ASTRAL ETC test spectra

(1) Photometric error (per resol) heavily filtered to produce relatively featureless reference distribution.

(2) Flux densities convolved with triangular profile comparable to local spectral resolution.

(3) Smoothed flux divided by similarly filtered photometric error (per resol), yielding monochromatic S/N.

(4) Spectrum resampled onto wavelength grid with uniform spacing 0.2 Å over range 1150−3150 Å. For each bin of output spectrum, flux density value is greater of actual flux, if S/N > 3, or heavily smoothed 1σ error, otherwise; taken over all original points falling within wavelength limits of output bin. Resampled spectrum traces envelope that always lies on, or above, statistically significant ("3σ") fluxes, even though highly under-sampled in wavelength compared to original. In resampling procedure, if original spectrum had gap, or all fluxes within output bin flagged as bad, then bin set to zero.

(5) Finally, resampled spectrum screened to retain the non-zero fluxes. Thus, gaps of varying size can occur. Compacted dataset, nevertheless, is appropriate for normal ETC simulations. IT IS NOT SUITABLE FOR SCIENTIFIC ANALYSIS.

Mind the gaps! Could underestimate global count rate if broad gaps are within free spectral range of simulated wavelength range. Also, local count rate of any target feature falling within, or partially in, gap will not be calculated properly. Such cases can be recognized by corresponding gaps in ETC output.

Compactified spectrum always lies above heavily smoothed 1σ photometric error, so in situations where intrinsic fluxes are closer to zero, and spectrum is dominated by few isolated emission features, global count rates could be overestimated. (Most such cases would not trigger global count rate limit, however.)

Nevertheless, local count rate calculation should always be correct (except in gaps) because resampled fluxes are peak values from original spectrum over each interval corresponding to resampled bin. At wavelengths of sharp absorption lines, clipped by bin-maximization procedure, calculated local count rate refers to surrounding continuum. Also, predicted total counts in emission feature will be overestimated if true profile is narrower than about 0.4 Å.