This is a follow on from a previous discussion. http://www.greenbuildingforum.co.uk/forum114/comments.php?DiscussionID=9445&page=1#Comment_153976
Thoughts are needed as to how this could be done.

Might do it on the basis of the number of sigmas or km/h over the (known imperfect) km-squares data? But I'd claim that as with temperature each building would have to find its own baseline rather than a regional or national value necessarily making sense.

Rgds

Damon

Posted By: SteamyTeaAs HDDs use the modal value above or below a reference value

Unless I'm misunderstanding something (which is quite possible) modal values are not what people like the Met Office and Degree Days.net use?
http://www.degreedays.net/calculation

Well I have just had a look at windspeed and Air Temp down here, seems the faster the windspeed the less temperature varies.
But is just one month in one year. Shall have a look at more later.

HDDs can be calculated using any average I shoudl think, each will have a limitation.
One advantage of using the mode is that it gets rid of the outliers right from the start and does, to a certain extend, take time into account (though that has to be checked as you may have 98 different readings and 2 the same). I shall ponder it at work later.

One thing to try and work around is that wind direction generally affects the amount of cloud cover, may be a bit different for us on the west coast, but then we are extremes. Then there is the effect of different building materials when subjected to solar radiation, glass and brick behave different from slate and timber.
A statistical approach can be taken to this i.e. what is the power in the sun for any given wind direction and what is the windspeed for any given wind direction, then plot the results against each other (could do any combination of bases, windspeed, direction, sun power).
I only have one internal dataset of internal temperature measurements (kitchen that faces north east), so any correlations would have to take that into account, easy enough to do as I can work out when the sun stops falling on the back wall (before noon). Also as this is in the lee of the prevailing wind, that would have to be taken into account.
Summer figures (non heating season anyway) are quite useful as they show what the weather is actually doing to the house i.e. temperature rise/fall.
As solar effects do not extend into night time, but air temperature, windspeed and direction do, it may be worth splitting the datasets into two, day and night (though a reference to how clear the day was may be useful as it may show a more rapid decline in night time temperature).

Not sure if the datasets should be done at the daily, weekly or monthly level, or if they should be presented as absolute or relative values. If relative values then what should the 'zero' value be, the daily, weekly or monthly value (and should it be mean, mode or median, or a time, say solar noon).
Just some thoughts so far.

Right, had a look at temperature now, Day is warmer than night, and the reason that the All Temperatures (red bar) is higher than the mid point between the Day and Night is because of the longer days (this will be tested when more months are showing.
Now to sort out the house temperatures.

The Chart.