Author Topic: Barometer error  (Read 8613 times)

0 Members and 1 Guest are viewing this topic.

Offline hankster

  • Forecaster
  • *****
  • Posts: 482
    • NFM Weather
Barometer error
« on: February 14, 2016, 03:18:13 PM »
Does anyone know if there is a way to calibrate the barometer reading? BloomSky is currently reading 30.33 and it should read 30.19

Been following the temp and humidity and it pretty much follows what My Davis Vue reads.

Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #1 on: February 14, 2016, 03:52:40 PM »
VaJim reported this to Bloom Sky some time ago, others too.

The problem is that so far, the Bloom Sky reports an "absolute pressure" based on their installed sensor and whatever in house calibration they do (if any).

So far, they do not provide a way to correct the barometer reading to match an elevation corrected value, the barometer altimeter reading that we all mostly use (as reported by a local airport and news agencies).

I have also suggested that Bloom Sky give us a way to adjust the numbers (calibrate) that they send to WU.  Or, that WU give us a way to calibrate sensor readings.

Of course a third option is that Bloom Sky gives us a way to program our individual stations (Bloom Sky firmware) with our own calibration constants.

There was a brief period as nincehelser noted some time back where there was difference between the Bloom Sky (different Bloom Sky pages) reported barometer values.  The difference for my station was very near the correct difference for my elevation.  So, I wonder if they were trying to do the elevation correction based on topo map data.  The problem is that my unit's absolute calibration also needs a calibrated offset (it appears to report the wrong numbers, even just for uncorrected absolute.)

It is too bad that Bloom Sky has not yet addressed this issue.  First, it hurts their credibility as providing a serious full wx station (as opposed to just a good webcam sky picture, which is nice).  Second, there is a risk they could corrupt amateur crowd sourced wx data (although most are so far off without calibration, perhaps the barometer data is automatically discarded).
« Last Edit: February 14, 2016, 04:05:23 PM by waysta »

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #2 on: February 14, 2016, 04:02:50 PM »
It's weird that they just don't sit down and fix this, or at least tell us that it's in the queue to be fixed.

The last time I mentioned it was when I noticed the units were in cubic inches.  They said they'd get right on fixing the unit issue, but no mention of correcting the adjustment.

Offline W3DRM

  • Forecaster
  • *****
  • Posts: 3360
    • Emmett Weather
Re: Barometer error
« Reply #3 on: February 15, 2016, 01:18:52 PM »
I'm very interested in this topic as my location is near 5000 feet ASL. As most of us who have a Davis VP2 and live significantly above sea level know, we have always had issues with accurate barometer readings. You can't simply crank-in an offset as it takes special handling of the data to make it track accurately and now sadly, it sounds like the Bloomsky may also have the same issue.
Don - W3DRM - Emmett, Idaho --- Blitzortung ID: 808 --- FlightRadar24 ID: F-KBOI7
Davis Wireless VP2, WD 10.37s150,
StartWatch, VirtualVP, VPLive, Win10 Pro
--- Logitech HD Pro C920 webcam (off-line)
--- RIPE Atlas Probe - 32849

Offline Maumelle Weather

  • Forecaster
  • *****
  • Posts: 1824
    • Maumelle Weather
Re: Barometer error
« Reply #4 on: February 15, 2016, 02:49:33 PM »
And the barometric pressure still reads in cubic inches.
GR2AE, GR3, Cumulus

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #5 on: February 15, 2016, 02:56:12 PM »
I'm very interested in this topic as my location is near 5000 feet ASL. As most of us who have a Davis VP2 and live significantly above sea level know, we have always had issues with accurate barometer readings. You can't simply crank-in an offset as it takes special handling of the data to make it track accurately and now sadly, it sounds like the Bloomsky may also have the same issue.

I'm only around 1000'.

The trends track fine.  They're just applying an incorrect adjustment.  In fact, they are applying a different adjustment depending on how you are accessing the data.

Offline VaJim

  • Virginia Beach Backyard Weather
  • Forecaster
  • *****
  • Posts: 760
    • Virginia Beach Backyard Weather
Re: Barometer error
« Reply #6 on: February 15, 2016, 07:48:48 PM »
I'm very interested in this topic as my location is near 5000 feet ASL. As most of us who have a Davis VP2 and live significantly above sea level know, we have always had issues with accurate barometer readings. You can't simply crank-in an offset as it takes special handling of the data to make it track accurately and now sadly, it sounds like the Bloomsky may also have the same issue.

I'm only around 1000'.

The trends track fine.  They're just applying an incorrect adjustment.  In fact, they are applying a different adjustment depending on how you are accessing the data.


