Error estimates

  The main source of errors in the deblending procedure are uncertainties in the estimated contamination c_ij from each contaminating spectrum due to the simplifying assumptions spelled out above. These errors depend on the structure in the contaminating object as a function of wavelength and are hard to quantify. The errors can be big and depend on the flux in the contaminating spectrum. For example, in the case of a faint object being contaminated by a bright object, the subtracted flux might be considerably larger than the remaining flux in the spectrum. As a simple rough estimate of the errors introduced by the deblending procedure, a user supplied fraction DEBLEND_ERR_MIN is used, i.e. the error is estimated as

 

$$\epsilon_{{\rm DEB},i} = \sqrt{\sum_{j=1,n_{ov}} ( \delta_{... ... {\tt DEBLEND\_ERR\_MIN} \times I_{i,j}\times F_{meas,j}, )^2}$$ (21)

where the $\delta_{i,j}$ assures that only the flux from the contaminating spectra is considered.

Realistic values of DEBLEND_ERR_MIN are on the order of 5 to 20%.