The outlier detection step accepts a weight_type parameter which is used:
for weighting during resampling
for weighting the input images when not resampling
The default weight_type is ivm which uses the var_rnoise to calculate the weight. However both coron and tso data lose their var_rnoise array during input conversion to outlier detection (where the input cube is converted to a model container). See #8473 where for tso data this lack of weighting was made more explicit.
For both of these modes the input cubes contain a var_rnoise array which could be used. However as they have not so far been used enabling their use wold change the output data.
Should these modes use input weighting and if so, should ivm be a valid weight type (based off the var_rnoise array)?
Issue JP-3684 was created on JIRA by Brett Graham:
The outlier detection step accepts a
weight_typeparameter which is used:The default
weight_typeisivmwhich uses thevar_rnoiseto calculate the weight. However both coron and tso data lose theirvar_rnoisearray during input conversion to outlier detection (where the input cube is converted to a model container). See #8473 where for tso data this lack of weighting was made more explicit.For both of these modes the input cubes contain a
var_rnoisearray which could be used. However as they have not so far been used enabling their use wold change the output data.Should these modes use input weighting and if so, should
ivmbe a valid weight type (based off thevar_rnoisearray)?