Validation of science data from the CU8 Apsis chain

Figure 1. Diagram of validation processes. Green boxes are MDB tables. Violet boxes are DPC tables available after the CU8COMMON version 17.0.1. Blue boxes indicate tasks that each WP manager should perform: (1) download "All Source Table", (2) generate "Validation Source Table", (3) update Apsis module in such as way to write outputs to "Validation Data", and (4) create a validation report. — **Figure 1**. Diagram of validation processes. Green boxes are MDB tables. Violet boxes are DPC tables available after the CU8COMMON version 17.0.1. Blue boxes indicate tasks that each WP manager should perform: (1) download "All Source Table", (2) generate "Validation Source Table", (3) update Apsis module in such as way to write outputs to "Validation Data", and (4) create a validation report.

**Figure 1**. Diagram of validation processes. Green boxes are MDB tables. Violet boxes are DPC tables available after the CU8COMMON version 17.0.1. Blue boxes indicate tasks that each WP manager should perform: (1) download "All Source Table", (2) generate "Validation Source Table", (3) update Apsis module in such as way to write outputs to "Validation Data", and (4) create a validation report.

During Gaia operations, CU8 APSIS software such as DSC and GSP-Phot will produce outputs including source classification (e.g. stars, galaxies, QSOs) and astrophysical parameters (e.g. temperatures, extinctions, etc). In order to assess these outputs, it is necessary to examine them in various ways such as plotting figures, calculating statistical quantities, etc., which is referred to as the internal validation. In addition there should be an external validation, which compares the Apsis results with other published results on the same objects.

Figure 1 summarizes the validation processes. The basic idea is that each module developer (or WP manager) produces a Validation Source Table listing the Gaia sources he/she would like to obtain Gaia data on (APSIS inputs and outputs) for validating his/her module. These data are then written to a dedicated table during the APSIS chain run. These output tables can then be downloaded by the developer and (manually) examined/compared with external catalogue. There is no automated validation and no separate validation chain.

Figure 2. One example of validation plot. Module developers or validation managers need to examine the output tables manually in order to assess the validity of those outputs. — **Figure 2**. One example of validation plot. Module developers or validation managers need to examine the output tables manually in order to assess the validity of those outputs.

**Figure 2**. One example of validation plot. Module developers or validation managers need to examine the output tables manually in order to assess the validity of those outputs.

Module developers should examine the output tables by calcuating statistics and/or plotting various figures. Figure 2 is one example of such validation figures showing a histogram of astrophyslcal parameters estimated using GSP-Phot.