NCL常见错误解决方案

2021年2月9日发(作者：transitional)

NCL Frequently Asked Questions (FAQ)

Basics

?|?

NCL core language

?|?

File input/output

?|?

Data handling

?|?

Data

a nalysis

?|?

Visualization

< p>
?|?

Array

functions

?|

Output

formats

?|?

Error

messages

?|?

Memory

issues

?|?

Document ation

?|?

Miscellaneous

?|

Basics

1.

Where can I download NCL

2.

I'm having problems downloading NCL from the ESG website.

3.

How do I set up my environment to run NCL

4.

How do I run NCL

5.

How do I determine the version of NCL

6.

Are there any email lists I can join

7.

Where can I see some examples of how to use NCL

8.

Are there any NCL command line options I can use

9.

Are there any editor enhancements for editing NCL scripts

10.

I

f I'm running NCL interactively, are there any command-line shortcuts I can

use

11.

I

f I'm running interactively, how can I load some existing NCL code

12.

I

s there anyway I can turn off the preceding

of values

13.

M

y NCL script is quitting suddenly, and I'm not sure where and/or why.

14.

H

ow do I tell if I have an OPeNDAP-enabled NCL

NCL core language

1.

How do I comment a line in NCL Is there a way to have a block of comments

2.

Is there a continuation character in NCL

3.

How do I easily convert a number to a string

4.

How do I create a new variable that doesn't contain a _FillValue attribute

5.

How do I create a double, long, short, or byte variable without using

one of the conversion routines, like

integertobyte

6.

How do I include the double quote (

7.

Can NCL prompt for user input

8.

Is it possible to have global variables in NCL

9.

How can I delete multiple NCL variables with one statement

10.

H

ow do I delete a named dimension

11.

I

s there a way to see the loaded variables and their dimensions in NCL similar

to

12.

I

'm getting an unexpected value when I use the power ('^') operator combined

with a negative value.

13.

H

ow

do

I

create

a

list

of

string

integers

with

leading

zeros

(

etc)

14.

H

ow do I get the current date into NCL as a string

15.

H

ow do I get a list of files into an array of strings in NCL

16.

H

ow can I recursively get a list of files in a directory tree into NCL

17.

H

ow do I concatenate strings and values in NCL

18.

H

ow do I retrieve an environment variable in NCL

19.

H

ow do I print nicely formatted, multiple columns of data in NCL

20.

C

an multiple variables be returned from a user created function

File input/output

1.

What's the easiest way to convert a GRIB1, GRIB2, HDF 4, or HDF-EOS 2 file to

a NetCDF file

2.

How do I dump the contents of a GRIB, HDF, or HDF-EOS file (or variable on the

file) to the screen

3.

How do I convert a large file to a NetCDF file

4.

How do I write a large (> 2 GB) variable to a netCDF file

5.

How do I get the latest tables for reading GRIB files

6.

Why have some of my GRIB2 variable names changed since I updated to NCL?How do

I write a scalar to a NetCDF file

7.

How

do

I

retain

metadata

for

a

variable

that

was

read

off

a

series

of

files

using

8.

How do I access an NCL file variable, attributes, or coordinate arrays if they

have hyphens/dashes ('-') in the name

9.

How do I read/write NetCDF, HDF4, HDF- EOS2, GRIB1, GRIB2, or CCM History Tape

files

10.

H

ow do I read/write ASCII files

11.

H

ow do I read CSV files

12.

H

ow do I write a formatted ASCII file

13.

H

ow do I read an ASCII file if I don't know how many lines it contains

14.

H

ow do I read an ASCII file that contains delimiters

15.

H

ow do I read/write binary files

16.

O

nce I've opened a file with

attributes it contains

17.

I

s there a way to control whether my binary file is read or written as little

endian or big endian

18.

I

'm having trouble reading a Fortran binary file with records.

19.

I

'm

trying

to

convert

a

file

to

a

NetCDF

file

using

but

getting

an error message about violating format constraints.

20.

W

hy

do

I

get

different

results

when

reading

my

HDF- EOS2

file,

depending

on

whether

I use the

21.

C

an I read shapefiles using NCL

22.

H

ow can I get at vgroup data on my HDF4 file

23.

W

hen I run my script, I'm getting an error message

Not a valid ID

24.

W

hen I try to access the Climate Diagnostics Center [CDC] OPeNDAP server via

NCL, I get

25.

I

'm running into a problem when opening lots of files with addfile or addfiles

(

Data handling

1.

How do I set a missing value for my data

2.

How do I determine if my data contains any missing values, or count how many

missing values it has

3.

How do I change the missing value to another value

4.

How do I deal with NaN (not-a-number) values in my data

5.

How do I deal with

Data analysis

1.

How do I average values in a multi- dimensioned array

2.

How

can

I

do

an

element-wise

minimum/maximum

between

two

arrays

of

the

same

size

3.

How do I interpolate my 2-dimensional random data to a 2-dimensional gridded

array

4.

How do I interpolate my 3-dimensional random data to a 3-dimensional gridded

array

5.

How

can

I

figure

out

if

a

particular

latitude/longitude

point

is

within

a

specific

area, like Europe

Visualization

1.

How can I create an

2.

How can I maximize the size of a plot in the frame

3.

How do I get multiple plots on a page

4.

What's a good way to generate a color map (table)

5.

My map projection doesn't look right.

6.

How do I change all of my fonts to be the same in my visualization

7.

How can I change the text function code to something other than a colon

8.

How can I get a degree symbol in a text string

9.

How can I get things like accents or umlauts over my text

10.

H

ow

can

I

get

special

characters,

like

Greek

symbols,

in

my

graphical

text

strings

11.

C

an I use real PostScript fonts in my PostScript file

12.

I

'm having trouble importing a PS file into an application like Adobe

Illustrator.

13.

H

ow can I get the length of my axes to be the same ratio as the range of my axes

14.

H

ow do I change the aspect ratio or size of my map or plot

15.

C

an I set color resources by color name rather than color index values

16.

H

ow can I speed up the drawing of filled contours in NCL

17.

C

an I change color index values or color maps in mid-frame

18.

H

ow do I get my PostScript/PDF output to cover most of an 8-1/2

19.

H

ow can I control the size and location of the X11 window that NCL brings up

20.

H

ow do I tell NCL to ignore missing values in my plot

21.

H

ow can I get more labels on my contour lines

22.

M

y X or Y axis values are not linear. How can I change this

23.

H

ow do I get log scaling on either of my plot axes

24.

H

ow

do

I

fill

my

contour

levels

with

various

shading

patterns

and/or

solid

colors

25.

H

ow do I overlay a contour plot (or a vector/streamline plot) on a map

26.

H

ow do I force the tickmarks to point inward

27.

I

'm trying to control the tickmarks on the top or right axis, and nothing is

happening.

28.

H

ow can I improve the resolution of my map outlines

29.

W

hen I draw a map using the high-resolution map database, I seem to randomly

get different resolutions.

30.

H

ow can I get more geographical outlines in my map, like the provinces of China

31.

I

f I am creating a color contour or vector plot, how do I select my colors such

that they are spread across my whole color map

32.

H

ow do I generate a wire- frame surface or isosurface plot in NCL

33.

H

ow do I draw weather symbols (wind barbs, weather front lines, high and low

pressure

indicators,

arrows,

dots,

weather

icons,

cloud

symbols,

lightning

bolts,

etc.)

on a map

34.

H

ow can I have a new line as part of a graphical text string in NCL

35.

H

ow

come

I

get

different

line

thicknesses

on

different

output

devices

using

the

same thickness value

36.

H

ow can I put a timestamp on my plot

37.

W

hy am I not getting the color I expect when I reference a named color

38.

W

hy is my land not being filled in the color I expect

39.

H

ow do I set up my NCL script so that it uses a resource file

40.

H

ow can I move and resize my labelbar and/or legend

41.

H

ow can I set the precision of the label bar labels

42.

H

ow can I generate some

Array functions

1.

How do I get the dimensions of an array

2.

How do I create unique random subscripts

3.

How do I eliminate missing values from a one-dimensional array

4.

