Strides, steps and stumbles in the annual march
Brian Mapes
NOAA-CIRES Climate Diagnostics Center, Boulder
The mean annual cycle at daily
resolution is examined in NCEP reanalysis, other observational climate data
sets, and two climate models. For each variable at each point, the 365-day time
series had been decomposed using harmonic and wavelet analysis. The great
"strides" are captured in annual and semi-annual Fourier harmonics.
Wavelet analysis captures shorter-period features, localized in time, with a
screening based on signal-to-noise ratio (statistical significance). These
wavelet features include robust and well-known "steps" such as monsoon
onsets, as well as more mysterious and often marginally significant
high-frequency "stumbles" (called 'singularities' in older climate
literature) like the northern U.S. 'January Thaw.'
The data set of Fourier and wavelet coefficients is still huge, so an interactive display tool has been written, allowing extensive exploration by dataset, field, region, level, time period, wavelet period, significance, etc. Examples and model comparisons will be presented, along with a live demonstration on an annual cycle issue of interest to a lucky member in the audience.