Extremely large wavelength difference of the detector map adjusted with few sky lines.

pfs_pipe2d version: w.2022.43

The detector map adjusted with very few sky lines shows the large wavelength difference (>1 nm for arm 'b' of the SuNSS data) from the wavelength solution in the original detector map before adjustment.

An example of the SuNSS data (arm b of visit=80906 obtained in 2022-10-05) is shown below.

Only a few sky lines are detected in the image of arm b. Especially, there is no sky lines seen in the blue side (lower side in the attached calExp image).

When this image is reduced with reduceExposure.py using the reduced calibration data prepared in 2022-06-30, the wavelength solution in the adjusted detector map shows a very large difference from that of the original detector map. Below image shows the wavelength difference between the detector maps (the original wavelength solution was subtracted from the adjusted wavelength solution). The wavelength difference is larger than 1 nm in the red side of the detector. 

When these two detector maps are compared with the Ar lamp image (visit=79997 obtained in 2022-09-25), the original detector map matches well with the detected Ar line positions. In the below image, the positions of Ar lines in the detector map are plotted with dots (red: the detector map adjusted with SuNSS data, blue: the original detector map) over the calExp image of Ar lamp data shown with gray scale. Only bottom quarter of the image, where the wavelength difference is quite large, is shown here. Other three quarter images are attached to this ticket.

Therefore, the adjustment of the detector map seems to fail in the region where very few or no sky lines are detected.