Spectral-trend estimation using incomplete data: application to Indian monsoon depressions

Thursday, 21 April 2016
Plaza Grand Ballroom (The Condado Hilton Plaza)
Naftali Y. Cohen, Yale University, New Haven, CT; and W. R. Boos, M. Gehne, and G. N. Kiladis

Monsoon depressions are synoptic-scale storms that form during the monsoon season in tropical regions such as South Asia and Northern Australia. At least half of continental monsoon rainfall is produced by monsoon depressions and other synoptic-scale low pressure systems, and extreme rainfall events associated with these storms also produce great floods and destruction.

Records from the India Meteorological Department show a significant downward trend in recent decades in the number of Indian monsoon depressions during summer. In particular, the years 2002, 2010, and 2012 were noted for having the first summers, in over a century, in which no depressions formed. Our certainty in this trend, however, is limited since recent analysis of satellite and reanalysis data show no detectable trend in the number of monsoon depressions in recent decades; satellite and reanalysis data also show significant synoptic-scale disturbances that satisfy traditional criteria for classification as monsoon depressions during the summers of 2002, 2010, and 2012.

Here, in order to validate or disprove the existence of trends in Indian monsoon depressions, we analyze long-term records of meteorological data from India obtained from surface weather stations and balloon soundings. We utilize the Lomb-Scargle periodogram and devise a method for estimating trends in spectral power. This allows for trend estimation based directly on in situ data that has not been altered by assimilation into a numerical model or by algorithms used to fill gaps in the data. We present trends in synoptic scale (2-14 day) and intraseasonal (20-90 day) variability of both winds and rainfall.

- Indicates paper has been withdrawn from meeting
- Submission entered in competition