Sky Brightness in LSST



In [1]:

    
%pylab inline









    



Populating the interactive namespace from numpy and matplotlib



In [14]:

    
import astropy.units as u

Read Opsim Cadence

See opsim db summary table docs.



In [4]:

    
import sqlite3



In [49]:

    
import astropy.table



In [5]:

    
conn = sqlite3.connect('/Data/DESC/minion_1016_sqlite.db')



In [6]:

    
c = conn.cursor()

Fetch 1000 rows of r-band observations.



In [50]:

    
c.execute("""SELECT airmass,dist2Moon,moonAlt,moonPhase,filtSkyBrightness
             FROM summary WHERE filter = 'r' LIMIT 1000""")
opsim = astropy.table.Table(
    np.array(c.fetchall()),
    names=('airmass', 'dist2Moon', 'moonAlt', 'moonPhase','filtSkyBrightness'))



In [ ]:

    
conn.close()

Predict Sky Brightness

See specsim atmosphere docs and speclite filter docs. The moon's contribution to the sky brightness is calculated using Based on Krisciunas and Schaefer, “A model of the brightness of moonlight”, PASP, vol. 103, Sept. 1991, p. 1033-1039.



In [3]:

    
import speclite.filters



In [8]:

    
import specsim.atmosphere
import specsim.config



In [9]:

    
cfg = specsim.config.load_config('desi')
atm = specsim.atmosphere.initialize(cfg)



In [31]:

    
def predict(airmass, dist_to_moon, moon_altitude, moon_illumination, filter_name='sdss2010-r'):
    """Predict the sky brightness 
    
    Parameters
    ----------
    airmass : float
        Airmass at the field center of the visit.
    dist_to_moon : float
        Distance from the field center to the moon's center on the sky, in radians.
    moon_altitude : float
        Altitude of the Moon taking into account the elevation of the site, in radians.
    moon_illumination : float
        Percent illumination of the Moon (0=new, 100=full).
    filter_name : str
        Name of the filter to use for calculating the sky brightness.
        
    Returns
    -------
    float
        AB magnitude of the sky in the specified filter.
    """
    
    atm.airmass = airmass
    atm.moon.moon_zenith = (np.pi / 2 - moon_altitude) * u.rad
    atm.moon.separation_angle = dist_to_moon * u.rad
    # Convert illumination percentage to temporal phase in the range [0,1]
    # with 0 = full moon, 1 = new moon.
    atm.moon.moon_phase = np.arccos(2 * (moon_illumination / 100.) - 1) / np.pi
    
    filt = speclite.filters.load_filter(filter_name)
    wlen = cfg.wavelength
    return filt.get_ab_magnitude(atm.surface_brightness * u.arcsec ** 2, wlen)

Compare OpSim with SpecSim



In [51]:

    
r_predicted = []
for obs in opsim:
    r_predicted.append(predict(obs['airmass'], obs['dist2Moon'], obs['moonAlt'], obs['moonPhase']))
r_predicted = np.array(r_predicted)



In [54]:

    
plt.plot(opsim['filtSkyBrightness'], r_predicted, 'x')
plt.xlim(20, 21.4);

Hmm, I was expecting better agreement. It looks like there are subsets of observations with tight correlations. What opsim parameters define these different subsets? Try coloring points using additional opsim vars.



In [66]:

    
plt.scatter(opsim['filtSkyBrightness'], r_predicted, lw=0., s=15., c=opsim['airmass'])
plt.colorbar().set_label('airmass')
plt.xlim(20, 21.4);



In [65]:

    
plt.scatter(opsim['filtSkyBrightness'], r_predicted, lw=0., s=15., c=opsim['dist2Moon'])
plt.colorbar().set_label('moon separation (rad)')
plt.xlim(20, 21.4);



In [64]:

    
plt.scatter(opsim['filtSkyBrightness'], r_predicted, lw=0., s=15., c=opsim['moonAlt'])
plt.colorbar().set_label('moon alt (rad)')
plt.xlim(20, 21.4);



In [62]:

    
plt.scatter(opsim['filtSkyBrightness'], r_predicted, lw=0., s=15., c=opsim['moonPhase'])
plt.colorbar().set_label('moon phase')
plt.xlim(20, 21.4);

It looks like the moon phase is the best predictor of discrepancies between specsim and opsim.



In [ ]: