Important updates from the ISI-MIP team // New data // Issues with existing data // Protocol changes

Posted by Lila Warszawski on May 23, 2012

We have some important information to convey to you about new and existing ISI-MIP climate data and changes in the ISI-MIP protocol.

New data

We are happy to let you know that the SSP data are available now for download from the ISI-MIP archive. The zip file available there contains the data and an associated description. Please use the data there instead of downloading them yourself from the SSP database, since we would like to avoid confusion due to multiple available datasets in the database.

In addition bias corrected data for all four RCPs for the HadGEM2 model (including 6.0 and 2.6) are now ready for download.

Issues with existing data
We have already received feedback from the LPJmL group at PIK concerning the ISI-MIP climate input data we provided last week for HadGEM2: There are still some problematic grid points showing extremely high daily precipitation values above 400 mm/day. Such values are clearly unphysical but the problem is sufficiently limited that we think that the provided data (precipitation and snowfall version 1) can be used. However, as the problem can be fixed relatively easily we decided to provide a new clean version for these data in this release (precipitation and snowfall version 2).

All new data will also include the fix. If you have already started your simulations based on the data we provided last week and if it is inappropriate to re-start them given the time constraints, that is fine. All original data will stay valid from now on. Only if possible we would recommend to use the data that will be available by tomorrow. This issue also effects the spin-up data. We also provide new spin-up data based on the new precipitation and snow fall data.

Due to other feedback from modelling groups we are already aware of another issue related to relative humidity that might be higher than one based on the specific humidity data we provide. The bias correction method does not ensure this consistency. Within the attached fact sheet we discuss that point and propose two options to fix it. Could you please comment on that point in case you need relative or specific humidity for you model runs. A quick response would help us to decide for the most appropriate option. Therefore please respond until Tuesday, May 29!

Unfortunately, we cannot exclude that you might find other issues related to the bias corrected climate input, that we have not noticed so far. Part of our quality control has to go in parallel to your application because of the time constraints. Your feedback is very important for the ISI-MIP process. Therefore we will initiate an interactive platform where you can comment on the data. We will initiate this next week and will let you know about how it will work. We will reply to your comments and propose ways to fix the problems while the original data will stay valid. All modelling groups should apply the fixes as long as they have not started their model runs or as long as there is enough time to re-run the model.

We see that approach as a compromise between getting "clean" harmonized input data and not delaying the process.

Within this process we need a system how we name the different versions as the old ones will stay available. We decided to count the variable specific updates. In case of precipitation and snowfall that means that the old data will be stay available as version 1, while the new will be called version 2. As the other variables have not changes the will be called version 1 from now on. If you want to download new data to run your model you always have to choose the variable set with the highest version number. Thus currently you have to combine version 1 temperature with version 2 precipitation data (see attached Table). We will document all modifications on our website by updating the attached table and will let you know via email.

Protocol changes

We would also like to make you aware of some changes to the ISI-MIP protocol.  The newest version has been uploaded to our website ( under 'Simulation Protocol') and all changes have been highlighted in yellow.  In particular, new spin-up data has been provided, based on de-trended and calibrated WATCH forcing data. Please consult the section on spin-up (p. 10, section 4.1.2) in the latest version of the protocol for detailed information and please be sure to use the latest version of the spin-up data where possible.

Most of the additional changes are very minimal, however some significant changes have been made in gridded biophysical models section (starting on p. 26, section 6.3.1) for those taking part in the agricultural intercomparison.  We ask you to be please consult the newest version of the protocol before continuing your model runs.

As always, a log of amendments to the protocol can be found under 'Simulation Protocol' on the website.