Supposedly, WeatherFlow is also utilizing Weather Underground’s dataset as references to calibrate pressure and UV. I personally wish they didn’t do this and end it asap. Weather Underground’s dataset is terrible and I think it undermines the performance of the Tempest station & WeatherFlow’s credibility when it’s in use. Why would you use competitors’s stations (some are cheap junk) as references to calibrate your sensors? It’s one thing to use ASOS/AWOS or the research grade stations from Campbell Scientific; but it is another thing to use Ecowitt/Ambient, Sainlogic, etc. as valid reference data. I know there are QC processes that goes into CL, BUT you can’t QC and math your way out of a bad dataset that’s full of cheap PWSs. CL should only use datasets from research/scientific grade stations like Oklahoma Mesonet, ASOS/AWOS, and CRN; not from CWOP or WU networks.
If I’m wrong on anything, please correct me @WFsupport @WFstaff I really like the things y’all are doing but I do believe that your Continuous Learning system is undermining the Tempest’s performance due to the reference data that it’s allowed to utilize.