consider validating station data in a method similar to NOAA's MADIS system
Aside from the current absolute checks, consider validating data (temperature, dewpoint, pressure, avg wind speed/direction) with a relative metric against five other stations within a radius (e.g. 25 miles). If the reported value is more than (n) delta from the average of the other five stations consistently over several reports, consider it to be suspect. Ultimately, develop a quality grade (e,g, A through F, Excellent through poor) for every station based on well data compares over a reasonably short historical average (one week?).
The idea is to help users choose better reporting stations, and also help station owners better recognize that there may be a potential problem with the data that their station is reporting.
If modeling and computation horsepower are a concern, perhaps consider using the quality report from MADIS, if possible.
N9PBY
shared this idea
This is how it works now. WE’re working on ways to improve the system, and also make station owners more aware of problems with their stations.
10 comments
-
@md54, n9pby: I've created a new item for this — please share any addition thoughts you have there:
-
md54
commented
I like the idea of "grading". How we do it (A-F, 1 star - 5 stars, good-bad- indifferent), it almost does not matter. But it would be great to have a grading system, which would help us to determine how trustworthy the station is.
-
N9PBY
commented
I think that an overall grade (e.g. A though F, or good - marginal - bad) would be great. Possibly taking it a step further, consider sub scoring or deviation/mean error figures for pressure, temperature and dew point, as such would help a PWS owner identify what might be contributing to a lower grade.
It would be awesome to attempt to point out easily identifiable problems to PWS contributors, such as: Your barometric pressure does not appear to be calibrated to sea level, see the wiki for help on fixing this, or, your daytime temperature reads abnormally high compared to nearby stations, etc. I am thinking of situations that can be easily identified mathematically, thus such messages/tips could be completely automated.
-
@md54: what do you think of some kind of "grading" system?
-
md54
commented
If possible, provide feedback to station operator on data quality. Or at least provide tools to verify how good "my station" is.
-
appatton
commented
Thank you, N9PBY. Unfortunately, the station I am interested in does not report to MADIS. I've sent a message to the station operator and will hopefully hear back shortly.
-
N9PBY
commented
I don't think that you can see WU's QC checks (yet). But if the station also reports into MADIS, you can see an analysis here:
http://weather.gladstonefamily.net/
The actual MADIS QC checks are also available, but those only report if the data is good, questionable or bad:
https://madis-data.noaa.gov/sfc_display/
Look for the MADIS ID on the right side of the WU station history page in the "About this station" pane.
-
appatton
commented
How do I access the QC reports for a station?
-
pwmeek
commented
This alerted me to a problem with my weather station. Thank you.
-
Thank you for the suggestion. Currently our WC system does this very thing, and performs better than the MADIS quality control is several areas. However, we're always working to improve it.
