Probably because it stores the url and session and then tries to access to RESTfull services.

This has 2 main drawbacks:
1) It causes many exception emails which might obfuscate the truly important ones
2) Google might be downgrading the page as it can't access it, and with it every other page in the server not sure though)

The solution might be to trim the robots.txt so that it may guide the bot a little better (see http://www.google.com/bot.html)

This is what I get per email:
User Agent: Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)
Method: GET
ESKE Session Id: 539eff98-cc14-40fd-a76a-6d7253d580b0
Server Port: 80
Server Name: albedo2.dkrz.de
Context Path: /esgcet
URL: /esgcet/browse/viewDataset.htm
Query String: datasetId=009a9947


org.springframework.web.bind.ServletRequestBindingException: Required UUID parameter 'datasetId' with value of '009a9947' is not a valid number; nested exception is java.lang.NumberFormatException: UUID has to be represented by the standard 36-char representation

(Thanks for adding the HTTP headers to the email, this helps a lot.)