Summary
Three independent techniques used in spectral analysis of time series, the conventional periodogram, the maximum entropy spectral analysis and the non-integer method, are applied to a drought index time series and the results are compared. The results reveal that in general the three approaches give similar estimates, especially in the high frequency domain.
The main drawback of the periodogram method is its poor frequency resolution, especially in case of short records. Consequently, while the position of the sinewave frequencies computed by the periodogram seems accurate, the peaks are broad mainly because of the limited length of the data sample. The maximum entropy spectral analysis on the other hand, gives better frequency resolution than either the periodogram or the non-integer method. This method is, however, sensitive to the number of terms of the filter and this has a great influence on the quality of the spectrum.
The non-integer spectra are comparable to higher order maximum entropy spectra. However, because the lowest frequency at which the non-integer spectra can be estimated is the inverse of the length of record, their spectra (for short records) may not be as reliable as those resolved by the maximum entropy spectral analysis in the low frequency domain.
Overall, it is suggested that the maximum entropy spectral analysis is preferable to either the periodogram or the noninteger method when one deals with short climatological time series. A good practice will be to compare the results of the maximum entropy with those of the non-integer method in order to strengthen inferences about the nature of periodicity in the analysed climatological time series.
Zusammenfassung
Drei unabhängige Techniken werden in Spektralanalysen von Zeitreihen verwendet: Die konventionelle Periodogramm-Methode, die Maximalentropie-Spektralanalyse und die Non-Integer-Methode werden auf Zeitserien des Trockenheits-Index angewandt und die Resultate verglichen. Diese Resultate zeigen, daß die drei Verfahren grundsätzlich zu vergleichbaren Ergebnissen führen, v.a. im Hochfrequenz-Bereich.
Der wesentliche Nachteil der Periodogramm-Methode ist ihre Frequenzungenauigkeit, besonders im Fall kurzer Aufzeichnungszeiträume. Daraus folgt, daß die im Periodogramm ausgewiesenen Positionen der Sinuskurven-Frequenzen genau zu stimmen scheinen, die Spitzen aber ungenau bleiben, v. a. auf Grund begrenzter Datenerhebungen.
Die Maximalentropie-Spektralanalyse wiederum weist höhere Frequenzgenauigkeit als die beiden anderen Methoden auf, reagiert jedoch äußerst sensibel auf die Anzahl der Filterterme wodurch die Qualität der Spektren stark beeinflußt wird.
Die Non-Integer-Spektren sind den Maximalentropie-Spektren höherer Ordnung vergleichbar. Da jedoch die niedrigste Frequenz, die auf den Non-Integer-Spektren gemessen werden kann, reziprok zur Aufzeichnungsdauer ist, sind diese Spektren (bei kurzer Aufzeichnungsdauer) im Niederfrequenzbereich nicht so zuverlässig wie jene der Maximalentropie-Spektralanalysen.
Daraus ergibt sich, daß bei Vorliegen kurzer klimatologischer Zeitserien der Maximalentropie-Spektralanalyse der Vorzug vor den beiden anderen Methoden zu geben ist. Ein Vergleich der Ergebnisse der Maximalentropie mit jenen der Non-Integer-Methode scheint empfehlenswert, da dies verstärkt Rückschlüsse auf die Art der Periodiziät in den untersuchten klimatologischen Zeitreihen erlaubt.
Similar content being viewed by others
References
Akaike, H., 1969: Power spectrum through autoregressive model fitting.Ann. Inst. Stat. Math. 21, 407–419.
Akaike, H., 1974: A new look at the statistical model identification.IEEE. Trans. Autom. Contr. AC-19, 716–723.
Anderson, N., 1974: On the calculation of filter coefficients for maximum entropy spectral analysis.Geophysics 39, 69–72.
Baggeror, A. B., 1976: Confidence intervals for maximum entropy spectral estimates. In:Aspects of Signal Processing Tacconi, G. (ed.). Dordrecht: D. Reidel Publishing Company, 617–630.
Barrodale, I., Erickson, R. E., 1980: Algorithm for leastsquares linear prediction and maximum entropy spectral analysis—Part I: Theory and Part II: FORTRAN PROGRAM.Geophysics 45, 420–446.
