Today I’m starting what may be a two or three part series having a detailed look inside the Eureka Weather Station and the data it produces. Thanks to the manager of the station Rai LeCotey, we have a lot of new information that had not previously been available on the web. Mr. LeCotey has been most gracious and forthcoming and I commend his openness, which as we have seen in Climate Science, is a rare quality. Here’s an aerial view of the station.
click for a larger image
Ecotretas and I looked at a number of what we identified as errors in data from Eureka, Nunavut, Canada weather station. Some errors are real, such as the January 1st 2007 METAR error (caused by transcription error).
We identified what we thought were errors on July 13th and 14th. The Station Manager Mr. LeCotey says that the July 14th new record high temperature “error” was real (meaning meteorological data, not a transcription error), and a function of wind direction bringing air from the North that has been warmed by terrain. He also says he’s working to get the error on Jan 1 2007 corrected in the record. He’s sent along some photo documentation of the July 14th 2009, event. We’ll get to that in a subsequent post, but first some background on the station itself.
Here’s what the Meteorological Instrument Complex looks like:
Looking Southwest at Eureka Met Instrument Compound -7/24/2004 - click to enlarge
The truck is interesting. Note the blowers on top of the Stevenson Screens for continuous aspiration.
Mr. LeCotey provided this official visitors guide to Eureka, which I have posted as a PDF, link below
He also graciously answered a number of questions. His answers are in blue.
1) The sensors in the Stevenson Screens are electronic it appears. Am I correct in assuming they are cabled to the met office? Of what type are they? Thermistors, RTD? Thermocouple?
We use a remote temperature and dew point measuring system type 2. The dry bulb temperature is measured with a thermistor and the dew point is measured with a dew cell.
2) I notice fan aspirators on the screens. Are these run continuously for the electronic sensors, or are they a holdover of earlier times when wet bulb DP/humidty readings required aspiration?
The sensors are housed in a ventilated Stevenson screen that runs continuously. Psychrometer comparisons between the remote system and our mercury thermometers (in a second Stevenson Screen right next to the remote screen) are done once a week with an thorough calibration done once a year.
3) How far away, in meters, is the Met Instrument compound from the nearest structure, such as the bright blue HQ building?
The Stevenson screens are approx. 40 meters
4) Why are there no readings in the first two hours of the day (00 and 01)?
We only do a surface weather program for 22 hours a day. We have a contract with Nav Canada that only wants aviation weather between 06Z to 03Z inclusive.
Surface weather observations are done on the side (with NavCan funding). Our primary function is that we send up 2 weather balloons a day as our commitment to the WMO. We stay on EST all year (there is no point to go to daylight savings time as an extra hour of daylight does mean anything to us when we have 24 hours of daylight in the summer anyway). Therefore, our last observation is at 22:00 EST (local) and we start observing again at 01:00 EST (local). The 23:00 & 00:00 observations are missing as NavCan does not pay us for those two observations.
5) Do the electronic displays have max/min memories?
Yes, our AES remote temperature and dew point (1987) system (RTD-87) measure the max and min temperatures and stores them in memory until cleared by the observer every six hours on the syno. The RTD measures the temperature & dew point every minute.
6) How often are the meteorological instruments calibrated and how is this done?
Psychrometer comparisons between the remote system and our mercury thermometers (in a second Stevenson Screen right next to the remote screen) are done once a week with an thorough calibration done once a year.
7) How are the hourly METAR reports made. Are they transcribed from the paper form to teletype or Internet data, or some other method?
The METARS are transcribe from the 2322 form into a WinIDE quality assurance software system (WinIDE version 3.0, is used as the principle data input system for human METAR observations within EC) that automatically checks for errors/discrepancies and gives a warning to the observer to make a correction before the observation data will be sent out. The WinIDE system is very good and follows the observing criteria of MANOBS very well. It will not allow an observation to be sent out over the met circuit with any errors or unnatural trends as in the case of the temperature being entered in as +23.0°C when it was supposed to be -23.0°C.
There will be subsequent posts on Eureka coming in the next day or two.
Visitor’s Guide to Eureka
1) The sensors in the Stevenson Screens are electronic it appears. Am I correct in assuming they are cabled to the met office? Of what type are they? Thermistors, PTD? Thermocouple?
We use a remote temperature and dew point measuring system type 2. The dry bulb temperature is measured with a thermistor and the dew point is measured with a dew cell.
2) I notice fan aspirators on the screens. Are these run continuously for the electronic sensors, or are they a holdover of earlier times when wet bulb DP/humidty readings required aspiration?
The sensors are housed in a ventilated Stevenson screen that runs continuously. Psychrometer comparisons between the remote system and our mercury thermometers (in a second Stevenson Screen right next to the remote screen) are done once a week with an thorough calibration done once a year.
3) How far away, in meters, is the Met Instrument compound from the nearest structure, such as the bright blue HQ building?
The Stevenson screens are approx. 40 meters
4) Why are there no readings in the first two hours of the day (00 and 01)?
We only do a surface weather program for 22 hours a day. We have a contract with Nav Canada that only wants aviation weather between 06Z to 03Z inclusive.
