NIRCam hot pixel flagging

grizli/aws/visit_processor.py

@@ -2309,7 +2309,9 @@ def cutout_mosaic(rootname='gds', product='{rootname}-{f}', ra=53.1615666, dec=-
                                      snowblind_kwargs=snowblind_kwargs,
                                      weight_type=weight_type,
                                      rnoise_percentile=rnoise_percentile,
-                                     calc_wcsmap=calc_wcsmap)
+                                     calc_wcsmap=calc_wcsmap,
+                                     **kwargs,
+                                     )
 
         outsci, outwht, header, flist, wcs_tab = _
 

grizli/jwst_utils.py

@@ -26,6 +26,21 @@
 # PP file WCS
 DO_PURE_PARALLEL_WCS = True
 
+JWST_DQ_FLAGS = [
+    "DO_NOT_USE",
+    "OTHER_BAD_PIXEL",
+    "UNRELIABLE_SLOPE",
+    "UNRELIABLE_BIAS",
+    "NO_SAT_CHECK",
+    "NO_GAIN_VALUE",
+    "HOT",
+    "WARM",
+    "DEAD",
+    "RC",
+    "LOW_QE",
+]
+
+
 def set_crds_context(fits_file=None, override_environ=False, verbose=True):
     """
     Set CRDS_CONTEXT
@@ -196,6 +211,41 @@ def set_quiet_logging(level=QUIET_LEVEL):
         pass
 
 
+def get_jwst_dq_bit(dq_flags=JWST_DQ_FLAGS, verbose=False):
+    """
+    Get a combined bit from JWST DQ flags
+
+    Parameters
+    ----------
+    dq_flags : list
+        List of flag names
+
+    verbose : bool
+        Messaging
+
+    Returns
+    -------
+    dq_flag : int
+        Combined bit flag
+
+    """
+    try:
+        import jwst.datamodels
+    except:
+        msg = f"get_jwst_dq_bits: import jwst.datamodels failed"
+        utils.log_comment(utils.LOGFILE, msg, verbose=verbose)
+        return 1
+
+    dq_flag = 1
+    for _bp in dq_flags:
+        dq_flag |= jwst.datamodels.dqflags.pixel[_bp]
+
+    msg = f"get_jwst_dq_bits: {'+'.join(dq_flags)} = {dq_flag}"
+    utils.log_comment(utils.LOGFILE, msg, verbose=verbose)
+
+    return dq_flag
+
+
 def hdu_to_imagemodel(in_hdu):
     """
     Workaround for initializing a `jwst.datamodels.ImageModel` from a
@@ -3492,4 +3542,169 @@ def time_corr_photom_copy(param, t):
     amplitude, tau, t0 = param["amplitude"], param["tau"], param["t0"]
     corr = amplitude * np.exp(-(t - t0)/tau)
 