How

do

I

generate

a

one- dimensional

array

of

equally-spaced

integers

or

floating

point numbers

5.

How do I find or print the minimum/maximum values of my data array

Output formats

1.

How do I change the default NCGM/PostScript/PDF file name to something else

2.

How do I convert my NCL graphics file to another format, like PNG, to put on

the web or in a presentation

3.

How do I remove all the white space from an image

4.

Is there a way to make the PostScript files generated by NCL smaller

5.

Can I produce a PNG image directly from NCL

6.

Is there a way to remove the

7.

How can I tilt an NCL graphic

8.

How can I convert a series of PS or EPS files to an animation

9.

How can I produce Google Earth kmz files

10.

H

ow do I convert my multi- frame NCL graphics file to an MPEG movie file

11.

H

ow do I include an encapsulated PostScript file from NCL in a LaTex document

Error messages

1.

When I run my script, I'm getting a

2.

When I run NCL, I get an error message about being unable to load the

Resource File

3.

When I run NCL, I get an error message about the X driver, like

display

4.

What's the deal with the error message

5.

How can I turn off NCL warning messages

6.

I'm trying to list some files, and getting an error

7.

When I try to run ncl, I get an error about

8.

I'm trying to compile my NCAR Graphics program using the Intel or gfortran

compiler, but getting some undefined references.

9.

When I run ncl version I get an error

specification string.

10.

W

hen I run ncl, I get an error about libpng not being current.

11.

I

'm getting an error from WRAPIT about being unable to find

12.

I

'm getting a strange

13.

I

am

trying

to

create

a

contour

plot

and

getting

an

error

about

the

scalar

field

being constant.

14.

I

am trying to create a contour plot and getting errors about sfXArray and

sfYArray having an incorrect dimension size.

15.

I

'm getting an

16.

I

'm getting the error message

FUNCTION CODE

17.

I

'm getting the error

18.

I

'm

getting

an

error

from

the

script

about

being

an

undefined

procedure.

19.

W

hen running a script with overlaid plots, I'm getting an error

Plot Member

20.

W

hen

I

use

NCL's

systemfunc

on

a

X/Cygwin

system,

I

get

the

error

create

child process

21.

W

hen I compile an NCAR Graphics program, I get errors about undefined symbols

like

22.

W

hen I use

23.

I

'm trying to assign a value to an NCL variable, but getting a

can't be coerced to type of left hand side

24.

I

'm calling NCL from a shell script and getting a fatal error on line -1.

25.

I

'm getting an

message.

26.

I

'm

trying

to

draw

primitives

(markers,

lines,

polygons)

on

my

plot,

and

getting

an error essage

27.

W

hen

I

call

one

of

the

gsn_csm_xxxx_map

plotting

scripts,

I

get

an

error

message

about a units attribute not being correct.

28.

I

'm trying to create a contour, vector, or contour plot, and getting a weird

error about spline approximation.

29.

W

hen

I

print

something

in

NCL,

I

get

an

error

about

my

PAGER

environment

variable.

Memory issues

1.

What's the biggest array that NCL can allocate

2.

I'm running into a memory issue when I create lots of strings.

Documentation

1.

Is there a way I can download the NCL website documentation so I can access it

locally on my machine

2.

Where can I see a list of all the NCL resources

3.

Where can I see a list of all the NCL functions and procedures

4.

Where can I see a list of links to dash patterns, marker styles, color tables,

etc

Miscellaneous

1.

How should I cite NCL

2.

How can I add a logo to an NCL plot

3.

How do I call a C or Fortran function or procedure from NCL

4.

How

do

I

make

sure

that

a

C/Fortran

routine

I

want

to

wrap

doesn't

conflict

with

a routine that might already be part of NCL

5.

How do I fix buffering issues with NCL

6.

If I obscure

an NCL X11

window with

another application, the

X11 window

doesn't

refresh when I bring it back to the foreground.

Basics

1.

Where can I download NCL

See the?

NCL download page

?for full instructions. You can download precompiled

binaries for a specific UNIX system, or source code.

2.

I'm having problems downloading NCL from the ESG website.

Please

see

our?

ESG

troubleshooting

guide

?for

help.

If

this

doesn't

help,

then

send

email to the?

ncl- install

?email list.

3.

How do I set up my environment to run NCL

o

Set the NCARG_ROOT environment variable and your search path to where

NCL/NCARG resides

In order to use the software, you must set your NCARG_ROOT environment variable to

the parent directory where the NCL executables and accompanying files were installed.

You also need to make sure that the directory where the NCL/NCARG executables reside

are on your search path. It is best to do this from one of your .* files in your home

directory. If you are not sure which shell you are running, you can do an

your home directory to see what kind of files you have that start with

look in these files to see how environment variables are set.

In the example above, we would set these with the following:

From C-shell (csh):

setenv NCARG_ROOT /usr/local

setenv PATH=/usr/local/bin:$$PATH

From bash or ksh:

export NCARG_ROOT=/usr/local

export PATH=/usr/local/bin:$$PATH

From sh:

NCARG_ROOT=/usr/local

PATH=/usr/local/bin:$$PATH

export NCARG_ROOT

export PATH

You

should

now

be

set

for

using

NCL

and/or

NCAR

Graphics.

If

you

have

any

problems,

or questions about installing NCL from source code or binaries, please send email

to??(you need to subscribe first).

o

graphics

This is not an NCL specific environment variable, but rather one that is used by

any X11 Windows application that you run. For NCL or NCAR Graphics, it is needed when

you display your graphics to an X11 window, or if you use?

ictrans

,?

ctrans

, or?

idt

?to

view your?

NCGM

?files.

This

variable

must

be

set

to

the

IP

address

or

name

of

the

machine

you

want

to

display

your

graphics

on,

for

example

.

If

you

are

not

running

on

a

remote

machine,

then

you

can

try

setting

it

to

Please

contact

your

system

administrator

if

you

are

not sure what to set it to.

For example, from C-shell (csh):

setenv DISPLAY :

setenv DISPLAY :0

Set the DISPLAY environment variable to indicate where to display

setenv DISPLAY localhost:

From bash:

export DISPLAY=:

export DISPLAY=:0

export DISPLAY=localhost:

o

Put a

In

order

to

better

customize

the

NCL

graphical

environment,

we

highly

recommend

that

you

copy

a

.hluresfile

?to

your

home

directory.

You

can

customize

this

file

to

your

liking,

but at a minimum, you want to

change the

default

font, the

default

color table, and

the

text function code.

o

Special instructions if you have an?

OPeNDAP

-enabled version of NCL

[You don't need to worry about these instructions if you are not using OPenDAP.]

OPeNDAP

-enabled binaries require some libraries to be installed on your host

computer:

1.

2.

Secure Sockets Layer

?(SSL) library

OPeNDAP

?libraries

Once you have these libraries, you may be required to set the environment variable

LD_LIBRARY_PATH (DYLD_LIBRARY_PATH for Macs) to contain the directory where they are

located

on

your

host

system

in

order

for

OPeNDAP-enabled

versions

of

NCL

to

work

properly.

4.

How do I run NCL

Once you have NCL installed, you can run it interactively by typing on the UNIX

command line:

ncl

This will echo some information to the screen, and give you a prompt:

Copyright (C) 1995-2008 - All Rights Reserved

University Corporation for Atmospheric Research

NCAR Command Language Version The use of this software is governed by a License

Agreement.

See for more details.

ncl 0>

To quit out of interactive mode, type

To run NCL on an existing NCL script called

ncl

5.

How do I determine the version of NCL

The version number is echoed whenever you run NCL. You can also type

the UNIX command line to get the version number.

6.

Are there any email lists I can join

Yes. For basic questions about installing NCL and setting up your environment to

use it:

all other questions on NCL:

questions relating specifically to NCAR Graphics:

7.

can I see some examples of how to use NCL

To see hundreds of NCL application examples, see the?

category list of application

pages

. These examples are full NCL scripts that you can download. Some of

the?

datasets

?are provided as a separate link.

If you are a complete beginner to NCL, then there's a?

tutorial with examples

?that

are described line-by-line.

8.

