[ncl-talk] Dimension size, for dimension number 0, of operands does not match
Vanúcia Schumacher
vanucia-schumacher at hotmail.com
Mon May 11 12:29:31 MDT 2015
NCL users,
I need to calculate the correlation between two variables with different lat and lon. But when I try to calculate the significance test this correlation appears this error:
Dimension size, for dimension number 0, of operands does not match, can't continue
I understand that the error is due to DOF1 and dof2 have different sizes of lat and lon, but how could I solve it in this case?
My code:
....
r = escorc( t, t1 ) ; correlationcor= r(:,:,0,0)copy_VarCoords(t,cor)
; Calculate degrees of freedom for each time series pair; This estimates the dof in autocorrelated series. siglvl = 0.05 dof1 = equiv_sample_size (t, siglvl,0) dof2 = equiv_sample_size (t1, siglvl,0)
; For each grid point time series pair, choose the lesser abs value dof = where(abs(dof1).le.abs(dof2), dof1, dof2) error line
copy_VarCoords(cor,dof)
rsig = rtest( cor, round(dof,3), 0) ; r test NCL...
Variable: tType: floatTotal Size: 1944000 bytes 486000 valuesNumber of Dimensions: 3Dimensions and sizes: [lat | 90] x [lon | 180] x [time | 30]
Variable: t1Type: floatTotal Size: 7920 bytes 1980 valuesNumber of Dimensions: 3Dimensions and sizes: [lat | 6] x [lon | 11] x [time | 30]Coordinates:
Variable: rType: floatTotal Size: 4276800 bytes 1069200 valuesNumber of Dimensions: 4Dimensions and sizes: [90] x [180] x [6] x [11]Coordinates: Number Of Attributes: 1
Variable: corType: floatTotal Size: 64800 bytes 16200 valuesNumber of Dimensions: 2Dimensions and sizes: [lat | 90] x [lon | 180]
Variable: dof1Type: integerTotal Size: 64800 bytes 16200 valuesNumber of Dimensions: 2Dimensions and sizes: [90] x [180]Coordinates:
Variable: dof2Type: integerTotal Size: 264 bytes 66 valuesNumber of Dimensions: 2Dimensions and sizes: [6] x [11]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20150511/262572d1/attachment.html
More information about the ncl-talk
mailing list