Berryman, J. G., 1978: Choice of operator length for maximum entropy spectral analysis.Geophysics 43, 1384–1391.
Bhalme, H. N., Mooley, D. A., 1980: Large-scale drought/floods and monsoon circulation.Mon. Wea. Rev. 108, 1197–1211.
Blackman, R. B., Tukey, J. W., 1959:The Measurement of Power Spectra from the Point of View of Communication Engineering. New York: Dover Publications, 190 pp.
Box, G. E. P., Jenkins, G. M., 1976:Time Series Analysis Forecasting and Control. Rev. Ed. San Francisco: Holden Day, 575 pp.
Burg, J. P., 1967: Maximum entropy spectral analysis. Paper presented at the 37th Annual Int. Meeting Soc. of Explor. Geophy. Oklahoma City.
Chen, W. Y., Stegen, G. R., 1974: Experiments with maximum entropy power spectra of sinusoids.J. Geophys. Res. 79, 3019–3022.
Childers, D. G., 1978:Modern Spectrum Analysis. New York: IEEE Press. 334 pp.
Courtillot, V., Le Mouel, J. L., Mayand, P. N., 1977: Maximum entropy spectral analysis of the geomagnetic activity index AA over a 107-year interval.J. Geophys. Res. 82, 1641–2649.
Fleer, H., 1981: Large-Scale tropical rainfall anomalies.Bonner Met. Abhndl., 26, 114 pp.
Fourgere, P. F., Zawalick, E. J., Radoski, H. E., 1976: Spontaneous line splitting in maximum entropy power spectrum analysis.Phys. Earth Planet Int. 12, 201–207.
Gilman, D. L., Fuglister, F. J., Michell, J. M., Jr., 1963: On the power spectrum of red noise.J. Atmos. Sci. 20, 182–184.
Goers, J. S., Koscielny, A. J., 1977: Monte Carlo estimates of confidence intervals for maximum entropy method. 5th Conf. on Prob. and Stat. in Atmospheric Sciences. 297–302.
Gottman, J. M., 1981:Time-Series Analysis—A Comprehensive Introduction for Social Scientists. London: Cambridge Univ. Press, 400 pp.
Gray, H. L., Houston, A. G., Morgan, F. W., 1978: On G-Spectra estimation. In:Applied Time Series Analysis Findley, I. D. G. (ed.). New York: Academic Press, 39–138.
Haykin, S., Kesler, S., 1979: Prediction-error filtering and maximum entropy spectral estimation. In:Non-Linear Methods of Spectral Analysis Haykin, S. (ed.). New York: Springer-Verlag, 9–72.
Jaynes, E. T., 1968: Prior probabilities.IEEE Trans. System Sci. Cybernetics. SSC-4, 227–241.
Jones, E. T., 1965: A reappraisal of the periodogram in spectral analysis.Technometrics 7, 531–542.
Kane, R. P., 1979: Maximum entropy spectral analysis of some artifical samples.J. Geophys. Res. 84, 965–966.
Kashyap, R. L., 1977: A Bayesian comparison of different classes of dynamic models using empirical data:IEEE Trans. Automatic Control AC-22, 715–727.
Katz, R. W., Skaggs, R. H., 1981: On the use of autoregressive-moving average processes to model meteorological time series.Mon. Wea. Rev. 109, 479–484.
Kaveh, M., Cooper, O. R., 1976: An empirical investigation of the properties of the autoregressive spectral estimator.IEEE Trans. Inform. Theory IT-22, 313–323.
Kay, S. M., Marple, S. L., 1981: Spectrum analysis—A modern perspective.IEEE Proc. 69, 1380–1419.
Kirk, B. L., Rust, B. W., Van Winkle, W., 1979:Time Series Analysis by the Maximum Entropy. Oak Ridge: National Lab. Pub. No. 1220, 206 pp.
Kromer, R. E., 1969:Asymptotic Properties of the Autoregressive Spectral Estimator. Tech. Rep. 13 Dept. of Statistics. Standford: Standford Univ., 188 pp.
Lacoss, R. T., 1971: Data adaptive spectral analysis methods.Geophysics. 36, 661–675.
