Instruments:ikf

From CedarWiki
Jump to: navigation, search

Inuvik CCD FPI

Contents


Contact Persons

Web Pages

Acknowledgments

The Inuvik fixed-gap imaging Fabry-Perot Spectrometer is operated by the Geophysical Institute of the University of Alaska in collaboration with Environment Canada. Initial funding came from the Aeronomy Program of the National Science Foundation (NSF), with additional support from the joint TIMED/CEDAR program of the National Aeronautics and Space Administration (NASA) and NSF.

Data Description

The fixed-gap imaging Fabry-Perot Spectrometer (FPS) or Interferometer (FPI) at Inuvik in the North West Territories (NWT) of Canada (68.33N, 133.50W) is located on a hill above the Mackenzie River close to the Arctic Ocean. It is 103.2 m above mean sea level and has been operating since late 1998, with reliable data after about February 2000. It is operated by the Geophysical Institute of the University of Alaska, in cooperation with Environment Canada. Chris Strube, and previously Jeff Allen, helps to keep things running.

Because of the extreme seasons in the Arctic, no data are collected May-Aug. The instrument is operated in a trailer using a lap top computer for raw data storage throughout the night. The daylight hours are used to run analysis programs on the previous night's data if there is sufficient time. time. The phone line to the trailer is used to upload the analyzed data to a computer at the Geophysical Institute, where a web page is populated with plots and access to the observations at:
http://gedds.pfrr.alaska.edu/inuvik_FPS/default.htm
The order in their .dbt files is: YYMMDD, UThrfrac*100, azimuth (0=N, 90=E), elevation (0=horiz, 90=vertical), Wn (m/s), Error in Wn, estimated background emission (not used), error in estimated background emission (not used), estimated relative emission, error in estimated relative emission, Tn (K), error in Tn.

The raw images are stored on CDRom and mailed to the Institute. The geophysical output of the FPS are relative emissions, line-of-sight neutral velocities, and neutral temperatures. Older red line data sets do not include the temperature but can be re-analyzed to add them.

The instrument was built by Roger Smith in the early 1990s, using a narrow field-of-view (fov) of 1.5 degrees. The dispersing element is an air-spaced, 150 cm diameter effective clear-aperture Queensgate etalon that is capacitance stabilized and maintains a fixed gap unlike most scanning FPIs where the etalon gap varies and is scanned in wavelength. Scanning FPIs use photomultiplier tubes to collect the signal, whereas here, the light is directed to a CCD imager for collection, and displays circular fringe patterns against a background that must be removed to determine temperature. Images from a frequency stabilized laser are used to characterize the etalon function and deconvolve the sky spectra, but the laser is not sufficiently reliable to correct for the etalon drift.

The FPS is equipped with a periscope to be able to look in any direction, but after 26 February 2000, it was pointed towards the zenith. Hence, the line-of-sight velocities are vertical winds. Horizontal winds are usually determined assuming that the vertical wind component along a non-zenith line-of-sight direction is small.

The original filter used was the atomic oxygen [OI] red line at 630.0 nm (kindat=17001) with airglow emission between about 210 and 300 km, peaking around 240 km. Auroral emissions are lower, with peak emissions as low as 180 km. Observations were summed over 200 seconds, with several seconds needed for storage of the results, resulting in an observational of just under 4 minutes.

After an upgrade in November 2001, the filter used was the atomic oxygen [OI] green line at 557.7 nm (kindat=10002) with an airglow emission peak range near 94-98 km. However, auroral emissions are much brighter and higher, with emission peaks around 110-120 km. The 557.7 nm emission is brighter than the 630.0 nm emission, so observations are integrated over 150 sec throughout the night.

The apex magnetic coordinates of Inuvik in December 2001 at 96 km were (71.25N, -85.60), with a magnetic declination of 30.82 deg (to NE), an inclination of 81.60 deg to SW, and 0 UT at 1310 MLT.

Observations are affected by moonlight and clouds, which usually increase the relative emission due to scattered light. With complete cloud cover, the velocity appears to be very small in all directions. The red line observations initially submitted to the CEDAR Database (DB) include only good data that have been verified by the contributors. However, even with clouds, thermospheric temperature observations are statistically OK as found by Smith and Hernandez (1995). The newer green line observations have be added to the DB as mostly unchecked data (kindat=10002,10012). Only conditions where the laser was obviously not pointing in the right direction all night (high I, low Tn and Wn) were removed. However, there will be conditions when the data are bad such as instrumental problems, snow on the observing dome, etc. Users should contact the data providers.

There is a meteorological office at the airport in Inuvik where an observer looks at the sky every hour. The sky is divided into eigths or octas, and for each eigth of the sky obscured by clouds, the cloud cover increases by 1 octa from 0 (clear skies) to 8 (completely obscured). A value of 9 means obscured by ground conditions (fog, blowing snow, etc). Generally, values of 3 or less for cloud cover are probably OK for velocities, although reflections from bright clouds can be a problem. Conversely, if stars can be seen through the clouds, the observations are probably OK, even if the clouds cover the entire sky (8 octas).

