I'm in the process of project shopping, but at the moment am starting to look at noise properties in the Lyman alpha forest. The goal is to quantify the effects of various components of noise on the power spectrum, P(k), derived from the Ly-alpha forest.
Although the results of the power spectrum are dependent on many physical properties (such as whether to include or how to deal with Damped Lyman Alpha systems, velocity broadening, metal-line systems, and continuum fitting), I plan to study the effects due to the data itself. Namely, four components add sufficient noise to affect our results.
1) Measurement noise -- read, sky & object noise (Poisson)
2) Sky-subtraction
3) Spectro-photometry -- effects due to the amount of light entering the fiber based on position on plate (?), which would largely affect the continuum fits of the Ly-A forest
4) Wavelength calibration
Not sure if this is correct, but it seems as though spectro-photometry and sky subtraction would cause the biggest source of error. The first test I'll conduct will be the effects of sky subtraction. I'll use the mock catalog which has 'perfect' data, unaffected by noise, and add only that noise due to sky subtraction. The mock catalog is made such that to emulate the BOSS survey (while simultaneously made with the correct distribution to give the correct power spectrum), so I'll use sky spectra from fibers closest (in ra/dec) to our mock spectra. After adding sky subtraction noise, I'll compute the correlation function on the sample to see how it deviates from the ideal case.
