Take the 2-minute tour ×
Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. It's 100% free, no registration required.

I am using sea surface temperature, chlorophyll, and other data parameters as covariants in a model, and I would like to average them over a particular time period.

I am currently doing this using a simple average of the values across the days in my time period, but this is flawed due to missing data from cloud-cover, etc. (and perhaps other reasons).

filledsst = np.average((datafile.variables["FilledSST"][idx1:idx2,:,:]), axis=0)

What technique is typically used for accounting for such missing data?

share|improve this question
You can apply a mask to your imagery in order to exclude areas where data is missing if they aren't already classified as 'no data'. –  Nick O Apr 23 '13 at 18:35
So I would create a cummulative mask from the datasets that I am averaging together, then apply that when doing the calculations? If I am missing values for just one day, then I would end up with a lot of missing data. Is it possible to weight it based on number of days where data is present, or something? –  shootingstars Apr 23 '13 at 18:40
How many total images are you working with? –  Nick O Apr 23 '13 at 18:55
The answer depends on how you will be using those averages. You might want to ask this question on the stats site, after searching it for "missing data" and "multiple imputation," where you will find plenty of food for thought. –  whuber Apr 23 '13 at 19:24
About 3-4 months worth (using daily files). My plan is to just use an average value for the entire sampling period, or perhaps for duration of each individual cruise performed within the sampling period. –  shootingstars Apr 23 '13 at 19:25

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.