How to store months from NETcdf time dimension (0 corresponding to jan)?

Question

I have a 3d NETcdf dataset and am trying to work with the precipitation variable. I would like to plot the monthly averages of it over 60 years, but am having difficulty producing a plot that is consistent with previous averaging calculations I did not involving for loops. Here is how I store the values for the january, february and take the mean:

janNES = np.empty_like(conprecip[0:720,6:-11,95:141])

for i in range(0,720,12):
     janNES[i]=(NESprecip[i,6:-11,95:141])          

mjanNES=(np.mean(janNES, axis=0))

febNES = np.empty_like(conprecip[0:720,6:-11,95:141])

for i in range(1,720,12):
     febNES[i]=(NESprecip[i,6:-11,95:141]) 

mfebNES=(np.mean(febNES, axis=0))

#values that are plotted, monthly 60 year averages
np.mean(mjanNES-mjan)
np.mean(mfebNES-mfeb)

Where 0-11 is jan to dec, so I store values from the other months similarly. I believe there is something wrong with the way I am storing the data as its producing weird results after I subtract the control(mjan and mfeb, store using this method) and take the mean again. Thank you for reading and helping.

Answer 1

The size of the first dimension of your janNES array is 720, but when you loop over time with:

for i in range(0,720,12):
    janNES[i]=...

You are only filling every 12th ( 0,12,24,... ) item, but in the end take the mean over the entire array. You probably want to create your array as:

janNES = np.empty_like(conprecip[0:720:12,6:-11,95:141])

And then loop over it with eg:

for ii,i in enumerate(range(0,720,12)):
    janNES[ii]=(NESprecip[i,6:-11,95:141])

ii now runs from 0,1,2,..,59 while i runs from 0,12,24,..,708 .

---

By the way, you don't need the for loops for this, Numpy can select the data directly by slicing the original array:

janNES = NESprecip[0:720:12, 6:-11, 95:141]
febNES = NESprecip[1:720:12, 6:-11, 95:141]

And you can even calculate the mean directly from this:

mjanNES = NESprecip[0:720:12, 6:-11, 95:141].mean(axis=0)