I'm trying to find and use data involving total cloud cover %. I have downloaded data from NCEI/NOAA specifically here and I'm plotting it like this:

import xarray as xr
import numpy as np
import matplotlib.pyplot as plt
import pygrib
import cartopy.crs as ccrs
import cartopy.feature as cfeature

#loading dataset ForecastTestDate.grb2
#filtering by atmosphereSingleLayer level
ds=xr.open_dataset('gfs_3_20220402_0000_000.grb2', engine='cfgrib', filter_by_keys={'stepType': 'instant', 'typeOfLevel': 'atmosphere'})
#showing level info

#tcc is one of the data variables for atmosphere
#tcc = total cloud cover
fig.colorbar(im, orientation='vertical')

I get this plot: map of world showing cloud cover

Obviously this is a lot of cloud and it is a lot more than is shown on zoom earth for the same day. I assume that that the total cloud cover % is inverted? Is this common practice and/or will it be the same for the other datasets from the same database?

  • 5
    $\begingroup$ Why do you think it's inverted? What is "zoom earth"? Could you add coastlines for easier interpretation? Did you compare with some satellite images to verify if you're interpreting the data correctly? On average, Earth is about 70% covered in clouds, so your plot does not look so bad to me. $\endgroup$
    – gerrit
    Commented Apr 5, 2022 at 7:26

2 Answers 2


I wasn't as quick to be convinced, but looks like this GFS model plot of simulated IR satellite over the previous hour for 1Z on the 2nd (so basically the same time as your image) verifies it's not inverted.

enter image description here From TropicalTidbits.com

Specifically this feature trailing over the Pacific and east Asia: enter image description here

I'm surprised its that cloudy on Earth. The one thing that does stand out about your image is almost all cloudy regions are at 100%... which would seem to contrast with the amount of times with partial cloudcover. So I'm not sure on that quite. Though when you throw in fog, stratus, marine layers, convective precip, etc... I suppose its feasible. Or perhaps a factor of data or plot resolution???

It seemed worthwhile to me to take a look at the GOES full disk with cloud top phase turned on to emphasize the different clouds to get a better idea how things really are in the full spectrum of clouds. Do hope you get it sorted, but does seem you might actually be more where you want to be than you thought.

  • 1
    $\begingroup$ Thanks that makes sense. I seems like although the data is supposed to be total cloud cover % they way they are measuring it seems to be quite granular meaning it appears like there is a lot more cloud compared to that shown in zoom earth. This is annoying but does not make the data completely useless $\endgroup$
    – Isabella
    Commented Apr 11, 2022 at 1:23
  • $\begingroup$ @Isabella I totally agree. If your work is something realtime rather than historic, wonder if it might even be better to use another source/higher resolution satellite data, or maybe even model plots. But depends upon your needs/situation :) $\endgroup$ Commented Apr 11, 2022 at 1:56

I assume that that the total cloud cover % is inverted?

No, it's not. As gerrit mentioned in a comment, about 70% of the Earth is covered by clouds at any point in time. Science fiction movies and TV shows up to the mid 1960s showed a cloud-free Earth. Science fiction movies and TV shows since then have tended to show a much more realistic image of the Earth, which is mostly covered by clouds. It was the onset of the satellite era that made those sci-fi movies change.

To make matters worse, your generated image appears to be using an equirectangular projection. It appears you have 0 as the north pole and 180 as the south pole (so co-latitude rather than latitude), with equal steps in co-latitude equally spaced on your generated image. This kind of projection exaggerates areas at high latitudes. High latitude regions are much more likely to be cloudy than are equatorial regions.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.