Minimize Time Domain Averaging

Time Domain Averaging processing is dedicated to Short Wavelength Infra-Red (SWIR) channels and aims to calculate the Time Delay and Integration (TDI) grid and all parameters of SLSTR Level-1B product provided on it. It reduces noise by applying an average to these channels.

This algorithm is computed only if the TDI stripe has been selected for one of the SWIR channels (S4 to S6). This is selected by users through processing parameters.

The SWIR channels are acquired using a two dimensional array, composed of two columns along the scan direction, i.e. in the across track direction. Each column is associated to four pixels of 500 m resolution, and the on-ground displacement along the scan between two acquisitions is 500 m. Hence, the swath is acquired twice in the SWIR channels; once by each column of the array at two successive times.

The processing comprises two main steps:

  1. Re-indexing of the SWIR calibrated reflectances: the first column is referenced as stripe A and the second column is stripe B. Since the re-indexing of detectors has already been carried out in source packet processing, this step is just the allocation of detectors 0 to 3 to Stripe A and detectors 4 to 7 to Stripe B.
  2. Averaging of the two columns: the average is applied to all pixels except those belonging to the first and last columns of each scan that will respectively take the values of the A column of the array for the first PIX10SYNC sample of the nadir swath, and of the B column of the array for the last PIX10SYNC sample of the nadir swath. (Note: this is contrary to the oblique view swath)
Logical Flow of Time Domain Averaging for SWIR Channels