I would say that there are two main limitations.
First of all, since the unemployment rate measures the percentage of unemployed people within the labour force (people over 15 who are willing and able to work), it doesn't actually account for a type of unemployment called 'hidden unemployment'. That is, if people are no longer actively seeking work, having been discouraged from doing so in some way, they are not actually counted as 'unemployed.
For example, if a 60 year old has been trying for a year to get a job, but has had all of their applications rejected, they may eventually become discouraged, and stop seeking a job. Therefore they are no longer included in the labour force. Although they would prefer to be working, and would happily accept a job offer, they are not counted as being unemployed.
Secondly, the unemployment rate doesn't account for 'underemployment' - that is, people who are technically employed as they do have a job and work at least one hour a week, but who would prefer to be working longer hours. This is a growing issue as the percentage of people undertaking casual and part-time jobs increases.
As for your question about what measures can overcome these limitations, do you mean in terms of data collecting, or in reducing the impact of these types of employment?