Let's say I'm extracting precipitation from a single pixel or region from a reanalysis (MERRA-2, ERA-5, Interim) dataset.

I can infer a distribution for the daily data (Gamma, Pearson-III, e.g. link), but I know that it will be conservative since I'm basing the distribution on a reanalysis product (smoothed, spatially averaged out).

Is there an appropriate method to approximate the "real/actual" local precipitation distribution from a reanalysis dataset?

  • $\begingroup$ you can only divide the pixel value by the grid box size and get an approximate value. $\endgroup$
    – user1066
    Commented Oct 18, 2019 at 4:55
  • 2
    $\begingroup$ I meant grid box area not grid box size. Reanalysis is not the correct way to get local precipitation distribution. For that you need rain gauges. $\endgroup$
    – user1066
    Commented Oct 18, 2019 at 5:02
  • $\begingroup$ Be aware that even rain gauges can have large errors, in particular automated ones. A leaf gets stuck and suddenly it "stops raining". It happens. $\endgroup$
    – gerrit
    Commented Oct 18, 2019 at 8:36


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.