[ncl-talk] Model Sub-sampling

[Melissa Lazenby]

Hi NCL Users

I am trying to set up a ncl script that loops through multiple analysis in which the 39-model ensemble is sub-sampled according to:

(i) random selection of models. I.e. Repeatedly sample sub-samples of varying size i.e. 5, 10, 15, 20, 25, 30, 35 models using a random selection. So I want to construct a random sample of 10 models, 50 times and a random sample of 15 models 50 times. Derive the mean, median, standard deviation, IQR and range of the dP each time. Plot the mean SD of dP over the 50 samples as a function of number of models in sample size (I.e. 5, 10, 15, 20…35) on the X axis and the mean SD of the dP on the Y axis. And do the same for mean, Inter quartile range and range

(ii) various combinations of performance metrics using a selection of 5, 10, 15, 25, 30 35 models same as above

If you could direct me as to which function I could use to generate the random model selection that would be most helpful.

I only have NCL version 6.0.0 and therefore cannot use the generate_sample_indices function.

Many thanks!

Kindest Regards
Melissa
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20151027/a4a278d5/attachment.html