periodicities in the time series of annual minimum temperatures for the united states gulf coast...

PAGEOPH, Vol. 138, No. 2 ( 1 9 9 2 ) 0033-4553/'92/020323-1151.50 + 0.20/0 �9 1992 Birkh/iuser Verlag, Basel

Periodicities in the Time Series of Annual Minimum Temperatures for the United States Gulf Coast Region

R. P. KANE 1 and D. GOBBI 1

Abstract--The minimum winter temperature series for the United States Gulf Coast for 1799-1988 (190 values) was subjected to Maximum Entropy Spectral Analysis. Significant periodicities in the QBO region (T = 2-3 years) and at T = ~3.7, 45, 5.5, 6.5, 7.5, 12.9, 15.5 and 22 years were detected. Some of these were present in the first half only ( 1799-1893) whiIe others in the latter half only ( 1894-1988), indicating a transient nature. Also, more than 50% of the variance was random. Many of the significant periodicities are seen in other geophysical parameters. Some may be harmonics of the ll-year sunspot cycle and the 22-year Hale magnetic sunspot cycle.

Key words: Minimum temperatures, USA Gulf Coast.

t. Introduction

The mean winter temperatures in the southeastern United States (adjoining the

Gul f o f Mexico) have declined considerably during the past few decades (DIAZ and

QUAYLE, 1980; BALLING and IDSO, 1989). There is also an increased incidence o f

extreme low temperature events in that region, causing considerable damage to the

citrus fruit industry (PARSONS, 1987). After establishment o f the Signal Service in

1870 and the U.S. Weather Bureau in 1891, regular data were available for several

locations. Fo r the earlier period, ERHARDT (1990) examined the data f rom several

locations in Louisiana, Mississippi, A labama and extreme western Florida and

produced a single series extending back to 1799 for the min imum winter tempera-

tures for the Gul f Coast, appropr ia te for an average location o f 31~ 90~ The

190-year series is reasonably s tat ionary and homogeneous , with an approximately

normal distribution, with a mean of - 8 . 3 ~ and a s tandard deviation o f 2.9 ~ For

its temporal patterns, Erhard t reported a significant downtrend in the last 30 years,

similar to that seen for Nor th America and the Nor the rn Hemisphere in general

(DIAz and QUAYLE, 1980; JONES et al., 1986). Also, he pointed out that unusual severe winter freezes seemed to be clustered in groups and, applying Pearson Type

I Instituto Nacional de Pesquisas Espaciais, INPE C.P. 5t5, Sg.o Jos6 dos Campos-SP 12201, Brazil.

324 R.P. Kane and D. Gobbi PAGEOPH,

III distribution, he gave the probable return periods of minimum temperatures at different extreme limits (e.g., recurrence interval 20 years for -14~ but did not investigate the presence of any systematic recurrence or periodicity. In this note, we report results of a power spectrum analysis of Erhardt's Gulf series of 190 years, using MESA (Maximum Entropy Spectral Analysis).

2. Method of Analysis

MESA was introduced by BURG (1967) and critically reviewed by ULRYCH and BISHOP (1975). For detecting periodicities (even those comparable to the data length), MESA is superior to the conventional BLACKMAN and TUKEu (1958) method as shown in studies with artificial samples (e.g., CHEN and STEGAN, 1974; ULRYCH and CLAYTON, 1976; KANE, 1977, 1979). For example, for a data sample of 100 data points, periods near T = 3 are detected with an accuracy of ~1%, T = 5 with ~2%, T = 10 with ~3%, while T = 80 can have errors as large as 20%. However, the amplitude estimates in MESA are not reliable (KANE 1977; KANE and TRIVEDI, 1982). Hence KANE (1977) suggested the following prescription, viz., use MESA only for locating the possible periodicities Tk (k = 1 to n) and then use these irk in the expression:

