== Noise mask of magnetic field strength for the inverted full-disk vector magnetic field data ==

The noise mask of the magnetic field strength is a full-disk map that represents the median of the quiet-Sun magnetic field strength over the Sun's disk. The noise values at a given location are used in some versions of the disambiguation code to optimize performance.

A [[http://hmi.stanford.edu/images/mag/HMI.BNoise.jpg|Noise Mask]] is derived from a group of FD10 full-disk magnetic field strength data observed at a specific HMI-Sun velocity. It has been demonstrated that the noise level of the measured field strength varies over the Sun's disk, and the pattern of this noise variation changes with the orbital velocity (Vr) of HMI relative to the Sun. Thus it is necessary to generate Vr-dependent noise masks. The method is described below. 

 . First, the field strength maps are grouped by the relative velocity of HMI to the solar disk center, Vr. Bins are constructed for each 50 m/s in velocity and span 100 m/s in range. For example, for Vr = -950 m/s, the field strength maps were observed when the velocity was in the range [-900, -1000] m/s.

 . Second, the median of the field strength for each on-disk pixel is calculated from the selected field strength data. Strong field pixels where field > 300.0 Gauss, roughly 3 sigma of field noise, are excluded. About 40 maps are considered in each velocity range.

 . Finally, a 15-order 2D polynomial is performed to fit the median map. The functions are Chebyshev polynomials of the first kind. The fitted polynomial is the noise mask for Vr_center. The coefficients of the polynomials, which are used to reconstruct the noise mask, are archived in JSOC2 with a series name of su_yang.cheby_coef or hmi_test.lookup_ChebyCoef_BNoise and [eventually] on JSOC at hmi.lookup_Cheby``Coef_BNoise.

Changes of instrument settings took place at several instants (tabulated below). When the settings change, the noise appears to change slightly. To reflect these changes, a noise mask needs to be generated specifically for each time period. The masks currently available in JSOC are generated using FD10 data from April-June 2012. Thus they can be used for data after 19 January 2012.  Based on comprison of samples at Vr=0 over a longer interval, no significant changes in the noise mask were observed to occur between instrument reconfigurations.


===== Inputs =====
 . During each time interval approximately 40 computations were used for each velocity interval when determinining the median.

===== Outputs =====
 . For each time interval we save sets of polynomial coefficients that can be used to reconstruct the noise mask for any velocity. Coefficients are calculated for each 50m/s step in Sun-HMI velocity.

===== Cadence =====
 . A new mask is needed when the instrument characteristics change.

 . '''Time Intervals'''
  || Time || Instrument Status ||
  || 2010.05.01 || Initial HMI Configuration ||
  || 2010.12.13_19:47 || Change tuning from 500 to 600 ||
  || 2011.07.13_18:35 || Change tuning from 600 to 620 ||
  || 2011.07.13_18:40 || Change exposure time from 120ms to 125ms ||
  || 2012.01.18_18:15 || Change tuning from 620 to 640 ||

===== Archive =====
 . The coefficients of the polynomials used to reconstruct the noise mask are archived in JSOC2 in the series su_yang.cheby_coef, hmi_test.lookup_ChebyCoef_BNoise, or [eventually on jsoc] hmi.lookup_Cheby``Coef_BNoise. The prime keys are Vr and an index for each time interval.
 . Samples for various velocities after 2012:01.18 are available (on jsoc2) at su_yang.cheby_image

===== Noise Mask Characteristics =====
 . [[http://hmi.stanford.edu/images/mag/HMI.BNoise.jpg|Noise Mask]]. Images show the variation in the median of the quiet-Sun magnetic field across the disk at HMI zero velocity relative to the Sun center and at the extremes of +-2200 m/s. The noise tends to a maximum near the limbs and a minimum near disk center. Values range between ~70 and ~140 Gauss.

 . [[http://hmi.stanford.edu/images/mag/HMI.BNoise.Variation.jpg|Noise Mask Velocity Variation]]. The variation through the orbit is not uniform, but varies from place to place. The velocity-dependent median value is shown for five areas around the disk. In most areas the total range is less than 30 Gauss. The west limb is more sensitive, changing by nearly twice as much through an orbit.

===== Issues =====

  * Not all time periods are currently available.
  * It takes some time after changing the instrument to reconstruct a new mask

===== Responsible scientist =====
 . Yang Liu


----
    * Back to [[VectorMagneticField]]
    * Back to MagneticField
    * Back to FrontPage