201.5. DiaSource table

201.5. DiaSource table¶

For the Rubin Science Platform at data.lsst.cloud.
Data Release: Data Preview 1
Container Size: large
LSST Science Pipelines version: r29.2.0
Last verified to run: 2025-09-02
Repository: github.com/lsst/tutorial-notebooks

Learning objective: To understand the contents of the DiaSource table and how to access it.

LSST data products: DiaSource

Packages: lsst.rsp, lsst.daf.butler

Credit: Originally developed by the Rubin Community Science team with feedback from Eric Bellm, Federica Bianco, and Tatiana Cuellar. Please consider acknowledging them if this notebook is used for the preparation of journal articles, software releases, or other notebooks.

Get Support: Everyone is encouraged to ask questions or raise issues in the Support Category of the Rubin Community Forum. Rubin staff will respond to all questions posted there.

1. Introduction¶

The DiaSource table contains measurements for sources--fading or brightening--detected in an individual difference_image with signal-to-noise ratio $\geq5$.

• TAP table name: dp1.DiaSource
• butler table name: dia_source
• columns: 87
• rows: 3,086,404

Related tutorials: The TAP and butler data access services are demonstrated in the 100-level "How to" tutorials. There are 200-level tutorials on difference_images and DiaObjects.

1.1. Import packages¶

Import standard python packages re, numpy, matplotlib and astropy.

From the lsst package, import modules for the TAP service, the butler, and plotting.

In [1]:

import re
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.patches as patches
from lsst.rsp import get_tap_service
from lsst.daf.butler import Butler
from lsst.utils.plotting import (get_multiband_plot_colors,
                                 get_multiband_plot_symbols)

1.2. Define parameters and functions¶

Create an instance of the TAP service, and assert that it exists.

In [2]:

service = get_tap_service("tap")
assert service is not None

Create an instance of the Rubin data butler, and assert that it exists.

In [3]:

butler = Butler('dp1', collections="LSSTComCam/DP1")
assert butler is not None

Define the colors and symbols to represent the LSST filters in plots.

In [4]:

filter_names = ['u', 'g', 'r', 'i', 'z', 'y']
filter_colors = get_multiband_plot_colors()
filter_symbols = get_multiband_plot_symbols()

2. Schema (columns)¶

To browse the table schema visit the Rubin schema browser, or use the TAP service via the Portal Aspect or as demonstrated in Section 2.1.

2.1. Retrieve table schema¶

To retrieve the table schema, define a query for the schema columns of the DiaSource table and run the query job.

In [5]:

query = "SELECT column_name, datatype, description, unit " \
        "FROM tap_schema.columns " \
        "WHERE table_name = 'dp1.DiaSource'"
job = service.submit_job(query)
job.run()
job.wait(phases=['COMPLETED', 'ERROR'])
print('Job phase is', job.phase)
if job.phase == 'ERROR':
    job.raise_if_error()
assert job.phase == 'COMPLETED'

Job phase is COMPLETED

Retrieve the query results and display them as an astropy table with the to_table() attribute.

In [6]:

results = job.fetch_result().to_table()
results

Out[6]:

Table length=87

column_name	datatype	description	unit
str64	str64	str512	str64
apFlux	float	Flux in a 12 pixel radius aperture on the difference image.	nJy
apFlux_flag	boolean	General aperture flux algorithm failure flag; set if anything went wrong when measuring aperture fluxes. Another apFlux flag field should also be set to provide more information.
apFlux_flag_apertureTruncated	boolean	Aperture did not fit within measurement image.
apFluxErr	float	Estimated uncertainty of apFlux.	nJy
band	char	Band used to take this observation.
bboxSize	long	Bounding box of diaSource footprint.
centroid_flag	boolean	General centroid algorithm failure flag; set if anything went wrong when fitting the centroid. Another centroid flag field should also be set to provide more information.
coord_dec	double	Fiducial ICRS Declination of centroid used for database indexing.	deg
coord_ra	double	Fiducial ICRS Right Ascension of centroid used for database indexing.	deg
dec	double	Position in declination.	deg
decErr	float	Error in declination.	deg
detector	long	Id of the detector where this diaSource was measured. Datatype short instead of byte because of DB concerns about unsigned bytes.
diaObjectId	long	Id of the DiaObject that this DiaSource was associated with.
diaSourceId	long	Unique identifier of this DiaSource.
dipoleAngle	double	Dipole orientation	deg
dipoleChi2	double	Chi2 per degree of freedom of dipole fit.
dipoleFitAttempted	boolean	Attempted to fit a dipole model to this source.
dipoleFluxDiff	double	Raw flux counts, positive lobe.	nJy
dipoleFluxDiffErr	double	Raw flux uncertainty counts, positive lobe.	nJy
...	...	...	...
scienceFluxErr	float	Forced PSF flux uncertainty measured on the direct image.	nJy
shape_flag	boolean	General source shape algorithm failure flag; set if anything went wrong when measuring the shape. Another shape flag field should also be set to provide more information.
shape_flag_no_pixels	boolean	No pixels to measure shape.
shape_flag_not_contained	boolean	Center not contained in footprint bounding box.
shape_flag_parent_source	boolean	This source is a parent source; we should only be measuring on deblended children in difference imaging.
snr	double	Ratio of apFlux/apFluxErr
ssObjectId	long	Id of the ssObject this source was associated with, if any. If not, it is set to 0
time_processed	char	Time when the image was processed and this DiaSource record was generated.
trail_flag_edge	boolean	This flag is set if a trailed source contains edge pixels.
trailAngle	double	Angle measured from +x-axis.
trailDec	double	Trail centroid declination.	deg
trailFlux	float	Trailed source flux.	nJy
trailLength	double	Trail length.	pixel
trailRa	double	Trail centroid right ascension.	deg
visit	long	Id of the visit where this diaSource was measured.
x	double	Unweighted first moment centroid, overall centroid	pixel
xErr	float	1-sigma uncertainty on x position	pixel
y	double	Unweighted first moment centroid, overall centroid	pixel
yErr	float	1-sigma uncertainty on y position	pixel

