JP-3444: update tso photometry to use ApertureStats

CHANGES.rst

@@ -83,6 +83,13 @@ stpipe
   added a `query_step_status()` function to use as an alternative to checking
   `self.skip`. [#8600]
 
+tso_photometry
+--------------
+
+- Replaced photutils.aperture do_photometry with photutils.ApertureStats to remove NaNs
+  when performing TSO photometry for non SUB64 WLP8 pupil data. For SUB64 WLP8 changed
+  summing data in the aperture to use np.nansum to ignore NaNs. [#8672]
+
 tweakreg
 --------
 

docs/jwst/tso_photometry/description.rst

@@ -31,6 +31,18 @@ the times will be computed from the exposure start time, the integration time,
 and the number of integrations.  In either case, the times are
 Modified Julian Date, time scale UTC.
 
+If NaNs exist in the source or background annulus, they are masked out and the value
+returned is the sum over the real-valued data.
+This is different from the convention for photometry in standard imaging mode:
+in that case, a NaN value is returned if it is present in any of the pixels in
+the source aperture. In the imaging case, the mostly likely cause of NaN pixels
+in the aperture is NaN-valued central pixels for a saturated source in an image
+with many sources. For TSO data, it is more likely that  single pixels are affected
+in a given integration and science analysis will be focused on variability of one source.
+If NaNs are present, the absolute flux will be underestimated, but the relative values may still
+be useful.
+
+
 The output table contains these fields:
 
 - MJD

jwst/tso_photometry/tso_photometry.py

@@ -6,7 +6,7 @@
 from astropy.table import QTable
 from astropy.time import Time, TimeDelta
 import astropy.units as u
-from photutils.aperture import CircularAperture, CircularAnnulus
+from photutils.aperture import CircularAperture, CircularAnnulus, ApertureStats
 
 from stdatamodels.jwst.datamodels import CubeModel
 
@@ -79,25 +79,28 @@ def tso_aperture_photometry(datamodel, xcenter, ycenter, radius, radius_inner,
                 'subarray.  No background subtraction was performed.')
 
         for i in np.arange(nimg):
-            aperture_sum.append(np.sum(datamodel.data[i, :, :]))
+            aperture_sum.append(np.nansum(datamodel.data[i, :, :]))
             aperture_sum_err.append(
-                np.sqrt(np.sum(datamodel.err[i, :, :]**2)))
+                np.sqrt(np.nansum(datamodel.err[i, :, :]**2)))
     else:
         info = ('Photometry measured in a circular aperture of r={0} '
                 'pixels.  Background calculated as the mean in a '
                 'circular annulus with r_inner={1} pixels and '
                 'r_outer={2} pixels.'.format(radius, radius_inner,
                                              radius_outer))
         for i in np.arange(nimg):
-            aper_sum, aper_sum_err = phot_aper.do_photometry(
-                datamodel.data[i, :, :], error=datamodel.err[i, :, :])
-            ann_sum, ann_sum_err = bkg_aper.do_photometry(
-                datamodel.data[i, :, :], error=datamodel.err[i, :, :])
-
-            aperture_sum.append(aper_sum[0])
-            aperture_sum_err.append(aper_sum_err[0])
-            annulus_sum.append(ann_sum[0])
-            annulus_sum_err.append(ann_sum_err[0])
+            aperstats = ApertureStats(datamodel.data[i, :, :],
+                                      phot_aper,
+                                      error=datamodel.err[i, :, :])
+
+            annstats = ApertureStats(datamodel.data[i, :, :],
+                                     bkg_aper,
+                                     error=datamodel.err[i, :, :])
+
+            aperture_sum.append(aperstats.sum)
+            aperture_sum_err.append(aperstats.sum_err)
+            annulus_sum.append(annstats.sum)
+            annulus_sum_err.append(annstats.sum_err)
 
     aperture_sum = np.array(aperture_sum)
     aperture_sum_err = np.array(aperture_sum_err)