- Farrow, Daniel J;
- Sánchez, Ariel G;
- Ciardullo, Robin;
- Cooper, Erin Mentuch;
- Davis, Dustin;
- Fabricius, Maximilian;
- Gawiser, Eric;
- Gebhardt, Henry S Grasshorn;
- Gebhardt, Karl;
- Hill, Gary J;
- Jeong, Donghui;
- Komatsu, Eiichiro;
- Landriau, Martin;
- Liu, Chenxu;
- Saito, Shun;
- Snigula, Jan;
- Wold, Isak GB
The construction of catalogues of a particular type of galaxy can be complicated by interlopers contaminating the sample. In spectroscopic galaxy surveys this can be due to the misclassification of an emission line; for example in the Hobby-Eberly Telescope Dark Energy Experiment (HETDEX) low-redshift [O ii] emitters may make up a few per cent of the observed Ly α emitter (LAE) sample. The presence of contaminants affects the measured correlation functions and power spectra. Previous attempts to deal with this using the cross-correlation function have assumed sources at a fixed redshift, or not modelled evolution within the adopted redshift bins. However, in spectroscopic surveys like HETDEX, where the contamination fraction is likely to be redshift dependent, the observed clustering of misclassified sources will appear to evolve strongly due to projection effects, even if their true clustering does not. We present a practical method for accounting for the presence of contaminants with redshift-dependent contamination fractions and projected clustering. We show using mock catalogues that our method, unlike existing approaches, yields unbiased clustering measurements from the upcoming HETDEX survey in scenarios with redshift-dependent contamination fractions within the redshift bins used. We show our method returns autocorrelation functions with systematic biases much smaller than the statistical noise for samples with at least as high as 7 per cent contamination. We also present and test a method for fitting for the redshift-dependent interloper fraction using the LAE-[O ii] galaxy cross-correlation function, which gives less biased results than assuming a single interloper fraction for the whole sample.