The table displayed above has been truncated.

Option to print every column name as a list.

In [7]:

# for col in results['column_name']:
#     print(col)

Option to use the regular expressions package re to search for column names that contain the string temp.

In [8]:

# temp = 'Err'
# temp = 'Flux'
# temp = 'psf'
# for col in results['column_name']:
#     if re.search(temp, col):
#         print(col)

Delete the job, but not the results.

In [9]:

del query
job.delete()

2.2. Key columns¶

Of the 87 columns of the DiaSource table, a $\sim$dozen are the most commonly used.

2.2.1. DiaSource Id¶

The long integer that uniquely identifies each row of the DiaSource table.

• diaSourceId

2.2.2. DiaObject Id¶

The long integer that identifies the Id of the DiaObject that this DiaSource is associated with.

• diaObjectId

2.2.3. SSObject Id¶

The long integer that identifies the Id of the ssObject that this DiaSource is associated with, if any. If no ssObject is associated with the DiaSource, then this value is set to 0.

• ssObjectId

2.2.4. Filter¶

The filter band used for the observation of the DiaSource.

• band

2.2.5. Visit Id and detector¶

The visit is a long integer that uniquely identifies the visit the DiaSources were detected. The detector is which of the 9 LSSTComCam detectors the DiaSources were detected on. Together, the visit and detector correspond to a processed visit_image.

• visit
• detector

2.2.6. Time of observation¶

Effective mid-exposure time for visit where the DiaSource was detected, expressed as Modified Julian Date, International Atomic Time.

• midpointMjdTai

2.2.7. Coordinates¶

The sky coordinates in decimal degrees for each DiaSource:

• ra, raErr
• dec, decErr

2.2.8. Flux¶

Fluxes are provided in nanoJanskys, which is preffered for difference-image photometry over magnitudes since negative flux measurements would be omitted when converting to magnituides.

PSF fluxes

A forced fit of the Point Spread Function (PSF) at the DiaSource's coordinates in each difference image. PSF fluxes are best to use for point-like sources.

• psfFlux, psfFluxErr

Note that in the DiaSource table, the psfFlux column contains fluxes measured on the difference image. In other tables psfFlux is a flux on the direct (science) image.

Other measurements

• apFlux: fluxes in a 12 pixel radius aperture on the difference_image.
• scienceFlux: Forced PSF flux measured on the direct visit_image.

2.2.9. Flags¶

Pixel flags

A variety of flags indicating whether pixels that are saturated, or affected by cosmic rays, contributed to the DiaSource's measurements.

• pixelFlags_*

Measurement flags

The flux (and shape) measurements mentioned above have associated flag columns suffixed with _flag.

isDipole flag

• isDipole

A boolean flag indicating whether or not the diaSource is classified as a dipole, a likely image-differencing artifact that appears as offset positive and negative lobes. Setting this to False will filter out poorly subtracted stars.

2.2.10. Reliability score¶

• reliability

The purpose of the reliability score in the DiaSource table is to provide a threshold measure to improve the purity of DIA detections. This score is computed using information from the source and image characterization, as well as the information on the Telescope and Camera system. However, given that the reliability score model was trained on a relatively small and limited dataset, some caution is warranted in interpreting the reliability score in DP1, in particular for variable stars. Significant improvements are expected on future releases with LSSTCam data.