f(t) = Ao + ~ [ak sin(21rt/Tk) + bk cos(2~t/Tk)] + E k = l

= A 0 + ~ rk sin(2rct/Tk +~k) +E (1) k = l

where f(t) is the observed time series and E is the Error factor. A Multiple Regression Analysis (MRA) (JOHNSTON, 1960; BEWNGTON, 1967) is then carried out which gives the best statistical estimates (and the standard errors) of the parameters Ao, (ak, bk) by a least-squares fit. From these, the amplitudes rk and their standard errors ar~ can be calculated. The variance explained by every rk is (r2/2). Hence the Percentage Variance Explained (PVE) by every rk is 50(r~/a 2) where a2= variance of the series f(t). This MRA with sinusoides looks like a Fourier Analysis but is not, because in Fourier Analysis, the various Tk are simple fractions (harmonics) of a fundamental period. In MRA, the various Tk may be unrelated.

3. Data

Figure la shows a plot of the yearly absolute minimum temperatures for the Gulf coast, reconstructed for 31 ~ 90~ as given in ERHARDT (1990). Some years

Vol. 138, 1992 Periodicities in Annual Minimum Temperatures 325

0

-5 -I0

-15

-6

~8

-I0

W

3

rr" I , i O_

W

-8

-9

1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 - - J T i i i i J i i - - i ~ i , , i i i ,

(G) YEARLY GULF WINTER TEMPERATURE MINIMA

I in r t rLEJ i rl n ! n mn i n n l n l n ~ r l n n [ ] n

. . . . . . . . . . . . . . . . . . . . . . . .

(b) IO-YEAR (~ MOVING -. AVERAGES .....x....~ . . . . . . . . . . xxxx~x xxx x lxxxx xx~xxx ,~ x~x

x~xx / xx �9

-_~.~ ..~'..." ....... "...~�9149 . . . . s ............. ~ NH ( L A N D + S E A l

!io, 3~ YEAR 2 MOVING (~ , L NH (LAND) _ f . . . . . . . - - AVERAGES - / . . . .

- 2 l - . . . . 5 / NH ( L A N D + S E A /

_1 k i J 1 i ~ i I i - . i i I F i ~ i i ,

1800 1820 1640 1860 1880 1900 1920 J940 1960 1980

Figure l (a) Yearly values of the minimum winter temperatures for the Gulf States (USA) for 1799-1988. Some years of extreme values are indicated. The full and hatched rectangles indicate occurrences of strong and moderate El Nifios (warm episodes on Ecuador-Peru coast) (ERHARDT, 1990). (b) 10-year moving averages of the Gulf temperature series and the Northern Hemisphere land temperature and (land + sea)

temperature series (JONES et al., 1986a). (c) 30-year moving averages, 10 years apart.

o f extreme variations (warm or cold) are indicated. Data from ~ 1880 onwards are more reliable. However, as pointed out by Erhardt, extension o f the series back from 1880 to 1799 yields mean values not appreciably different from the 1900-1950 period. For 1950-1980, the average seems to be slightly colder than the entire rest o f the record. The year-to-year fluctuations are very large ( - 1 to - 18~ Often, alternate years are almost opposite extremes, indicating a large contribution of an irregular, transient quasibiennial oscillation QBO (T = 2 - 3 years), which is strong during some intervals and weak during others.

Figure lb shows moving averages o f the Gulf series over 10 consecutive years. The short-term fluctuations (T less than 10 years) are practically eliminated and many long-term fluctuations (T exceeding 10) are revealed, with amplitudes as large as 3~ However, these also seem to be irregular, i.e., having different amplitudes in different intervals. For the Northern Hemisphere, JONES e t a l . (1986a) have given an average for land-based stations data and fixed position weather ship data. In Figure lb, the 10-year moving averages o f these data N H ( L A N D ) are also shown,


