The ZTF survey has been running since 2018, and from the beginning has used similar technology to the Rubin alert stream – public alerts with Kafka transport and AVRO messages – and has therefore been a very useful as a prototype to Rubin alerts. The Lasair project has two websites: lasair-ztf.lsst.ac.uk for ZTF and lasair-lsst.lsst.ac.uk is for the Rubin-LSST alerts. There is also a web address lasair.lsst.ac.uk that currently redirects to ZTF and soon will redirect to LSST. We note that (1) lasair-ztf will continue operating and (2) you will need a new login for the LSST website.
Specific guidance on porting your existing ZTF resources is here.
Lasair concepts retained
The basic concepts are the same between ZTF and LSST:
- Filtering with attributes that are tied together with SQL, as well as geometric criteria
- “Active” filters so that results are immediately transmitted via email or kafka
- intelligent crossmatch from Sherlock
- Watchlists and watchmaps; when “active” there is real-time matching
- Annotation by users and other brokers to add to the information portfolio aboout an object
- A REST API and a python client
Key differences
- The LSST survey will produce about 30 times as many alerts as ZTF, up to 10,000,000 per night
- LSST is six bands (ugrizy) and the public ZTF is two bands (gr). In the latter case it is reasonable to compute features for g and r separately, and the filter would have criteria that can be satisfied by one of the two monochrome lightcurve features. But with six bands, we have tried to use the two-dimensional wavelength-time surface rather than six separate monochrome lightcurves.
- Difference Flux replaces Difference Magnitude. The flux of a source is clearly positive, and it is reasonable to take the logarithm to get magnitude. But the difference in flux between two times can be positive or negative, and ZTF used an awkward scheme with the logarithm of the absolute value of the flux difference, together with a separate sign bit. The LSST alerts are simply difference flux in nanoJanskies: 10,000 nJ is magnitude 21.4 and 1,000,000 nJ is magnitude 16.4. Conversion table and formula here.
- Pairs of detections. If a source is rapidly brightening, it is difficult to measure colour if detections in different bands are at different times. But the LSST cadence detects in different bands, separated by only 30 minutes (essentially the same time), from which a reliable colour measure can be made.
New features for filters
- More information is here.
- Jump filter in two different bands: For each band, we take the mean and variance of flux between 70 days and 10 days in the past, then compute the number of standard deviations (sigma) the current flux is from that baseline. The quantity “jump1” is the largest of those, and “jump2” is second largest, from another band.
- Extinction correction: Lasair computes the interstellar dust reddening/extinction corresponding to the sky position of an object, using the Schlegel, Finkbeiner & Davis (SFD) dustmap, and the E(B-V) is reported as. the attribute
ebv. - BazinBlackbody, a 2D temporal spectro parametric fit: Each point of the panchromatic lightcurve is considered to be a flux at given time and wavelength; we try to fit two models and take the best. Each model has a blackbody spectrum in wavelength, and the time part of the model is either exponentional flux (linear rise in magnitude) or Bazin flux, with a rise rate then a fall rate. The fit is made with extinction-corrected flux values.
- Temperature: For ZTF, it is reasonable to use a colour index (magnitude difference) for example “g-r” for the colour of a source, but the paired observations that LSST makes could be any of u+g, u+r, g+r, r+i, i+z, z+y, making it very difficult to understand a colour index. Therefore Lasair makes a blackbody fit to the two fluxes in two bands, and reports an effective temperature. The fit is made with extinction-corrected flux values.
- Absolute magnitude, extinction corrected: If the Sherlock system reports a host galaxy for the object (SN or NT classification), and the redshift of that galaxy is known, then Lasair computes an extinction-corrected absolute magnitude for the object.
Website improvements
- Community resources: While any filter, watchlist, or watchmap can be public, some are emphasised and listed as “community resources”, which you can see by clicking the “Resources” in the left margin of lasair-lsst. One looks for flux jumps in white dwarfs, another finds supernova-like light curves with a host galaxy. There is a watchlist of Abell clusters and RR Lyrae; there is a watchmap of galaxies within 5 Mpc and 50 Mpc.
- More comprehensive documentation: The improved documentation includes a cookbook, new videos, how Lasair is different from other brokers, the API client, and other topics.
- Example Notebooks: These illustrate the use of the API for SQL queries and cone search, for fetching the kafka stream from and active filter, and the implementation of the new lightcurve features described above.
- Multimessenger support: Lasair is now ready to search for the optical counterpart of a graviational-wave event, by sorting LSST alerts by likelihood. If an object has a host galaxy with distance, then the 3D gravitational-wave skymap is used for priority, otherwisr the 2D skymap is used.