1) Since the official release of the ISI-MIP data we have put a lot of effort into testing and documenting the bias correction method. We have found that the algorithm currently used does not do exactly what we want: In some cases, due to a technical problem, the width of the frequency distribution of daily data is not corrected. That means that in these cases, the variability on the daily timescale is closer to the original GCM data. The problem is related to the precipitation data and to a minor extent to the other variables that are corrected in a multiplicative way. Temperature data are not affected.
The problem does not affect the correction of the monthly
means, and the input data that is currently being used still represents
plausible climate scenarios. The current ISI-MIP data set will remain
valid.
We have corrected for the error and included some
improvements in the code. Once we have finished testing and documenting
this improved version, we will provide you with relevant plots and
details in order to compare the results of both versions; however, we
wanted to share our findings with you already now, so that you are aware
of the shortcomings of the method that is currently used. The technical
problem, its implications and the corrected method will be described in
a paper we are currently working on. It should become the reference for
the ISI-MIP data. We will also provide the updated data. They can be
used for a sensitivity analysis but will not replace the current ISI-MIP data set that will stay the reference set for the intercomparisons.
2) To those who are attending the workshop in Reading next week, we
would like to encourage you once more to bring with you any plots, data
tables or other material, even very preliminary, that might potentially
be useful for the discussions. Looking forward to meeting you there!
