= VALD3 Workshop in Moscow =

We had a very productive workshop in Moscow on 21-28 of September 2012 dedicated to the improvement and finalization of the user interface, to data quality control and preparation of the final VAMDC version of VALD nodes.

Participants locally in Moscow: Nikolai Piskunov (UU), Tanya Ryabchikova (INASAN), Yuri Pakhomov (INASAN), Luydmila Mashonkina (INASAN), Nastya Titarenko (MSU) and Tanya Sitnova (INASAN).

Remote participants: Eric Stempels (UU), Thomas Marquart (UU) and Samuel Regandell (UU).

The topics and achievements of the workshop are listed below.

== User interface ==

The main change takes advantage of the support for multiple units built in VALD3 extraction tools. The new web interface brings the specification of the extraction range in consistency with the units requested for the output wavelengths. For example, if a user selects 1/cm as wavelength units the wavelength range must be specified in 1/cm. Other alternatives are Angstroems in air or vacuum, nanometers in air or vacuum.

These changes required modifications and extensive testing of the web forms, request parsing tools, [[preselect3|PRESELECT]], [[presformat3|PRESFORMAT]] and [[select3|SELECT]]. select tools.

The selection in any units is carried out using the "observed" wavelength stored in VALD3 and NOT using the Ritz wavelength computed from the energy differences by the PRESELECT on the fly. This is connected to the issue with the level energies described below.

== Detecting and Fixing Level Energies ==

Before the change of the interface beta testing of VALD3 revealed a flaw in selection. It manifested itself through truncation of the selected data. The problem was traced to the fact that the range selection was based on the Ritz wavelengths that for several hundred of thousand transitions did not match the stored wavelengths and did not even follow the same sequence as the "observed" wavelengths. We have traced all the cases to the errors in the energies assigned to the levels involved. In many cases we managed to trace to source of confusion and fix it. In other cases we had to get in touch with data producers. In one case, we were forced to pull a (small) line list from the current data collection until we can fix it. Note, that many of energy errors were inherited from VALD2!

== Fixing Term Designation ==

We have noted a few cases of undefined coupling when it should be clearly defined. This was traced to non-robust interpretation of energy level translation table. This is fixed now.

== Improving the Import Tool ==

The IDL tool vald3_ascii2bin.pro was extended to verify the reference labels against the current VALD wiki page. It was also equipped with addition check of energy consistency with the transition wavelength.

== Preparation of the Final Version of the VAMDC VALD Node ==

We have prepared a new dump of the merged VALD data (complete and observed only). The new, improved term parsing tool was updated to handle special case of Hydrogen lines where level energies are actiauuly averaged over the fine splitting.