Generating a multi-parameter space and sampling from it

Hi all,
I have a long list of combination of parameters which covary. I would like to sample from it but not exactly the same values. What I'd like to do is to create a multi-parameter space that accounts for the covariation and then sample from it the parameter values to create new combinations that resemble the original ones.
To make it easier, let's think to only two variable. I would sample within the range of the cloud of points.

A		B		C		D		E		F		G
371		272		1532		831		2		1460		735
12		9		668		821		3		716		810
50		66		653		547		3		645		540
31		31		669		365		5		660		362
A possible resamble could produce something like this (done by eye):
A		B		C		D		E		F		G
100		81 		1000		800		2 		900 		750
I'm not sure it makes sense but I hope I was clear enough in the scope.
Do you know any statistical model that would make the job?
Any suggestion or comment?
Thanks in advance
Last edited: