The Gleissberg (~85 year) and other periodicities in the flood cycles of the Brahmaputra River past present and future; implications for possible global mechanisms

Asten and McCracken (AGU 2021, paper H45Z-12) note strong ~85 and ~50 year periodicities in flood data and lake level data in NSW, south-east Australia.  We now compare data sets for the Brahmaputra River (latitude 25⁰N) 1800-2000CE with Lake George (latitude 35⁰S)  levels 1820-2020CE.  Both data sets show a pair of dominant spectral maxima at 80 and 50 year periods.

A study by Rao et al (2020) of observed and reconstructed discharge of the Brahmaputra shows only limited correlation of discharge rates with recorded floods.  We use a record of Oceanic Nino Index 1870-2021CE (McNoldy, 2021) to compare with floods and find that for time 1875-2010, 14 of 17 observed floods associate with La Nina events. However there were 27 La Nina events in this interval hence as a working hypothesis LaNina events are close to being a necessary condition (~82%) for floods but not a sole determinant. We use spectral analysis to locate multi-decadal natural cycles which also influence discharge levels and flood frequency.

The Brahmaputra discharge rate data extends back to year 1309CE (Rao et al, 2020).  The power spectrum   shows a series of strong maxima, especially at 242, 132, 90, &75year periods similar to those in ¹⁴C and ¹⁰Be records for the Holocene.  The entire record can be fitted using a model of 8 sinusoids, leaving only a 20% residual variance.  The model allows extrapolation of the discharge rate into the future and predicts an above-average discharge for years 1995-2040CE, peaking ~2020.  This predicted time-span of above-average discharge is based on natural frequencies embedded in the record and does not include any possible influences from 21^st-century global warming.  The prediction appears closer to the observed increase post-2000, than does a prediction based on CMIP5 models as provided in the Rao etal (2020) paper.

A further test of the efficacy of the discharge curve fitting method is provided by limiting the observed data to years 1309-1900CE, then projecting the model to 2200.  The projected curve from 1900 replicates the observed dry period 1950-1995 and validates the hypothesis that the dry period was not an unusual event but was part of the natural cycles as reconstructed since 1309.  The projected curve from 1900 also closely follows the model based on all data to 2010 in predicting the above-average discharge rates 1995-2040.

  As noted above both data sets show a pair of dominant spectral maxima at 80 and 50year periods. The similarity between the spectra invites a hypothesis that the long-period natural cycles at both locations have a common origin, possibly solar-related rather than being of local atmospheric/oceanic origin.  A key difference is that the phases of the spectral maxima are reversed for the two sites.  Physical mechanisms producing these dominant periods for the 19^th and 20^th centuries, and the phase difference between the northern and southern hemisphere sites are not yet known. They could be related to variations in solar insolation, cosmic-ray ionization of cloud cover, or mode changes in global ocean current systems driven by unknown external forcing.