Are there any NCL command line options I can use

Yes, see the

NCL command line options

NCL Reference Manual

.

9.

Are there any editor enhancements for editing NCL scripts

Yes, see the?

support page

?for a list.

10.

I

f I'm running NCL interactively, are there any command-line shortcuts I can

use

There are shortcuts that allow you to navigate through previous commands and edit

them. For more information, go to the

NCL

Reference Manual

.

11.

I

f I'm running interactively, how can I load some existing NCL code

You can use the?

loadscript

?procedure, where you give it the name of the file that

contains NCL commands.

12.

I

s there anyway I can turn off the preceding

of values

If you are using?

print

?to print some values, and you see output like this:

Variable: x

Type: float

Total Size: 80 bytes

20 values

Number of Dimensions: 1

Dimensions and sizes: [20]

Coordinates:

(0) 0

(1)

(2)

(3)

(4)

(5)

. . .

then you can get rid of the

option

?when you invoke

Variable: x

Type: float

Total Size: 80 bytes

20 values

Number of Dimensions: 1

Dimensions and sizes: [20]

Coordinates:

0

. . .

13.

M

y NCL script is quitting suddenly, and I'm not sure where and/or why.

If you are running an NCL script and getting a

(which

technically

shouldn't

happen,

but

does

on

occasion),

then

try

commenting

out

your

command line option

:

ncl -x

This will cause every line of your script to be echoed to the screen, and you will

be able to see where it is quitting suddenly.

14.

H

ow do I tell if I have an OPeNDAP-enabled NCL

A quick way is to do an

references libraries like

with?

OPeNDAP

?capabilities enabled:

ldd `which ncl`

You should see output like:

=> /usr/local/lib/ (0x0000002a)

=> /usr/local/lib/ (0x0000002a956ee000)

=> /usr/local/lib/ (0x0000002a958b9000)

Here's

short

test

to

see

if

your

OPeNDAP-enabled

NCL

is

working.

Note

that

you

must

be on a system that allows OPeNDAP access:

url =

filename =

exists =

isfilepresent

(url+filename)

if(. then

print(

print(

servers, or NCL does not have OPeNDAP capabilities on this system

status_exit(1)

else

print

(

f =

addfile

(url + filename,

vars =

getfilevarnames

(f)

vars_out = (/

sqsort

(vars)

sqsort

(vars_out)

print

(vars) These two lists should

print

(vars_out) be the same.

end if

NCL core language

1.

How do I comment a line in NCL Is there a way to have a block of comments

The

comment

character

in

NCL

is

a

semicolon

(';').

You

can

use

it

on

a

line

by

itself

or after an NCL command:

; Here is a comment before a command that assigns a variable.

x = 5.

x = 5. ; This comment can be at the end of a line

Unfortunately,

there

is

no

way

to

designate

a

block

of

comments,

other

than

putting

semicolons in front of all the lines:

;

; This is a block of comments,

; each of which must be

; preceded by its own semicolon

;

2.

Is there a continuation character in NCL

Yes, you can use the '' character:

x = (/

; Can use for arithmetic expressions

output(i,1:n-2) = (p4)*(var( i-1, 1 : n-2 ) + var( i, 2 : n-1) +

var( i+1, 1 : n-2 ) + var( i, 0 : n-3)) +

(q4)*(var( i-1, 0 : n-3 ) + var( i-1, 2 : n-1) +

var( i+1, 2 : n-1 ) + var( i+1, 0 : n-3))

3.

How do I easily convert a number to a string

Use the '+' string concatenator:

x =

str =

Use?

sprinti

< p>
?or?

sprintf

?for formatted strings.

4.

How do I create a new variable that doesn't contain a _FillValue attribute

5.

x =

new

(5, float,

6.

How do I create a double, long, short, or byte variable without using

one of the conversion routines, like

integertobyte

7.

8.

9.

x = double, x = 1d (also acceptable)

l = 1l long

s = 5h short

10.

b = 1b ; byte

11.

T

he long, short, and byte literals were added in?

do I include the double quote

(

Since NCL uses double quotes to denote strings, you have to add an extra step to

get them into a string. As of we have a new function?

str_get_dq

?for retrieving the

double quote character:

quote =

str_get_dq

()

print(

print(

If you have an older version, you can do it this way:

quote =

inttochar

(34)

print(

print(

12.

Can NCL prompt for user input

No, not directly, but you can use NCL's?

systemfunc

?function or its?

command line

option

?capability to do this.

print

(

ans =

toint

(

systemfunc

(

print

(

Command line argument examples

A more involved way is to write a shell script that queries for input values. You

can then

use the

NCL

command

line argument

?capability

to

pass

these

values as arguments

on the NCL command line.

Shell script example to prompt for a float and a string

1.

2.

3.

4.

5.

6.

7.

Create a shell script called

#!/bin/sh

echo -n

read x

echo -n

read title

8.

9.

10.

eval ncl x=$$x 'title=

Create an NCL script called

11.

;---You can provide a default value for

12.

if(.(

13.

print(

14.

exit

15.

end if

16.

17.

print(

18.

print(

19.

20.

y = x * 100.

21.

print(

22.

24.

./

This should prompt you for a value and a title and then print some information.

C-Shell script example to prompt for a filename and a level value

25.

27.

28.

echo -n

29.

set fname = $$<

30.

31.

echo -n

32.

set level = $$<

33.

34.

eval ncl level=$$level 'fname=

35.

Create an NCL script called

Create a c-shell script called

26.

#!/bin/csh -f

Make sure

23.

chmod 775

36.

if(.not.

isvar

(

37.

fname =

38.

end if

39.

40.

if(.not.

isvar

(

41.

level = 850. Set a default

42.

end if

43.

44.

print

(

45.

print

(

46.

47.

;---Do some more stuff in your NCL script...

48.

a =

addfile

(fname,

49.

T = a->T({level},:,:) Read a 3D variable at a particular level.

50.

52.

./

This will prompt you for a filename and level value, and pass this information to

the NCL script. This particular NCL script will assign these values to normal NCL

variables, and then use them in the script as usual.

You

can

do

this

kind

of

thing

with

any

scripting

language

you

want,

like

tcsh,

bash,

or even Python.

13.

I

s it possible to have global variables in NCL

NCL does not have an explicit way of defining

scoping

rules,

which

are

similar

to

Pascal's,

can

be

used

to

simulate

variables

For more information, see the?

NCL reference manual

, specifically, the

NC L?

statement

?documentation.

In a nutshell, if you define a variable?

above

?a function, procedure, or the main

code, you can reference it in that function, procedure, or main code:

PI = Use some naming convention for

; global variables [here, all caps]

function area_circ(r:numeric)

begin

return(PI*(r^2))

end

; Main code

begin

print(

radius =

area = area_circ(radius)

print(

end

14.

H

ow can I delete multiple NCL variables with one statement

In version and later, you can delete multiple variables of mixed types and sizes

by

using the?

delete

procedure and

the special syntax

the variables

you want to delete:

lat =

fspan

(-90,90,64)

lon =

fspan

(-1 80,180,128)

data =

random_uniform

(-10,10,(/10,20,30/))

varnames = (/

Make sure

51.

chmod 775

. . .

delete

([/data,varnames,lat,lon/])

print

(lon(0)) This will produce an error that

This is currently in alpha-test mode.

15.

H

ow do I delete a named dimension

Unfortunately,

this

is

not

possible.

If

we

get

a

lot

of

requests

for

this

capability,

then

we'll

consider

adding

this

functionality.

You

can

set

a

named

dimension

to

but

this is not quite the same thing:

x = (/1,2,3/)

x!0 =

print(x)

x!0 =

print(x)

16.

I

s there a way to see the loaded variables and their dimensions in NCL similar

to

See t he?

list_vars

?procedure.

< p>

17.

I

'm getting an unexpected value when I use the power ('^') operator combined

with a negative value.