expressed as deviations from the mean of the reference period 1951-1970. These data seem to show variations considerably different, both qualitatively and quanti- tatively, from the variations of the Gulf series. For example, during 1915-1955, the Gulf values (10-year moving averages) were nearly constant while the NH (LAND) values showed a steady increase of ~0.4~ (Watch the difference in scales.) The Gulf data seem to have peculiar characteristics of their own. Using UK Meteoro- logical Office (UKMO) data, FOLLAND et al. (1984) obtained similar series for marine air temperature data. However, those data contained some problems of inhomogeneities resulting from nonclimatic factors. JON~S et al. (1986b) used the more copious Comprehensive Ocean Atmospheric Data Set (COADS), made several corrections and produced global and hemispheric annual mean variation series for 1861-1984. In Figure lb, the crosses represent their Northern Hemisphere (land + sea) series. The NH (land) and NH (land + sea) series are very similar to each other but differ considerably from the Gulf series, which, incidentally, is not an annual mean temperature series but a minimum winter temperature series and has much larger amplitudes.

Figure lc shows the 30-year moving averages, 10 years apart. Here, the differences between the Gulf series on one hand and the NH (land) and NH (land + sea) series on the other are very clearly seen. A power spectrum analysis of the latter series was reported by KANE and TEIXEIRA (1990) and a periodicity of T = ,-~55-80 years was indicated, besides a linear warming trend of ,-~0.3~ - tury. We now present the results of MESA for the Gulf series.

4. Maximum Entropy Spectral Analysis

The 190-year series (1799-1988) for the Gulf States minimum winter temperatures was subjected to MESA. Corresponding to the lag m in the Blackman and Tukey method, MESA has an adjustable variable called Length of the Prediction Error Filter (LPEF). Very low LPEF reveals only very low periodicities. Larger LPEF reveal larger periodicities but then lower periodicity peaks show peak splitting. There are no clear criteria for choosing the LPEF, though ULRu and BISHOP (1975) recommend LPEF ,-~50% of the data length. KANE (1977) suggested that spectra may be obtained for several LPEF (e.g., 25%, 33%, 50%, 75% of the data length) and low periodicities be picked up from the lower LPEF plots and larger periodicities from the larger LPEF plots.

Figure 2 shows the spectra of the 190-year Gulf series for LPEF = 63, 95, 127, 152, 180 corresponding to ~33%, 50%, 66%, 80%, 95% of the data length 190. Conventionally, the abscissa scale is frequency from 0 to 0.5 (folding frequency). However, our experience shows that on such a plot, smaller frequencies (larger periodicities) become overcrowded. Also, for geophysical studies, periodicities T (= l / f ) are more relevant than the frequency f Hence, for abscissa, we have used


2.0

0.0

-2.0 I

2.0

0.O

-2.0

n- 2.O bJ

0 0.0 n

0 - 2.0

g 2.0 g

(3.0

-20

2.0

O0

-2.0

80 5 �9 13 40 8 L PEF = ( ~ / 15 2~ ' L27 .30 4+9

21 85

^

1.5.5 21.4 @ 2[8 30 41 89

15.5 21.4 @ 7.8 12. 29 40 98 "~ 12. 29 40 ~e~ _

~Omrto 7 ai'~,..: . 3. . 48 5. 667.6 t5.5 21.4 ~eJ~-,i~N 4.5 5.8 7.6 12~ --'- ~"~

l I J LOGIo r ..4~. Ol 4 0.6 0% 1.0 L2 14 116

I I i I 1 T= --,," 3 5 I0 20 50

@ J

2.4

1 L.__ ~ 0 0 200 YEARS

Figure 2 Maximum Entropy Power Spectra of the Gulf minimum winter temperature series (1799-1988) for LPEF = 63, 95, 127, 152, 180 (33%, 50%, 66%, 80%, 95% of data length 190). The numbers indicate

periodicity peaks in years.

log~o T. In the Blackman and Tukey method, power can be studied at only certain fixed frequencies f = r/2rn, where m = the chosen lag and, r = 1, 2 , . . . , rn. In MESA, power can be calculated for any frequency. Hence, we calculated power for frequenciesfcorresponding to periodicities T for which log10 T varied from 0.300 to 2.500 in steps of 0.005, i.e., T studied were 2.00, 2.02, 2.04 . . . . 309, 312, 316 years.

