suppose u have the variable u and v components of wind vector, you may write:

'd skip(u,7);skip(v,7)'

d means display

7 gives the size of the arrow. 

bigger the number is the less you will count arrows and the smuller the number is u will count many arrows

You can use the "maskout" function in conjunction with "skip" to plot vectors at greater intervals at high latitudes.

First, generate a file (e.g., called "mask.dat") on the same grid as your data, but with values=1 for latitudes higher than some threshold (say, 75 deg. lat.), and values =-1 for latitudes less than that.

Next, create a new .ctl file (mask.ctl), with DSET mask.dat, using same XDEF, etc. as your real data, but with just 1 variable, e.g.

VARS 1

  mask 0 99 mask

ENDVARS

Then when running grads, "open mask.ctl".  This makes the mask array available as mask.2 .  First, plot the low-latitude vectors without any "skip" - excluding the high-latitude region with "-mask.2" (which has negative values at high lats):

d maskout(u,-mask.2);maskout(v,-mask.2)

Then plot the high-latitude vectors using "skip", excluding the low-lat region with "mask.2" (where neg. values are at low lats):

d skip(maskout(u,mask.2),4,4);skip(maskout(v,mask.2),4,4)

You will probably need to experiment a bit with "skip" values, and with threshold latitude.   

Note that "maskout" only applies to a gridded scalar field (like U, or V), not to the vector field, so the "skip" function must apply after any "maskout" used.

If you really wanted, you could use 3 or 4 different "masks" to select 3 or 4 different vector densities.

I hope you can make sense of all that. I just tried this myself on just a regular lat-lon grid, and at least it works on that okay.

Thank you for your answer Kpaikpai Batebana. I usually use 'd skip(u,2);v' in grads because it is same as 'd skip(u,2);skip(v,2)'. Also, if I try 'd skip(u,7);skip(v,7)' as your kind answer, then there is hardly arrow anymore in Arctic map with 'set mproj nps', 'set lat 60 90' field. So, secondly I will try Enda William O'Brien's answer and then I will upload if it will be successful. Thank you very much:)

Dear Enda William O'Brien

I'm very happy and thank you for your helpful kind comments.

I perfectly solve my problem from your answer.

Thank you very much!

Hello Hyunggyu,

Very glad this worked out for you!

Onwards & upwards!

Cheers,

Enda