Reprinted from Aeronautics and Astronautics, Vol. 13, No.4, pp. 10-19, April 1975, with the permission of the publisher, the American Institute for Aeronautics and Astronautics, and of the author. Sun-Weather Relationships ByJ. W. KING Appleton Laboratory, Slough, England As they profoundly influence civilization, and have not been well explored, Sun-weather relationships should become a major field of research in the decade being ushered in by GARP and the IMS Many people have suggested in the past latitudes a negative correlation existed oscillations associated with the ll-year that the weather is influenced by the before about 1915 A.D. and a positive sunspot cycle exhibit opposite phases at 11- and 22-year sunspot cycles. I believe one after that. Bowen (1974) drew 17 deg Sand 43 deg S (see F-2). that the accumulated evidence is so attention to the fact that in the Moreover, the sunspot cycle appears to compelling that it is no longer possible southern hemisphere the rainfall influence the rainfall in opposite ways to deny the existence of strong con- at 55 deg Nand 35 deg N (King, 1973). nections between the weather and The solar-cycle-induced rainfall radiation changes (electromagnetic oscillations referred to range from and/or corpuscular) associated with 'I about 3 to some 50% of the normal whole range of solar phenomena. Eve" annual total. Obviously, a reduction of the most skeptical scientist who in- rainfall by 25% in each of several years vestigates the literature thoroughly will around one of the extremes of the be forced to concede that important sunspot cycle is of considerable im- aspects of lower-atmosphere behavior portance. .Anillyses comb in in data are associated with solar phenomena ~ zones exhibiting opposite solar- ranging from short-lived events such as cycle effects will inevitably lead to the solar flares, through 27-day solar erroneous conclusion that no solar- rotations to the l I-year, 22-year, and cycle effect exists, as will an ana ySIS of even longer solar cycles. The com- rainfall data from regions situated plicated pattern of sun-weather between zones in which the sunspot- relationships undoubtedly needs much cycle effect is opposite. further clarification, but progress in The annual rainfall totals at For- this field will be hindered if the view taleza, Brazil, and at three sites in prevails that such relationships should South Africa (F-3 and F-4) were not be taken seriously simply because positively correlated with the "double" the mechanisms involved in explaining JOE W. KING received a Ph.D. at sunspot cycle for considerable periods Cambridge University, England, in them are not yet identified. 1955 and was awarded the of time. The modulation associated What must be sought now are the Hamilton Prize by the University with the double cycle amounted to mechanisms responsible for the for his dissertation. He is an about 35% of the average annual total complicated sun-weather relationships "individual merit senior principal at Fortaleza and to about 25% of the scientific officer" at the Appleton that exist and greater insight into their Laboratory. At Appleton he average rainfall at the South African practical consequences for agriculture, received the 1966 Wolfe Award stations. Data from Fortaleza are energy consumption, and national from the U.K. Ministry of Technol- available from 1865 onwards. After the economies. ogy for his upper-atmospheric first 60 years the relationship between research. For four years Dr. King chaired the European Space the rainfall and the double cycle at Weather and the Solar Cycle Research Organisation's ion group Fortaleza changed phase. The SO years F-l summarizes a rainfall analysis by and is now vice-president of the of data available from South Africa ICSU special committee on solar- show a consistent positive correlation Xanthakis (1973). He observed that at terrestrial physics and chairman- high northern latitudes (70-80 deg) the elect of the URSI commission on with the double cycle. The African ll-year solar cycle was positively the ionosphere. He is a member of rainfall data for latitudes south of those correlated with a lO-cm oscillation in three British national committees where the double-sunspot-cycle in- the annual rainfall total; at latitudes (for radio science, geomagnetism fluence is observed exhibit a and aeronomy and solar-terrestrial between 60 and 70 deg N the opposite physics). pronounced oscillation having a period behavior occurred while at still lower of about 10 years (Tyson et al, 1974) in 109 anti-phase with the sunspot cycle (King 25070"· BO° N 157"W-srE et al, 1974). U)200 12 STATI F-5 gives particularly impressive ~ evidence of an association between :f 150 rainfall and the double sunspot cycle. It !z 100 <l shows the date within each year from X 1844 to 1944 by which one quarter of n:: ~~;X?;~E the year's rainfall at Adelaide (35 deg S, t 300 22S ATIONS 139 deg E) had fallen. This "quartile <l, 250 date" (Cornish. 