In Figure 2, larger LPEF do reveal larger periodicities. For example, L P E F = 152 and 180 reveal a possible periodicity near T = ( 1 9 0 + 10) years. However, lower periodicities show peak-splitting. Hence, we picked lower periodicities from LPEF = 63 and larger periodicities from LPEF = 152 and 180. For a Multiple Regression Anatysis (MRA), we picked the foItowing 23 periodicities: T = 2.24, 2.40, 2.64, 2.76, 3.10, 3.72, 4.28, 4.80, 5.26, 5.77, 6.62, 7.60, 8.53, 9.80, 11.0, 12.6, 15.5, 21.4, 29.0, 40.0, 49.0, 85.0, 190.0. In this method, the standard error crr~ of the amplitude r k is the same for all rk and was 0.27~ The periodicities which had r k exceeding 2ark (i.e., 0.55~ were T = 2.24, 2.40, 4.28, 4.80, 6.62, 7.60, 15.5, 21.4 and 85 years. Amongst these, T = 15.5 and 21.4 years were significant at a 3a,~ level. Thus, this series has the most prominent periodicity at T = 15.5 years, which


explains ~5.5% variance. The next prominent periodicity is at T = 21.4 years, explaining ~4.6% variance. Periodicities T = 4.28, 4.80, 6.62, 7.60 and 85 years together explain ~ 13% variance while T = 2.24, 2.40 years (QBO) together explain ~ 4 % variance. There is no single periodicity explaining even 10% variance, indicating that the series is highly multispectral. Also, all the 23 periodicities together explain only 38% variance, indicating a large random component (62%).

From Figures la and b, it seems that some periodicities are transient. To examine this apsect, the 190-year data were divided into two equal portions of 95 years each (1799-1893, 1894-1988) and each portion was subjected to MESA. Seventeen periodicities were chosen in each for MRA. For the first half (1799- 1893), periodicities significant at a 2o- level were T = 2.09, 2.76, 3.7, 4.2, 4.9, 5.8, 7.8, 12.9 and 15.5 years. The most prominent periodicities were T = 15.5 (8.5% variance), T = 4.2 (6.5% variance) and T = 2.76 (6.4% variance) years. For the latter half (1894-1988), periodicities significant at a 2o level were T = 2.24, 2.58, 2.76, 3.1, 3.8, 4.8, 7.3 and 22 years. The most prominent periodicities were T = 22 (7.8% variance), T = 7.3 (7.1% variance), T = 3.1 (6.6% variance) and T = 4 . 8 (6.4% variance) years. Figure 3 shows the amplitudes for all these periodicities, Figure 3a for the first 95 values (1799-1893), Figure 3b for the latter 95 values, (1894-1988) and Figure 3c for the entire period (190 values, 1799-1988). As can be seen, QBO (T = 2-3 years), T = ~3.7, T = 5.8 are present in both the groups. T = 15.5 years is the strongest periodicity in the first half but negligible in the second half. In contrast, T = 22 years is very prominent in the second half but negligible in the first half. Thus, a transient nature for these periodicities is clearly

indicated. Figures 3d and e show similar spectra for the Northern Hemisphere annual

mean land temperature series NH (land) and ( l and+sea) temperatures NH (land + sea) as reported by KANE and TEIXEIRA (1990). Here, the amplitudes are smaller by almost a factor of 10 but some periodicities are similar to those of the Gulf series, e.g., QBO (T = 2-3 years), T = ~(4.5 +_ 0.3), T = ~5.5, T = 7.5 years. Periodicity T = 15.5 years seems to be prominent only in the Gulf series during the last century. T = --~ 22 years (the Hale magnetic sunspot cycle?) seems to be present

only in the present century.

