Orbital drift of the EIS wavelength scale#
There is a shift of the spectral line position during the 98.5 minute Hinode orbit that is due to the thermal changes occuring across the instrument during the orbit. The effect is clearly seen if a velocity map is made from an EIS raster, such as the example below.
Wide, alternating bands of red and blue shift are seen that have an amplitude of about 35 km/s and thus mostly dominate the real solar Doppler shifts.
The amplitude is approximately fixed in wavelength/pixel space to 0.0223 angstroms/1 pixel, and thus the velocity amplitude varies with wavelength. E.g, for Fe XII 195.12 it is 35 km/s, while for Fe XV 284.16 it is 24 km/s.
Two methods are available to users for correcting the orbital drift: one uses instrument housekeeping data, while the other uses the spectral data themselves. Tests are being performed to compare the two methods but generally inexperienced users should use the housekeeping data method.
NOTE: the accuracy of EIS absolute velocities obtained with these methods is no better than 4 km/s.
Instrument housekeeping data correction#
This method is described in detail by Kamio et al. (2010) and basically makes use of temperature readings within the EIS instrument to model how the Fe XII 195.12 line drifts on the detector over the course of the mission. Unlike the correction method based on measured centroid positions (see below) there is no need to make an assumption about the large scale velocity structure in a single raster.
Measured centroid correction#
This method uses line centroid positions measured from a single raster to define the orbit correction for that same raster. It is somewhat risky since it requires an assumption about large-scale velocity flows within the raster. E.g., often one makes an assumption that a quiet Sun region within the raster is at rest.
The method is implemented through the process of performing Gaussian fits to the emission lines in a data-set, and so the reader is referred to the documents listed below:
The method essentially requires an initial guess to be made for the orbit variation using the routine eis_wave_corr, and then refinements are made using Gaussian fits to the line of interest.