Thanks for the responses, everyone. The reason I asked is because there are several stations in my CWA (El Paso/Santa Teresa) that report only to WeatherLink.com or Weather Underground. For several years, we had about 4-6 WU-only stations up in the Sacramento Mountains (7500-10,000' elevation), so there was no data from that area getting into the MADIS feed, so no MesoWest plots, and the RTMA analysis was warm-biased as a result. Finally I sent a message to one of the most reliable WU stations asking him to consider sending data to CWOP (he had a web page, so it was easy to get in touch with him).
MADIS does not currently have an agreement with WxUnderground, though I wish they did. Even if they go that route in the future, I'm not sure what the terms of the agreement would be. For example, AWS/Weatherbug sends data to MADIS, but they have restrictions on data re-dissemination (they want you to go to their website for data, and see the ads that help keep them in business). So, no AWS stations show up in MesoWest or WeatherUnderground plots, but apparently the data does get into the RTMA analysis.
My gut instinct was that most people sent their data to WU simply because it provides nice plots and a quick place to view the data, whereas the CWOP site, and FindU.com is a little more... clunky? For the average person, there isn't much incentive to upload to CWOP, other than knowing that your data is going to a wider audience.
However, I was unaware of the quality control issue irking the more savvy users. I'll need to look into this some more myself, but the past two posts brought up excellent points. My first question is what is the "analysis" field they're using? Is it simply based on surrounding stations, or is it a from a gridded model analysis of some sort (e.g., LAPS, RTMA, RUC)? Either method is likely to have several potential pitfalls.
My guess is that this was a well-intentioned but poorly-executed method of quality controlling observations that go into the MADIS feed. MADIS provides datasets that are either filtered or unfiltered, but their NWS/AWIPS feeds are not QC'd at all. MesoWest, for example, does not eliminate data, but it does have a system of QC "caution flags" ... I'm not sure if this is from the MADIS QC or something home-grown.
Looking at the various websites, the MADIS QC is indeed presented a little too strongly, and the wording suggests that it is more of a measuring stick than merely "guidance." There's no mention of the potential fallibility of the 'almighty analysis' and if that is alienating potential observers, then perhaps it needs to change. In the meantime, all I can say is don't take it personally. I'm not sure who (if anyone) even notices the QC flags.That said, those red X graphics would tick me off too!
I'd like to see the NWS become more involved with CWOP, potentially addressing some of the network's issues. While 'Big Daddy Government' shouldn't get to say who can and cannot upload data to CWOP, perhaps we could do more outreach, suggesting the best siting possible, diagnosing (real) problems with observations (e.g., maybe all a warm-biased site needs is a simple 'beehive' radiation shelter, perhaps a barometer reading is off because an observer was confused by the 3 billion different ways of measuring pressure), and ensuring that users are aware of available software/firmware updates that ensure the data being sent meets the CWOP specs (2-minute Average Wind, 10-Minute Gusts). Perhaps stations that have had some sort of inspection could get a special designation? (Just thinking out loud here).
Those are some of the biggest problems I see with sites in my area. For example, some sites do not upload 24hr Precipitation (and MADIS, for whatever reason, does not accept Since-Midnight Precip. though it is often more useful), others are using station pressure instead of altimeter. Some do not upload wind gusts. Other do upload the wind gusts, but there's no way to know if their software uses the proper 2/10 spec. There's also sites that may be following the 2/10 spec, but have a 15-30 minute upload frequency (meaning "peak" wind gusts can fall through the cracks). In the case of the precip and wind gusts, sometimes software or firmware upgrades can fix the problem, but not everyone is comfortable doing upgrades...especially for firmware.
Finally, there needs to be a way where humans are helping guide quality control for all observations going into the RTMA analysis. Automated means can be used to "flag" and even temporarily black-list a station, but if a forecaster or HMT at a WFO has looked into the potential problem, and has diagnosed it to be OK, then the flag is overridden. Again, just thinking aloud.
Thanks again for the comments.
-Mike