5. Relationship with Southern Oscillation/El Ni~o

In Figure la, the rectangles immediately below the gulf series show the occurrences of E1 Nifio events. Full rectangles represent strong events and hatched rectangles represent moderate events (QU~NN et al., 1987). E1 Nifios are warm water episodes occurring off the Ecuador-Peru coast early in the year, at irregular intervals. These sea-surface temperature (SST) anomalies increase in area and magnitude during the next few months, reaching maximum values near the coast


T = 5 5 I0 20 50 I00 200 YEARS J I t [

I I ",~ YEARS (O) 1.2 .z. I~ ~ GULF ( 1 7 9 9 - 1 8 8 5 )

2 ea

0 II2 ,~. ( b )

~ ~ ~:~ GULF ( 1884 - 1988 ) ,.0 ~.J~.~ 95 VALUES

4 b

2 I I " 0 (c)

GULF ( 1799 -1988 ) I.O "~# ,~'.~' 190 VALUES

8 ~ i f ~

s 0 I ( d ) I-- .12 -- I~

- - I NH (LAND) d .10 - 1851- 1984 ::E .08 ! 154 VALUES <~ .06 ,t.r ~'~ ~,~' ~,

o2 I I b o - - I I

( e ) I0 ~'~ NH (LAND+SEA) 08 ~ 1861- 1984

124 VALUES

h 1 o ,Ill II I (f)

u3 2 ~ SO INDICES I-- ~ 1851 - 1974

0 ~ ,,.1 ~ ~dJ ~ 124 VALUES

i \\' \9 4~ 2

~ o I b <

Clot ~-0,4 06 I 0.8 liO 12 I 1.4 1.6 I 1.8 2.[0 22 I 24

- " 5 I0 20 50 I00 200 YEAR!

Figure 3 Amplitudes of different periodicities obtained by Maximum Entropy Spectral Analysis for (a) Gulf series first half (1799 1893), (b) Gulf series second half (1894 1988), (c) Gulf series total (1799-1988), (d) Northern Hemisphere land temperature NH (land), (e) Northern Hemisphere (land + sea) temperature NH ( land+sea) , (f) Southern (SO) index for 1851-1974. The hatched portion indicates the 2 sigma

limit.


around April-June. The warm water spreads westward along the equator and, by September-October, covers the entire eastern and central equatorial Pacific. E1 Nifios are strongly related to the minima of southern oscillation indices like the sea-level pressure difference between Tahiti (18~ 150~ and Darwin (12~ 131~

E1 Nifio effects are known to extend to various parts of the world. WRIGHT (1984) prepared a SST index using data for locations in the region 6~176 180~176 and showed that this index was intimately related to his SO index (WRIGHT, 1975, 1977) and both SST and SO indices were well correlated with E1 Nifio events. The relationship between E1 Nifio occurrences and rainfall in some lowqatitude regions is discussed in KANE (1989). for the Central American- Caribbean region, HASTENRATH (1976) reported that E1 Nifios were associated with droughts. For Florida, DOUGLAS and ENGELHART (1981) reported that E1 Nifios were associated with excess rainfall.

From Figure la, the Gulf winter temperature minima do not seem to be related to E1 Nifio events. In fact, some very large minima (e.g., 1807, 1835, 1852) seem to have occurred at times when no E1 Nifio events occurred. The temperature max ima

also do not seem to have any clear relationship with E1 Nifio occurrences. Table 1 shows the temperature values for strong E1 Nifio years El(0) as also for two preceding years E l ( -2 ) , E l ( - 1) and two succeeding years El( + 1), El(+2). Large as well as small values seem to be spread all over and the averages for the five columns are all very near the mean of the series, viz. -8 .3~ It seems, therefore, that E1 Nifios have no effect on the Gulf temperature extremes.