2.3. Descriptions and units¶

For a subset of the key columns show the table of their descriptions and units.

In [10]:

col_list = set(['diaSourceId', 'diaObjectId', 'ssObjectId', 'band', 'visit',
                'midpointMjdTai', 'ra', 'dec', 'psfFlux', 'psfFluxErr', 'isDipole'])
tx = [i for i, item in enumerate(results['column_name']) if item in col_list]
results[tx]

Out[10]:

Table length=11

column_name	datatype	description	unit
str64	str64	str512	str64
band	char	Band used to take this observation.
dec	double	Position in declination.	deg
diaObjectId	long	Id of the DiaObject that this DiaSource was associated with.
diaSourceId	long	Unique identifier of this DiaSource.
isDipole	boolean	Source well fit by a dipole.
midpointMjdTai	double	Effective mid-exposure time for this diaSource, expressed as Modified Julian Date, International Atomic Time.
psfFlux	float	Flux derived from linear least-squares fit of PSF model.	nJy
psfFluxErr	float	Flux uncertainty derived from linear least-squares fit of PSF model.	nJy
ra	double	Position in right ascension.	deg
ssObjectId	long	Id of the ssObject this source was associated with, if any. If not, it is set to 0
visit	long	Id of the visit where this diaSource was measured.

Clean up.

In [11]:

del col_list, tx, results

3. Data access¶

The DiaSource table is available via the TAP service and the butler.

Recommended access method: TAP.

3.1. Advisory: avoid full-table queries¶

Avoid full-table queries. Always include a constraint on the coordinates or visit.

The DiaSource table is a large set of all the measurements made in all the difference images of all detected sources.

The DP1 data release DiaSource table is relatively small, however, skipping spatial constraints is not a good habit to form, because future data release DiaSource tables will contain trillions of rows.

3.2. Advisory: not recommended for light curves¶

The ForcedSourceOnDiaObject table is recommended for light curves.

The DiaSource table contains measurements of difference-image sources with $\geq5\sigma$ detections, which mean lower-level variable/transient behavior will not be captured.

The ForcedSourceOnDiaObject table, which contains forced PSF photometry for all difference images as well as non-difference (i.e. visit) images, is recommended for variable and transient object light curves.

3.3. TAP (Table Access Protocol)¶

The DiaSource table is stored in Qserv and accessible via the TAP services using ADQL queries.

Include spatial constraints: Qserv stores catalog data sharded by coordinate (RA, Dec), so ADQL query statements that include constraints by coordinate do not requre a whole-catalog search and are typically faster (and can be much faster) than ADQL query statements which only include constraints for other columns. Use either an ADQL cone or polygon search for faster queries (do not use WHERE ... BETWEEN statements to set boundaries on RA and Dec).

3.3.1. Demo query¶

Define a query to return the "key columns" from Section 2.3.

Impose spatial constraints: search within a 2 arcsecond radius of a known transient in the Extended Chandra Deep Field South (ECDFS) field, RA, Dec = $53.125, -27.740$.

In [12]:

ra = 53.124768
dec = -27.739815

In [13]:

query = """SELECT diaSourceId, diaObjectId, ssObjectId, band, visit,
        midpointMjdTai, ra, dec, psfFlux, psfFluxErr, isDipole
        FROM dp1.DiaSource
        WHERE CONTAINS(POINT('ICRS', ra, dec),
        CIRCLE('ICRS', {}, {}, 0.00056)) = 1
        ORDER BY diaSourceId ASC""".format(ra, dec)
print(query)
job = service.submit_job(query)
job.run()
job.wait(phases=['COMPLETED', 'ERROR'])
print('Job phase is', job.phase)
if job.phase == 'ERROR':
    job.raise_if_error()
assert job.phase == 'COMPLETED'

SELECT diaSourceId, diaObjectId, ssObjectId, band, visit,
        midpointMjdTai, ra, dec, psfFlux, psfFluxErr, isDipole
        FROM dp1.DiaSource
        WHERE CONTAINS(POINT('ICRS', ra, dec),
        CIRCLE('ICRS', 53.124768, -27.739815, 0.00056)) = 1
        ORDER BY diaSourceId ASC

Job phase is COMPLETED

Fetch the results as an astropy table.

In [14]:

results = job.fetch_result().to_table()
print(len(results))

268

Option to display the table.