Lacoss, R. T., 1976: Autoregressive and maximum likelihood spectral analysis methods. In:Aspects of Signal Processing: Tacconi, G. (ed.). Dordrecht: D. Riedel Publishing Company, 591–615.
Livezey, R. E., Chen, W. Y., 1983: Statistical field significance and its determination by Monte Carlo techniques.Mon. Wea. Rev. 111, 46–59.
Marple, L., 1980: A new autoregressive spectrum analysis algorithm.IEEE Trans. Acoust. Speech Signal Proc. ASSP-28, 441–454.
Neil, J. C., 1980: Prediction of rainfall trends. Final Rept. Phase II for NSF Grant AIM 76-11379. Illinois State Water Survey. Urbana 103 pp.
Newton, H. J., 1982: Using periodic autoregressive for multiple spectral estimation.Technometrics. 24, 109–116.
Oladipo, E. O., 1985: A comparative performance analysis of three meteorological drought indices.J. Climatology 5, 655–664.
Parzen, E., 1969: Multiple time series modelling. In:Multivariate Analysis, Vol. 2: Krishnaiah, P. R. (ed.). New York: Academic Press, 389–409.
Priestley, M. B., 1981:Spectral Analysis and Time Series. New York: Academic Press (2 Vol.), 890 pp.
Radoski, H. R., Zawalick, E. J., Fourgere, P. F., 1976: The superiority of maximum entropy power spectrum techniques applied to geomagnetic micropulsations.Phys. Earth Planet. Int. 12, 208–216.
Rao, K. N., George, C. J., Morey, P. E., Metha, N. K., 1973: Spectral analysis of drought index (Palmer) for India.India J. Met. Hydro. Geophys. 24, 231–270.
Reed, R. J., Campbell, W. J., Rasumussen, L. A., Rogers, D. G., 1962: Evidence of a downward propagating annual wind reversal in the equatorial stratosphere.J. Geophys. Res. 66, 813–818.
Reid, J. S., 1979: Confidence limits and maximum entropy spectra.J. Geophys. Res. 84, 5289–5301.
Rhode, H., Virji, H., 1976: Trends and periodicities in East African rainfall data.Mon. Wea. Rev. 104, 307–315.
Robinson, E. A., 1967:Multichannel Time Series Analysis with Digital Computer Programs. San Francisco: Holden-Day, 288 pp.
Ross, C. H., 1975: Maximum entropy analysis. Unpublished Manuscript. Meteorological Office. Bracknell 47 pp.
Rovelli, A., 1982: On the information time scale of geophysical records in maximum entropy spectral analysis.IEEE Trans. Geos. Remote Sensing. GE-20, 158–161.
Schickedanz, P. T., Bowen, E. G., 1977: The Computation of climatological power spectra.J. Appl. Met. 16, 359–367.
Steel, R. G. D., Torrie, J. H., 1960:Principles and Procedures of Statistics. New York: McGraw-Hill Book Co. Inc., 479 pp.
Swingler, D. N., 1979: A comparison between Burg's maximum entropy method and a non-recursive technique for the spectral analysis of deterministic signals.J. Geophys. Res. 84, 679–685.
Thompson, D. J., 1977: Spectrum estimation techniques for characterization and development of WT4 waveguide-1.Bell Syst. Tech. J. 56, 1769–1815.
Ulrych, T. J., 1972: Maximum entropy spectrum of truncated sinusoids.J. Geophys. Res. 77, 1396–1400.
Ulrych, T. J., Bishop, T. N., 1975: Maximum entropy spectral analysis and autoregressive decomposition.Rev. Geophys. Space Phys. 33, 183–200.
Ulrych, T. J., Clayton, R. W., 1976: Time series modelling and maximum entropy.Phys. Earth Planet Int. 12, 188–200.
Wilkinson, R. H., 1981: Will the real MEM please stand up? Proc. IEEE ASSP, 508–511.
Author information
Authors and Affiliations
Additional information
With 3 Figures
Rights and permissions
About this article
Cite this article
Oladipo, E.O. Spectral analysis of climatological time series: On the performance of periodogram, non-integer and maximum entropy methods. Theor Appl Climatol 39, 40–53 (1988). https://doi.org/10.1007/BF00867656
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF00867656