The SO index is known to have some periodicities (WmGHa', 1975, 1977; KANE, 1989). In Figure 3, the bottom plot (Figure 3f) shows the spectra for the SO index for 124 yearly values (1851-1974). The QBO (T = 2-3 years) is present in the SO series also. In addition, T = ~ 3 . 7 , T = ~ ( 4 . 5 + 0 . 3 ) , T = ~ 5 . 5 , T = ~ 6 . 5 , T = -~7.5 years seem to be common between SO series and Gulf series. The solar cycle line ( T = 11 years) is present in SO but absent in Gulf series. The T = 15.5- year peak in gulf series (last century) seems to be near the T = 14.9-year peak in SO series. The T = ~22-year peak is prominent in the SO series, the NH (land) and NH (land + sea) series and in the recent Gulf series (1894-1988). A T = ,-~ 28-year peak seems to be present in the SO and NH series but absent in the Gulf series.

6. Conclusions and Discussion

The minimum winter temperature series for the U.S. Gulf Coast States for 1799-1988 (190 values) has many significant periodicities but some of these are transient, i.e., present in the first half (1799-1893) but absent in the latter half (1894-1988) or vice versa. Significant periodicities are in the QBO region (T =-2-3 years) as also at T = ~3.7, 4.5, 5.5, 6.5, 7.5, 12.9, 15.5 and 22 years. Some of these


Table 1

Minimum winter temperature (~ for the Gulf States for strong El Nifio years El(O) indicated in column 1 and for two preceding years El(--l), E l ( -2) and two succeeding years El(4-1) and El(+2).

YEARS El ( -2) El( - 1) El(0) El( + 1) El( + 2)

1804 - 0 6 - 0 4 - 0 8 - 0 9 - 12 1814 - 0 7 - I1 - 0 4 -08 - 0 9 1828 - 0 9 - 0 7 -01 - 0 8 -03 1845 - 0 7 - 0 3 - 0 6 - 0 9 - 0 9 1864 - 0 3 - 0 4 - 0 9 - 0 8 - 0 9 t871 - 0 9 - 0 6 - 1 2 - 0 7 - I O 1877 - 0 8 - 0 4 - 0 9 - 0 5 - 1 2 1878 - 0 4 - 0 9 - 0 5 - I2 - 0 6 1884 - 0 4 - 0 7 - 1 2 - 0 8 - 1 5 1891 - 0 6 - 0 6 - 0 4 - 0 8 - 0 9 1899 --09 - 0 8 -18 --09 --04 1900 --08 - 1 8 - 0 9 --04 --11 1911 --06 --07 - 0 9 --08 --04 1912 - 0 7 - 0 9 - 0 8 - 0 4 - 0 7 1917 --06 --07 --08 - 1 2 -- 10 1918 - 0 7 - 0 8 -- 12 - 10 --09 1925 - 0 7 - 11 --06 - 0 9 --07 1926 -11 --06 - 0 9 - 0 7 - 0 9 t932 - 0 9 - 0 6 --06 - 0 9 --06 1940 - 0 7 - 0 7 --13 -07 - 0 9 1941 - 0 7 - 1 3 --07 --09 --07 1957 - 0 8 - 0 9 - 0 9 -- It --08 1958 - 0 9 - 0 9 -11 - 0 8 - 0 8 1972 -- 11 --08 - 0 8 --08 - 0 7 1973 --08 - 0 8 -08 --07 - 0 6 1982 --09 - 0 9 --16 --04 -13 1983 - 0 9 - 1 6 - 0 4 --13 - 16 Sum -201 -220 -231 -221 -235 /27= -7 .4 -8.1 -8 .6 -8 .2 -8 .7

a re seen in the S O index as a l so in the a v e r a g e l a n d a n d sea t e m p e r a t u r e s o f the

N o r t h e r n H e m i s p h e r e .

