Abstract
Empirical Mode Decomposition (EMD) is used to examine the relationship
between precipitation and surface temperature from six regions. Three
regions are defined by physiography: world, ocean, and land. The other
three regions are defined by averaged precipitation: dry, normal and
wet. Monthly averaged daily precipitation rate from the Global
Precipitation Climatology Project are compared with average monthly
surface air temperature anomalies from the Goddard Institute for Space
Studies using EMD. The EMD process produces component time series
referred to as intrinsic mode functions (IMFs). Theses IMFs are ordered
by frequency from high to low. Eight IMFs were produced for each the
time series. The first three IMFs corresponded to seasonal, semi-annual
and annual variations, respectively. IMF 4 to 6 corresponded to a
biennial, pentennial and decadal climate signals, respectively. IMF 7
was related to the broad 20-30 year period, with the trend being
revealed in IMF 8. The time series spanned the period from January 1980
to December 2015 at monthly intervals. Temperature and precipitation
time series from six sampling regions were analyzed for evidence of
correlation. Results from the analysis reveal the following: (1) The EMD
process reveals both linear and non-linear trends. The trends are not
entirely consistent between regions though they are highly correlated.
(2) Apparent wave-to-wave interactions between high and low frequency
components appear to be observed in the IMF 1 and 2. These distortions
appear to correspond to the troughs and peaks in the decadal cycle
captured in IMF 6 and may related to the solar cycle. (3) The
correlation between precipitation and temperature increases with
increasing IMF number.