#
#FitBaseline_demo_feb2011.py: instructions to run FitBaseline
#    Task.
#
# v0.0 31/Jan/2011 ACAB: created for Feb 2011 HIFI DWPS IPAC
#

#For most up-to-date information on baseline removal, see 
#Chapter 13 in "The HIFI User's Manual". 
#As described, FitBaseline is the main task to fit
#and subtract or divide instrumental baseline structures using
#polynomials, but the "sub_base" option on FitHifiFringe
#offers an alternative way of doing it. Care should be
#taken with the latter as it is a smoothed version of the
#line-clipped intensity spectrum.

#load band 7B C+ DBS raster map observation 
#from ~/.hcss/lstore/
#Small offsets are apparent between different sky positions,
#which may be instrumental and should be removed before the
#map cubes are constructed.
obs=getObservation("1342205481",poolName="1342205481_map_dbs_7b")

#Click on obs, and on fitBaseline under 'Applicable Tasks'.
#For this demo, we will run it in command line mode, however.
#The important options are:
#
#movemasks=1 : transfer line masks between spectra, but not sidebands
#
#domask=1 : mask lines by clicking
#
#interactive=True : ask for confirmation before applying polynomial fit.
#                   Most useful if one is not sure about polynomial order
#                   or whether it varies between scans.
#
#doglue=True : combine sub-bands. Use False if there are sub-bands jumps
#              or for non-contiguous HRS modules
#
#order=2 : polynomial order
#
#basemode="subtract" : for strong continuum source one may used "divide"
#
#Note the instructions on the console!
#
#Finish masking for each spectrum by clicking outside of plot box.
#
#Double-click to remove existing mask
#
#By default products WBS-H-LSB and WBS-H-USB are both processed.
#One only needs to define masks once, as these are the same data.
fitBaseline(obs=obs,backends=['WBS-H'],order=2,domask=1,movemasks=1,doglue=True,basemode="subtract",interactive=True,restart=True,smoothResolution=2.0)

#Baseline-subtracted spectra are stored in 'obs', replacing the
#original spectra. Note the:
#
#linemask: contains all masks
#
#fitproduct: contains before/after spectra, masks, and polynomials as well
#            as polynomial coefficients and uncertainties
#

#The linemask can be re-used if one decides to re-do the
#baseline fitting, e.g. after running a more recent pipeline.
#For this, enter previsously basline subtracted obs in the 
#'obsContext' input parameter.
#Suppose that this time we want to divide out the baseline,
#instead of subtract.

obsnew=getObservation("1342205481",poolName="1342205481_map_dbs_7b")
fitBaseline(obs=obsnew,obsContext=obs, backends=['WBS-H'],order=2,domask=1,movemasks=1,doglue=True,basemode="divide",interactive=True,restart=True,smoothResolution=2.0)

#to overplot the WBS-H spectra in obs and obsnew:
#-double-click on 'obs'
#-right-click on WBS-H-USB-->create new variable
#-same for 'obsnew'
#-double click on obs_level2_WBS_H_LSB
#-double click on 'All' twice
#-drag obsnew_level2_WBS_H_LSB to plot
#-double click on 'All' twice

#Example of using FitBaseline on SScan observation
#(it is important to remove baselines before applying
# doDeconvolution task)
#With respect to previous examples, set:
#
# -movemasks=2: once a line is clicked, an algorithm determines
#               whether it is in the USB or LSB. Subsequently
#               the mask is moved between LO settings. This
#               greatly reduces amount of clicking.
#
# -interactive=False: reduce interactively, i.e. no
#               modification of polynomial orders possible.
#               This reduces amount of clicking.
obs=getObservation("1342181161",poolName="1342181161_sscan_dbs_1b_r4")
fitBaseline(obs=obs,backends=['WBS-H'],order=2,domask=1,movemasks=2,doglue=True,basemode="subtract",interactive=False,restart=True,smoothResolution=2.0)