HI! Not that difficult.

Add both variables and calculate the daily mean and multiply the mean with the number of seconds per day.

(in general the output is on kg per seconds which is equal to mm/seconds.)

A day has 86400 seconds, so type:

cdo -mulc,86400 -daymean InFile Outfile

Afterwards you will have daily sums of precipitation.

Can't CDO subtract two .nc files?

For example, given wrfout_d02_2010-09-08_00:00:00, wrfout_d02_2010-09-08_06:00:00, wrfout_d02_2010-09-08_12:00:00, wrfout_d02_2010-09-08_18:00:00, wrfout_d02_2010-09-09_00:00:00.

Ignoring time coordinates, to get total precipitaton for that day, Isn't it simply just subtracting RAINC/RAINNC of wrfout_d02_2010-09-08_00:00:00 and wrfout_d02_2010-09-09_00:00:00?

To do this automatically, maybe CDO coupled with shell sripting is necessary.

Sorry if this info is not so useful.

HI! Not that difficult.

Add both variables and calculate the daily mean and multiply the mean with the number of seconds per day.

(in general the output is on kg per seconds which is equal to mm/seconds.)

A day has 86400 seconds, so type:

cdo -mulc,86400 -daymean InFile Outfile

Afterwards you will have daily sums of precipitation.

try

cdo daysum infile_6hour_precip outfile_day_precip [mm/day]

try again

I do not know what went wrong doing cdo monavg (your output unit would be like your input unit)

cdo monsum infile_6hour_precip outfile_mon_sum_precip

this should work, since you like accumulated data [mm/Month]

I voted for Tim's for Tim's answer, because the first step in dealing with totals, such as precipitation, evaporation and radiations, is to make sure one has the right unities. After that, the procedure is simple.

Thanks all for your responses. I have converted the 6 hourly data to daily ones by using CDO command (e.g. cdo -settaxis, 1999-01-01-01:00:00:00,1day -dayavg ifile ofile). Tim and Nisia, the precipitation (rainc and rainnc) in WRF output is already in mm and I think no need to multiply it by the number of seconds in a day or so. I have now realized that the precipitation from both rainc and rainnc are accumulated precipitation, so to find for example today's precipitation you have to subtract the accumulated precipitation of yesterday from today (i.e. rainnc today= rainnc(t)-rainnc(t-1). This actually works and the result is now realistic.   Please forward me if I have made some mistakes in my procedures above that compromise the results and my analysis. 

ok! I am no using WRF, so I am surprised that there is some accumulation during the run. But I have seen that you are able to change this in the namelist, so you can reset precipitation to 0 after each output. But you are absolutely right, there is no average precipitation rate like mm/seconds, something I would prefer.

Coming back to cdo: you can duplicate the dataset temporally and shift it by one day. Afterwards you can subtract both time series and remove the tmp-data. (In just one cdo command it is like this:

cdo -selhour,xyz -sub -infile -shifttime,+1day infile outfile (hope I am wrong with +)

all values from infile will be set to the next day, so on the timestep i you will have the data from (i-1). Subtracting both files (cdo subtracts values from the same date) gives you 24h precipitation sums (even periods spanning over two days, e.g.: 12:am to 12am). Select the tilmestep xyz you are interested in (most interesting 00:00) and you are done without any temp. file to remove.

Use free package called ncks to extract the data you want, and then concatenate the individual  files into monthly averages, which can also be done on ncks or another package like matlab

Hi. How did you come up with the computation/program for 6 hourly rainfall for

WRF?

Hello

Can someone show me  shell script or how best to use cdo to get daily rainfall totals from WRF output. I have run WRF ,output is 6hourly and  I need to get daily total values since WRFOutput rain is cumulative