Re: design for multiple time series

On 12/13/2013 4:46 AM, rob stone wrote:
> The only fly in the ointment with this is a rain gauge. If you don't
> empty it each day the actual rainfall is the difference between
> readings.

(somewhat off topic)

The electronic rain gauges I've seen have all been tip-bucket. they
measure each 0.01" (or equiv metric unit) at a time, and the rain total
is counter based, the weather station software automatically rolls over
daily, weekly, monthly totals, also tracks 'last 24 hours' which is a
rolling total.

the one I have has a 'teeter totter' rocker, each side of it is a
'bucket', when 0.01" of precip. comes down the funnel into the currently
high side of the teeter, it flips over to the other side, dumping that
0.01" and a magnet on the side of the teeter registers a pulse on a
nearby coil.  now the other side fills, and it tips back

we've had 1.40" total at my house since the rainy season started in
August 1.    we should be getting that much or more weekly this time of
year.    this will be the 3rd year of draught on the central coast ;(((

best design for multiple time series *I've* seen is RRDTOOL, which is an
updated version of the classic MRTG.   this is /not/ SQL.



--
john r pierce                                      37N 122W
somewhere on the middle of the left coast