In [15]:

# results

Filter out the DiaSources that exhibit a dipole

In [16]:

tx = ~results['isDipole']
print(len(results[tx]))

256

Plot the RA and Dec offsets of every DiaSource that does not exhibit a dipole from the search coordinates, in arcseconds.

In [17]:

results['ra_offset_arcsec'] = (results['ra'] - ra) * 3600.0
results['dec_offset_arcsec'] = (results['dec'] - dec) * 3600.0

In [18]:

fig, ax = plt.subplots(figsize=(4, 4))
circle = patches.Circle((0, 0), radius=0.00056*3600.0,
                        facecolor='None', edgecolor='blue')
ax.add_patch(circle)
ax.plot(results[tx]['ra_offset_arcsec'],
        results[tx]['dec_offset_arcsec'],
        'o', ms=2, mew=0, alpha=0.3, color='black')
ax.set_aspect(1)
plt.xlabel('RA offset [arcsec]')
plt.ylabel('Dec offset [arcsec]')
plt.tight_layout()
plt.show()

Figure 1: The offset in RA and Dec, in arcseconds, of every source from the search coordinates. A 2" radius circule marks the search region.

Clean up.

In [19]:

job.delete()
del query, results, tx

3.3.2. Joinable tables¶

DiaObject Table

The DiaSource table can be joined to the DiaObject table on the column containing the unique diaObjectId.

The DiaObject table contains statistical information from DiaSource properties associated with the DiaObject.

Query the same coordinates as the previous subsection (with a smaller 1-arcsecond search radius and band = r) and add a table join to the DiaObject table to retrive the r_psfFluxMax, the maximum r-band difference-image PSF flux, and nDiaSources, the number of diaSources associated with the DiaObject.

In [20]:

query = """SELECT dias.diaSourceId, dias.psfFlux, diao.r_psfFluxMax, diao.nDiaSources
        FROM dp1.DiaSource AS dias
        JOIN dp1.DiaObject AS diao ON dias.diaObjectId = diao.diaObjectId
        WHERE CONTAINS(POINT('ICRS', dias.ra, dias.dec),
        CIRCLE('ICRS', {}, {}, 0.00028)) = 1
        AND dias.band = 'r'
        ORDER BY dias.diaSourceId ASC""".format(ra, dec)
print(query)
job = service.submit_job(query)
job.run()
job.wait(phases=['COMPLETED', 'ERROR'])
print('Job phase is', job.phase)
if job.phase == 'ERROR':
    job.raise_if_error()
assert job.phase == 'COMPLETED'

SELECT dias.diaSourceId, dias.psfFlux, diao.r_psfFluxMax, diao.nDiaSources
        FROM dp1.DiaSource AS dias
        JOIN dp1.DiaObject AS diao ON dias.diaObjectId = diao.diaObjectId
        WHERE CONTAINS(POINT('ICRS', dias.ra, dias.dec),
        CIRCLE('ICRS', 53.124768, -27.739815, 0.00028)) = 1
        AND dias.band = 'r'
        ORDER BY dias.diaSourceId ASC

Job phase is COMPLETED

Fetch the results as an astropy table.

In [21]:

results = job.fetch_result().to_table()

Show the maximum r-band PSF flux and the number of diaSources associated with the DiaObject which was retrieved from the DiaObject table.

In [22]:

print(np.unique(results['r_psfFluxMax'][0]))
print(np.unique(results['nDiaSources'][0]))

[5481.84472656]
[253]

Show that this is consistent with the maximum psfFlux measurement from the diaSources.

In [23]:

np.max(results['psfFlux'])

Out[23]:

np.float32(5481.84)

In [24]:

del ra, dec

SSObject Table

The DiaSource table can also be joined to the SSObject table on the column containing the unique ssObjectId.

Query a 1-degree cone search of the center of the Extended Chandra Deep Field South (ECDFS) field (RA, Dec = $53.13, -28.10$) and add a table join to the SSObject table to identify ssObjects.

In [25]:

ra_cen = 53.13
dec_cen = -28.10

In [26]:

query = """SELECT dias.diaSourceId, dias.ra, dias.dec, sso.ssObjectId
        FROM dp1.DiaSource AS dias
        JOIN dp1.SSObject AS sso ON dias.ssObjectId = sso.ssObjectId
        WHERE CONTAINS(POINT('ICRS', dias.ra, dias.dec),
        CIRCLE('ICRS', {}, {}, 0.5)) = 1
        ORDER BY dias.diaSourceId ASC""".format(ra_cen, dec_cen)
print(query)
job = service.submit_job(query)
job.run()
job.wait(phases=['COMPLETED', 'ERROR'])
print('Job phase is', job.phase)
if job.phase == 'ERROR':
    job.raise_if_error()
assert job.phase == 'COMPLETED'

SELECT dias.diaSourceId, dias.ra, dias.dec, sso.ssObjectId
        FROM dp1.DiaSource AS dias
        JOIN dp1.SSObject AS sso ON dias.ssObjectId = sso.ssObjectId
        WHERE CONTAINS(POINT('ICRS', dias.ra, dias.dec),
        CIRCLE('ICRS', 53.13, -28.1, 0.5)) = 1
        ORDER BY dias.diaSourceId ASC

Job phase is COMPLETED

Fetch the results as an astropy table.