The negation operator ('-') has the highest precedence in an NCL expression,

potentially producing unexpected results. For example, the following expression:

x = - 3^2

yields a value of 9 because it behaves as:

x = (-3)^2

Note how the use of parentheses can change the results:

print(- (3+2)^2) ----> yields 25

whereas:

print(- ((3+2)^2)) ----> yields -25

Finally, if the '-' symbol appears between two expressions or numbers, then it is

treated as a minus rather than negation. Hence, the following yields a value of 9:

x = 0 - 3^2 equivalent to x = (0) - (3^2)

18.

H

ow

do

I

create

a

list

of

string

integers

with

leading

zeros

(

etc)

Use the?

sprinti

?function with

a format

like

strings with leading zeros where possible). For example:

ii =

ispan

(0,100,1)

iis =

sprinti

(

19.

H

ow do I get the current date into NCL as a string

Use the?

systemfunc

?function, which executes a UNIX system call and returns the

output as a string:

datestr = systemfunc(

20.

H

ow do I get a list of files into an array of strings in NCL

Use the?

systemfunc

?function with the UNIX

filenames = systemfunc(

21.

H

ow can I recursively get a list of files in a directory tree into NCL

Use the?

systemfunc

?function with the UNIX

filenames =

systemfunc

(

If you are looking for a specific set of files, like ones that end in

the

filenames =

systemfunc

(

22.

H

ow do I concatenate strings and values in NCL

You can use the

operating on strings. You can also concatenate strings with other NCL variable types,

like float and integers. For example, the following NCL code snippet:

month = 1

day = 21

year = 2002

date_str =

will produce the string

You can use the functions?

sp rinti

?and?

sprintf

?to generate formatted strings. For

example, if you want to construct filenames on the fly with names like

your code would look something like this:

do i=1,10

filename =

sprinti

(

...do something with

filename

...

end do

23.

H

ow do I retrieve an environment variable in NCL

Use the function?

getenv

:

env_str =

getenv

(

24.

H

ow do I print nicely formatted, multiple columns of data in NCL

There

are

several

functions

that

will

nicely

format

2D

arrays

using

a

user- specified

format

string.

The

original

procedure

was?

write_matrix

.

This

can

write

to

either

a

text

file

or to standard

out.

It is

limited to

one

variable

of the same

type. It use fortran

style format syntax. Two new procedures were introduced in NCL

version

?

p rint_table

?and?

write_table

.

These

can

handle

variables

of

mixed

type.

Please

see the documentation and examples for these procedures.

Here's an example on how to use?

write_matrix

:

N = 5

M = 7

ave =

std =

x =

random_normal

(ave, std, (/N,M/))

write_matrix

(x,

This example will produce the following output:

A more complicated method is to write a C or Fortran subroutine to print the data

as you want it, and then use?

wrapit77

?as described in the section

using the wrapit77

application

NCL Reference Manual

. There's an example on?

2-dimensional array

printing

?in the?

WRAPIT

?guide.

The functions?

sprinti

?and?

sprintf

?can also be used to generate formatted output.

25.

C

an multiple variables be returned from a user created function

Yes. However, it not as direct as (say) Matlab. In NCL, the return argument must

be of type list. This can contain variables of different types and sizes.

A schematic example:

undef(

function foo(...)

begin

:

:

x = .. type float (:,:)

y = .. type double (:,:,:,:)

z = .. type string (1) a scalar

i = .. type integer (:)

return(

[/

x, y, z, i

]

) return as variable of type

list

end

ans = foo(...)

printVarSummary(ans) variable of type list

x = ans

[0]

explicitly extract each return variable

y = ans

[1]

; [..] syntax is for variable of type list

z = ans

[2]

i = ans

[3]

delete(ans)

Note: There is no need to explicitly extract the list variables.

They could be accessed as components of 'ans'

Please see the Mini-Language Manual: Sections and

Click

input/output

1.

What's the easiest way to convert a GRIB1, GRIB2, HDF 4, or HDF-EOS 2 file to

a NetCDF file

On the UNIX command line:

ncl_convert2nc

where?

xxx

?is

the

file

suffix,

like

or

(other

file

extensions

are recognized).

If you need to write an NCL script to do the conversion, see the?

output netcdf

applications page

.

For more information on supported file formats in NCL, see the

Information on

supported data formats

NCL Reference Manual

.

2.

How do I dump the contents of a GRIB, HDF, or HDF-EOS file (or variable on the

file) to the screen

Use?

ncl_fi ledump

?on the UNIX command line. Some examples:

ncl_filedump ncl_filedump -v

PRES_6_SFC

-ftime

-h

for

a

usage

message.

3.

How do I convert a large file to a NetCDF file

If

you

are

using?

ncl_convert2nc

?to

convert

a

supported

file

(like

GRIB)

to

a

NetCDF

file, and getting the error message:

ncendef: ncid 0: NetCDF: One or more variable sizes violate format

constraints

then try the?

?option:

ncl_convert2nc -L

4.

How do I write a large (> 2 GB) variable to a netCDF file

In

order

to

write

large

variables

to

a

NetCDF

file,

you

have

to

set

the

NetCDF

option

to

or

to

This

requires

calling?

setfileoption

?and

setting the

setfileoption

(

or

setfileoption

(

or

setfileoption

(

The

reason

you

have

to

set

this

option

is

that

the

NetCDF

library

requires

that

this

decision be made when the file is first created, before there is any knowledge of the

size of variables that will be written. We may eventually make this this default, but

the current recommendation from Unidata is to maintain backwards compatibility where

possible for apps that can only deal with the original small file format of NetCDF.

This

will

make

NCL

create

a

NetCDF4

file,

which

has

no

limitation

on

variable

size.

NOTE:

For any of the above three options to function properly,

called prior to

addfile

(

5.

How do I get the latest tables for reading GRIB files

GRIB1

You can use your own text-based tables that you can update yourself. Start with

text-based

tables

that

are

downloadable

from

our

website

at

on

any

of

the

built-in

tables

and you will find a link to a downloadable text-based version that you can use if you

set?

NCL_GRIB_ PTABLE_PATH

?appropriately. Also read the next section entitled

User-defined GRIB1 parameter tables

GRIB2

Except for parameters in

files

created for the TIGGE

project,

GRIB2 parameters are

all defined

in text

files contained in

the grib2_codetables directory tree. By default

this directory is found at $$NCARG_ROOT/lib/ncarg/grib2_codetables. If you want to add

parameters or modify the name, description, or units of any number of parameters, copy

the complete directory tree to your own location, and then set the environment

variable?

NIO_GRI B2_CODETABLES

?to the path of this location. The files that relate

specifically to parameter tables can be found in the parallel subdirectories ecmwf/4/

and

ncep/4.

They

have

names

of

the

form?

.*.*.table

.

Just

be

sure

to

maintain

the

format

of these files when modifying or adding to them. Basically each line contains 5 fields

separated by colons. The parameter number is repeated:

param

number:param

number:description

(long_name

attribute):units:short

name

(must

not contain spaces or non-alphanumeric characters)

Example:

0:0:Temperature:K:TMP

Why

have

some

of

my

GRIB2

variable

names

changed

since

I

updated

to

NCL?

necessary,

you can change them back the way they were by setting the environment

variable

NIO_GRIB 2_CODETABLES

?to the path: $$NCARG_ROOT/lib/ncarg/. See the release

notes for?

GRIB Updates

?for an explanation of why they have changed.

6.

How do I write a scalar to a NetCDF file

The first way is to use

spherical = new(1,

spherical!0 =

spherical =

tochar

(1)

ncid->spherical = spherical

The second way is to use?

filevardef

:

filevardef(ncid,

ncld->spherical = tochar(1)

Note

that

for

either

of

these

methods,

if

you

look

at

this

file

in

NCL

it

will

appear

as if there is a dimension named

dimension will not appear, so the variable?

is

?a true NetCDF scalar.

7.

How

do

I

retain

metadata

for

a

variable

that

was

read

off

a series

of

files

using

Use a special function from??called?

addfiles_GetVar

. For example:

load

fils =

systemfunc

(

f =

addfiles

(fils,

T =

addfiles_GetVar

(f,fils,

printVarSummary

(T)

8.

