2
$\begingroup$

I starting a weather project where I'm using ERA5 hourly dataset (ref: https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-single-levels?tab=form). Using this dataset in python, I have now a dataframe as below.

As you can see, for one State I have n observations considering latitude|longitude and Time of observation. So, now I would like to aggregate this dataset daily, considering Admin2 and Admin1, in order to have Max Temp, Min Temp and Total precipitation. Is there a right way to do this aggregation or is this simple the max / min values from 2m temp and sum of precipitation? As you can see, by the content of my question, I'm not a meterologist. I'm just trying to build a reliable database for past historical weather data from some cities around the world.

enter image description here

Improve this question
$\endgroup$
1
  • $\begingroup$ I don't know much about ERA5 values at all. But normally max\min temperatures aren't the max\min hourly temperatures. For instance today Orlando's high today was 87 and low 63... but it's hourly max\min were 86 and 65. This is a typical result, unless you happen to be at max\min on an hour. So I'd certainly question results that extrapolate these hourlies into max\min and tried to use compare them to the actual max\mins. $\endgroup$
    – JeopardyTempest
    Apr 15, 2021 at 3:07

2 Answers 2

Reset to default
1
$\begingroup$

I don't know if ERA5 data has some weird format, as I'm really not familiar with working with such data directly. But it's vital to realize that the greatest hourly temp is rarely the daily max, and the same idea with mins. Orlando's data today: enter image description here

The max today was 87°F and the min 63°F (found from this using the 6 hour max\mins... there can be rare issues with this when those don't fall at midnight... station reports or metars give 24 hour max\min midnight to midnight... though it's only an issue when a max or min is near midnight, which is rare, typically only during airmass changes)

Whereas the top hourly was 86°F and the bottom one was 65°F. And that's a fairly typical thing to occur... they only match when the extreme temperatures happen to occur near the end of the hour (values are typically taken about 7 minutes before the hour in the US, and I believe that standard varies a bit internationally)

So I'd certainly question any work that tried to derive daily actual max\min values or compare to them using such hourly results, unless it turns out ERA5 does something unique. It's tough to imagine it's valid given it you only get one hourly value.

Precipitation measurements on the other hand are over a time period, so those values should be hourly values, and summing them should give the results for the day.

When it comes to temperature (assuming ERA5 isn't somehow strangely taking care of the discontinuity), you could try to apply some sort of rough multiplier to give a very crude better guess... but thinking you won't be able to get legitimate daily max\mins without looking for actual site data, which are for example available here for basic US historic or recent data searches, or at NCDC. Of course those are only going to work for places they have observation sites. But it doesn't seem that you can directly compare the ERA5 maxima results to actual station maxima easily.

Improve this answer
$\endgroup$
1
  • 1
    $\begingroup$ Hi Jeopardy, thanks for you kindly answer! Yes, you're right, I'll not be able to get the real max and min and I'm ok with that. Actually, I was wondering, since I not working with meterological data, if what I'm doing makes some sense or if I should do some calculations considering also the area coverage by the stations.. $\endgroup$
    – user9585960
    Apr 15, 2021 at 11:40
1
$\begingroup$

Yes, daily minima, maxima and precipitation totals are calculated simply as you suggested in your question. Take care that there aren't missing values,, but that's more of an issue calculating means. Also check whether the reported values for precipitation are in fact a null value, it's unlikely that it could be measured in such small units.

Improve this answer
$\endgroup$
2
  • $\begingroup$ There are by definition a lot of missing values when you're only taking hourly temperatures? $\endgroup$
    – JeopardyTempest
    Apr 15, 2021 at 3:44
  • $\begingroup$ It may be the best you can do with ERA5 and other reanalysis tools. Maybe a multiplier or some alternative formula may be able to better approximate a natural look. But directly using these values should likely give diurnal ranges that are systematically a bit smaller than they are in reality. $\endgroup$
    – JeopardyTempest
    Apr 15, 2021 at 4:00

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.