In [27]:

results = job.fetch_result().to_table()

Show the SSObjects associated with diaSources within the 1-degree cone search.

In [28]:

np.unique(results['ssObjectId'])

Out[28]:

<MaskedColumn name='ssObjectId' dtype='int64' length=9>

21163620217073748
21163637482928473
21163645768577093
21164728252512342
21164741002015298
21164745447978322
23133931615301680
23133931615301681
23133931615301682

Plot the RA and Dec offsets of every diaSource from the search coordinates (in degrees) for all diaSources in the search region that are associated with a SSObject.

In [29]:

results['ra_offset_deg'] = (results['ra'] - ra_cen)
results['dec_offset_deg'] = (results['dec'] - dec_cen)

In [30]:

marker_styles = ['o', 's', 'D', '^', 'v', '<', '>', 'P', '*', 'X']

fig, ax = plt.subplots(figsize=(8, 8))
for i, ssobj in enumerate(np.unique(results['ssObjectId'])):
    fx = np.where(results['ssObjectId'] == ssobj)[0]
    ax.plot(results['ra_offset_deg'][fx],
            results['dec_offset_deg'][fx],
            marker_styles[i], ms=8, alpha=0.7, label=ssobj)
ax.set_aspect(1)
plt.xlabel('RA offset [Degree]')
plt.ylabel('Dec offset [Degree]')
plt.legend(loc='lower right')
plt.title('DiaSource RA, Dec offsets from ECDFS center, per SSObject')
plt.tight_layout()
plt.show()

Figure 2: The RA and Dec offsets of diaSources associated with 9 different SSObjects, when those moving objects were within 1 degree of the ECDFS field center.

Clean up.

In [31]:

job.delete()
del query, results

3.4. Butler¶

TAP is the recommended way to access the diaSource table, but the butler is a convenient way to retrive all diaSources detected in a given tract.

Show that the dimension for the dia_source table is just the skymap's tract, and that it is required.

In [32]:

butler.get_dataset_type('dia_source')

Out[32]:

DatasetType('dia_source', {skymap, tract}, ArrowAstropy)

In [33]:

butler.get_dataset_type('dia_source').dimensions.required

Out[33]:

{skymap, tract}

3.4.1. Demo query¶

Retrieve the dataset references (refs) for dia_source tables for tracts that overlap the coordinates near the center of the ECDFS field. There will be only 1.

In [34]:

refs = butler.query_datasets("dia_source",
                             where="tract.region OVERLAPS POINT(:ra, :dec)",
                             bind={"ra": ra_cen, "dec": dec_cen},
                             order_by='tract')
print(len(refs))

1

Define a subset of columns to retrieve from the butler, and retrieve the table data.

In [35]:

col_list = set(['diaSourceId', 'band', 'midpointMjdTai',
                'ra', 'dec', 'psfFlux', 'psfFluxErr'])

results = butler.get(refs[0],
                     parameters={'columns': col_list})
print(len(results))

401814

Subset the results to r-band fluxes between 20 and 25 mag.

In [36]:

tx = np.where((results['psfFlux'] > 360.0)
              & (results['psfFlux'] <= 36000.0)
              & (results['band'] == 'r'))[0]
print(len(tx))

57952

As an example, plot the sky coordinates of DiaSources, and their flux vs. uncertainty.

In [37]:

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(7, 3))
ax1.plot(results['ra'][tx], results['dec'][tx],
         'o', ms=2, mew=0, alpha=0.1, color='grey')
ax1.set_xlabel('Right Ascension')
ax1.set_ylabel('Declination')
ax2.plot(results['psfFlux'][tx], results['psfFluxErr'][tx],
         'o', ms=2, mew=0, alpha=0.1, color='grey')
ax2.set_xlabel('PSF Flux')
ax2.set_ylabel('Error')
plt.tight_layout()
plt.show()

Figure 3: At left, the RA vs. Dec of DiaSources as grey overlapping points, appearing darker for the locations of DiaObjects with many detections. At right, the PSF flux in the difference image, vs. its uncertainty.

In [ ]: