Project

General

Profile

Defintion of growing season length in eca_gsl differs from the one used by ECA

Added by Stefan Fronzek over 13 years ago

Dear colleagues,

the website of the European Climate Assessment (ECA) gives a definition of the growing season length (GSL) at http://eca.knmi.nl/images/indices/GSL.png. There, the start of the growing season is defined to be the 1st occurrence of at least 6 consecutive days with TG > 5oC.

This definition differs from the one implemented in cdo (https://code.zmaw.de/embedded/cdo/1.4.5/cdo.html#x1-6530002.16.9), where the start of the growing season is restricted to be within the first 6 months of the year.

cdo gives a missing value for the growing season length for a time series of TG that exceeds 5oC only after the end of June, whereas with the ECA definition the growing season could be several weeks in July and August. Such a situation is quite often the case in higher latitudes or altitudes.

I am struggling how to best deal with these missing values when calculating GSL for northern Europe and would prefer a definition of the start of the growing season that is not restricted to the first 6 months.

What is the reason why GSL is apparently defined differently in cdo? Since the climate indices' names in cdo all start with "eca_", I assumed they would be implementing exactly what has been done by ECA.

Best regards,
Stefan


Replies (5)

RE: Defintion of growing season length in eca_gsl differs from the one used by ECA - Added by Ralf Mueller over 13 years ago

Hi Stefan,
thanks for the hint. Unfortunately I did not used the official definition for rechecking the implementation. Instead I added support for southern hemisphere, where the end lies in the first 6 month of the year. I created an issue (#396) for your report.

To compute the correct GSL, you should propably use the consecsum operator. It has been developed to gain deeper insight into periods of consecutive timesteps. See #14 for usage help. Here is a little workflow for manually computing the GSL beginning:

  1. Mask the data for temperatures above 5oC, e.g. with
    cdo gtc,5 inFile maskFile
    for the start of the period.
  2. Compute the lenght of each period of consecutive timesteps with value 1:
    cdo consecsum maskFile periodsFile
  3. Check for the first timestep with value 6:
    cdo eqc,6 periodsFile startOfGSLFile

regards
ralf

RE: Defintion of growing season length in eca_gsl differs from the one used by ECA - Added by Stefan Fronzek over 13 years ago

Hi Ralf,

thanks for pointing to these operators. I couldn't find a way in cdo to check for first timestep where a conditions is met (i.e. the first occurrence of 1 in the above file startOfGSLFile). Is there such an operator?

regards,
Stefan

RE: Defintion of growing season length in eca_gsl differs from the one used by ECA - Added by Ralf Mueller over 13 years ago

Hi Stefan,
such an operator would produce a file with grids and time axes similar to the input file; data would be a single '1' at the corresponding timestep and missval otherwise. Is that, what you intend?

Could I send you a patched cdo tar ball for checking the eca_gsl operator?

regards
ralf

RE: Defintion of growing season length in eca_gsl differs from the one used by ECA - Added by Stefan Fronzek over 13 years ago

Hello Ralf,

I would want an output similar to the output from eca_gsl with a single time step that holds the growing season length and the day number of growing season start.

What would be included in the patch that you offered to send? I would be very keen on getting that, if you managed to implement a version of eca_gsl that calculates the growing season start without restricting it to the first 6 months.

Thanks for your help,
Stefan

RE: Defintion of growing season length in eca_gsl differs from the one used by ECA - Added by Ralf Mueller over 13 years ago

Stefan Fronzek wrote:

Hello Ralf,

I would want an output similar to the output from eca_gsl with a single time step that holds the growing season length and the day number of growing season start.

That's allready there. eca_gsl computes 2 variables: growing season lenght and growing season start (day of the year)

What would be included in the patch that you offered to send? I would be very keen on getting that, if you managed to implement a version of eca_gsl that calculates the growing season start without restricting it to the first 6 months.

I'd like to send you a complete tar file, which you can compile an install just like the regular release.

Further work will be logged to #396.

    (1-5/5)