T h e o r ig in o f these per iod ic i t i e s is n o t ve ry clear . W e l l - k n o w n g e o p h y s i c a l

pe r iod ic i t i e s a re T = 11 years ( s u n s p o t cycle) and T = 22 years ( H a l e m a g n e t i c

s u n s p o t cycle). T = 3.7, 5.5 c o u l d be h a r m o n i c s o f the 11-year cycle and T = 4.4,

7.3 years c o u l d be h a r m o n i c s o f the 22-yea r cycle. T h e Q B O ( T = 2 - 3 years) seems

to be p r e sen t in m a n y p a r a m e t e r s a n d its ex is tence in 50 m b t r o p o s p h e r i c winds has

been t heo re t i c a l l y e x p l a i n e d by HOLTON a n d LINDZEN (1972). I t is t e m p t i n g to

d i smiss m a n y per iod ic i t i e s as r a n d o m . H o w e v e r , these are n o t i c e d in severa l

g e o p h y s i c a l p a r a m e t e r s , e.g., g e o m a g n e t i c field v a r i a t i o n s ( C U R R m , 1976; KANE,

1988), l eng th o f the d a y ( L . O . D . ) f l uc tua t i ons ( C U R R m , 1973) a n d s u n s p o t n u m b e r s

( K A N E and TRIVEDI, 1985) a n d c a n n o t be i g n o r e d easily.

332 R.P . Kane and D. Gobbi PAGEOPH,

The climatological implications of these results are not very obvious. The large number of peaks precludes any prediction possibilities. For USA air temperature and other climate variables, CURRIE and O'BRIEN (1990) and CURRIE (1991) indicated peaks at 1 l and 19 years which had recognized forced functions and some prediction potential and implications for U.S. agriculture and economic science. (Also see HANDLER, 1990.)

7. Acknowledgement

This work was partially supported by FNDCT, Brazil under contract FINEP- 537/CT.

REFERENCES

BALLING, R. C., Jr., and IDSO, J. B. (1989), Historical Temperature Trends in the United States and the Effect of Urban Population Growth, J. Geophys. Res. 94, 3359-3363.

BEVINGTON, P. R., Data Reduction and Error Analysis for the Physical Sciences (McGraw-Hill Book Co., New York 1969) pp. 164-176.

BLACKMAN, R. B., and TUKEY, J. W., The Measurements of Power Spectra (Dover, New York 1958) p. 190.

BURG, J. P. (1967), Maximum Entropy Spectral Analysis, Paper presented at 37th meeting, Soc. of Explor. Geophys'., Oklahoma City.

CHEN, W. Y., and STEGAN, G. R. (1974), Experiments with Maximum Entropy Power Spectra of Sinusoids, J. Geophys. Res. 79, 3019-3022.

CURRLE, R. G. (1973), The 60-year Spectral Line in Length of Days Fluctuations, S. Afr. J. Sci. 69, 180-182,

CU~RIE, R. G. (1976), Long Period Magnetic Activity, 2 to 100 years. Astrophys. Space Sci. 39, 251 - 254.

CURRIE, R G. (1991), Deterministic Signals in Tree-rings from North America, Ind. J. Climatol. 1I, 866-876.

CURRIE, R G., and O'BmEN, D. P. (1990), Deterministic Signa& in USA Preeipttation Records, Ind. J. ClimatoL 10, 795-818.

DIAZ, H F., and QUAYLE, R. G. (1980), The Climate of the United States since 1895: Spatial and Temporal Changes, Mon. Wea. Rev. 108, 249-266.

DOUGLAS, A. V., and ENGELHART, P. J. (1981), On a Statistical Relationship between Autumn Rainfall m the Central Equatorial Pacific and Subsequent Winter Precipitation in Florida, Morn Wea. Rev. 109, 2377-2382.

ERHARDT, R. D., Jr. (1990), Reconstructed Annual Minimum Temperatures for the Gulf States, 1799-1988, J. Climate 3, 678-684.

FOLLAND, C. K., PARKER, D. E., and KATES, F. E. (1984), Worldwide Marine Temperature Fluctua- tions, 1856-1981, Nature 310, 670-673.

HANDLER, P. (1990), USA Corn Yields, the El Ni~o and Agricultural Drought: 1867-1988, Ind. J. Climatol. 10, 819-828.

