red line trolley The Red Line Fabry-Perot Interferometer (FPI) at Millstone Hill monitors the upper thermospheric 6300 OI emission on a nightly basis, weather permitting. The instrument provided thermospheric wind and temperature data from 1989-2001 using photomultiplier detection and FPI pressure scanning. The FPI was upgraded in 2008-2009, and is now a Doppler imaging FPI. The instrument returned to regular service in October of 2009.

The Millstone Hill Doppler Imaging FPI is at latitude 42 36 42.82 N, longitude 71 29 4.70 W, and an altitude of 340 feet above sea level. The magnetic declination is 15 2 W changing 4/year W, and the magnetic inclination is 68 10 down, changing by 6/year up. (These magnetic coordinates are as of January 22, 2010.)

The imaging FPI is an f/13.8 system, with 10.16 cm clear aperture and a 140 cm effective focal length. The 50.8mm diameter 6300 interference filter is 4.0 broad (FWHM), in a telecentric optical configuration that minimizes image artifacts caused by cosmetic filter defects. Spectral filter defects (like fringing) are removed by flat-field calibration. The etalon system is an air-gap, with 1.0525 cm spacing. This provides a 0.1886 free spectral range at 6300.3 , and a typical spectral resolution of 0.021. Seven orders of the 6300 airglow emission are sampled simultaneously with an Andor ikon-M 934-BV camera featuring a 1024x1024 E2V back illuminated CCD with a 13.3 mm x 13.3 mm image area and 13 m2 pixel size . This detector provides a quantum efficiency of nearly 95% at 6300 , with dark noise of 0.00013 e- s-1 pixel-1 at -90OC. Typical instrument performance produces 6300 wind vectors with accuracies of approximately 1 m/s and temperatures with accuracies of 150 K using four-minute exposure times. Exposure times can be enhanced to reduce temperature errors, or reduced to improve time resolution of wind vectors.

A twin-mirror pointing system with 0.5 degree pointing accuracy mounted beneath a clear plexiglass dome is used to point to any direction in the sky, and to point downward to a flat-field white light calibration box. The standard operating mode is to make measurements in the four cardinal directions (North, South, East, and West) at a 45 zenith angle, with a fifth measurement in the zenith. Using 4-minute exposures in each direction, approximately 70 meridional and zonal wind vectors are collected each night.

The Millstone Hill Imaging FPI can be configured to automatically operate in 6300 observing modes (look directions, integration times) specified by the experimenter. An unlimited number of modes can be scheduled by the user each night, and for weeks in advance. Full remote operation is also integrated into the system, via PC-to-PC communication. The FPI instrument function is sampled using a frequency stabilized HeNe laser transmitting at 6328.165 . Laser, CCD bias, and dark current calibrations can be easily accomplished via remote control, day or night. Similarly, flat-field calibrations can be completed by remote control, these only during the night.

Ring image processing proceeds by assembling multiple bias, dark, and flat-field images. Flat-field images need not be gathered each night, only at times when some aspect of the instrument configuration has been altered. Bias and dark images are assembled each night. In reality, the CCD dark current is so low that the dark images are dominated by the CCD bias, and dark current is negligible even for exposure times exceeding five minutes. The bias, dark, and flat-field images are median filtered in time, to remove cosmic ray induced high-charge pixels. Each image is also passed through a 8 by 8 pixel running standard-deviation filter to remove hot or dark pixels. Airglow (sky) images and laser images are also filtered with the standard-deviation filter. The dark images are averaged in time, and subtracted from the airglow and laser images. Finally, an averaged flat-field image is normalized to unity at its highest ADU value, and divided into each airglow and laser image to account for variable response across the CCD. The flat-field removal also serves to remove image artifacts that are the consequence of spectrsal defects in the interference filter or other optical elements. Because defects in the pointing mirrors can contribute to uneven response across the optical path, the flat-field calibration source is sampled through the pointing mirror system.

Ring images processed in this manner are then analyzed for extraction of the geophysical parameters, including line-of-sight winds in each cardinal direction, meridional and zonal winds, meridional and zonal gradients, temperatures, and relative brightness. The imaged FPI ring pattern is analyzed by summing electrons in radial annuli, or radial bins, whose radial distances (bin number) from ring-pattern center are linear in wavelength. Because the annuli mapped over square pixels consequently contain slightly different numbers of pixels, the unit of merit becomes signal/pixel. These bins are then aligned linearly to produce the 6300 Doppler spectra.

