Petabytes? Are you serious? WTF where they storing? If you have 256 sites, each logging 100MB a day of climate data (a ridiculous amount of data), that's a bit more then 108 years of storage per petabyte.
I would also point out that its not as simple as getting the data from the original source.
1) First of all, there are a lot of original sources, which ones exactly did they get data from?
2) The data was on magnetic tape and paper form, who knows what is still readable or available
3) Data gets skipped, overlooked, entered in error, etc.
4) Often times there are several steps involved in post processing the data.
This is why you keep your raw data every step along the way. Do you really think a person now, 20+ years after the fact, can go through the process ( accumulating, organizing, ingesting, analyzing, and processing data from disparate data sets) and come up with the same data set? Really?
I would seriously doubt UEA would be able to reconstruct 90% of the original data set with a high degree of certainty