TypeError when creating netcdf file

Question

I am trying to creat a netcdf file. In the first step all I want is to read the data from a netcdf file and write it in a new netcfd file I have a data-array, that has the dtype float64 and three dimensions (time, latitude, longitude).

When I run the script I get the following error:

TypeError: ufunc 'subtract' did not contain a loop with signature matching types dtype('U32') dtype('U32') dtype('U32')

I understand, that somehow the datatypes don't match. But I don't see where the problem is.

fid = Dataset(Dir, "r")
t2m = np.array(fid.variables["t2m"])

#create new netcdf file
new_file = "ERA_t2m_2011to2017"
dataset = Dataset(new_file, "w", format="NETCDF3_64BIT_OFFSET")

#create dimensions
longitude = dataset.createDimension("longitude", 1200)
latitude = dataset.createDimension("latitude", 101)
time = dataset.createDimension("time", None)

#create variables
newvar = dataset.createVariable("longitude", np.float32, ("longitude"))
newvar.long_name = "longitude"
newvar.units = "degrees_east"
newvar.axis = "longitude-axis"
newvar[:] = np.array(fid.variables["longitude"])

newvar = dataset.createVariable("latitude", np.float32, ("latitude"))
newvar.long_name = "latitude"
newvar.units = "degrees_north"
newvar.axis = "latitude-axis"
newvar[:] = np.array(fid.variables["latitude"])

newvar = dataset.createVariable("time", np.int32, ("time"))
newvar.long_name = "time"
newvar.units = "hours since 1900-01-01 00:00:0.0"
newvar.calendar = "gregorian"
newvar.axis = "time-axis"
newvar[:] = np.array(fid.variables["time"])

newvar = dataset.createVariable("t2m", np.float64, ("time", "latitude", "longitude"))
newvar.long_name = "2 meter temperature"
newvar.units = "K"
newvar.FillValue = "-32767"
newvar.scale_factor = "0.00137038771743"
newvar.add_offset = "238.393472216"
newvar.missing_value = "-32767"
newvar[:] = t2m

Maybe an important information is also, that

fid.variables["t2m"]

yields the following output:

class 'netCDF4._netCDF4.Variable'

int16 t2m(time, latitude, longitude)

scale_factor: 0.00135205961432

add_offset: 238.864955074

_FillValue: -32767

missing_value: -32767

units: K

long_name: 2 metre temperature

unlimited dimensions: time current shape = (2920, 101, 1200) filling off

So the 2tm variable seems to be the dtype int32 at first, but it turns to be float64 when I read it in. However both creating the variable 2tm as float64 or as int32 leads to the same error.

Can you help me out here? Thank you.

Seems odd that the 2-meter temperature is an integer. But anyhow, please add the full trace, at least the part which mentions which line results in the TypeError. So far I don't see anything wrong with your code, usually, Numpy/NetCDF4 is quite flexible in casting things to the correct data type. — Bart
– Bart, Commented Nov 21, 2017 at 7:39

Community · Accepted Answer · 2020-06-20 09:12:55Z

1

The problem is in how you specify:

newvar.FillValue = "-32767"
newvar.scale_factor = "0.00137038771743"
newvar.add_offset = "238.393472216"
newvar.missing_value = "-32767"

scale_factor and add_offset should be numerical values, not strings. FillValue and missing_value can be strings (at least, in my small test), but you probably want to make them numerical as well.

From the documentation:

When data is written to a variable it is packed using::

data = (data - self.add_offset)/self.scale_factor

This is where the error comes from, for example:

np.array(1.)-'string'

Results in the exact same error as you get.

edited Jun 20, 2020 at 9:12

CommunityBot

11 silver badge

answered Nov 21, 2017 at 10:55

Bart

10.4k5 gold badges54 silver badges83 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

TypeError when creating netcdf file

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related