How do I access an NCL file variable, attributes, or coordinate arrays if they

have hyphens/dashes ('-') in the name

If you have an NCL file variable with hyphens in the name (for example,

in->Data- Set-14

variable directly. This is because NCL treats the hyphens as minus symbols:

ncl>

print

(in->Data-Set-14)

fatal:Either file (in) isn't defined or variable (Data) is not a variable in the

file

fatal:Execute: Error occurred at or near line 24

To

get

around

this

problem,

turn

the

name

of

the

variable

into

a

string,

and

enclose

the string in dollar signs ('$$'):

print

(in->$$

or

var_string =

print(in->$$var_string$$)

You can also use $$$$ for attribute names:

lname = data@$$

For

more

information

on

using

'$$'

to

reference

file

variables,

see

the

section

File

variable string references

NCL Reference Manual

.

9.

How do I read/write NetCDF, HDF4, HDF-EOS2, GRIB1, GRIB2, or CCM History Tape

files

To open any one of these types of files, use the?

addfile

?function, which opens a

data file and returns a pointer to the file. Once you have a pointer to the file, you

have

access

to

all

of

its

global

and

variable

attributes

and

coordinate

arrays.

You

can

also

use?

addfile

?to open

a NetCDF or HDF file for

writing

(not GRIB1,

GRIB2, HDF-EOS2,

or CCM files, since you can't write these files in NCL).

For example, to open a GRIB file and write some of its variables to a NetCDF file:

begin

grib_file = addfile(

ncfile_out = addfile(

; Write some data to NetCDF file.

ncfile_out->PRES = grib_file->PRES_6_SFC

ncfile_out->HGT = grib_file->HGT_6_SFC

ncfile_out->A_PCP = grib_file->A_PCP_6_SFC_acc

ncfile_out->NCPCP = grib_file->NCPCP_6_SFC_acc

ncfile_out->ACPCP = grib_file->ACPCP_6_SFC_acc

ncfile_out->CAPE = grib_file->CAPE_6_SFC

end

For some more complex examples, see the?

file I/O application examples

.

For

detailed

information

on

the

file

formats

that

NCL

supports,

see

the

Information

on supported data formats

NCL Reference Manual

.

10.

H

ow do I read/write ASCII files

Use the?

asciiread

?function to read an ASCII file and

the?

a sciiwrite

?or?

write_matrix

?procedures to write an ASCII file.

Here's an example:

;

; Assume the

; 16471 of them.

; the data, and in this case we are passing dimensions (/91,181/), so

;

; be verified by printing the dimsizes of

;

ocean1 = asciiread(

print(dimsizes(ocean1))

;

; Write out part of

; 15 integers, since (7:3:2,0:4) is a 3 x 5 array.

;

asciiwrite(

For more complex examples, see the ASCII read/write examples in the list of?

file

I/O application examples

.

11.

H

ow do I read CSV files

See the

Reading CSV files application

12.

H

ow do I write a formatted ASCII file

Use a function called?

write_matrix

?that will nicely format 2D arrays using a

user-specified format string.

Here's an example that writes a 2D array to a file called

N = 5

M = 7

ave =

std =

x =

random_normal

(ave, std, (/N,M/))

res = True

res@fout =

write_matrix

(x,

This example will produce the following output in

13.

H

ow do I read an ASCII file if I don't know how many lines it contains

There are two ways you can do this:

1.

2.

Use

asciiread

:

z =

asciiread

(

If

you

have

multiple

columns

of

data,

and

you

know

how

many

columns

there

are,

then

you

can

use

onedtond

?to

restructure

the

one-dimensional

array

returned

from?

asciiread

.

In the example below, assume there are 4 columns of data:

z =

asciiread

(

ncol = 4 ; number of columns

nrow =

dimsizes

(z)/ncol ; get number of points

znew =

onedtond

(z,(/nrow,ncol/)) ; znew will be nrow x ncol

3.

4.

5.

6.

7.

Use?

systemfunc

?to count the number of lines in the file before you read

ncol = 4

nrow =

toint

(

systemfunc

(

; of lines in a file

z =

asciiread

(

it. Assuming again that you have a data file with 4 columns of data:

14.

H

ow do I read an ASCII file that contains delimiters

There's

an?

applications

page

?that

describes

a

script

that

reads

an

ASCII

file

with

delimiters, and creates a netCDF file that contains all the data. This script assumes

that the

first line of your

ASCII

file

contains the names of

the fields, and uses these

names as variable names on the netCDF file.

15.

H

ow do I read/write binary files

Use the

app ropriate?

cbinread

,?

cbinwrite

,?

craybinrecread< /p>

,?

fbindirread

,?

fbindirwrite

,?

fbinre

ad

,

fbinrecread

,?

fbinrecwrite

,

or?

fbinwrite

?function ,

depending

on

what

type

of

binary

file you have. In addition, you may also want to use the

functions?

craybinnumrec

?or?< /p>

fbinnumrec

?for getting the number of unformatted Fortran

records in a file.

For some complex examples, see the binary read/write examples in the list of?

file

I/O application examples

.

See the?

setfileoption

?procedure for ways to change how your binary file is

read/written (big endian versus little endian, for example).

16.

O

nce I've opened a file with

attributes it contains

Use the functions?

getfilevar atts

,?

getfilevardims

< br>, and?

getfilevarnames

?to

retrieve the list of attribute names for a file variable, the list of dimension names

for a file variable, or the list of variable names for a file, respectively.

17.

I

s there a way to control whether my binary file is read or written as little

endian or big endian

Use the?

setfileoption

?procedure:

setfileoption

(

v =

cbinread

(

setfileoption

(

cbinwrite

(

18.

I

'm having trouble reading a Fortran binary file with records.

It's possible you are running into a problem where the record marker was written

with 8 bytes, and NCL is expecting the marker to be 4 bytes.

This happens if the Fortran program that created the binary file was compiled with

a

version

of

that

sets

the

marker

size

to

whatever

is

on

that

system.

As a work-around, you can recompile the code that created the binary file with the

option

to

force

a

4-byte

record,

and

run

the

code

again

to

regenerate

the file.

It is on our list to create a new option in?

setfileoption

?that will allow you to

specify the record marker size that the file was written with.

19.

I

'm

trying

to

convert

a

file

to

a

NetCDF

file

using

but

getting

an error message about violating format constraints.

If

you

are

trying

to

convert

a

file

to

NetCDF

using?

ncl_convert2nc

,

and

getting

the

following error message:

ncendef: ncid 0: NetCDF: One or more variable sizes violate format

constraints

then

you

may

have

what

qualifies

as

a

file

You

need

to

use

the?

?option

with ncl_convert2nc.

20.

W

hy

do

I

get

different

results

when

reading

my

HDF- EOS2

file,

depending

on

whether

I use the

If

you

use

the

extension,

then

NCL

uses

the

underlying

HDF4

interface

to

read

your data, whereas if you use the

the HDF-EOS2 library interface.

Using the

since the HDF-EOS2 software provides software to calculate additional lat/lon

information.

Side tip: if your HDF-EOS2 file has an

read this file as an HDF-EOS2 file simply by appending an

file

name

in

the?

addfile

?call

(you

don't

need

to

rename

the

file).

For

example,

if

your

file is called

a =

addfile

(

21.

C

an I read shapefiles using NCL

NCL

currently

does

not

have

the

ability

to

read

shapefiles.

The

NCL

development

team

is looking into this format to determine the best way these files might be supported.

Meanwhile, some users have emailed us with suggestions which we'll include here:

GDAL

?- Geospatial Data Abstraction Library

This package includes a set of command lines tools that are very handy to work

with georeferenced data format (including GeoTiff, HDF and NetCDF):? particular,

you

can

rasterize

vector

data

(like

shapefiles)

with

the

command

There's an example in this post on how to use these tools:? can also extract the

polygon

from

the

vector

file

into

an

ascii

file:?

FWTools

?packag e

-

an

open

source

GIS binary kit for Windows and Linux

You

can

use

this

package

to

convert

shapefiles

to

text.

The

command

line

utility

allows

for

type

conversion

between

a

variety

of

GIS

related

formats

(it hasn't been verified whether NetCDF or HDF is one such format).

