I made a bunch of changes for the 1.2 release over the weekend.
Mostly ncwa was affected. I patched the ncwa logic and var_avg()
so that non-linear operations can make use of weights. So now you
can get the correct weighted average of the square of a field, etc.
This was especially important for computing spatial standard deviations.
I'm not quite sure how rmssdn should be defined when weights are
involved. Subtracting 1 from the normalization in the denominator
implicityly assumes that the weights are all 1.
I need a statistics book or guru to figure out the correct formula.
Help?
Also I moved the mathematical definition of averaging to the new
operations section of the manual, and explicitly added definitions
of all -y switches. Please check all this out and you will see
what I mean about the rmssdn problem.
Also I cleaned up ncap a little, am thinking about starting to
work on it again after 1.2 is released but can't decide whether
to move to complete C++ framework to simplify parsing/lexing,
or to keep existing bison yacc components. Anybody else
interested in this?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I made a bunch of changes for the 1.2 release over the weekend.
Mostly ncwa was affected. I patched the ncwa logic and var_avg()
so that non-linear operations can make use of weights. So now you
can get the correct weighted average of the square of a field, etc.
This was especially important for computing spatial standard deviations.
I'm not quite sure how rmssdn should be defined when weights are
involved. Subtracting 1 from the normalization in the denominator
implicityly assumes that the weights are all 1.
I need a statistics book or guru to figure out the correct formula.
Help?
Also I moved the mathematical definition of averaging to the new
operations section of the manual, and explicitly added definitions
of all -y switches. Please check all this out and you will see
what I mean about the rmssdn problem.
Also I cleaned up ncap a little, am thinking about starting to
work on it again after 1.2 is released but can't decide whether
to move to complete C++ framework to simplify parsing/lexing,
or to keep existing bison yacc components. Anybody else
interested in this?
Hi Charlie,
Shall have a look at 1.2 ASAP. Sorry I've not been able to contribute much this week but I had some paid work.
Regards Henry