These bin summed spectra covering five spectral orders are fit with a multiple Gaussian function using the Levenberg-Marquardt non-linear least squares algorithm. The data supplied to the community uses a quadratic background functional form and statistical weighting in this process. After the Gaussian fit a Fourier decomposition is used to separate the instrument function from the sky emission function in each order. After the sky emission functions in each order are isolated in frequency space,, an inverse transform establishes the airglow spectrum free of the instrument function,is now centered at its Gaussian line center and redefined on a 1 free spectral range (FSR) grid. Now the emission function for each order can be summed, fully realizing the field widening advantage of an imaging FPI. A single Gaussian is fit to the summed profile, and geophysical parameters are extracted.

There is no calibrated low brightness source currently available at Millstone Hill, so only relative brightness data are reported. The relative brightness is useful for comparing brightness values during a single night of data, or across several nights wherein the instrument configuration remains unchanged. The relative brightness values may not useful for comparison across seasons, years, or the solar cycle. Temperatures are determined from the FWHM of the summed emission profile.

Winds are extracted by comparing the line center of measurements in the cardinal direction to the line center of the zenith measurements. It is assumed that vertical winds are insignificant relative to the horizontal winds. The zenith line center position is monitored throughout the night, and does not remain precisely constant, due either to the presence of small vertical winds, or to slight thermal instrument drift, or both. Frequency stabilized HeNe laser calibrations are also distributed throughout the night to monitor instrument drift. Any drift is corrected by normalizing the time series of laser samples to their mean, and applying these corrections to all airglow emission line centers measured throughout the night, in any direction.

The analysis also assumes that the wind is constant within the instrument field-of-view. The five-order airglow images used for analysis has a full field-of-view of 1.8, or 7.85 km for an emission at 250 km altitude measured in the zenith. In the cardinal directions, observed using a 45 zenith angle, the field-of-view covers an 11.11 km diameter region. We assume that the wind is constant within these fields-of-view. Line-of-sight winds are calculated by subtracting the drift-corrected line center bin position in a cardinal direction from the zenith center position interpolated in time to match the observation time at the cardinal point. These line-of-sight vectors are cosine corrected to lie in the emitting layer, and use the convention of positive northward, and positive eastward. Meridional [zonal] winds are calculated by summing the North-South [East-West] line-of-sight vectors and dividing by two. Each North or South [East of West] measurement creates a meridional [zonal] vector by interpolating the corresponding South or North [West of East] vector corresponding to the time of the North or South [East or West] observation. Meridional [zonal] gradients, South-to-North [West-to-East] are calculated by subtracting the North [East] line-of-sight vector from the South [West] line-of sight vector, and dividing by the distance between the two emission regions, which is established by the emission height and the observation zenith angle. We now use a smaller zenith angle (45) at Millstone Hill than was used in the past, or is commonly used at other facilities due to increasing light contamination is Westford, MA. Using 45 zenith angle, the distance between meridional or zonal emission points is near 500 km. The gradients are produced as m s-1 per 500 km, and the calculated gradient per km is normalized to 500 km.

Before extracting the geophysical parameters, data quality filtering is performed to prevent transmission of contaminated data, and to identify data that may be of questionable quality. Data points are rejected if any of any of several conditions are met. If individual data values, extracted from a single airglow exposure, survive the initial quality control, these are further classified using a quality code of 0,1 or 2 in order of decreasing quality. A complete description of the data analysis technique and quality control filtering is available from the madrigal database and at

The imaging FPI at Millstone Hill is a community asset currently operated by Scientific Solutions Inc. of Chelmsford, MA. Technical and Scientific assistance, and building maintenance is provided by the Atmospheric Science group at the MIT Haystack Observatory. Data from the common-mode measurements are deposited directly to the Madrigal data system. Interested community users should use the contact information below to use the instrument for special experiments. Guest users can choose to operate the red line FPI themselves, on-site or by remote control, or they can work with SSI and Haystack staff to complete a desired experiment. Similarly, guest users can collaboratively use existing data analysis algorithms, or develop their own.

Bob Kerr
Scientific Solutions Inc.
55 Middlesex Street #210
Chelmsford, MA 01863
(978) 251-4554 ext. 7018