1936 and 1954) E, ~ !200 NV ~ oscillates between about April 15 and ~ OO~ May 25 in phase with the double ~ I~ z sunspot cycle (Whipple. 1936). The ~ ~~ correlation coefficient. +0.79. between ~ ~O 0 g the 101 dates calculated by Cornish and ~ ~,r~70~Ww~_I;~;-'-------------------------------'~----------~ ~ the corresponding normalized sunspot ~ 500 TATIO 0 numbers plotted in Fig. 5 is highly Z W significant (>99. 9%). Cornish showed ~ 450 ~ ~ 100~ that the dates of the other rainfall ~ 400 50 ~ "octiles" exhibited similar oscillations. =w 350 Z Oscillations of up to six weeks in the Z 0 i'5300 date by which various fractions of the :;;;: annual rainfall total have fallen ob- o 250 viously affect agriculture. w ~ 200 As another striking illustration of the 8 --- SUNSPOTNUMBER influence ofthe double sunspot cycle on :;;;150: --RAINFALL U) the weather. the July temperature in central England during the period 1098S80~--~~--~19~OO~--~B~10~--~19~2~0~--~19~30~--~19~4~0~~19~50~--~I96'O 175O-18SO exhibited an oscillation of YEAR AD nearly 1 C in phase with the double F:1 Smoothed mean "excess" annual rainfall in three northern-hemisphere sunspot cycle (F-6. King et al, 1974). zones (after Xanthakis. 1973) compared with the corresponding "normalized" The temperature curve is somewhat annual sunspot numbers. The normalized sunspot number for a particular year is anomalous during the years 1840-1855 given by 100 (Ny-NN)/(NX-NN) where Ny is the annual mean sunspot number for the year concerned and NX and N N are respectively the annual mean sunspot when the temperature extremes oc- numbers for the maximum and minimum years between which the particular year curred around sunspot minima instead falls. It should be noted that data from the American, European and Asian sectors of near sunspot maxima. have been analysed, but that, in the 500-600N zone, data from only the American The double sunspot cycle appears to sector are shown; the rainfall in the European and Asian sectors of this zone behaved differently. influence the climate of the United States in several ways. Thompson (1973) and Roberts (1974) have shown that droughts in various parts of the (j) 2r----,----.----.----~----.----.----.----.----, country occur around every second W I sunspot minimum. Newman (1965) has <..) shown that winter temperatures in Z Boston exhibit a 22-year periodicity. ..J ..J Thompson (1973) reported a ~ "remarkable" correlation between the ;?; o July/August temperature in the corn « belt ofthe United States and the double 0:: ..J sunspot cycle. Striking evidence by « -I Mather (1974) relating the double cycle :2-: 0:: and temperatures in the U.S. will be o described toward the end of this article. z Possible relationships between the :2-: weather and the "SO-year sunspot ~ cycle" have been discussed by Willett LL o W (1 %1). <..) A sunspot-cycle-induced meteoro- Z W logical variation may suddenly reverse 0:: -I W phase (e.g .• see Fig. 1). Two classical LL LL HOBART(43°S. 14JOE) examples of such phase changes should + + + + + + ~~~t~~T be mentioned: o -2~---L----~--~----~--~L----L----~--~----~ 1. In his review of the relationships 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 between solar and meteorological phe- YEAR AD nomena. Brooks (1951) observed: "The F-2 Difference between the actual and normal annual rainfalls at two latitudes one fact which appears to be firmly in Australia, The curves were obtained by applying an 8-15-year fi Iter to the established is that over the world as a annual differences, but it is significant that the extremes of the resulting whole. and especially in the tropical oscillations occur very close to sunspot minima and maxima. After Bowen (1974). 110 F-3 CORRELATION OF RAINFALL AND "DOUBLE" SUNSPOT CYCLE A. Smoothed annual rainfall totals at Fortaleza in South America derived from data pub Iished by Markham (1974). B. Annual sunspot numbers plotted in the conventional "double sunspot cycle" form; the period shown includes six sunspot minima. 200 A FORTALEZA (4°S.39°W) E <J 175 -.J -.J Lt z <! 150 0::: DUNDEE -.J (28·ZOS.30·3°E) <l: ::::J z z 125 <l: 0::: 100 w B CD :2 ::::J Z 0::: 100 I- 0 w o0.. CD :2 (f) ::::J Z Z ::::J (f)-IOO ~ (f) tSUNSPOTi Z I MINIMA ::::J r i r (f) -100 -2001~9~10~~B~20~~19~3~0~~19~4~0--~19~5~0---1~9L60--~1970 YEAR AD. r r I i F-4 Smoothed annual rainfall totals at three places in -2°98~6~5~~1~87=5~~18~8~5--~18~9~5---=19~0~5--~B~15~--~19South25 Africa (after Tyson, 1974) compared with the double sunspot cycle; the period studied includes six sunspot YEAR AD minima.