Continuum fitting is an important aspect of Ly \alpha forest science , since errors in the estimated optical depths scale with the fractional continuum error .
However , traditional methods of estimating continua in noisy and moderate-resolution spectra ( \mathrm { S / N } \lesssim 10 \mathrm { pixel } ^ { -1 } and R \sim 2000 , e.g .
SDSS ) such as power-law extrapolation or dividing bythe mean spectrum , achieve no better than \sim 15 \% RMS accuracy .
To improve on this , we introduce mean-flux regulated/principal component analysis ( MF-PCA ) continuum fitting .
In this technique , PCA fitting is carried out redwards of the quasar Ly \alpha line in order to provide a prediction for the shape of the Ly \alpha forest continuum .
The slope and amplitude of this continuum prediction is then corrected using external constraints for the Ly \alpha forest mean-flux .
From tests on mock spectra , we find that MF-PCA reduces the errors to 8 % RMS in \mathrm { S / N } \sim 2 spectra , and < 5 \% RMS in spectra with \mathrm { S / N } \gtrsim 5 .
The residual Fourier power in the continuum is decreased by a factor of a few in comparison with dividing by the mean continuum , enabling Ly \alpha flux power spectrum measurements to be extended to \sim 2 \times larger scales .
Using this new technique , we make available continuum fits for 12,069 z > 2.3 Ly \alpha forest spectra from SDSS DR7 for use by the community .
This technique is also applicable to future releases of the ongoing BOSS survey , which is obtaining spectra for \sim 150 , 000 Ly \alpha forest spectra at low signal-to-noise ( \mathrm { S / N } \sim 2 ) .