Flux and Error Bars

  Once an objects has been identified on the direct image and the location of its spectrum in the grism/spectrum image has been determined, the spectral extraction can proceed. The spectrum is constructed using

$$F_l= \sum w_{i,l} \cdot g_{i,l}$$ (4)

where the sum is over pixels with flux g_i,l and weight w_i,l at a given wavelength l.

Individual error estimates for each pixel are taken from the array 'err' of the input grism image, which should include all sources of uncertainty for the charge in each individual pixel. This errors are propagated to compute the error at each wavelength in the extracted spectrum as

$$\epsilon_{{\rm STAT},l} = \sqrt{ \sum (w_{i,l}\cdot\epsilon_{i,l})^2}.$$ (5)

This statitical error is one component of the final error estimates (see section 3.5.7).