Article

 

Evidences for a shift in barometric pressure, air temperature and rainfall patterns circa 1920, and its possible relation to solar activity

 

Evidencias de un cambio en los patrones de la presión atmosférica, la temperatura del aire y la precipitación pluvial alrededor de 1920 y su posible relación con la actividad solar

 

Norma Sánchez Santillán, Alfonso Esquivel Herrera y Rubén Sánchez-Trejo

 

Depto. El Hombre y su Ambiente. UAM-Xochimilco. e-mail: santilla@cueyatl.uam.mx

 

Recibido: 19 de julio de 2001
Aceptado: 1 de abril de 2002

 

Abstract

A survey of 300 meteorological stations worldwide located showed that nearly half of the stations presented a pattern change in the correlations between the sunspot number and barometric pressure, air temperature and precipitation, around 1920. This change was mainly from a negative correlation to a positive correlation. Furthermore, we also found strong correlations between sunspots and pressure and precipitation in both tropical and extratropical latitudes; for the temperature the strong correlations appear mainly in extratropical zones.

Keywords: Climatic reversal, climatic variability, sunspots.

 

Resumen

Se analizan los registros de 300 estaciones meteorológicas distribuidas mundialmente, encontrando que en la mitad de éstas se presentaba un cambio en el patrón de las correlaciones entre el número de manchas solares y la presión barométrica, la temperatura y la precipitación pluvial, que se reportó alrededor de 1920. Este cambio consistió principalmente en un cambio en el signo de la correlación, de negativo a positivo. Además, encontramos fuertes correlaciones entre el número de manchas solares y la presión barométrica y la precipitación pluvial, tanto en latitudes tropicales como extratropicales; las correlaciones fuertes con la temperatura sólo se presentan en las zonas extratropicales.

Palabras clave: Rompimiento climático, variabilidad climática, manchas solares.

 

Introduction

Recently, Mendoza and Ramírez (1999) showed that during times of low solar activity sunspots dominate over bright features within the time scale of solar cycles and viceversa. They identified 4 periods: from circa 1530 to 1715, and from circa 1780 to 1915 the sunspots dominated over bright features, the associated cycle lengths were longer (L > 10.941) and the Earth's temperature was relatively low. From circa 1725 to 1789, and from circa 1920 up to the present time the solar cycle has been dominated by bright features, the cycle lengths are shorter (L < 10.94) and the temperatures have been relatively higher.

It is this last change of the Sun's behavior, which occurred around 1920, the one that has been obviously most documented. In fact several papers have reported that sign reversals in meteorological variables such as temperature, rainfall, water level, etc., occurred in various locations around 1920: Schuurmans (1975) and Kullmer (1943) observe this situation in 1922 for the barometric pressure, dominant winds and tropical storm frequency. Starr and Oort (1973), King (1973, 1975), Eddy (1976) and King et al. (1974) noticed this change in the temperature pattern during the period from 1921 to 1930. King et al. (1974), King (1975) and Cornish (1982) noticed a shift in precipitation patterns within the period from 1922 to 1925. Clayton (1934) observed a similar shift in the apparent dependence of water levels in Lake Victoria after 1920.

It is the purpose of the present paper to assess if the claimed change in patterns of the main meteorological variables, namely barometric pressure, temperature and precipitation, ocurred around 1920.

 

Material and methods

We surveyed 300 meteorological stations worldwide located for barometric pressure, temperature and precipitation data (Clayton, 1927, 1934 and 1959). From these, a time series of five-year moving averages for each of the variables was computed; this was performed in order to smooth the original series. We plotted each of these smoothed series versus the R_z, sunspot estimates time series, where R_z, is Wolf's sunspot number, also known as Zurich's relative sunspot number (Herman & Goldberg, 1978). Fig 1a shows an example for the barometric pressure in La Habana Cuba; Fig. 1b displays an example for the temperature in Bombay India; and Fig. 1c presents an example for precipitation in Curitiba, Brasil.

