<p><strong>The noise level in my data cube strongly increases towards the edge of the field. Can SoFiA deal with this situation?</strong></p>
</dt>
<dd>
<p>Yes, SoFiA can handle data cubes with variable noise levels. There are two basic approaches that one could follow. One method would be to provide a so-called <em>weights cube</em> (parameter <code>import.weightsFile</code>) to SoFiA that describes the noise variation across the entire cube (or rather the inverse thereof). SoFiA will then multiply the data cube with the specified weights cube to create a cube with constant noise level before proceeding with source finding. Parameterisation of the extracted sources will then be carried out on the original data cube again. If the cube in question is a mosaic created in Miriad, a weights cube can be easily generated using the Miriad task <code>mossen</code>.</p>
<p>Another possibility would be to use SoFiA’s built-in <em>noise-scaling</em> functionality (parameter <code>steps.doScaleNoise</code>). This will attempt to automatically measure the rms noise across the specified number of axes of the cube using a user-defined method (standard deviation, median absolute deviation or Gaussian fit to flux density histogram; parameter <code>scaleNoise.statistic</code>). The data cube will then be normalised with respect to the measured rms to achieve a constant noise level prior to source finding.</p>
<p>If the nature of the noise variation is exactly known (e.g. in the case of Miriad mosaics), we recommend using the first method, while the second method is useful in cases where the exact behaviour of the noise variation cannot be reconstructed.</p>
