Did DCT reduce correlations among values in the sample block?

Here's a correlogram of the sample data block:

and here's a correlogram of the DCT residuals:

The correlograms show autocorrelation results. The data file is first correlated against a copy of itself. The resulting value is, of course, a perfect 1.0 correlation, and is plotted above the X (horizontal axis) value 0. Then the first value of the copy is deleted, and the correlation is repeated. That result is plotted above, or below, the X value 1. And so on...

The X-axis, then, represents the number of values deleted from the copy, or the offset for each repeated correlation.

The DCT process clearly DID remove internal correlations, at least when the data are taken row-wise from left to right. The column-wise correlations would probably be similar, but not identical.

Did DCT increase the unevenness of the data distribution?

Here's a frequency graph of the sample data block:

and here's a frequency graph of the DCT residuals:

The DCT process clearly DID force most of the data values into a relatively small region around zero.