Sky conditions are routinely observed every hour, and are assumed to persist within the hour indicated (e.g. 0 UT applies to 0:00 to 0:59 UT). The actual surface observations where sky conditions were available have been condensed into files of several years which are located at /instruments/ikf.html.

Before addition to the Data Support Section (DSS) of the National Center for Atmospheric Research (NCAR), the cloud cover remarks from stations were first converted to a standard format: octas (eighths). After the institution of the METAR format in 1997, sky cover estimates were converted as follows:

  • 0 = SKC = clear
  • 2 = FEW = a few clouds
  • 3 = SCT = scattered clouds (used 4 in kindat=10002)
  • 6 = BKN = broken clouds
  • 8 = OVC = completely overcast with clouds
  • 9 = VV = no vertical visibility
    These are conservative numbers since FEW=1-2, SCT=3-4, BKN=5-6, OVC=7-8.

REMARKS (RMK) also includes notes of aurora sighting sometimes (AURBO). The aurora were reported as a flag in kindat=10002 (old parameter code 441) which was no report (0), aurora (1), or missing (9 or -32767), where missing can be no surface observation was taken, or when there is ground obscuration (VV) when the sky cover is 9. The aurora were also assumed to persist for 1 hour. The aurora was sighted 24 times on 14 days in 150 days from Nov 2001 to Mar 2002. This flag was removed in the present set since the REMARKS were not easily available. This should not be a serious lack since aurora will also increase the brightness observed. The aurora were also not noted much at the location of Inuvik by the airport observers.

Near real-time hourly surface conditions (e.g. 'Mostly Cloudy', 'Light Snow', 'Drifting Snow', 'Clear', etc) are available at http://www.wunderground.com. Put in 'Canada' in the 'Fast Forecast'. Then click on 'Canada', then 'Northwest Territory', and then 'Inuvik' to get current conditions. Use 'Historical Conditions' to get previous hourly surface conditions. The time is in GMT (UT).

The sky cover estimates from Jan 2000 to April 2005 were taken from the METAR hourly surface observations as kept in the dss461 NCEP (National Centers for Environmental Prediction) surface obs reports and the airways dss472 data for the US and Canada from the Data Support Section (DSS) of NCAR. The dss461 data only keeps the cloud or sky cover at the lowest cloud level, while the cumulative or total cloud cover in up to 6 cloud layers is in the dss472 airways data. The low cloud cover data in dss461 was provided by Doug Schuster, while the low cloud cover and the cumulative cloud cover in the dss472 airways data was provided by Joey Comeaux, both of DSS/NCAR. The cloud cover at the lowest level was the same as from the dss461 data set, except for 6 time in 2001, where 4 of them were missing in the airways set. The cloud cover can increase with height as clouds from other levels are added into the estimate. Hence, parameter code 440 was used to give the biggest estimate of cloud cover using all available levels, while code 441 was revised to list only the sky cover from the lowest level of clouds. The airways data were available through Jan 2005, while the surface obs data were available through April 2005. There were also hourly or longer gaps in the airways data that were filled with the dss461 data. Where the dss461 data only were used, the cumulative cloud cover was set to be the same as the cloud cover from the lowest level. Users should use the cumulative values (code 440) since the FPS has to see through all the cloud levels.

The 2 Nov 2001 to 31 Mar 2002 METAR data used in April 2002 were compared to the low level cloud cover and cumulative airways values. The lowest leval (new parameter code 441) cloud cover was exactly the same 69% of the time, and less 26% of the time, while the other times were marked missing. Comparing the cumulative cloud cover (parameter code 440), showed the cloud cover to be exactly the same 95% of the time and less than the older near real-time estimates only 2% of the time with the remainder times missing. Thus, the cumulative cloud cover (code 440) is nearly the same as the the previous estimates of cloud cover used in kindats 10002.

To summarize, sky cover is a subjective evaluation of the skyward visibility which conservatively characterizes actual FPS viewing conditions at Inuvik. FPS winds are most trustworthy under clearest conditions (0-2 octas), perhaps still okay under partly cloudy conditions (up to 4 octas), and unreliable when cloudier (more than 4 octas).

All etalon air-spaced observations must be corrected for the effects of temperature and pressure changes in the air space, which leads to drifts in the instrument. Since the winds are determined from the Doppler shift, it is important to determine the zero line as it drifts throughout the night as the ambient temperature in the trailer drops. One way to correct for the instrument drift is to assume that the vertical velocities average to zero over the night. This was done in earlier analyses of the data, but after about February 2000, the pressure and temperature values in the trailer were used as basis functions for an initial drift correction. The final correction was made by fitting a weak polynomial (low-pass filter to eliminate outliers) to the vertical velocity which effectively gives it a zero average over the night. The ealier analysis of the vertical velocity used before the temperature was estimated from the image employed a correction filter that forced the end point velocities to zero. Some red line images have been re-analyzed to calculate the temperature, so the velocity analysis has been revised, but most red line days retain the original velocity analysis.

