Earthdata Forum

Under what circumstances should I use the missing_value attribute in my Earth Science data product?

Hi @pleonard Could you please provide us with a little more information? I.e. which dataset are you using and what are you trying to accomplish?

I am producing a gridded soil-moisture product in netCDF.

The data in each grid cell can be either high quality, low quality or missing.

There are many reasons for why data can be missing from a grid cell, and I think it is worth noting why the data are missing in each case.

How should I represent these different cases of missing data in my product?

Hi @pleonard, in general the netCDF CF convention (http://cfconventions.org/) provides two attributes that you can use to specify a single missing value, one is _FillValue and the other is missing_value. But _FillValue is preferred as your data will get better support from the netCDF library. Here are some example cases that _FillValue can be use in: 1) you want to assign a _FillValue to grid cells falling within water bodies; and 2) if your soil moisture data is a time series but you couldn't produce data at certain time periods due to sensor failure or other reasons.

To specify quality level of your data, the CF convention suggests you define ancillary variables. Here are some examples for your reference:
- Ancillary quality flag data: http://cfconventions.org/Data/cf-conventions/cf-conventions-1.10/cf-conventions.html#quality-flag-ex
- A flag variable, using flag_values: http://cfconventions.org/Data/cf-conventions/cf-conventions-1.10/cf-conventions.html#flag-variable-flag-values-ex

Hope this helps.