A

user

reported

that

he

converted

a

shapefile

to

a

Mapinfo

file

(MIF

format)

which is a text format file of lon/lat pairs for each defined region. You could

then read in this text file using NCL's?

asciiread

< br>?function.

22.

H

ow can I get at vgroup data on my HDF4 file

NCL

cannot

(yet)

read

vgroup

data

on

an

HDF

file.

If

you

install

HDF4

on

your

system

(available

from

then

you

can

use

the

HDF

dumper

program,

hdp

to

get

at

the

vgroup

data

you need. You first

need

to use

on the file to

find out what

your data is in:

hdp dumpvd

This will produce output like this:

Vdata: 0

tag = 1962; reference = 34;

number of records = 1; interlace = FULL_INTERLACE (0);

fields = [Product_ID, Date_Time_at_Granule_Start,

Date_Time_at_Granule_End,

Initial_Subsatellite_Latitude,

Initial_Subsatellite_Longitude,

Final_Subsatellite_Latitude,

Final_Subsatellite_Longitude, Lidar_Data_Altitudes];

record size (in bytes) = 1746;

name = metadata; class = metadata;

number of attributes = 0

- field index 0: [Product_ID], type=4, order=80

number of attributes = 0

. . .

You

can

see

that

the

variable

is

in

vdata

0,

so

use

this

with

the hdp command to get the data only:

hdp dumpvd -i 0 -d -f Lidar_Data_Altitudes

An alternative approach to extracting the Lidar_Data_Altitudes is to invoke

the?

hdp

?command line operator from with an NCL script. For example:

; input hdf4 file name and the directory where it resides

diri =

fili =

; specify desiredvariable

vari =

; output text file name and the directory where it is to be placed

dirh =

filh =

; use NCL's

system

procedure and pipe output to file

system

(

; read the file created by the

hdp

hgt =

asciiread

(dirh+filh, -1,

hgt!0 =

hjt@long_name =

hgt@units =

hgt = hgt(::-1) ; if height need to be reordered

If you are not sure where?

hdp

?resides, from your home directory, try:

find . -name hdp -print

23.

W

hen I run my script, I'm getting an error message

Not a valid ID

You

should

only

see

this

error

if

you

are

using

version of

an

OPeNDAP-enabled

NCL.

It usually occurs when you exit the NCL script.

We

verified

this

is

a

bug

created

by

recent

changes

to

NCL's

handling

of

its

internal

cleanup

and

exit

procedures.

This

bug

has

been

fixed

and

will

be

available

in?

that

this

does not cause a problem in reading NetCDF files. When you write a file, however, you

may have to delete the file handler variable before you exit the script, by using

the?

delete

?procedure:

begin

...

f = addfile(

...write stuff to file pointed to by 'f'.

delete

(f)

end

24.

W

hen I try to access the Climate Diagnostics Center [CDC] OPeNDAP server via

NCL, I get

It is likely that the CDC OPeNDAP server does not recognize the IP address of the

machine being used. The CDC server was shut down due to security concerns. They have

opened

the

OPeNDAP

server

on

a

host

by

host

basis.

If

you

wish

to

access

the

CDC

OPeNDAP

server, please e-mail??with:

o

The IP address of the machine from which you wish access. This should

be a static IP address (not generated via DHCP). It would better if this address could

successfully undergo reverse DNS lookup to produce a fully qualified hostname.

o

o

The name of your institution.

A one-line blurb about your intended use of the data.

See the??URL for the complete story.

After registration with CDC, an NCL script might look like:

CDC_URL =

CDC_FILP =

filp = CDC_URL + CDC_FILP

if (

isfilepresent

(filp)) then

fp =

addfile

(filp,

else

print(

print(

print(

print(

print(

exit

end if

x = fp->FOO

25.

I

'm running into a problem when opening lots of files with addfile or addfiles

(

If

you

need

to

open

lots

of

files

with?

addfiles

,

then

call?

setfile option

?as

follows,

before you call?

addfiles

:

setfileoption(

The number of files you can have open at any time is dependent on your system. A

general limit seems to be around 1024.

Data handling

1.

How do I set a missing value for my data

For the variable that contains your data, set the attribute

_FillValue

missing value. For example, the following NCL code:

x = (/1,2,3,-999,5,6,7,-999,-999,10/)

x@_FillValue = -999

sets the missing value of?

x

?to -999.

The

thus

the

computations

will

behave

differently

when

they

encounter

missing

values

(like

ignore

them

or

return

with

an

error

message).

For

example,

if

you

pass

the

above?

x

?array

to the NCL function?

sum

, you will get a return value of 34.

2.

How do I determine if my data contains any missing values, or count how many

missing values it has

Use the?

ismissing

?function combined with the?

any

?function to return a single True

or False indicating whether your data contains missing values:

is_msg =

any

(

ismissing

(data)))

if(is_msg) then

print(

else

print(

end if

To count the number of missing values, use the?

num

?function:

num_msg =

num

(< /p>

ismissing

(data)))

print(

3.

How do I change the missing value to another value

If

your

data

contains

a

attribute,

simply

set

it

to

the

value

you

want,

but be sure to use the same type (float, double, etc).

For example, to change your data's missing values from 0 to -999, use:

data@_FillValue = -999

If

your data contains a

as

an

official

missing

value.

You

want

to

set

the

attribute

to

this

value:

data@_FillValue = data@missing_value

delete

(data@missing_value) ; not necessary, just cleaning up

There are other ways to assign a _FillValue attribute. See

the?

assignFillValue

?function.

4.

How do I deal with NaN (not-a-number) values in my data

The

computational

and

plotting

functions

in

NCL

cannot

handle

NaN

values

directly.

You must first convert the NaN values to missing values.

To

do

this,

use

the?

replace_ieeenan

?function

to

convert

the

NaNs

to

missing

values.

You can optionally use

isnan_ieee

?to first determine if any of your values are equal

to NaN:

If

if (any(isnan_ieee(x))) then

replace_ieeenan (x, x@_FillValue, 0)

end if

If

if (any(isnan_ieee(x))) then

x@_FillValue = or whatever value you want to use

replace_ieeenan (x, x@_FillValue, 0)

end if

5.

How do I deal with

As with NaN values, the computational and plotting functions in NCL cannot handle

To do this, use the?

where

?funct ion:

If

x = where(abs(x)., x@_FillValue, x)

Here's some more generic code to first check if x has a _FillValue attribute:

if(.(x,

x@_FillValue = default_fillvalue(typeof(x))

end if

x = where(abs(x)., x@_FillValue, x)

Data analysis

1.

How do I average values in a multi-dimensioned array

Use the?

avg

,?

dim_avg_n

?or?

dim_avg

?functions.

For

example,

in

the

following

NCL

code

snippet

where?

x

?is

a

2

x

3

x

2

array,?

y

?will

be a scalar value equal to :

x = (/(/(/2,3/), (/4,5/), (/1,2/)/), (/(/3,8/), (/1,4/), (/0,2/)/)/)

y =

avg

(x)

If you have a multi-dimensional array, say?

T(ntim,klev,nlat,mlon)

?a nd you wish to

obtain the temporal mean at each level, latitude and longitude then:

T_time_mean =

dim_avg_n

(temp, 0) (klev,nlat,mlon)

or, to retain meta data:

T_time_mean =

dim_avg_n_Wrap

(temp, 0) requires

If

you

have

a

multi-dimensional

array,

say?

prc(ntim,nlat,mlon)

< br>?containing

3-hourly

precipitation rates (eg, mm/hour) and it is desired to calculate the average

precipitation rate for an each day then:

dimprc =

dimsizes

(prc)

ntim = dimprc(0) total number of time steps

ntJump = 8 ; eight samples per day (=24/3)

prcDay = prc(::ntJump,:,:) create array with meta data

ntStrt = 0

ntLast = ntJump-1

do nt=0,ntim-1,ntJump

prcDay(nt/8,:,:) =

dim_avg_n

(prc(ntStrt:ntLast,:,:), 0)

ntStrt = ntStrt + ntJump

ntLast = ntLast + ntJump

end do

printVarSummary

(prcDay)

If the original data had been total precipitation over a 3-hour span and the daily

total

(sum)

precipitation

had

been

desired,

the

approach

would

be

exactly

the

same

except

that?

dim_s um_n

?would be used:

prcDay(nt/8,:,:) =

< p>
dim_sum_n

(prc(ntStrt:ntLast,:,:), 0)

2.

