I will do that and post results here Dne sreda, 18. april 2018 15.22.25 UTC+2 je oseba Greg Troxel napisala: > > Damjan Hajsek <hajsek...@gmail.com <javascript:>> writes: > > > So we can't do nothing to make UV index show right value, because with > > calibrating I can not get real data? > > This can probably be solved, but "calibration" is very likely not the > approach. > > Let me try to summarize what we know. > > Your station seems to show reasonable UV values on the console. > > Your station reports some encoded values that a weewx driver reads. > > Currently, the bits for UV are extracted in some straightforward way > and assumed to be the UV index. This is leading to values that are > not even close to correct. > > "Calibration", in weewx and generally, is a mechanism that makes small > changes to values to make them closer to the correct values. > Calibration is by definition a very minor change, like subtracting 0.7 > hPa of pressure, or adding 0.5 C of temperature. Or perhaps 2 C, but > not 20 C. Calibration as a concept depends on having a close-to-right > value with a simple relationship, so that "V = R + c" holds fairly > closely for readings R, and calibration value c, and true value V. > > So far, we have no reason to believe that attempting something like > calibration above with the values being read by weewx for your station > will give reasonably correct answers. > > And then making assumptions: > > Probably, the way the weather station sends UV index to the computer > is not just an integer in a field. We don't understand how that is > done yet. > > So the path forward is, more or less: > > 1) Ensure (assume, and retry if not, because absent the manufacturer > documenting it, and documenting it correctly, we have to assume) that > the UV field really is the UV field. > > 2) Ensure that the path from bits in the UV field to the number that is > recorded is straightforward and reversible. > > 3) Keep records of pairs of displayed UV value and > reported/recorded-in-weewx UV values. A simple chart with two columns > would be fine (but I might add date/time in case it is useful). > > 4) Put this chart in a spreadsheet or csv format > > 5) Make a scatter plot of console vs weewx values, where each point is > an > observation pair, with console on x axis and weewx-read value on y > axis. > > 6) Figure out if there is a clear relationship and what it is. Try to > come up with a formula, especially one that we can explain after we > create it how it works. > > Change the driver to use the formula. > > I am guessing these steps seem too complicated, but I expect others can > help. > > So I would suggest that you do steps 3 and 4, for at least several days > and preferably a week, at night, and at various points during the day, > both sunny and cloudy days, and post a link to the spreadsheet, or > preferably mail csv format to the list. > > Without steps 3 and 4, then I don't see a way to make progress. > > Greg > > >
-- You received this message because you are subscribed to the Google Groups "weewx-user" group. To unsubscribe from this group and stop receiving emails from it, send an email to weewx-user+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
