IUE Dynamic Spectrum Browser
******************* DISCLAIMER *********************
This browser and its accompanying data files are provided for your
convenience, and are not original data products of the IUE project.
They were developed as part of an
program during 2001, and MAST is not responsible for errors which may have
occurred during the transcription of the original data to reduced form.
For detailed work, it is recommended that the original data products be used.
These browser files are for NEWSIPS data only.
How the Series Were Identified
First, we searched a data base of all the homogeneous object IDs of all
IUE observations which weren't moving targets (solar system objects)
or wave cals, flats or IGM pointings. The resulting sets were saved for
each camera/dispersion combination. Next, duplicate EIDs (as occurs for
multiple settings in the aperture or pointings in which both large and
small aperture data were obtained) were culled from the list. If the culled
set contained resulted in more than 10 observations, then the total number
of observations were counted in the list of objects containing 10 or
more spectra in a set. The results are a reasonable estimate of how many
time series exit. The results are listed in the following table.
Time Series Statistics
Camera Disp Sets Num Mean obs
Obs per set
SWP low 462 16955 36.3
SWP high 316 12515 39.6
LW low 352 12640 35.9
LW high 251 8106 32.3
Thus, there are 1381 potential time series containing 50,216 spectra.
This implies that ROUGHLY HALF of the IUE observations are members of
The Main Browser Page
Each browser page displays a normalized dynamic spectrum. This is
a set of spectra which have been ordered in time and normalized by a mean
spectrum. A few remarks are needed to clarify how this was determined
Currently have analyzed ONLY large aperture low resolution data.
- Dynamic spectrum: The main portion of the page is a GIF image
of a dynamic spectrum. This is a 2-dimensional image of time ordered,
normalized spectra. The ordinate is time, and the abscissa is wavelength.
Each horizontal strip in the image is a single spectrum which has been
normalized by the sample mean. The dynamic spectra in the browser pages
differ from those normally used in one respect. Because the objects were
often sampled very irregularly in time, we elected not to display them with
respect to a linear time axis (as is normally done). Instead, they are
simply stacked sequentially. This avoids problems displaying series with a
few short periods of intense monitoring of the object separated by large
time gaps. In such a case, a linear time scale would cause all of the data
in the dynamic spectrum to be squashed into a few small strips and the rest
of the image to be blank.
- Temporal plot: Because the ordinate of the dynamic spectrum is
sequential instead of linear, we needed to devise another means to display
the temporal sampling. Therefore, we plot the times at which the individual
spectra were obtained to the right of the dynamic spectra. This plot is
aligned with the dynamic spectrum. It has time from the first observation
as the abscissa and the ordinate is also the sequence number. In this way,
the researcher can immediately see the progression of the time sampling and
which spectra in the main image were taken at which time. When the time
sampling curve is nearly vertical for several spectra, this means that they
were obtained over a very short period of time.
- Data Quality Bar: Directly attached to the right side of the dynamic
spectrum and to the left of the time scale, is a strip which gives the
exposure level of each spectrum relative to the best exposure in the set.
This turns out to be a useful means for quickly determining whether a
particular spectrum which does not appear to fit into the sequence may be
a poor exposure and, therefore, not to be trusted.
- Color Bar: Each dynamic spectrum has its own color bar across the
top of the figure. This shows the range of normalized values used in the
dynamic spectra. This varies from one to the next, as some cover a rather
large range of values and others only a few percent. Automatically finding
the proper range for a given object is a challenging task, and we have used
an approach which keys off of percentiles and has provided very good
results, although we are still experimenting.
- Mean Spectrum: Attached to the bottom of the dynamic spectrum is the
mean spectrum of the set which was used in the normalization process. While
this has little meaning in a highly variable object, it typically contains
all of the prominent spectral features and enables the researcher to
identify what aspects of the spectrum are changing.
- Links: Below the dynamic spectrum and its supporting plots, is
a set of links. These are:
- Information on the contents of the dynamic spectra: A link to
- List of spectra used to construct the dynamic spectrum:
This is a link to a detailed list of the spectra used to construct the
dynamic spectrum, along with details about the time they were obtained, the
length of the exposure and the continuum and background levels for the
exposure (extremely important in evaluating the quality of the exposure).
- List of spectra rejected for use in the dynamic spectrum:
This link goes to a simple list of the spectra which were automatically
rejected from the dynamic spectrum. These links are useful for determining
whether the automated rejection scheme functioned properly. Currently
rejection from the sample is done only for spectra which have less than 20%
of the data points usable. This is determined by having the software
inspect quality flags within the files for each spectrum. Spectra with less
than 80% of the data usable or with a mean signal-to-noise ratio less than
2 are included in the plots, but excluded from the calculation of the mean
- Order the spectra from MAST: This link populates a MAST
retrieval form to download the spectra used to construct the dynamic
- This reduced data product was developed under the NASA AISRP:
An acknowledgment to NASA's AISRP program for support.
Send comments to: