How low of trajectory starting height is appropriate?

[sethnlyman]

I am using HYSPLIT to generate back-trajectories in very stable conditions under strong, multi-day, wintertime temperature inversions. I recognize that meteorological models generally perform poorly under these conditions, but investigating the performance of models in these conditions is the purpose of the HYSPLIT work we are doing. We have prepared our own HYSPLIT met data from some WRF runs. We have been using starting heights that are between 5 and 50 meters above ground. If I go higher than 20 or 30 meters above ground I get radically different results than when I start lower, which is what we expect, since strong inversions are strongly vertically stratified and wind conditions change dramatically in the different layers. A colleague brought up the possibility that starting trajectories at these very low heights could lead to calculation errors or artifacts. the only possibility I'm aware of is that my trajectories may hit the ground, but if they do they should continue their horizontal transport anyway. Are there any other worries or problems I should be aware of when doing trajectories this low to the ground? Thank you. -Seth

[ariel.stein]

When trajectories hit the ground they might loose information and their reversibility can be questioned. For an example and a short discussion about this topic please follow this link:
http://ready.arl.noaa.gov/documents/Tut ... _isob.html