How

can

I

do

an

element- wise

minimum/maximum

between

two

arrays

of

the

same

size

Use the vers atile?

where

?function:

xmin =

where

xmax =

where

3.

How do I interpolate my 2-dimensional random data to a 2-dimensional gridded

array

There are lots of functions available for doing different kinds of interpolations

and regridding. For a list, go to the?

NCL functions and procedures page

?and look in

the?

interpolation

?or?

regridding

?sections.

The?

Random Data to Grid

?example URL has several examples. As a specific example,

two interpolation functions are?

natgrid

?or?

dsgrid2

. Here's an example:

NX = 21

NY = 21

; Define the input data arrays.

x = (/, , , , ,

y = (/, , , , ,

z = (/, , , , ,

; Define the output grid.

xc = 1./(NX-1.)

xo =

ispan

(0,NX-1,1) * xc

yc = 1./(NY-1.)

yo =

ispan

(0,NY-1,1) * yc

; Do the gridding using natgrid.

; dimensioned 21 x 21.

natout =

natgrid

(x, y, z, xo, yo)

print

(natout)

; Do the gridding using dsgrid2.

; dimensioned 21 x 21.

dsout =

dsgrid2

(x, y, z, xo, yo)

print

(dsout)

4.

How do I interpolate my 3-dimensional random data to a 3-dimensional gridded

array

First, see the question above on 2-dimensional grids.

Two functions for handling 3D random data are?

< br>triple2grid

?and?

dsgrid3

.

Here's an example of using?

dsgrid3

:

NX = 21

NY = 21

NZ = 21

x = (/, , , , ,

y = (/, , , , ,

z = (/, , , , ,

u = x*x + y*y + z*z

xo =

new

((/NX/),float)

yo =

new

((/NY/),float)

zo =

new

((/NZ/),float)

;

; Create the output grid.

;

xmin =

ymin =

zmin =

xmax =

ymax =

zmax =

ii =

fspan

(0,20.,21)

xo = xmin + (ii/(NX-1)) * (xmax- xmin)

yo = ymin + (ii/(NY-1)) * (ymax-ymin)

zo = zmin + (ii/(NZ-1)) * (zmax-zmin)

;

; Interpolate.

;

output =

dsgrid3

(x, y, z, u, xo, yo, zo)

print

(output)

5.

How

can

I

figure

out

if

a

particular

latitude/longitude

point

is

within

a

specific

area, like Europe

There's no ready-made function that will do this based on just the area name, like

the lat/lon coordinates that define

the particular area

that you're interested in, then you can use the

gc_inout

?func tion.

Otherwise, assume you have a?

crude

?regional approximation that indicates the area

you're interested in, say:

latS = 36

latN = 70

lonW = -10

lonE = 60

Further assume you have a data array

eMask

):

work = x(0,:,:)

LAT = conform( work, work&lat, 0)

LON = conform( work, work&lon, 1)

eMask = new ( (/nlat,mlon/),

eMask = where .and. .and.

.and. True, False)

copy_VarCoords(work, eMask) Copy coordinate arrays

; Clean up

delete(work)

delete(LAT)

delete(LON)

To plot an area using your new mask variable to exclude the areas you're not

interested in, see example 5 on the?

mask

?applications page.

Visualization

1.

How can I create an

Use

as

an

output

format,

when

you

open

a

workstation

for

sending

the

graphics

to.

For example, if you're using the?

gsn_open_wks

?function:

wks =

gsn_open_wks

(

2.

How can I maximize the size of a plot in the frame

If you are using one of the?

gsn_

xxx

?pl otting scripts, then set the special

re source?

gsnMaximize

?to True.

Run the following simple script and compare the two frames:

load

data =

generate_2d_array

(10, 10, , , 0, (/129,129/))

wks =

gsn_open_wks

(

plot =

gsn_con tour

(wks,data,False)

res = True

res@

gsnMaximize

= True

plot =

gsn _contour

(wks,data,res)

3.

How do I get multiple plots on a page

If your plots are all the same size, then you can use the procedure?

gsn_panel

. If

your plots are not the same size, then you can use

the?

viewport

?resources?

vpXF

,?

vpYF

,?

vpWidthF

, and?

vpHeightF

.

For some

panel examples,

see the?

panel applications page

. This page also shows you

tricks

on

paneling

plots

that

are

not

the

same

size

(

#9

?and?

# 14

?are

two

such

examples).

4.

What's a good way to generate a color map (table)

See

the

page

on?

creating

your

own

color

map

,

or

the

section

Creating

your

own

color

map using HSV valuesHSV

?color wheel and how it helps you create a color map.

For an example, try modifying the HSV values in the file

NCL to see the different results you get.

You

can

use

the

function?

gsn_ draw_colormap

?to

draw

the

current

color

map

associated

with the workstation:

load

wks =

gsn_open_wks

(

gsn_draw_colormap

(wks) Draw the default colormap

gs n_define_colormap

(wks,

gsn_draw_colormap

(wks) ; draw it.

5.

My map projection doesn't look right.

If you're plotting a subset of longitudes and getting a strange looking map, then

you may need to set the resource?

mpCenterLonF

. This will tell NCL where you want the

center longitude of the plot to be. Usually, this should be set to the halfway point

between?

mpMinLonF

?and?

mpMaxL onF

:

res@mpCenterLonF = (res@mpMinLonF + res@mpMaxLonF)/2.

6.

How do I change all of my fonts to be the same in my visualization

First

select a font

from

the NCL?

font tables

?and remember the

name, .

If

you

always

want

to

use

this

font

in

all

of

your

NCL

scripts,

then

put

the

following

line in a file called

.hluresfile

*Font : helvetica

(We?

highly

?recommend using the?

helvetica

?font for all NCL scripts, because

the?

default font

?doesn't look good when you print it.)

If NCL sees a

running any NCL scripts.

If

you want this font to apply only to a particular NCL

script, first

create a file

called?,

where?

xxx

?is

the

second

argument

to?

gsn_open_wks

.

Then,

include

the

above

line

in that file.

Any

fonts

that

you

are

explicitly

setting

in

your

NCL

script

will

override

resources

set in ~/.hluresfile or the NCL script- specific resource file.

Here's an example (with the

load

wks =

gsn_open_wks

(

y = (/1,3,-1,-2,0,2,3/)

x =

ispan

(0,

dimsizes

(y)-1,1)

res = True

res@

tiMainString

=

res@

tiXAxisString

=

res@

tiYAxisString

=

xy =

gsn_xy

(wks,x,y,res) Create and draw an XY plot.

The

! Change all fonts to times-bold

*Font : times- bold

Try

changing

in

the

file

to

other

fonts

to

see

the

results.

Use

either

a font name or a font index as listed in the?

font tables

.

7.

How can I change the text function code to something other than a colon

If

you want your?

text function

code

?permanently changed

(to, say, a '~') for every

time you run NCL, then in the ~./

.hluresfile

?put the following line:

*TextFuncCode : ~

TextFuncCode

can

be

set

to

whatever

character

you

want

(the

default

is

a

colon,

':').

This special resource can?

only

?be set in a resource file and not in an NCL script.

To change the text function code to another character inside of an NCL script, use

any one of the?

*FuncCode

resources that corresponds with the resource you are using

to set the text string.

For example, if you are setting the?

txString

?resource:

res@

txString

=

then

you

change

the

function

code

associated

with

this

resource

to

another

character

(say '~') with the

txFuncCode

?resource:

res@

txFuncCode

=

res@

txString

=

If you are setting the X axis label with?

tiXAxisString

, then the corresponding

function code resource would be

tiXAxisFuncCode

, and so on.

8.

How can I get a degree symbol in a text string

You can use?

font table 34

?with the character

load

begin

wks =

gsn_open_wks

(

res = True ; plot mods desired

res@

tiMainString

=

res@

tmXBMode

=

res@

tmXBValues

= (/ , , ,

res@

tmXBLabels

=

(/

res@

tmXBLabelFontColor

=

plot =

gsn_blank_plot

(wks,res)

end

9.