In your opinion, what do you think Bloomsky needs to do to correct it.  I hope the fix is something that is done automatically through an update to the app.

Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #7 on: February 15, 2016, 08:02:06 PM »
They also need to calibrate the barometer readings they report to WU.

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #8 on: February 15, 2016, 08:07:33 PM »
In your opinion, what do you think Bloomsky needs to do to correct it.  I hope the fix is something that is done automatically through an update to the app.

Since their target is the average consumer, probably the most straightforward solution is present altimeter as that is what most people are used to seeing.  They should have the altitude from the GPS reading at installation, or through a mapping service.

Originally they were presenting absolute pressure, then in a later app update it suddenly shifted... closer to a sea-level reading, but still off.  Then when the web access came about, there is a different shift.

So I'm assuming there must be two groups of software developers, one focused on the web, the other on the app, and they aren't doing things the same way.

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #9 on: February 15, 2016, 08:20:52 PM »
They also need to calibrate the barometer readings they report to WU.

I expect that will happen when they get it straightened out. 

In the end, the pressure they report in the app should match what they are reporting to WU.

Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #10 on: February 15, 2016, 11:34:32 PM »
There might be more serious problems, maybe calibration is not a simple offset, or even y=mx+b with slope and offset.

This is a graph of what I think are same time scaled values of RainWise barometer (blue, verified against the nearby mil airfield) and BloomSky (orange) from 2/15/16 between about 5am and 8:30pm.  The temperature range was from about -6F to +17F (the temperature has been slowly ramping up all day).

There is not a simple offset (like a cal screw on a mechanical barometer, or imagine sliding one curve over the other.  If it was just an offset error (e.g. the elevation correction), they would match identically once overlapped).  Once the equation gets more complicated, there will be a question of is it just a matter of applying one equation with the Bloom Sky Baro as input to convert Bloom Sky to altimeter?  Or, is there going to be an additional temperature correction, and hopefully not something else as well, such as a humidity correction for the Bloom Sky baro sensor.

The size of the steps is not an issue.  Bloom Sky has a more course analog to digital conversion, but that's okay.  It's how two best curves drawn through the data would overlap, or not.

The data is from both WU pages, "Table" and "download" as CSV, graphs in Excel, I'm new to iMac, best I can do as a first pass look.
http://www.wunderground.com/personal-weather-station/dashboard?ID=KNYWHITE24#history/tgraphs/s20160215/e20160215/mdaily RainWise
http://www.wunderground.com/personal-weather-station/dashboard?ID=KNYWHITE25#history/s20160215/e20160215/mdaily BloomSky

Just a very first look ... at a very small set of data ... no conclusions yet.
« Last Edit: February 16, 2016, 12:09:42 AM by waysta »

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #11 on: February 16, 2016, 12:53:32 AM »
There might be more serious problems, maybe calibration is not a simple offset, or even a y=mx=b with slope and offset.

This is a graph of what I think are same time scaled values of RainWise barometer (blue, verified against the nearby mil airfield) and BloomSky (orange) from 2/15/16 between about 5am and 8:30pm. 

There is not a simple offset (like a cal screw on a mechanical barometer).  Once the equation gets more complicated, there will be a question of is it just a matter of applying one equation with the Bloom Sky Baro as input to convert Bloom Sky to altimeter?  Or, is there going to be an additional temperature parameter, and hopefully not something else as well, such as a humidity correction for the Bloom Sky baro sensor.

Just a very first look ... at a very small set of data ... no conclusions yet.

Yes, conversion to sea-level is more than just a simple offset (though you can sometimes get by with that).

The most common method is altimeter, which just takes the absolute pressure and runs it through a standard equation with your elevation.  Since Bloomsky knows your elevation, this should be trivial to implement.  Not everyone is always happy with the results, though, as it starts to break down at high altitudes (as someone mentioned earlier).

There are more complex methods used by meteorologists that also take temperature and humidity into account.  Most people aren't familiar with that, so I doubt Bloomsky will implement it.

Simply put, there doesn't seem to be anything particularly wrong with the baro sensor itself.  It just reports the "true" pressure of the atmosphere pushing down on it.

It's when you try to convert that reading to sea-level equivalent that things can get sticky.

« Last Edit: February 16, 2016, 03:48:46 AM by nincehelser »

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #12 on: February 16, 2016, 03:41:32 AM »
OK.... after some analysis I'm thinking something different is going on.

