An analysis of Indian tide-gauge records

Abstract

The paper presents an analysis of four Indian tide-gauge records. The stations were: Bombay, Madras, Cochin and Vishakhapatnam (Vizag). They were selected because of their reliability. There was no evidence of a monotonic rising trend at all four stations. The test by Mann and Kendall (loc. cit.) showed a rising trend at Bombay from 1940 to 1986 and at Madras from 1910 to 1933. The other records did not reveal a significant trend. The records reveal evidence of long-period cycles (50-60 year period), with shorter cycles (4.5 to 5.7-year period) riding on them. Spectral peaks corresponding to shorter cycles passed a false alarm probability test at 95% level of significance. The peaks were identified by computing periodograms and by maximizing the entropy of the time series. ARIMA models suggest a third order autoregressive model for Bombay and Madras (1953-1986). The remaining records only had a moving average component. Monthly tide-gauge data of Bombay reveal a 13.4-month cycle which was statistically significant. This was close to the 14.7-month Chandler wobble. But, an interaction between a 13.4-month and an annual cycle could not fully explain the observed short period cycles. Finally, the paper summarizes evidence to indicate that a pattern exists between fluctuations of monsoon rain and relative sea level at Bombay.