The Swift UVOT grism calibration is still progressing. In the past years a new way to anchor the wavelength scale was developed by using a mapping from the target sky position to an anchor point in the first order grism spectrum. The calibration is being done for the whole detector, not just the centre. Recently, the sensitivity, or flux calibration has been worked on. The flux calibration was completed for the uv grism. One essential ingredient had to be a correction for the coincidence loss.
The UVOT grism calibration in the HEASARC CALDB and the FTOOLS uvotimgrism software apply only when the source is located close to the boresight of the instrument. Only recently we started to understand the level of contamination of second order overlap in the uv grism better. Cooler calibration stars were used to extend the uv grism flux calibration to about 5000A. The accuracy of the current calibrations wavelength scale depends on how well the source position is determined from the zeroth order positions, and is estimated to be of order of 2-3 pixels for the UV grism, and slightly better for the V grism. For about 4% of the grism images no aspect solution can be found by the FTOOLS. The flux calibration was mainly done using white dwarf standards. The current wavelength calibration for the uv clocked grism is not optimal and will be upgraded in the near future.
The new wavelength calibration is based on a scaled version of the Swift UVOT/XMM OM optical model, is valid for the whole image, and is based on mapping the position from a source in one of the lenticular filters directly to the first order position in the grism image. If the grism image was taken in an observing mode without a lenticular filter directly before or after the exposure, the position can be derived from the grisms zero-order aspect solution also. That is the case for most of the early grism exposures, and the accuracy is still not well understood.Some personal information has been included, but it is a bit dated.