My questions are:
1) Is there any other difference between these same variables from different data sets?
2) If I look at these variables associated with specific extreme event, what data sets for these variables are more appropriate to use? I am also simultaneously looking at the large-scale meteorological features from “M2T1NXSLV” (for example, time-averaged single-level temperature and wind speed). Does this justify the use of time-averaged TQV from “M2T1NXSLV” and “RH” from “M2T3NPCLD”?
Any suggestion is appreciated.
The simple answer is that sometimes, a variable is duplicated for convenience to the user downloading files and other times, the data is saved at different temporal characteristics for specific purposes.
You can find part of the answer to this in the MERRA-2 File Specification document. https://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/docs/
Take TQV for example. As you say, it is saved in two different file collections. If you look at the File Spec Doc, you can learn how to parse the unique alphanumeric identification for each file, and you will find that one is saved instantaneously, and the other is time averaged. Instantaneous is needed in several purposes, for example, computing the total time rate of change of the vertically integrated water vapor, while time average is more useful in comparing with the weather with other time average variables. Which you choose to use depends on your purpose. Also, sometimes RH is needed to compare to dynamic quantities (ASM) or cloud data analysis (CLD), so just a convenience and time saver for the user.