and the pressure difference between Roseburg and Baker seem to have little significance. During August, however, all the “wide-spread” and “scattered ” storm August, the lf ay is relatively safe (fig. 2). Not a single OCTOEHB 1934 MONTHLY WEATHER REVIEW 369 0.04 ' 0 -0.04 -0.08 -0.1 2 -DANGEROUS I ROSEBURG PRESSURE SEATTLE PRESSURE BELOW 30.04 40'lb08 -0.04 0 004 0.08 0.12 0.16 SEATTLE PRESSURE MINUS Rg. 7. KAMLOOPS PRESSURE &. 2. I SEAlTLE PRESSURE ABOVE 30.02 0.12 v) 3 z z 0.08 W % con E 5 0 ==I 5 E 0.04 W W w w v) m g 3-0.04 w v) *%08 -0.04 0 0.04 0.08 0.12 JULY SEATTLE PRESSURE MINUS Rg. 9. KAMLOOPS PRESSURE SEATTLE PRESSURE ABOVE 30.05 I n SEATTLE PRESSURE BELOW 30.05 SEATTLE PRESSURE BELOW 30.04 I SEATTLE PRESSURE ABOVE 30.02 v) a 5g W v ) =W v) =a con z g L s w w dco Y v) sg AWUST SEATTLE PRESSURE MINUS AUGUST SEAlTLE PRESSURE MINUS JULY SEATTLE PRESSURE MINUS Rg. IO. KAMLOOPS PRESSURE Rg. 11. KAMLOOPS PRESSURE Rg. 12. ROSEBURG PRESSURE 7 SEATTLE PRESSURE BELOW 30.05 I tn 3 E =W n a W E ;s P W JULY ROSEBURG VAPOR PRESSURE AUGUST ROSEBURG VAPOR PRESSURE AUGUST ROSEBURG VAPOR PRESSURE e. 4. I m.5. I &. 6. 0.08 0.04 0 -0.08 0 0.04 0.08 0.12 0.16 0.20 0.24 370 MONTHLY WEATHER REVIEW OCTOBEII 1934 days that occurred when the pressure was above 30.05 inches at Seattle came when the vapor pressure was above 0.34 inch at Roseburg, ancl the Roseburg pressure was from 0.02 to 0.10 inch lower than a t Baker (fig. 5). I n fact, when such a situation prevails, the probability is strongly in favor of dangerous storms over the area under consideration. During the same month, when the pres- sure is below 30.05 inches at Seattle, vapor pressure above 0.35 inch at Roseburg, and the pressure at Roseburg not more than 0.12 inch higher than Baker, the lightning fire hazard is great. There was a total of 53 cases during the August months when the pressure was below 30.05 inches at Seattle, but did not meet the other requirements, and only four “scattered” and no “wide-spread” storm days occurred (fig. 6). METEOROLOGICAL CONDITIONS IN MORNING AND THUNDER- STORM ACTIVITY NEXT DAY Studies of the same character as the foregoing were made to find relationships between meteorological elements observed at 8 a. m. E. S. T. and the occurrence of thun- derstorms the following day. The relationships described hereunder for July are intended for consideration in connection with the weather maps from June 30 to July 31, inclusive. It was found that dumg July when the pressure is higher than 30.02 inches a t Seattle and the pressure is higher at Roseburg than at Baker conditions are relatively safe (fig. 7). Out of a total of 98 observations no “wide- spread” and only two “scattered” storm days occurred under these conditions. There is no well defined correla- tion between these factors and the occurrence of thunder- storms during August. During July conditions are relatively dangerous (thun- derstorms occurring in about 50 percent of the cases) when the morning pressure at Seattle is below 30.04 inches and not more than 0.12 inch higher than a t Kam- loops, and in addition the pressure at Seattle is higher than a t Roseburg or not more than 0.04 inch lower. On the other hand relatively safe conditions prevail when the pressure at Seattle is below 30.04 inches and the other observations do not come within the above classification. Out of a t,ot,al of 41 such cases, there were 2 “wide-spread” and 1 “scatt,ered” storm day (fig. 8). During the same month, thunderstorms occurred in about 50 percent of the cases when the pressure a t Seattle was above 30.02 inches and higher than a t Roseburg, and in addition the Seattle pressure was not more than 0.08 inch higher than a t Kamloops. Only 1 “wide-sprend” and 3 “scattered” storm days occurred out of the 119 days when the pressure at Seattle was above 30.02 inches, and the other observations did not come within the above classification (fig. 9). During August conditions are relatively safe when the pressure a t Seattle is below 30.04 inches and more than 0.14 inch higher than a t Kamloops, or if the Seattle pressure is either higher or not more than 0.04 inch lower than a t Roseburg (fig. 10). Forty-five cases came within this classification and only one “ wide-spread ” and no “scattered” storm days occurred. With pressure higher than 30.02 inches a t Seattle and more than 0.20 inch higher than a t Kamloops, conditions are relatively safe (fig. 11). One “wide-spread” and one “scattered” storm day occurred under these circumstances out of a total of 35 observations. There were no “wide-spread” or “scattered” storm days during July out of 57 cases with pressure at Seattle above 30.02 inches and the pressure a t Seattle the same or lower than at Roseburg (fig. 12). Under these condi- tions during August, 1 “wide-spread” and 3 “scattered” storm days occurred out of 47 cases. CONCLUSION It is regretted that the record available for study is short and the conclusions, as mentioned above, must not be considered as final. However, it is believed that there are sufficient data to justify development of work- ing hypotheses a t the present time. LIGHTNING STORMS AND FIRES ON THE NATIONAL FORESTS OF OREGON AND WASHINGTON By WILLIAM G. MORRIS [Paciflc Northwest Forest Experiment Station, Portland, Oreg. Summarized by W. R. Stevens, Weather Bureau, Washington] Lightning causes more than one-half of all the fires on the national forests of Oregon and Washington, where an average of 750 fires annually is att,ributed to this one cause. These lightning-caused fires cost hundreds of thousands of dollars to extinguish; they destroy an enor- mous amount of timber, imperil entire wnt,ersheds by destroying the cover at the headwaters of important streams and wreak hea\y damnge in recreat,ional arens of these two States. Unlike man-caused fires, which are potentially pre- ventable, lightning fires can never be prevented. For lightning fires, the forest protectionist has recourse only to prompt detection and suppression. A single storm may start so many fires that the prot.ective forces are strained to the ut.most to reach and extinguish every fire before any attains devastating size. Since, on most nntional forests of the region, many of the lightning fires are at considerable distances from the areas of everyday man- caused risk, special steps must be taken to protect the lightning fire zone whenever lightning storms are expected. The study here reported on was made (1) to discover the fundamental characteristics of lightning storms andthe fires they start so as to assist in planning the best possible systems of lightning fire control, and (2) to supply some of the basic information needed for effective lightning storm forecasting. BASIC DATA This study is based on more than 6,000 systematic reports describing lightning storms seen by United Stat,es Forest, Service fire lookouts in Oregon and Washington during the summer months froin 1925 to 1931, inclusive. During this period an average of about 200 lookouts have submitted reports each year. Each report shows the following three points concerning the location nnd move- ment of nn individual lightning storm: (1) Location of the storm and the time when it wits first seen by the look- out; (2 ) location of the storm (nnd in many cases the time) when it was nearest the lookout; (3) location and time when the storm was last seen by the lookout. The territ,ory for which these reports were made includes all of the Cnscade Range from southern Oregon to the British Columbia boundary, the Coast Range in western