Missing Values for long term weather statistics

Hi,

I have some 7+ years of ‘every’ minute weather data, but occationally I miss some data due to problems of different sorts. Mostly it is just a minute or two here and there, f ex switching batteries during winter, when we have (almost) no sun for weeks. Sometimes there are missing values because of a board struck out by lightning.

Well, this makes it pretty awkward comparing f ex March of every year. Or comparing year by year. Of course it affects averages and such only by a small margin but nevertheless.

I have Statistical Methods in the Atmospherical Sciences, third ed, by DAniel S. Wilks, and it is wonderful in many ways but lacks and input on the Missing Values/Missing Data and what methods would be best implemented.

I have browsed the net for any articles that would suit my needs, but seen none. I have read papers on sociological studies and missing values et c, but these methods apparently aren’t fit for my kind of missing data.

Have you seen any good info regarding this, or use some algorithms you’ve found to be good practice?

Regards,

IngemarS