How can I get things like accents or umlauts over my text

It's not trivial, but it is doable, using?

text function codes

. Here's an example:

load

begin

wks =

gsn_open_wks

(

res = True

res@

txJust

= 4

res@

txFont

= 25

res@

txFontHeightF

= .1

;

; Explanation of the text string function codes:

;

; :F22: - Select font 22 (Helvetica bold)

; :H-15F35: - Move 15 digitized font units left and select font 35.

; The character

; :H4F22: - Move 4 digitized font units right and select font 22.

; :H-14F35: - Move 14 digitized font units left and select font 35.

; :H3F22: - Move 3 digitized font units right and select font 22.

;

str1 =

;

; Explanation of the text string function codes:

;

; :F22: - Select font 22.

; :H-16V6F35: - Move 16 digitized units left, 6 digitized units

; up, and select font 35.

; :V-6H5F22: - Move 6 digitized units down and select font 22.

; :H-13F35: - Move 13 digitized units left and select font 35.

; :F22H2: - Select font 22 and move 2 digitized units right.

;

str2 =

gsn_text_ndc

(wks,str1,,,res) ; Draw the text string

gsn_text_ndc

(wks,str2,,,res)

frame

(wks)

end

10.

H

ow

can

I

get

special

characters,

like

Greek

symbols,

in

my

graphical

text

strings

First

browse

the?

font

tables

?to

find

the

one

that

contains

the

character(s)

you

want.

Clicking on any one of the font names, like

characters for that particular font.

Once you find the character you desire, note the font table index (

math_symbols

is index 18, for example), and the corresponding character, then you can use the?

text

function code

?'F' to change the font.

For example, to get a bell symbol, you would see that character 'n' in the

symbol_set1

any text string resource to indicate you want a bell symbol. The

to the original font.

For example, to use it in a title:

res@

tiMainString

=

Note

that

we're

assuming

the

default

function

code

(a

colon,

':'),

has

been

changed

to a tilde, '~'. See the question above on changing function codes.

11.

C

an I use real PostScript fonts in my PostScript file

Essentially no, except in one situation that is discussed below. When you use

NCL?

font number 4

, you will get the stroked simplex_roman font, no matter what kind of

workstation you are going to, even a PostScript workstation. If you change

then

you

will

get

NCL?

font

21

?which

is

a

Helvetica

font

that

is

drawn

using

filled areas, . filled areas will

be put

into

the output

PostScript and not

code to use

the built-in PostScript Helvetica font.

However, there is one circumstance which will allow you to use a PostScript font

directly and not have filled areas inserted into the output PostScript. This can be

accomplished

by

setting

the

resou rce?

txFontQuality

?to

There

are

some

significant

downsides to this. First, no matter what font you select in NCL, the font used in the

output

PostScript

will

be

Helvetica.

If

you

select

a

character,

such

as

a

degree

symbol,

that is not in the Helvetica font, it will not be drawn correctly. Secondly, all text

strings

will

be

monospaced.

Thirdly,

when

plotting

to

workstations

other

than

PostScript

or PDF, you will get a very low-level stick font.

The

primary

upside

to

this

is

a

reduction

in

the

size

of

the

output

PostScript

file.

Another advantage is that you will be using a built-in PostScript font that will be

produced with high fidelity, even at very small sizes.

Here is an example of using font quality

begin

wks =

gsn_open_wks

(

txres = True

txres@

txFontQuality

=

txres@

txFontHeightF

=

txres@

txFuncCode

=

gsn_text_ndc

(wks,

frame

(wks)

txres@

txDirection

=

gsn_text_ndc

(wks,

frame

(wks)

txres@

txDirection

=

gsn_text_ndc

(wks,

.5,.5,txres)

frame

(wks)

end

12.

I

'm having trouble importing a PS file into an application like Adobe

Illustrator.

You can try

PostScript using an

application like

or

fro m?

ImageMagick

:

convert

13.

H

ow can I get the length of my axes to be the same ratio as the range of my axes

You can use the?

vpWidthF

?and?

vpHeightF

?resources to control the width and height.

You will need to calculate the range of each of your axes, and then figure out the

normalized values (from 0 to 1) to use.

Or, if you are using any of the non-map gsn_

xxxx

?plotting functions, then you can

set the special resource

gsnShape

?to True:

load

load

begin

ny = 100

mx = 50

; Generate some dummy data.

z =

gener ate_2d_array

(20,20,-136.,450.,0,(/ny,mx /))

; Open an X11 window.

wks =

gsn_open_wks

(

; Generate contour plot with default axes.

res = True

res@

gsnMaximize

= True

plot =

gsn_csm_contour

(wks, z ,res)

; Set resource to reshape axes.

res@

gsnShape

= True

plot =

gsn_csm_contour

(wks, z ,res)

end

14.

H

ow do I change the aspect ratio or size of my map or plot

For a plot that doesn't contain a map, you can use

the?

viewport

?resources?

vpWidthF

?a nd?

vpHeightF

. The viewport coordinates are

normalized; that is, they take values from to (inclusive).

For example, if you want your plot to be wider than it is high, then you might

set?

vpWidthF

?to and

vpHeightF

?= :

load

load

wks =

gsn_open_wks

(

y = (/1,3,-1,-2,0,2,3/)

x =

ispan

(0,

dimsizes

(y)-1,1)

res = True

res@

vpWidthF

=

res@

vpHeightF

=

xy =

gsn_csm_xy

(wks,x,y,res) Create and draw an XY plot.

In order to change the aspect ratio of a map, you can still use the

as above, but you must also set the resource?

mpShapeMode

?to

load

wks =

gsn_open_wks

(

mpres = True

mpres@

mpShapeMode

=

mpres@

gsnMaximize

= True

mpres@

vpWidthF

=

mpres@

vpHeightF

=

map =

gsn_map

(wks,

In

the

above

script,

to

maximize

the

plot

in

the

frame

but

retain

the

aspect

ratio,

the special resource

gsnMaximize

?is set to True.

15.

C

an I set color resources by color name rather than color index values

Yes, as long as the color you want to reference exists in your current color map.

Any place that you would normally use a color index value, you can use a?

color

name

?enclosed in double quotes. An ASCII list of the valid color names is in the file

$$NCARG_ROOT/lib/ ncarg/database/

and

a?

graphical

representation

?is

also

available

For

example,

if

color

index

2 is

red

in

your

color

map,

and

you

want

to

set

the

main

title font color to red, then you can do it one of two ways from an NCL script:

res@

tiMainFontColor

= 2 ; index value

res@

tiMainFontColor

=

Whenever

a

named

color

is

used,

the

current

color

map

is

searched

for

the

color

that

is the closest match to that color. This means that if you access a color that doesn't

have a close match in your color map, you may not get the color you are expecting. To

get around this problem, make sure the color map you use contains all the named colors

you want by adding them explicitly to your color map.

For example, if you want your general color map to have the following RGB values:

(/(/,,,(/,,,(/,,,(/,,,

(/,,,(/,,,(/,,,(/,,,

(/,,,(/,,,(/,,,(/,,,

(/,,,(/,,,(/,,,(/,,,

(/,,/)

and

you

also

want

to

include

the

named

colors

and

...

cmap = (/

wks =

gsn_open_wks

(...)

gsn_define_colormap

(wks,cmap)

...

Note:

?in order to mix RGB values with color names, you must enclose each entry in

double quotes. More information about named colors is available in the?

color fill

applications page

.

16.

H

ow can I speed up the drawing of filled contours in NCL

You can try setting the?

cnFillMode

?resource to

and

further

< p>
set

cnRasterSmoothingOn

?to

True

if

you

want

smoother

contours.

17.

C

an I change color index values or color maps in mid-frame

The main consideration in doing this is changing the background color. If the

background color is changed in mid- picture (color index 0), then in most cases at the

time you change the background color your entire frame will be overwritten with that

color. As long as you do not change the background color in mid-picture, your results

should be as desired. See the simple example below.