> Why are there small differences in my trajectories when I run more than one at a time versus running just one trajectory?
Why are there small differences in my trajectories when I run more than one at a time versus running just one trajectory?
To save computer time in processing a large number of trajectories on the
web, the trajectory computation is done using a sub-grid that
moves along with the trajectory. There is no point in processing all of
the meteorological data, especially the global data (360x181), if you are
interested in only a few points. However, the diagnostic vertical motion
options (all but the model vertical motion) require a computation of the slope
(horizontal and vertical) of the field in question. The slope is computed using
a centered difference in the sub-grid interior and a simple
difference along the edges. When the trajectory endpoint approaches the
edge of the sub-grid, the sub-grid is re-centered and then the data fields are
recomputed. So when you add additional trajectories the sub-grid
calculation (size and location) may not be the same, so each trajectory
of the multiple trajectory calculation may approach the edge at a
different time than in a previous calculation with a different number of
trajectories, which means that for one trajectory at a particular point
the vertical motion may have been computed with a simple difference or a
centered difference, resulting in a slightly different trajectory. This is not
an issue if you choose the default vertical motion option.
On the PC, the simple solution to this problem is to increase the size of the
subgrid in the namelist configuration file (SETUP.CFG).