1.  The Bloomsky data portal pressure is absolute.
2.  The pressure being fed to Wunderground is absolute.
3.  The pressure reported by the iOS app is the absolute pressure converted to altimeter.
4.  The pressure reported by the Bloomsky Map web site is absolute pressure converted to altimeter.

This is based on the assumption they my particular Bloomsky is reporting 8mb high.  That seems a lot of error to me, but it makes the numbers work out.

Does this make sense given your data?

Offline W3DRM

  • Forecaster
  • *****
  • Posts: 3360
    • Emmett Weather
Re: Barometer error
« Reply #13 on: February 17, 2016, 02:04:07 PM »
I'm watching the difference between my VP2 and my Bloomsky barometer. So far, there appears to be a 3.4 mb difference between the two. Will continue watching for a while to see if it changes any. We have an incoming storm at the moment so this should be a good time to monitor the readings and see how they compare.

My temp values are also fairly close with the VP2 at 54.6F and the Bloomsky at 55F. Humidity is VP2 = 45% and B = 42%.
Don - W3DRM - Emmett, Idaho --- Blitzortung ID: 808 --- FlightRadar24 ID: F-KBOI7
Davis Wireless VP2, WD 10.37s150,
StartWatch, VirtualVP, VPLive, Win10 Pro
--- Logitech HD Pro C920 webcam (off-line)
--- RIPE Atlas Probe - 32849

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #14 on: February 17, 2016, 08:42:37 PM »
I looked at the specs of my other barometers to see what their error range is.  They range from +/- 1 to 3 mb for absolute readings.  (relative readings are much closer)

I was surprised to see my iPhone 6 is +/- 1 mb.  Apparently it uses a Bosch BMP280.  (https://www.adafruit.com/datasheets/BST-BMP280-DS001-11.pdf)

A bit off-topic for this thread, but I found this write-up interesting: http://www.extremestorms.com/iphone_6_barometer.htm

Anyway, is someone brave enough to open up their Bloomsky and figure out what baro sensor they use?  I haven't been able to find any tear-downs on the Internet yet.

Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #15 on: February 18, 2016, 04:58:12 PM »
Here is a spot check of my data 2/18/16 about 4:45pm:

iOS 30.81 in Hg
WU BloomSky 30.18
BloomSky Web data portal 1022 mb (30.18)
WU RainWise 30.58
WU Rome, NY AFB 30.59 (turns out while the RME reports are delayed through NWS, their data on the Wunder Map is near real time)

I agree that Bloom Sky seems to be trying to do the altimeter correction by location.  However, the uncalibrated instrument error, here .23 in Hg, is too large to be directly useful for crowd sourced wx data. 

It may be they need one or two additional user calibration constants to calibrate the absolute data from our individual sensors.


OK.... after some analysis I'm thinking something different is going on.

1.  The Bloomsky data portal pressure is absolute.
2.  The pressure being fed to Wunderground is absolute.
3.  The pressure reported by the iOS app is the absolute pressure converted to altimeter.
4.  The pressure reported by the Bloomsky Map web site is absolute pressure converted to altimeter.

This is based on the assumption they my particular Bloomsky is reporting 8mb high.  That seems a lot of error to me, but it makes the numbers work out.

Does this make sense given your data?
« Last Edit: February 18, 2016, 05:00:30 PM by waysta »

Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #16 on: February 18, 2016, 05:08:34 PM »
Once a barometer reads as "altimeter", it seems most instruments just offer a simple add or subtract to "set" the barometer (only an offset).  Is there more going on first from absolute to altimeter?
« Last Edit: February 18, 2016, 05:10:06 PM by waysta »

Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #17 on: February 18, 2016, 05:12:46 PM »
I forgot my iPhone has a barometer!  A quick free download correcting altimeter for GPS altitude (I'll check the alt more closely later) gives 30.61, much closer to my RainWise 30.59 now.

I looked at the specs of my other barometers to see what their error range is.  They range from +/- 1 to 3 mb for absolute readings.  (relative readings are much closer)

I was surprised to see my iPhone 6 is +/- 1 mb.  Apparently it uses a Bosch BMP280.  (https://www.adafruit.com/datasheets/BST-BMP280-DS001-11.pdf)

A bit off-topic for this thread, but I found this write-up interesting: http://www.extremestorms.com/iphone_6_barometer.htm

Anyway, is someone brave enough to open up their Bloomsky and figure out what baro sensor they use?  I haven't been able to find any tear-downs on the Internet yet.

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #18 on: February 18, 2016, 05:31:07 PM »
I agree that Bloom Sky seems to be trying to do the altimeter correction by location.  However, the uncalibrated instrument error, here .23 in Hg, is too large to be directly useful for crowd sourced wx data. 

It may be they need one or two additional user calibration constants to calibrate the absolute data from our individual sensors.

Hmmm... when I convert mb to inHg, I am off .236 to your .23.  We differ in error by 6/1000th.

Maybe it's coincidence, but it's odd we're both off almost the same large amount of error.






Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #19 on: February 18, 2016, 05:35:14 PM »
It would be pretty funny (and easy for them to correct) if they just got a calculation wrong.  ... unless it goes back to their firmware ...
« Last Edit: February 18, 2016, 05:37:20 PM by waysta »

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #20 on: February 18, 2016, 05:53:58 PM »
Once a barometer reads as "altimeter", it seems most instruments just offer a simple add or subtract to "set" the barometer (only an offset).  Is there more going on first from absolute to altimeter?

The equation doesn't look linear, but I suspect it may be close enough for some.  http://www.srh.noaa.gov/images/epz/wxcalc/altimeterSetting.pdf

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Barometer error
« Reply #21 on: February 18, 2016, 06:03:22 PM »
It would be pretty funny (and easy for them to correct) if they just got a calculation wrong.  ... unless it goes back to their firmware ...

If someone opens up their Bloomsky and can get a part number from the sensor, we could look up the specs and see what the absolute error range is.

Given Apple's is +/- 1 mb, I can't see why Bloomsky should be as far out as 8 mb.

Depending on how the baro sensor they use presents the data, a calculation error wouldn't be unheard of.  The baro sensor in the Acurite bridge sends out 13 or so parameters that have to run through an ugly vendor-supplied equation before you get the absolute pressure. 

Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #22 on: February 19, 2016, 10:16:10 AM »
There are a couple of odd things about the NOAA equation (it may be fine)

The complex looking term just after the 1 is really just a constant.  They may have presented it that way to show its derivation.  Or, did they miss a variable?

Also, in the limit where hm=0 (sea level, zero height), the Altimeter is the absolute pressure in mb - 0.3.  Is this correct?

Lets try some experiments with the Davis equation for now:
http://www.davisnet.com/product_documents/weather/app_notes/AN_28-derived-weather-variables.pdf
A = (P^N + K*Z)^(1/N), where P is the raw station pressure (in. Hg), N = 0.1903, K = 1.313E -5,Z is elevation (feet).


Once a barometer reads as "altimeter", it seems most instruments just offer a simple add or subtract to "set" the barometer (only an offset).  Is there more going on first from absolute to altimeter?

The equation doesn't look linear, but I suspect it may be close enough for some.  http://www.srh.noaa.gov/images/epz/wxcalc/altimeterSetting.pdf
« Last Edit: February 19, 2016, 11:05:48 AM by waysta »

Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #23 on: February 19, 2016, 11:11:02 AM »
Spot check about 11am 2/19/16: (the pressure is changing, there might be time discrepancies on some of this data)

iOS 30.63 in Hg
WU BloomSky 30.00 (probably Bloom Sky absolute with some uncorrected sensor error)
BloomSky station web page 1037 mb (30.62) (probably Bloom Sky attempted altimeter based on my estimated elevation from a map source)

BloomSky Web data portal 1016 mb (30.00)

Bloom Sky abs to altimeter by Davis equation (calculated for my estimated 538 ft AGL) 30.30

WU RainWise 30.43
WU Rome, NY AFB 30.44 (turns out while the RME reports are delayed through NWS, their data on the Wunder Map is near real time)

Bloom Sky error about  ~+.20 in Hg (to my RainWise)
« Last Edit: February 19, 2016, 11:19:23 AM by waysta »

Offline waysta

  • Contributor
  • ***
  • Posts: 148
Re: Barometer error
« Reply #24 on: February 19, 2016, 11:40:03 AM »
This is going to take some more investigation.  My initial thought is that there are one or more typos in the NOAA equation.

Here is an interesting find from some googling:
http://www.av8n.com/physics/cat.cgi/barometry.pl

excerpt (which looks like the Davis formula):

I asked the folks who manufacture AWOS machines to tell
me what algorithm they use for computing Altimeter Setting.

They replied:
> it strictly follows the FAA algorithm for AWOS.
> The algorithm for the Altimeter is attached for your information.

I quote from the attachment:  In normal operation:

   AS = (Pa^N + K Ha)^(1/N)
where:
      Pa = Field pressure
      N = 0.1903
      K = 1.313e-5
      Ha = Field elevation of the airport
           in feet above mean sea level

« Last Edit: February 19, 2016, 06:22:00 PM by waysta »

 

anything