2
$\begingroup$

I'm in doubt of some values for future scenarios of climate change that were downloaded from http://www.worldclim.org/version2. For example, the NCAR circulation model for bio_5 variable at optimistic scenario (Max Temperature of Warmest Month) to 2050 is 335.33.

Other similar calculations with other scenarios, circulation models and another future year, returns values ranging from 55 to 103 for the max temperature of warmest month. This appears wrong. Am I reading the data correctly?

Improve this question
$\endgroup$
8
  • 6
    $\begingroup$ sure would be nice if you cited or linked to the source of information you are intrepreting. $\endgroup$
    – f.thorpe ♦
    Commented May 4, 2017 at 3:14
  • 7
    $\begingroup$ Those numbers are impossible and I'm quite sure you're reading the data wrong. Without source data, as @farrenthorpe points out, your question is meaningless. $\endgroup$
    – userLTK
    Commented May 4, 2017 at 4:17
  • 2
    $\begingroup$ Also, 335.33 as a temperature is far too specific. No climate model is likely to predict temperature to 1/100th of a degree C for a specific year. 335.33 is likely a measurement of some kind, not a temperature prediction. $\endgroup$
    – userLTK
    Commented May 4, 2017 at 6:21
  • 4
    $\begingroup$ @userLTK I suspect that the OP is quoting raw numbers from CMIP5-like outputs, in which case the precision is just because they're stored as 32-bit floats. They're also possibly quoting an individual daily max temperature from some grid box on some day, rather than a max monthly mean, so values > 50ºC are possible. For example, a random dip into the CMIP5 data gives me 57ºC for CCSM3 and 95ºC for HadCM3 under RCP4.5. $\endgroup$
    – Deditos
    Commented May 4, 2017 at 10:11
  • $\begingroup$ @kingledion I think the question can be useful. It is common for scientific values to be stored as scaled integers, in case of NetCDF with scale_factor defined as a variable attribute. Emanuel is not the first and won't be the last with this problem. $\endgroup$
    – gerrit ♦
    Commented May 5, 2017 at 10:54

1 Answer 1

Reset to default
3
$\begingroup$

My professor gives the solution, as we can see at the site were the climatic variables came from http://www.worldclim.org/formats1:

Please note that the temperature data are in °C * 10. This means that a value of 231 represents 23.1 °C. This does lead to some confusion, but it allows for much reduced file sizes which is important as for many downloading large files remains difficult. The unit used for the precipitation data is mm (millimeter).

(Edited in by Gerrit): storing (climate) data with offsets and scale factors to reduce file size is common practice, as scaled integers can easily be less than half the size of floats. When data are stored in standard formats sch as NetCDF or HDF-5, there is probably a variable attribute documenting this. When data are stored in a homegrown format, check the documentation carefully for any mention of offsets and scale factors that need to be applied.

Improve this answer
$\endgroup$
2
  • 1
    $\begingroup$ Finally somebody that can read metadata! Oh, the wonders of knowing how to read the metadata in a data file before overreacting! $\endgroup$
    – arkaia
    Commented May 5, 2017 at 11:31
  • $\begingroup$ @aretxabaleta Déjà vu... $\endgroup$
    – Pont
    Commented May 5, 2017 at 12:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.