Wrong TypeOfData data about sdslite

[vassilyl]

Hi Auberginn, as far as I understand, the bits of each float32 were written to the file as two 16-bit words. You will need to read the variable as Int16, an then to manually convert each pair of ints to a float, i'm afraid. Class BitConverter may help you with this.

[Auberginn]

Hi vassilyl, I'm not sure I understand the pairing Int16 that you mention. The table data in the image corresponds to a partial printscreen cut of the grid cells values shown by the DataSet Viewer Table. When I use the GetData in C#, it returns the same values, which -19725 corresponds to the first cell of the grid located at index [0,0,0] of the Int16 array (where the structure is [latitude, longitude, value]). The value should be in Kelvin and around 245°C.

[vassilyl]

I may have misunderstood you.

... but the data is actually a float32

-- what does it actually mean? Can you show actual metadata on the variable (sds meta <ncfile> <var>)?

[Auberginn]

My mistake, just float.
DataFlow Shell version 1.3.14517.0
DataSet Core version 1.3.14517.0

$dataset = Y2010M01.nc
Attached new DataSet at $dataset with id=1
show $dataset meta t2m
$dataset net.pipe://localhost/microsoft/research/science/data/StorageService/bebd021e-5c80-40cd-8f7f-d6c654d83492/
msds:nc?file=E:\era5land\Y2010M01.nc
[-6] t2m of type Int16 (time:744) (latitude:231) (longitude:441)
Name = t2m
_FillValue = -32767
units = K
long_name = 2 metre temperature
add_offset = 260.533688291066
scale_factor = 0.000845463766003984
missing_value = -32767

[SravaniBeeram]

I am having the same issue..@Auberginn can you please let me know if you were able to resolve this?

[vassilyl]

SDSLite is transparent in regards to attribute conventions except for "missing_value".
In this particular case the library returns int16 values stored in the file, e.g. -19725. The scale_factor and add_offset metadata on the variable indicates how to convert the stored values to "physical" ones: -19725 * 0.000845463766003984 + 260.533688291066 = 243.8569155066374.
See Appendix A: Attribute Conventions at https://www.unidata.ucar.edu/software/netcdf/docs/attribute_conventions.html.