The `errors' given in the data are uncertainties resulting from errors in the fit to the spectra peak position, and in errors from the drift due to the pressure and temperature changes in the air spacing of the etalon. Additional errors arise from fitting the width of the spectra to determine the neutral temperature of the [OI] emitting species. However, fits to the sky spectra are usually good, resulting in small error bars. The image analysis does not account for all the errors, so the actual error is much larger. Including geophysical error bars due to uncertainties in knowing the altitude of the emission range results in temperature errors of about +/-50 K.

Occasionally, the image analysis of the temperature results in suspect values. The elevated red line temperatures found at the beginning of the nights of 9 Dec 2000 and 4 Jan 2001 are not reliable. The unchecked green line data have some bad nights, such as the temperatures on Mar 23, 2002.

Green line temperatures are also subject to 'order hops' where the temperature jumps early in the night such as on 7-10 Jan 2001. These jumps can be over 500 K on 8-11 Dec 2002 and 1-4 Jan 2003. The reason for the jumps is a hardware-related problem where the etalon's orders are shifted and broadened after an order hop. The temperature is good before an order hop, but not later. The etalon's order hop is easily correctable in the baseline or zero-velocity reference with software so Wn is not really affected. Users should be cautious, particularly with the temperature, and contact the data providers for ways to determine if an order hop occurs.

References for the instrument and data processing procedures

Conde, M., Deriving wavelength spectra from fringe images from a fixed-gap single-etalon Fabry-Perot Spectrometer, Applied Optics, 41(14), 2672-2678, 2002.
Conde, M., Analysis of Fabry-Perot spectra of lidar backscatter echoes, In ANARE Reports, No. 146, Morris, R. J., and P. J. Wilkinson, eds., 91-114, 2001.
Hernandez, G., 'Fabry-Perot Interferometers', Cambridge University Press, 343 pp., 1988, second printing with corrections.
Smith, R. W. and G. Hernandez, Upper thermospheric temperatures at South Pole, Adv. Space. Res., 16(5), 31-39, 1995.

Data Files for Inuvik cloud cover in octas and surface obs

URLs for cloud or sky cover

Summary Plots for Inuvik CCD FPI

Summary plots of the relative brightness, line-of-sight neutral wind (vertical velocity), and neutral temperatures from the CCD FPI at Inuvik, NWT, Canada are shown using the atomic oxygen [OI] red (630.0 nm) line, with an emission peak near 240 km. During bright aurora, emissions are lower (down to 180 km). The elevated red line temperatures found at the beginning of the nights of 9 Dec 2000 and 4 Jan 2001 are not reliable.

Red Line (~240 km) Summary Plots

Green Line (~96 km) Summary Plots

Data is available from web site given above.
NOTE: Original data were for 2002 1 Feb to 31 Mar. These data are the same except Feb 1 and 3 were changed , and Feb 15-17 were added. However, the data now go from Nov 2001 to Apr 2005.

Summary plots of the relative brightness, line-of-sight neutral wind (vertical velocity), and neutral temperatures from the CCD FPI at Inuvik, NWT, Canada are shown using the atomic oxygen [OI] green (557.7 nm) line, with airglow emission peak around 96 km. During aurora, emissions are higher (up to 120 km). Cloud cover is represented by 2 lines, the lowest is the lowest level of clouds as seen by an observer, while the highest is the total cloud cover including clouds at higher levels.

These data are not quality checked, and good data could be as low as 20% due to bright moonlight, clouds, poor instrument performance, snow on the observing dome, or other reasons. Check the cloud cover conditions and the data providers before using the data.

Conditions where the temperature is low, the vertical wind is low, and the intensity is relatively high correspond to times when the laser is not pointed correctly. Nights under these conditions (e.g. 2002 Jan 6 or Apr 24), were removed. Old analyses were retained for 2 days in the initial period in 2001 (Dec 5 and 26).

Statistically, temperatures are less affected by clouds than the velocity, which make the Doppler shift small. However, many days show 'order jumps' in the temperature which are unrelated to any changes in brightness or velocity such as in 7-10 Jan 2002. These jumps can be over 500 K at times (e.g. 8-11 Dec 2002). The temperature is good before the jump, but not after, while the vertical velocity is corrected with software and is OK (except with cloud cover). Users should be cautious and consult with the data contact persons. Other temperatures are just poor, like the ones on 23 Mar 2002, being both too low and too high.

The vertical winds relatively large, but there is no reason to think that they are therefore wrong. Again, users should consult with the data providers.


-Revised 31 May 2005 by Barbara Emery