How to add time dimension and merge TRMM daily data using cdo or any other software?

I have downloaded the TRMM daily data which doesn't contain time dimension and I tried cdo mergetime *nc4 out.nc4 but the file size is too small if I compare to original files as combined and there is no time dimension to it and also in outfile global attributes it is showing all inputfiles name. I have 5 years data which is around 1825 files.

So, anyone can help me with this?

Thanks, Utkarsh

It might be late but I was struggling with this issue for a long time, so:

1. I have downloaded TRMM_3B42_Daily data. Each NetCDF file contains one-day data without a time attribute or axis. Therefore Climate Data Operators (CDO) will not work.
2. This code in R should read each file and obtain timely information from the file name. and then write and new NetCDF file with the time attribute.

#1.LIBRARIES=================================================================
library(ncdf4)
library(ncdf4.helpers)
library(stringr)
library(stringi)

# Set directory where TRMM data are.
setwd()

#list all .nc4 files
nc.files<- list.files(pattern = ".nc4$")

# open first file to obtain lon and lat information 
ncin<- nc_open(nc.files[1])

lon<- ncvar_get(nc = ncin,"lon")

lat<- ncvar_get(nc = ncin,"lat")

Time<-vector(length = length(nc.files))

#it better to close NetCDF files in R especially when writing them.

nc_close(ncin)

for ( i in 1:length(nc.files)){

 ncin<-nc_open(nc.files[i])
     
 med<-ncvar_get(nc = ncin,"precipitation")
    
 raw_time<-unlist(str_split(nc.files[i], "[.]"))[2]
    
 Time[i] <- as.Date(strptime(raw_time, "%Y%m%d"))
    
  nc_close(ncin)

 #=======================================================================
 # Writing NetCDF file of the  
    
 dimLON  <- ncdim_def('lon', "degree",
                                             longname='longitude',
                                             vals=lon)
    
 dimLAT  <- ncdim_def('lat', "degree",
                                             longname='latitude',
                                             vals=lat)
    
 dimTime <- ncdim_def('time', "days since 1970-01-01 00:00:0.0",
                                             longname='time', 
                                             calendar="standard",
                                             vals=Time[i],unlim=TRUE)
 # define variables
 fillvalue <- 1e32
    
 dlname <- "Precipitation"
    
 var_def <- ncvar_def(name = "precip",
                                             units = "mm day -1",
                                             list(dimLAT,dimLON,dimTime),
                                             fillvalue,dlname,prec="double")
    
    
 #netcdf file name 
 ncname<-paste0("precip_TRMM_",raw_time,"_time_.nc")


 #netCDF file location 
    
 # where to write the netcdf file
 ncpath <- ""
    
 ncfname <- paste0(ncpath, ncname)
    
 #create netCDF file and put arrays

 ncoutput<-nc_create(ncfname,list(var_def),force_v4=TRUE,verbose =F) 

 #put variable 
    
 ncvar_put(ncoutput,var_def,med)
    
 #put additional attributes into dimension and data variables

 ncatt_put(ncoutput,"lon","axis","X") 

 ncatt_put(ncoutput,"lat","axis","Y")

 ncatt_put(ncoutput,"time","axis","T")
    

 nc_close(ncoutput)
    
 # I Use R CMD BATCH and this just for control
 print(c(i,raw_time))
    

 rm(var_def,ncoutput,med)

    
}

3. After this Climate Data Operators (CDO) can be implemented and shall work :) (e.g., to merge all the file in a folder: cdo mergetime *.nc outfile.nc