6
$\begingroup$

I have temperature data from a weather station, but there are missing data (not more than 80% of the daily temperature observations). The World Meteorological Organization has some rules to validate estimations of monthly temperatures, etc. (for example to have 80% of the data). But since my periods of time are related to the lifecycle of the migratory species I am studying, for example: 1) arrival to the breeding ground and egg laying is from 20-May to 20-Jun (31 days) 2) incubation is from 21-Jun to 15 July (24 days) 3) brood rearing is from 16-July to 15-Aug (30 days)

Do you know if there are some other rules? for example when my period of time is <30 days? Does the same rule would apply for the incubation period, even if it is shorter than one month?

thanks a lot for your advice/suggestions

Improve this question
$\endgroup$
3
  • $\begingroup$ Random uninformed thought: you might try fitting a Fourier series (sine wave) through your data and seeing how well the mean of the Fourier curve matches with your actual mean. Temperatures don't follow a pure sine curve long-term, but this may work short-term. $\endgroup$
    – user967
    Commented Jan 7, 2017 at 17:08
  • $\begingroup$ Are you missing 80% during your critical periods or missing 20%? Do you have any other stations near-by that you can correlate with and estimate the temperatures for missing days? Are the days missing for long periods or is it just a day here or there? If the missing periods are short you would be justified in interpolating between days on either side. What ever you do to fill in, try to do a sensitivity analysis to see how your assumptions for treating missing data affect the result. $\endgroup$
    – haresfur
    Commented Jan 10, 2017 at 23:17
  • $\begingroup$ For what it's worth, when I upscale rainfall data from 30 min to daily, I take the average of the available data and multiply it by 48 so I'm essentially filling missing values with the average for the day. $\endgroup$
    – haresfur
    Commented Jan 10, 2017 at 23:20

0

Reset to default

Your Answer

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

Browse other questions tagged or ask your own question.