Linear correlations for the five-year averages -of each of the three meteorological variables considered- versus R_z were performed and the correlation coefficients were computed. These analysis were performed separately for the data from 1869 to 1925 and for those from 1926 to 1977. Histograms of the correlation coefficients that included data of the stations considered were plotted for each of the former periods and a non-parametric test (Kruskal-Wallis' ANOVA) was performed. This test was chosen because a prior assumption of normal distribution is not needed.

 

Results

In nearly half of the stations the plots showed a change in the patterns of the meteorological variables within the period 1920 to 1925; therefore the data were divided in two periods 1869 to 1925 and 1926 to 1977, and the correlations between the meteorological parameters and the sunspots numbers were computed separately for each period. The results appear in Tables 1, 2 and 3 for the barometric pressure, the temperature and the precipitation respectively. These tables contain the localities and countries, their geographical coordinates, their correlation coefficients for both periods and the number of five-year average data pairs considered for each locality. These tables show that the change of pattern occurs either as a sign reversal of the correlation coefficient -positive to negative or viceversa- or as a variation of the amplitude of the correlation coefficients, i.e. their absolute value.

These results are suminarized in Table 4 as percentages of the stations where changes occurred. For barometric pressure almost half of the stations (47%) show a change in the amplitude of the correlation, of the remaining half, a larger number of stations (31%) presented a shift from negative correlation to a positive correlation, those with the opposite trend were a mere 22%.

For temperature around half of the stations did not present a significant change (54%), most of the rest presented a change from a negative to a positive correlation (34%) and the inverse trend accounted only 12%. Finally, for the precipitation about half of the stations (45%) did not change. 30% presented sign reversals from positive to negative and 25% shifted from negative to positive.

The results of tables 1, 2 and 3 are also depicted in Figs, 2, 3 and 4 respectively, and a summary is presented in Table 5 which contains the sign of the correlation coefficients -either positive or negative- between the meteorological parameter and the sunspot numbers for two periods (I from 1869 y 1925 and II from 1926 to 1977). The results -converted to percentual correlation coefficients- were divided into intervals for each of the three meteorological variables and a code was employed in order to show whether the values ocurred within the tropical zone or in northern or southern extratropical locations. Also in agreement with Figs. 2, 3, and 4, the intervals of the correlation coefficients appear as percentages.

We notice that for the barometric pressure during period I the strongest negative correlation occurred in the tropical zone (zone I) and a weak correlation exists in the southern extratropical zone (2S), while the strongest positive correlation occurred in the northern extratropical zone (2N). For the period II both correlations hold in the same zones but they have weakened.

For the temperature in period I the strongest negative correlation is found in both southern and northern extratropical zones, while the strongest positive correlations are found in the northern extratropical zones. For period II the strongest negative correlations occurred in the southern extratropical zone, and a weak correlation is in the northern extratropical zone, while a positive correlation is found in the tropical zone.

For precipitation, during the period I the negative correlation is strongest in both northern and southern extratropical zones and there is a weaker positive correlation in the tropical zone. For period II the strongest negative correlations are found in the tropical and northern extratropical regions.

For the three variables here considered, correlation coefficient histograms (Figs. 5a, 5b and 5c) show significant differences between both periods, where those for the second period (1926 to 1977) show a fairly good fit to a normal distribution with the mean close to 0.0. On the other hand, the histograms for the first period are close to a rectangular distribution and so higher frequencies of high correlations -in absolute values- occurred during this period.

Even though histograms point to differences in correlation coefficient distributions during both periods, Kruskal-Wallis' results for correlation coefficients only show significant differences between both periods for the temperature series (p=0.003) (Table 6). Thus, according to this latter test, the claimed change in climate variables only holds for the correlation coefficients of air temperature vs sunspot numbers.

