As with all data, care must be taken not to assume too much with regards to what information reanalysis products can provide. In this section, I provide a few reading assignments that revisit the pros and cons of how reanalysis data sets can be useful.
In this lesson, we have been playing around with reanalysis data sets as a means of learning about NetCDF and GRIB file formats. All along I have cautioned you about not treating reanalysis data as an observational data set -- they are not. Yet, reanalysis data sets can (and do) provide us with remarkably accurate records of the past atmospheric state, in a temporally coherent and spatially gridded format. While I wouldn't use reanalysis output to construct a daily record of temperature for a certain location, I might consider using such data to reconstruct a reasonably accurate climatology, especially when observations are lacking.
The Climate Data Guide from NCAR/UCAR [1] has a wealth of global data sets, including an excellent series of pages on reanalysis data sets. Rather than reinvent their thoughts and findings, I will send you there in the following reading assignment.
From the Climate Data Guide from NCAR/UCAR read...