HASTENRATH, S. (1976), Variations in Low Latitude Circulation and Extreme Climatic Events in the Tropical Americas, J. Atmos. Sci. 33, 202-215.

HOLTON, J. R., and LINDZEN, R. S. (1972), An Updated Theory for the Quasibiennial Oscillation of Tropical Stratosphere, J. Atmos. Sci. 29, 1076-1080.

JOHNSTON, J., Econometric Methods (McGraw-Hill Book Co., New York 1960) pp. 134-135.


JONES, P. D., RAPER, S. C. B., BRADLEY, R. S., DIAZ, H. F., KELLY, P. M., and WIGLEY, T. M. L. (1986a), Northern Hemisphere SurfaCe Air Temperature Variations, I851-1984, J. Climate Appl. Meteor. 25, 161-179.

JONES, P. D., WlGLEY, T. M. L., and WmGHT, P. B. (1986b), Global Temperature Variations between 1861 and 1984, Nature 322, 430-434.

KANE, R. P. (1977), Power Spectrum Analysis of Solar Geophysical Parameters, J. Geomag. Geoetect. 29, 471-495.

KANE, R. P. (1979), Maximum Entropy Spectral Analysis of Some Artificial Samples, J. Geophys. Res., 84, 965-966.

KANE, R. P. (1986), Power Spectrum Analysis of Geomagnetic Indices, Proc. Ind. Acad. ScL (Earth Planet. Sci.) 95, 1-12.

KANE, R. P. (1989), Relationship between the Southern Oscillation/El Ni8o and Rainfall in Some Tropical and Midlatitude Regions, Proc. Ind. Acad. Sci. (Earth Planet. Sci.) 98, 223-235.

KANE, R. P., and TRIVEDI, N. B. (1982), Comparison of Maximum Entropy Spectral Analysis (MESA) and Least-squares Linear Prediction (LSLP) Methods for Some Artificial Samples, Geophysics 47, 1731 1736.

KANE, R. P., and TRIVEDI, N. B. (1985), Periodicities in Sunspot Numbers, J. Geomag. Geoelect. 37, I071 1075.

KANE, R. P., and TI~IXE~RA, N. R. (1990), Power Spectrum Analysis of the Time-series of Annual Mean Sulface Air Temperatures, Climatic Change 17, 121-130.

PARSONS, L. R., Impact of recent severe freezes on Florida citrus, In Proc. of the Syrup. on Climate Change in the Southern United States." Future Impacts and Present Policy Issues (May 28-29), New Orleans, LA (ed. MeG, M.) (University of Oklahoma Science and Public Policy Program, Norman 1987) pp. 161-168.

QUINN, W. H., WEAL, V. T., and ANTUNES DE MAYOLO, S. E. (1987), El Ni~o Occurrences over the Past Four and a Half Centuries, J. Geophys. Res. 92, 14449-14461.

ULRYCH, T. J., and BISHOP, T. N. (1975), Maximum Entropy Spectral Analysis and Autoregressive Decomposition, Rev Geophys. 13, 183-200.

ULRYCH, T. J., and CLAYTON, R. (1976), Time Series Modeling and Maximum Entropy, Phys. Earth Planet. Inter. 12, 188-200.

WRIGHT, P. B., An Index of the Southern Oscillation (Climatic Research Unit CRU RP4, Univ. East Anglia, Norwich, England 1975) 22 pp.

WRIGHT, P. B., Thc Southern Oscillation--Patterns and Mechanisms of the Teleconnections and the Persistence (Hawaii Inst. of Geophys. Univ. of Hawaii (HIG-77-13), Honolulu 96844, 1977) 107 pp.

WRIGHT, P. B. (1984), Relationships between Indices of the Southern Oscillation, Man. Wea. Rev. 112, 1913 1919.

(Received January 24, 1992, revised April I, 1992, accepted April 6, I992)

periodicities in the time series of annual minimum temperatures for the united states gulf coast...

Documents