Hi Martin,
Thanks for the report about sampleij; we'll look into how it interacts
with the missing-value flag. I've created an example of this and it
does look like a bug. Until we can fix this, a workaround would be to
explicitly give the correct missing-value when you define your variable
that calls sampleij:
yes? LET/BAD=`temp,RETURN=bad` tempij = SAMPLEIJ( temp, ipts,
jpts)
For your question about the valid_range attribute, Ferret doesn't
automatically do anything with this attribute, but you can use it
yourself in Ferret commands using the attribute-handling syntax in
Ferret:
yes? let temp_range = temp.valid_range
yes? list temp_range
VARIABLE : TEMP.VALID_RANGE
FILENAME : my_test_data.nc
SUBSET : 2 points (X)
1 / 1: -10.0
2 / 2: 100.0
Martin Schmidt wrote:
Hi,
I have a data file that has missing values defined like
temp:missing_value = -1.e+20f ;
Ferret 6.1 deals with this definition almost corrrectly. One exception
is sampleij.
Selecting data at some points with samplexy and writing them to a file
gives correct missing values.
Using sampleij instead, missing and fill value is set to
TEMP_IJ:missing_value = -1.e+34f ;
TEMP_IJ:_FillValue = -1.e+34f ;
The original missing values are used as valid data now and additional
measures are needed for a correct data treatment.
I think this is a bug?
The original data have also an attribute
temp:valid_range = -10.f, 100.f ;
Does ferret read this?
Best,
Martin Schmidt
|