Climate data - what metric describes "coldest temp. a few days in a row"?

[MartinLe]

When designing biogas plants, One thing I want to know is the lowest temperature to size my heating for (so that at said temp. I get enough power into the substrate to compensate for losses through the tank walls). K Values etc. are all known.
To size my heating system I usually take the average from the coldest month and subtract another 10 - 15 °C (more in mountains and far inland, less if a plant is near a coast).

I don't care for short term cold-spike, I have tanks with 2-3000m³ mostly water, this is enough thermal mass that I don't worry about a cold night. But what common metric describes "the average over the coldest few days in a row in month x" that I can look up somewhere?

How have others solved this problem? Also, what climate databases do you use? I mostly used

http://klimadiagramme.de/

so far (monthly averages and precipitation).

 

[Drew08]

Not an official term, but I'd call it something similar to 7Q10 for low stream flows, like 7T10 meaning the 7-day average low temperature that has a 10-year return period or 10% annual chance of occurrence. Could be 3T2, 3-day average low temperature with a 2-year return period, or whatever combination you are looking for. It is easy to calculate in excel with historic temperature data. Calculate a running X-day average temperature, and count how many times that average falls below a certain value divided by the number of years of record. There is even a program called EPA DFLOW that does this for stream flow data. Data is data though; the program doesn't consider units of measure so it should work equally for temperature data and stream flow data.

DFLOW:

http://water.epa.gov/scitech/datait/models/dflow/

Places to look for temperature data (at least for US):

http://waterdata.usgs.gov/usa/nwis/nwis

http://www.ncdc.noaa.gov/

http://www.raws.dri.edu/

http://www.wunderground.com/