| Histogram of maximum potential temperature amplitude
changes for cyclones and anticyclones |
|
TPV intensity changes (growth and decay; 1 K bins) |
|
Plots of TPV strong amplitude change locations. For each vortex, the location in which the maximum growth and decay occurred was
saved. Those vortices in which the growth and decay was more than 30 K were retained in
the following plot. The plots shows the locations where the number of vortices (normalized by cos(latitude))
within a 2.5 degree latitude by 7.5 degree longitude box had more than a 30 K tropopause potential
temperature amplitude change. |
|
TPV high amplitude change density plot |
| However from the histogram, it is suspicious that
amplitude changes of these intensities were occurring. Since the mean and median are significantly different for the
cyclones, it is apparent that at the very least there are some strong outliers. Upon eye examination of the datasets,
two startling discoveries were made. (1) In some vortex tracks (data include latitude, longitude, date, theta, theta
amplitude) that were physically adjacent to each other, there were sometimes exact
repeats! Further investigation revealed that there were some tracks that had different lengths, different staring
times, and/or different ending times, but that clearly the data in the middle were the same as the adjacent track. To
take care of this problem, I made a filter that compares adjacent vortex tracks by looking at their correlations. If
the correlation is unity, then they are repeats. But what about the ones with different track lengths that were not
exact repeats? Here, I took half of the first track centered about the middle, then took the second track and
phase compared it with the previous track. There had to be at least 4 (24 hours of data) phase matches (correlation of unity) before
eliminating one. (2) There were sudden "spikes" in potential temperature amplitudes. It is reasonable to assume that
a spike of more than 30K is not physically realistic and can be thrown out. So to eliminate these spikes, I found each
time that the amplitude changed by more than 30 K. If this amplitude change was followed by an immediate amplitude
change of the opposite sign at the next data point of more than 90% of the original amplitude change, it seems
reasonable to replace the corresponding large amplitude with NaN. |
|
Unfiltered cyclone growth |
|
After filtering out the repeats only |
|
After filtering out the spikes only |
|
Filtering out both the repeats and the spikes |
| So what percentage vortices with growth over 50K get removed?
|
|
Vortex before and after correlations |
| TPV density plots: Comparisons of filtered vs. unfiltered
|
|
Cyclones and anticyclones |