-    return corr
+    return corr
+
+
+def flag_nircam_hot_pixels(data='jw01837039001_02201_00001_nrcblong_rate.fits', hot_threshold=7, max_filter_size=3, hot_filter_sn_max=5, plus_sn_min=4, corner_sn_max=3, jwst_dq_flags=JWST_DQ_FLAGS, verbose=True):
+    """
+    Flag isolated hot pixels and "plusses" around known bad pixels
+
+    Parameters
+    ----------
+    data : str, `~astropy.io.fits.HDUList`
+        NIRCam image filename or open HDU
+
+    hot_threshold : float
+        S/N threshold for central hot pixel
+
+    max_filter_size : int
+        Size of the local maximum filter where the central pixel is zeroed out
+
+    hot_filter_sn_max : float
+        Maximum allowed S/N of the local maximum excluding the central pixel
+
+    plus_sn_min : float
+        Minimum S/N of the pixels in a "plus" around known bad pixels
+
+    corner_sn_max : float
+        Maximum S/N of the corners around known bad pixels
+
+    jwst_dq_flags : list
+        List of JWST flag names
+
+    verbose : bool
+        Messaging
+
+    Returns
+    -------
+    sn : array-like
+        S/N array derived from ``file``
+
+    dq : array-like, int
+        Flagged pixels where ``hot = HOT`` and ``plus = WARM``
+
+    count : int
+        Number of flagged pixels
+
+    Examples
+    --------
+
+        .. plot::
+            :include-source:
+
+            import numpy as np
+            import matplotlib.pyplot as plt
+            import astropy.io.fits as pyfits
+
+            from grizli.jwst_utils import flag_nircam_hot_pixels
+
+            signal = np.zeros((48,48), dtype=np.float32)
+
+            # hot
+            signal[16,16] = 10
+
+            # plus
+            for off in [-1,1]:
+                signal[32+off, 32] = 10
+                signal[32, 32+off] = 7
+
+            err = np.ones_like(signal)
+            np.random.seed(1)
+            noise = np.random.normal(size=signal.shape)*err
+
+            dq = np.zeros(signal.shape, dtype=int)
+            dq[32,32] = 2048 # HOT
+
+            header = pyfits.Header()
+            header['MDRIZSKY'] = 0.
+
+            hdul = pyfits.HDUList([
+                pyfits.ImageHDU(data=signal+noise, name='SCI', header=header),
+                pyfits.ImageHDU(data=err, name='ERR'),
+                pyfits.ImageHDU(data=dq, name='DQ'),
+            ])
+
+            sn, dq_flag, count = flag_nircam_hot_pixels(hdul)
+
+            fig, axes = plt.subplots(1,2,figsize=(8,4), sharex=True, sharey=True)
+
+            axes[0].imshow(signal + noise, vmin=-2, vmax=9, cmap='gray')
+            axes[0].set_xlabel('Simulated data')
+            axes[1].imshow(dq_flag, cmap='magma')
+            axes[1].set_xlabel('Flagged pixels')
+
+            for ax in axes:
+                ax.set_xticklabels([])
+                ax.set_yticklabels([])
+
+            fig.tight_layout(pad=1)
+
+            plt.show()
+
+    """
+    import scipy.ndimage as nd
+    from jwst.datamodels.mask import pixel as pixel_codes
+
+    if isinstance(data, str):
+        is_open = True
+        rate = pyfits.open(data)
+    else:
+        rate = data
+        is_open = False
+
+    bits = get_jwst_dq_bit(jwst_dq_flags)
+
+    mask = (rate['DQ'].data & bits > 0) | (rate['ERR'].data <= 0)
+    mask |= (rate['SCI'].data < -3*rate['ERR'].data) | (~np.isfinite(rate['SCI'].data))
+
+    if 'MDRIZSKY' in rate['SCI'].header:
+        bkg = rate['SCI'].header['MDRIZSKY']
+    else:
+        bkg = np.nanmedian(rate['SCI'].data[~mask])
+
+    indat = rate['SCI'].data - bkg
+
+    if 'BKG' in rate:
+        indat -= rate['BKG'].data
+
+    indat[mask] = 0.    
+    sn = indat / rate['ERR'].data
+    sn[mask] = 0
+
+    ##########
+    # Isolated hot pixels
+    footprint = np.ones((max_filter_size, max_filter_size), dtype=bool)
+    footprint[(max_filter_size - 1) //2, (max_filter_size - 1) //2] = False
+
+    snmax = nd.maximum_filter(sn, footprint=footprint)
+
+    hi = sn > hot_threshold
+    hot = hi & (snmax < hot_filter_sn_max)
+
+    ###########
+    # Plus mask
+    plus = np.array([[0,1,0], [1,0,1], [0,1,0]]) > 0
+    corner = (~plus)
+    corner[1,1] = False
+
+    sn_up = sn*1
+    sn_up[mask] = 1000
+
+    dplus = nd.minimum_filter(sn, footprint=plus)
+
+    dcorner = nd.maximum_filter(sn, footprint=corner)
+
+    plusses = (dplus > plus_sn_min) & (dcorner < corner_sn_max)
+    plusses &= (rate['DQ'].data & bits > 0)
+
+    plus_mask = nd.binary_dilation(plusses, structure=plus)
+
+    dq = (hot * pixel_codes['HOT']) | (plus_mask * pixel_codes['WARM'])
+
+    msg = f'flag_nircam_hot_pixels : hot={hot.sum()} plus={plus_mask.sum()}'
+    utils.log_comment(utils.LOGFILE, msg, verbose=verbose)
+
+    if is_open:
+        rate.close()
+
+    return sn, dq, (dq > 0).sum()

