1

I am using data on total precipitation in January 2015 for an area in Nigeria from the ERA5, but I am getting very different values using each of the following two methods below. What am I doing wrong?

  1. Daily mean statistic of ERA hourly data on single levels

  2. Monthly average reanalysis using ERA5-Land monthly averaged data

For 1, I get data as follows: one total precipitation value for each of Jan 1, Jan 2, Jan 3..., Jan 31, which I can then sum up to get the sum of total precipitation in Jan.

For 2, I get data as follows: one total precipitation value for Jan, which scaled to have an "effective" processing period of one day (see section on Monthly means in https://confluence.ecmwf.int/display/CKB/ERA5%3A+data+documentation#heading-Monthlymeans). I then multiply this by 31 to get the sum of total precipitation in Jan.

I am not sure if my methodology is incorrect, or if there is something else I should be doing

Giorgio Maarrawi
  • 11
  • 1
  • SO is a coding website, not a science website. So far this is a question about scientific methodology, not code. Can you tell us what the code issue is? – Robert Wilson Mar 13 '23 at 09:08

1 Answers1

0

ERA-land is a 9km product, whereas ERA5 standard reanalysis is a 30km product, the two are not the same, and with the different resolution you can't expect the values to be identical.

For ERA5: The hourly values are accumulated over one hour, so you need to sum them both over a day and then over a month to get monthly totals

so for those you need

cdo monsum hourly_tp.nc montotal1.nc

whereas the monthly mean value is in units of m/day and so you need to do

cdo muldpm monmean_tp.nc montotal2.nc

of course both have units of meters which you need to scale using mulc,1000 if you want mm

Instead, for ERA5-land, hourly variables from short forecasts and their accumulation are handled like seasonal forecasts and ERA-Interim, quoting from the wiki:

"the convention for accumulations used in ERA5-Land differs with that for ERA5. The accumulations in the short forecasts of ERA5-Land (with hourly steps from 01 to 24) are treated the same as those in ERA-Interim or ERA-Interim/Land, i.e., they are accumulated from the beginning of the forecast to the end of the forecast step. For example, runoff at day=D, step=12 will provide runoff accumulated from day=D, time=0 to day=D, time=12. The maximum accumulation is over 24 hours, i.e., from day=D, time=0 to day=D+1,time=0 (step=24)."

Please see the online documentation here for further information on accumulated variables in ERA5-Land.

Finally, if you wish to reconvert to hourly data from the accumulated variable you can use my answer here.

ClimateUnboxed
  • 7,106
  • 3
  • 41
  • 86
  • This answer should be correct if the data were as it should be. But I take a random point and month to test (March 1969, Latitude -40, Longitude -69). ERA5 monthly data give the value: 0.0009562549. If I take the hourly data ONLY for the 00UTC and then do the mean it gives 0.000944277, which is very similar but different... And if I take all the hours, sum them over a day and then calculate the mean, it gives 0.0119375 two orders higher. This is because ERA5 hourly data accumulates precipitation from 01 UTC by hour till 00UTC. – Santiago I. Hurtado Mar 21 '23 at 14:49
  • strange, when I checked this I got identical results and everythign worked out - I'll do a check again and get back to you. – ClimateUnboxed Mar 21 '23 at 16:14
  • The server is really slow, so I am still waiting for the monthly to download. But I did a quick check on the hourly, January 2020, extracting a point for 0/0 and I get 94 mm for the month, which seems reasonable using this: cdo -b f32 -mulc,1000 -monsum -remapnn,lon=0/lat=0 tp_hr.nc tp_hr0.nc - not sure where your point is but your values seem REALLY small so I think you are doing something wrong? – ClimateUnboxed Mar 21 '23 at 16:23
  • Hmm, but you are correct that they are not exactly the same... 94.72475 versus 94.84029, need to run, maybe ask the CDS people on the WIKI? I'm not sure why they don't match, for other variables it worked for me... Need to run – ClimateUnboxed Mar 21 '23 at 16:26
  • My value is low because is the daily mean in meters... But I have done some checks with other points and months, and I keep founding discrepancies, little ones in most cases but there are discrepancies... Normally about 0.1 to up to 10% differences. – Santiago I. Hurtado Mar 21 '23 at 16:33
  • I found that ERA5 single levels have the precipitation accumulated by hour, as you stated. But ERA5-Land has the accumulation from the start of the day. So, the last value (and biggest) of the day must be used. Nevertheless, the discrepancy persist. – Santiago I. Hurtado Mar 21 '23 at 16:41
  • yes I realized your values were in meters, but even converted to mm, they seemed small...BUT, I realize I also missed completely that you were referring to ERA-LAND, my comparison was between era5 hourly and era5 monthly, which also didn't match. I need to read up a bit more on era-land it is something I have never used myself, ah - so it seems it is higher resolution, so you can't possibly expect it to be identical to era5 – ClimateUnboxed Mar 22 '23 at 08:38
  • But I compared ERA-LAND hourly with ERA-LAND monthly. And so, they should be identical. Regarding the number is about 28mm (31 days * 1000 mm/m) is a point in Patagonia and march is the dry season. – Santiago I. Hurtado Mar 22 '23 at 12:27
  • then you should clarify that in the question, for #1 you write " ERA hourly data on single levels", no mention is made of land, so the default assumption is you are referring to ERA5 – ClimateUnboxed Mar 22 '23 at 17:45