Surface weather observations are done on the side (with NavCan funding). Our primary function is that we send up 2 weather balloons a day as our commitment to the WMO. We stay on EST all year (there is no point to go to daylight savings time as an extra hour of daylight does mean anything to us when we have 24 hours of daylight in the summer anyway). Therefore, our last observation is at 22:00 EST (local) and we start observing again at 01:00 EST (local). The 23:00 & 00:00 observations are missing as NavCan does not pay us for those two observations.
5) Do the electronic displays have max/min memories?
Yes, our AES remote temperature and dew point (1987) system (RTD-87) measure the max and min temperatures and stores them in memory until cleared by the observer every six hours on the syno. The RTD measures the temperature & dew point every minute.
6) How often are the meteorological instruments calibrated and how is this done?
Psychrometer comparisons between the remote system and our mercury thermometers (in a second Stevenson Screen right next to the remote screen) are done once a week with an thorough calibration done once a year.
7) How are the hourly METAR reports made. Are they transcribed from the paper form to teletype or Internet data, or some other method?
The METARS are transcribe from the 2322 form into a WinIDE quality assurance software system (WinIDE version 3.0, is used as the principle data input system for human METAR observations within EC) that automatically checks for errors/discrepancies and gives a warning to the observer to make a correction before the observation data will be sent out. The WinIDE system is very good and follows the observing criteria of MANOBS very well. It will not allow an observation to be sent out over the met circuit with any errors or unnatural trends as in the case of the temperature being entered in as +23.0°C when it was supposed to be -23.0°C.
Gerald Machnee (19:58:45) : “You state that hourly temps are not important – maybe for yourself…”
My point, which I guess I didn’t make very well, is that I don’t think hourly observations that are used primarily for aviation are indicative enough for use in climate forecasting—except that we can see that the spikes up- or downward were anomalies that lasted for a short time. Look at the debate about station sitings near airports; temperatures from those stations record the effects of the local environment not necessarily climate trends. So the question is how to use readings that are prone to wild swings because of where the stations are situated? For Eureka in particular, because of its terrain of rock and gravels and nothing else except Musk oxen and wolves should one apply UHI adjustment to its readings because clearly something abnormal is going on there.
Gerald Machnee
April 27, 2010 7:33 pm
RE:
Sean Peake says:
April 27, 2010 at 5:46 am
**I don’t think hourly observations that are used primarily for aviation are indicative enough for use in climate forecasting—except that we can see that the spikes up- or downward were anomalies that lasted for a short time.**
The climate forecasters do not really use hourly readings – they will use longer term averages. The spikes you note are probably too complicated and numerous to be taken into account worldwide for long range forecasting. With a station like Eureka you may have to live with what you have. There should not be a UHI adjustment if the instruments are sited properly. The water and slope effects have occurred over a long period of time so long term changes are what should be looked for. There are UHI effects in and near many cities that are not accounted for properly.
Climate forecasting unfortunately uses too many assumptions in the models which causes errors. They are using infilling between stations which changes means.
I have tried to stay away from trying to do climate or seasonal forecasts because of too many built in errors.
With respect to checking errors such as have been found at Eureka and other sites, Budget cuts have resulted in less manpower being available to do detailed verification. At one time everything was checked manually. This has changed to more automated checks and less checks.
I discussed the two Eureka errors or possible errors with a co-worker. On one of them the wind was fairly steady from the north and the temperature still seemed to zoom up 7 degrees, then down again. This is possible in a variable foehn situation, but it seemed quite steady. I think there is also an ice cap or glacier some miles to the north.
Gerald Machnee (19:58:45) : “You state that hourly temps are not important – maybe for yourself…”
My point, which I guess I didn’t make very well, is that I don’t think hourly observations that are used primarily for aviation are indicative enough for use in climate forecasting—except that we can see that the spikes up- or downward were anomalies that lasted for a short time. Look at the debate about station sitings near airports; temperatures from those stations record the effects of the local environment not necessarily climate trends. So the question is how to use readings that are prone to wild swings because of where the stations are situated? For Eureka in particular, because of its terrain of rock and gravels and nothing else except Musk oxen and wolves should one apply UHI adjustment to its readings because clearly something abnormal is going on there.
RE:
Sean Peake says:
April 27, 2010 at 5:46 am
**I don’t think hourly observations that are used primarily for aviation are indicative enough for use in climate forecasting—except that we can see that the spikes up- or downward were anomalies that lasted for a short time.**
The climate forecasters do not really use hourly readings – they will use longer term averages. The spikes you note are probably too complicated and numerous to be taken into account worldwide for long range forecasting. With a station like Eureka you may have to live with what you have. There should not be a UHI adjustment if the instruments are sited properly. The water and slope effects have occurred over a long period of time so long term changes are what should be looked for. There are UHI effects in and near many cities that are not accounted for properly.
Climate forecasting unfortunately uses too many assumptions in the models which causes errors. They are using infilling between stations which changes means.
I have tried to stay away from trying to do climate or seasonal forecasts because of too many built in errors.
With respect to checking errors such as have been found at Eureka and other sites, Budget cuts have resulted in less manpower being available to do detailed verification. At one time everything was checked manually. This has changed to more automated checks and less checks.
I discussed the two Eureka errors or possible errors with a co-worker. On one of them the wind was fairly steady from the north and the temperature still seemed to zoom up 7 degrees, then down again. This is possible in a variable foehn situation, but it seemed quite steady. I think there is also an ice cap or glacier some miles to the north.