 

Discussion

For most of the stations that presented a change of pattern around 1920 (see Table 4), the change was from negative to positive; this effect is particularly noticeable in temperature. It should be remembered that several authors have reported regional changes of pattern from positive to negative around 1920 in various meteorological variables (King, 1973; 1974; King et al., 1974). Furthermore in a global context it has also been shown that a change of pattern from positive to negative should have occurred around 1920 (Mendoza and Ramírez, 1999).

It has been suggested that only in the equatorial regions there is a significant correlation between the sunspot cycle and the meteorological variables, specially the temperature, while from subtropical to polar latitudes the magnetic cycle masks the sunspot cycle (Willett 1974). Table 5 indicates that the strongest correlations -either positive or negative- are present in both extratropical and tropical zones for temperature, and also appear in both extratropical regions for precipitation. In the present study the correlations for temperature found in the tropical zones are weak and this trend is opposite to the results of previous works (Schostakowitsch, 1933). However, both for barometric pressure and for precipitation the correlations with the sunspots cycle are very strong in the tropical zone even though the regions which present such correlations are less extensive in longitude and latitude than the areas with strong correlations in the extratropical latitudes.

 

Conclusions

• Around 1920 we found a change of pattern in the correlation between sunspots and pressure, temperature and precipitation.

• Most of the stations that presented a change did it from negative to positive correlation.

• We also found strong correlations between sunspots and pressure and precipitation in both tropical and extratropical latitudes. For the temperature the strong correlations appear mainly in extratropical zones.

 

References

CLAYTON, H. H., 1927. World Weather Records. Smithsonian Institute. no page numbers.         [ Links ]

CLAYTON, H. H., 1934. World Weather Records. Smithsonian Institute. no page numbers.         [ Links ]

CLAYTON, H. H., 1959. World Weather Records. Smithsonian Institute. no page numbers.         [ Links ]

CORNISH, E. A., 1982. On the secular variation on rainfall at Adelaide, Australia. Journal Physical, 7, 334.         [ Links ]

DOBROVOLSKI, S. G. and B. J. CHOUDHRY, 1996. Temporal climatic variabilities of global atmospheric, oceanic, and land surface parameters. Atmósfera, 9 (4): 237-289.         [ Links ]

EDDY, J. A., 1976. The Maunder minimum. Science 192, 11.         [ Links ]

HERMAN, J. R. and R. A. GOLDBERG, 1978. Sun, Weather and Climate. National Aeronautics and Space Administration. U.S.A. 358 p.         [ Links ]

KING, J. W., 1973. Solar radiation changes and the weather. Nature 245, 443.         [ Links ]

KING, J. W., 1975. Sun-weather relationships. Astronautic and Aeronautic, 10 p., April.         [ Links ]

KING, J. W., E. HURST, A. J. SLATER, P. A. AMITH and B. TAMKIN, 1974. Agriculture and Sunspots. Nature 252, 2.         [ Links ]

KULLMER, C. J., 1943. A remarkable reversal in the distribution of storm frequency in the U. S. in double Hale solar cycles. Smithsonian Misc. Collection 103, 20.         [ Links ]

MENDOZA, B. and J. RAMÍREZ, 1999. Can the low-activity sun become dimmer at maximum? Solar Physics 199: 201-209.         [ Links ]

SCHOSTAKOWITSCH, W. B., 1933. Sunspots and the Weather. Bulletin American Meteorological Society 14(3): 65-69.         [ Links ]

SCHUURMANS, C. J. E., 1975. On climate changes related to the 22-year solar cycle. In: BANDEEN, W. R. y S. P. MARAN (Eds.), Goddard Space Flight Center Special Report SP-366. 161 p.         [ Links ]

STARR, V. P. and A. H. OORT, 1973. Five-year climatic trend for the northern hemisphere. Nature 242, 310.         [ Links ]

WILLETT, H. C., 1974. Recent Statistical Evidence in Support of the Predictive Significance of Solar-Climate Cycles. Monthly Weather Review 102: 679-686.         [ Links ]