Quick fix to solve most data related problems.....

[bordersweather]

Hi Steve,

I wonder if I could offer some programming advice to solve a number of data related problems I have seen occuring with the software?
If you were to add a "quick-fix" solution, this might suit a lot of people.

I have been working on Scotweather site doing as much as I can with the cumuluswebtags.php and realtime.txt files...
Unfortunately the only problem I cannot programmatically solve is the issue with sudden errornous data?

So far I have seen the Scotweather site suddenly report many hundred mm of rain, 50'C temps and -17'C temps.

Could I impart a little advice, that some sort of user-configurable (or even hard coded in v1 and user configured in v2) code along this lines of:

if the temp has increased more than 5'C in 1 min ignore it,
if the temp has decreased more than 5'C in 1 min ignore it
if the rain has increased more than 10mm in 1 min ignore it.

this sort of quick fix should solve most problems??

Please let me know if I can be of any assistance... I am really looking forward to making the move to cumulus

Andy

[steve]

So far I have seen the Scotweather site suddenly report many hundred mm of rain, 50'C temps and -17'C temps.

Yes, in this case the cause is the USB adapter. Causes garbage data for some people (with La Crosse stations), and works fine for others. So sadly I don't support their use with Cumulus.

if the temp has increased more than 5'C in 1 min ignore it,
if the temp has decreased more than 5'C in 1 min ignore it

Cumulus already has something very similar this, as an ini-file option, and the threshold is configurable. Set the value of EWTempDiff in the [Station] section in cumulus.ini to the required threshold. Consecutive readings greater than the threshold are ignored.

if the rain has increased more than 10mm in 1 min ignore it.

Cumulus also has filters for excessive rainfall (and for most other variables) but because of the way Cumulus stores and uses its rainfall data, and has to support different types of weather station, 'rogue' rainfall can still appear in the data. And you can't just 'ignore it', because of the way that some weather stations supply the rainfall data. It's too hard to do anything about this in Cumulus 1.

I'm looking at options for making Cumulus 2 more resistant to the weird readings that some stations produce.