Orville Wideband Imager

Description

The Orville Wideband Imager is a realtime GPU-based all-sky imager for the output of the Advanced Digitial Processor (ADP) broadband correlator that runs at LWA-SV. Orville receives visibility data from ADP for 32,896 baselines, images the data, and writes the images to the disk in a binary frame-based format called "OIMS". The imaging is performed using a w-stacking algorithm for the non-coplanarity of the array. For each image, the sky is projected onto the two dimensional plane using orthographic sine projection. To reduce the number of w-planes needed during w-stacking, the phase center is set to a location approximately 2 degrees off zenith that minimizes the spread in the w coordinate. The gridding operation is based on the Romein gridder implemented as part of the EPIC project. Every 5 seconds, the imager produces 4 Stokes (I, Q, U and V) images in 198 channels, each with 100 kHz bandwidth.

Data Archive

Orville data with reduced spectral resolution (six 3.3 MHz channels) are available at the LWA data archive.

Reading OIMS Files

You can use the OrvilleImageDB.py Python module to read the data stored in an OIMS file:

from lsl_toolkits.OrvilleImager import OrvilleImageDB

db = OrvilleImageDB(oimsFile, 'r')

# Get parameters from the input file

ints = db.nint # number of integrations
station =  db.header.station # station info
stokes = db.header.stokes_params # Stokes parameter info
inp_flag = db.header.flags # flag info
file_start = db.header.start_time # file start time
file_end = db.header.stop_time   # file end time
ngrid = db.header.ngrid # image size (x-axis)
psize = db.header.pixel_size # angular size of a pixel (at zenith)
nchan = db.header.nchan # number of frequency channels

# There are two main ways to read data from the OIMS file:
#  - From a single integration
#  - From all integrations at once

# 1. Reading a particular image integration
#    (Usually 720 integrations (each 5 seconds) in an hour) 

hdr, dat = db.__getitem__(710) # collecting header and data from the 710 th integration

# The dat array contains the image data in a four dimensional array of the form [nchan,stokes,xgrid,ygrid]
# where nchan = 198 frequency channels, stokes = 4 [stokes (I, Q, U,V)], xgrid = grid size in the x direction,
# ygrid = grid size in the y direction (typically xgrid = ygrid = ngrid)
# Copy over to numpy arrays for further processing (e.g. image subtraction and transient searches).

# Individual header items can be accessed from the hdr dictionary:

t = hdr['start_time'] # starting time in MJD
int_len = hdr['int_len'] # length of each integration
lst = hdr['lst'] # starting LST time of observation
start_freq = hdr['start_freq']/1e+6 # starting Frequency in MHz
stop_freq = hdr['stop_freq']/1e+6  # stopping Frequency in MHz
bandwidth = hdr['bandwidth']/1e+6 # bandwidth of the observation
cent_ra = hdr['center_ra'] # phase center RA
cent_dec = hdr['center_dec'] # phase center Dec
cent_az = hdr['center_az']   # phase center coordinates, azimuth, Ideally towards zenith, changed due to w-projection 
cent_alt = hdr['center_alt'] # phase center coordinates, elevation, Ideally towards zenith, changed due to w-projection 

# 2. Reading all integrations at once

hdr_list, data_all = db.read_all()

# data_all contains the image data in a 5d masked array of the form [integration,nchan,stokes,xgrid,ygrid]
# where integration is the number of integrations (typically 720 per hour) and the other axes are the same
# as above. hdr_list is a list of dictionaries, one per integration. This can easily be converted to a 
# pandas dataframe object by doing:

import pandas as pd
hdr_df = pd.DataFrame(hdr_list)

# Close the database file
db.close()