[wxqc] Looking for help with Station - Odd shift in analysis

Chris Miller c_miller_1 at hotmail.com
Thu Feb 18 17:43:24 CST 2016

I had been using my phone as a comparison for a couple of weeks and every time I checked it seemed to track, however that sample is small and scattered so may have had the luck of the draw.  Discovering the correction error in WUHU did account for a couple of mb.  The station does track the nearby stations and has, when i get one of those artic highs, all seem to gather together nicely so, crossing fingers, I hope to see the model come back soon.  Mind you, in another month or so it is libel to shift back due to the arrival of spring.  I think there are a few SLP stations from the government side that may shift the model line average??  
Oh!  While I think of it, is it possible to find out what stations are used for the L3 checks and the model line?  Mostly curious as to how close they are and such.    

From: Ted Lum 
Sent: Thursday, 18 February, 2016 07:50
To: Chris Miller ; wxqc at lists.gladstonefamily.net 
Subject: Re: [wxqc] Looking for help with Station - Odd shift in analysis

No, the MADIS analysis line gets so crazy sometimes since it's so sensitive to individual samples/readings.

You may actually have a faulty sensor that is drifting. I know it's not easy to do, but, I'd start by comparing the raw readings from the station with another local, uncorrected, pressure sensor. Nothing needs to be calibrated since you're looking at the delta over time; if they differ by 2mb today, they had better differ by 2mb 5 days and 10 days from now... if not, you've got a very fundamental problem.

On 2/18/2016 7:37 AM, Chris Miller wrote:

  Thanks Ted,  I had been playing with the analysis period to escape the shifted data when I made corrections.  The best I have so far, with a short analysis period of 3 days, is still the 4 to 6 mb below the model line.  I was starting to wonder if the model has a large fixed period that it has to work through before I would see it shift down toward my setting.

  From: Ted Lum 
  Sent: Wednesday, 17 February, 2016 16:12
  To: wxqc at lists.gladstonefamily.net 
  Subject: Re: [wxqc] Looking for help with Station - Odd shift in analysis

  The graphs are what they are in terms of date/time, but the default analysis period is 28 days. You can select a different period from the links at top of the page, or by specifying a specific number in the url.

  On 2/17/2016 4:04 PM, Chris Miller wrote:

    Ok.  I have set the pressure on the PEET station to 23.46 mb over my local surface pressure.  I have let the station run for a while and cross checked the device used to get the local pressure to the station AV065 and CYYB and any measurement error corrected for.  I discovered that WUHU, when set to QNH pressure to CWOP with the altitude of the sensor set to 650 feet (643 feet plus sensor height), that there is a correction that gets applied to the sent data that is not visible on the display locally in the software.  It appears that it has a temperature value in the math, if I read the WUHU news group correctly.  I switched back to weather display with no correction value in the software field and no altitude set in the software and so far the data seems to match what is reported to CWOP.  

    I have let the data run for a few days and my setting tracks the output of AV065, CYYB, AR943 and CYWA with the expected stagger due to weather pushing through.  The models for those stations show them to be on, yet my model still shows me as being 4 to 6 mb low.

    Is there anything else that could be coming from the data supplied that will skew the model so high?  
    Also, is there a long delay, 3 to 6 months for example, before the model would return to track the station after a burst of bad data?

    (I cropped down the email as it was getting a little long.)  

    My station, model and nearest neighbours

    CYYB, its model and same stations

    CYWA, its model and same stations


This message has been scanned for viruses and 
dangerous content by MailScanner, and is 
believed to be clean. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://server.gladstonefamily.net/pipermail/wxqc/attachments/20160218/41c324a8/attachment.html>

More information about the wxqc mailing list