Question about forcing/dynamic variables

[Shikamaru-Nara1729]

I would like some clarification about the rationale behind incorporating weekday as a forcing variable in Aurora. The setup utilizes dow_cos and dow_sin as dynamic inputs; I would like to understand the contribution these variables are expected to make towards enhancing the accuracy of Aurora's predictions.

[wesselb]

Hey @Shikamaru-Nara1729! Thanks for opening an issue. We incorporated these variables for the air pollution version of the model, as the day of week could be informative for anthropogenic emissions. E.g., people go to work on Mon-Fri, but don't on Sat and Sun, and that could make a difference. We did not do a careful ablation study to investigate the impact of these variables, so how much they really help is unclear.