HYSPLIT Forum: hysplitbbs.arl.noaa.gov

Hello,
I’m running Hysplit for backward trajectories over the Atlantic and Pacific ocean.
I was looking in the BAMS paper (https://doi.org/10.1175/BAMS-D-14-00110.1) for the temporal error between the calculated and actual source location for calculations of 240 hrs over oceans, but I couldn’t find it (i might have missed it).
I’ve been running the model in the Ensemble option to have an idea of the dispersion of the back trajectories, but I was looking for a spatial error as a function of hour from the source point.
Thanks,
Michel

The exact spatial error as a function of time is not known.  Running the ensemble option as you are doing gives you an estimate of the uncertainty - if all the trajectories are close together the error is relatively low, if they are far apart, the error is larger.  The trajectories only track the mean wind, not the effects of turbulence. We know there are errors in the gridded meteorology at any one time (doesn't perfectly give the meteorology at every location, and modeling what is happening at smaller scales is excluded).  Further we don't know the meteorology between the time periods provided.  The trajectory calculation is dependent on the 3-d gridded meteorology with time.  There is some explanation at https://www.arl.noaa.gov/documents/work ... ajerro.pdf.   

Thank you very much for your answer!
Given these uncertainties, how far back will you think is reasonable to calculate the back trajectories, especially over oceans? I was asked by a reviewer to provide 240hr back trajectories instead of the 120 hrs i have now in my analysis. Does it make sense to go that far back?

What would the 240-h back trajectories show that your 120-h doesn't? Obviously they cover more territory. People use 240-h back trajectories without knowing exactly how good they are.