Merge pull request #2 from drlaw1558/drl_combine1d

Add median filtering to background spectrum

jwst/master_background/master_background_step.py

@@ -2,6 +2,7 @@
 import numpy as np
 from stdatamodels.properties import merge_tree
 from stdatamodels.jwst import datamodels
+from scipy.signal import medfilt
 
 from jwst.datamodels import ModelContainer
 from jwst.stpipe import record_step_status
@@ -25,6 +26,7 @@ class MasterBackgroundStep(Step):
         save_background = boolean(default=False) # Save computed master background
         force_subtract = boolean(default=False) # Force subtracting master background
         output_use_model = boolean(default=True)
+        median_kernel = integer(default=1)
     """
 
     def process(self, input):
@@ -51,6 +53,10 @@ def process(self, input):
             has already been applied and, if so, the master background step is skipped.
             If set to True, the step logic is bypassed and the master background is subtracted.
 
+        median_kernel : integer, optional
+            Optional user-supplied kernel with which to moving-median boxcar filter the master background
+            spectrum.  Must be an odd integer.
+
         Returns
         -------
         result : `~jwst.datamodels.ImageModel`, `~jwst.datamodels.IFUImageModel`, `~jwst.datamodels.ModelContainer`
@@ -140,6 +146,12 @@ def process(self, input):
                         exptime_key='exposure_time',
                     )
 
+                    # If requested, apply a moving-median boxcar filter to the master background spectrum
+                    if (self.median_kernel > 1):
+                        self.log.info('Applying moving-median boxcar of width %i',self.median_kernel)
+                        master_background.spec[0].spec_table['surf_bright'] = medfilt(
+                            master_background.spec[0].spec_table['surf_bright'], kernel_size=[self.median_kernel])
+
                     background_data.close()
 
                     result = ModelContainer()