As far as both variables are quantitative, you can use Pearson correlation. From my experience, it is better to use individual observations. Both variables (parameters) for which you're checking the relationship should be quantitative/continuous.

Make sure that data is normally distributed and there is no outlier in the data set.

Thank you Wabi Bajo It gives sense. The issue, however, is many people (reviewers) recommend to use mean values instead of individual observations.

You are right. Either way is possible. The good thing is that the assocation between the two variables doesn't change if you use individual observation or means of the two parameters. It gives you almost the same result

Yes, the nature of the relationship will not change, but when you use mean values, the number of observations will reduce and the correlation may become non-significant. That is what I experienced.

Just use mean values. I think using individual observations is time consuming.