grizli/utils.py

@@ -21,6 +21,7 @@
 from sregion import SRegion, patch_from_polygon
 
 from . import GRIZLI_PATH
+from .jwst_utils import JWST_DQ_FLAGS
 
 KMS = u.km/u.s
 FLAMBDA_CGS = u.erg/u.s/u.cm**2/u.angstrom
@@ -5852,7 +5853,10 @@ def drizzle_from_visit(visit, output=None, pixfrac=1., kernel='point',
                        calc_wcsmap=False,
                        niriss_ghost_kwargs={},
                        snowblind_kwargs=None,
-                       get_dbmask=True):
+                       jwst_dq_flags=JWST_DQ_FLAGS,
+                       nircam_hot_pixel_kwargs={},
+                       get_dbmask=True,
+                       **kwargs):
     """
     Make drizzle mosaic from exposures in a visit dictionary
 
@@ -5921,7 +5925,15 @@ def drizzle_from_visit(visit, output=None, pixfrac=1., kernel='point',
     snowblind_kwargs : dict
         Arguments to pass to `~grizli.utils.jwst_snowblind_mask` if `snowblind` hasn't
         already been run on JWST exposures
-
+
+    jwst_dq_flags : list
+        List of JWST flag names to include in the bad pixel mask.  To ignore, set to
+        ``None``
+
+    nircam_hot_pixel_kwargs : dict
+        Keyword arguments for `grizli.jwst_utils.flag_nircam_hot_pixels`.  Set to
+        ``None`` to disable and use the static bad pixel tables.
+
     Returns
     -------
     outsci : array-like
@@ -5947,6 +5959,7 @@ def drizzle_from_visit(visit, output=None, pixfrac=1., kernel='point',
 
     from .prep import apply_region_mask_from_db
     from .version import __version__ as grizli__version
+    from .jwst_utils import get_jwst_dq_bit, flag_nircam_hot_pixels
 
     bucket_name = None
 
@@ -6077,6 +6090,20 @@ def drizzle_from_visit(visit, output=None, pixfrac=1., kernel='point',
 
             #bpdata = 0
             _inst = flt[0].header['INSTRUME']
+
+            # Directly flag hot pixels rather than use mask
+            if (_inst in ['NIRCAM']) & (nircam_hot_pixel_kwargs is not None):
+                _sn, dq_flag, _count = flag_nircam_hot_pixels(
+                    flt,
+                    **nircam_hot_pixel_kwargs
+                )
+                if (_count > 0) & (_count < 8192):
+                    bpdata = (dq_flag > 0)*1024
+                    extra_wfc3ir_badpix = False
+                else:
+                    msg = f' flag_nircam_hot_pixels: {_count} out of range'
+                    log_comment(LOGFILE, msg, verbose=verbose)
+
             if (extra_wfc3ir_badpix) & (_inst in ['NIRCAM','NIRISS']):
                 _det = flt[0].header['DETECTOR']
                 bpfiles = [os.path.join(os.path.dirname(__file__),
@@ -6276,7 +6303,11 @@ def drizzle_from_visit(visit, output=None, pixfrac=1., kernel='point',
 
                 # JWST: just 1,1024,4096 bits
                 if flt[0].header['TELESCOP'] in ['JWST']:
-                    dq = flt[('DQ', ext)].data & (1+1024+4096)
+                    bad_bits = (1 | 1024 | 4096)
+                    if jwst_dq_flags is not None:
+                        bad_bits |= get_jwst_dq_bit(jwst_dq_flags, verbose=verbose)
+
+                    dq = flt[('DQ', ext)].data & bad_bits
                     dq |= bpdata.astype(dq.dtype)
 
                     # Clipping threshold for BKG extensions, global at top