Summation along a dimension

Suppose you have a netcdf file with temporal data of say evaporation (in meters). And you want to know how much water evaporates during a year. And also you do not want to loose time writing a program for this. Fortunately we have NCO. I did the procedure for the ERA-Interim reanalysis file downloaded from the ECMWF site. I picked the year 1985. The header of the file looks like this:

	~# ncdump -h evap-1985.nc
	netcdf evap-1985 {
	dimensions:
	longitude = 480 ;
	latitude = 241 ;
	time = UNLIMITED ; // (730 currently)
	variables:
	float longitude(longitude) ;
	longitude:units = "degrees_east" ;
	longitude:long_name = "longitude" ;
	float latitude(latitude) ;
	latitude:units = "degrees_north" ;
	latitude:long_name = "latitude" ;
	int time(time) ;
	time:units = "hours since 1900-01-01 00:00:0.0" ;
	time:long_name = "time" ;
	short e(time, latitude, longitude) ;
	e:scale_factor = 1.59076922019745e-07 ;
	e:add_offset = -0.00268823914932497 ;
	e:_FillValue = -32767s ;
	e:missing_value = -32767s ;
	e:units = "m of water" ;
	e:long_name = "Evaporation" ;
	// global attributes:
	:Conventions = "CF-1.0" ;
	:history = "2012-01-06 16:13:54 GMT by mars2netcdf-0.92" ;
	}

view raw ncdump_output.txt hosted with ❤ by GitHub

Here is the description of the procedure that I used:

1. Calculate the mean of the variable along the specified dimension (time in my case).

   ~# ncra -d time,0,729 -v e evap-1985.nc mean_evap_1985.nc
   

   Here I was not sure whether I should take care of the scale_factor and add_offset. The answer is no, because nco takes care of it for you.
2. Then, since sum = mean * ntimes, I use the mean to get the sum. Minus is because of the model convention, they consider the flux negative if it is upwards.

   ~# ncap -O -s "e=-730*e" mean_evap_1985.nc sum_evap_1985.nc
   

3. So we have our resulting field in the file sum_evap_1985.nc, which can be viewed using ncview. The field looks as shown below. I do not know how to save the legend in ncview, so the figure is not very informative. The result corresponds to the means given by ECMWF: E-P and P.
   
   
   

Data source: "ECMWF ERA-Interim data used here have been obtained from the ECMWF data server."

ECMWF variables description

The table of variable names and units in the ECMWF files: http://www.ecmwf.int/publications/manuals/d/gribapi/param/