 Problem #1  The meteorology error tutorial section showed results only using the NARR and WRF27UW data. Add the ERA40 and WRF27 data to the graphic.
 Hint  Retrieve traj_fwrd_control.txt to get started.
 Solution  Note that the first trajectory in the display list determines the image domain. The ERA40 is the northern most trajectory while the NARR is the southern most trajectory.
 Problem #2  Determine the sensitivity of the trajectories to the integration time step by comparing the calculation using the default time step with an arbitrary one set to a fixed value equal to the meteorological data interval. Computational differences can be enhanced using the 3 km resolution WRF data which is provided at a 15 min temporal resolution.
 Hint  Define the WRF03 data file and compute the trajectory (without using SETUP.CFG). Check the MESSAGE file to determine the timestep used in the calculation (1‐5 min). Then open the advanced menu and force the time step to 15 minutes. After changing the name of the trajectory endpoints file, rerun the model.
 Solution  Although trajectories using the smaller domain 3 km data terminate near the OhioPennsylvania border, it is already clear that the midboundary layer trajectories are starting to become different from each other. The faster trajectory (red) was calculated with the variable time step while the slower trajectory (blue) used the fixed 15 min time step. The longer fixed time step forces the calculation to skip over grid points. Unlike other modeling approaches, Lagrangian calculations will not fail if the time step criterion is violated.
 Problem #3  Repeat problem #1, but this time only include the four variations of the wrf27km meteorology files. These would be wrf27, wrf27uw, wrf27mc, and the four daily wrfout files in the hysplit/datem directory.
 Hint  Retrieve traj_fwrd_control.txt to get started.
 Solution  Three of the four wrf27km trajectories are very similar and almost overlap entirely, except for the one using the timeaveraged winds (wrf27mc), which is much further north than the ones computed using the instantaneous data, more similar to the previous trajectory using ERA40 data. All the others also contain a distinct kink in the trajectory around 2100 UTC of the 27th. A closer examination of the wind field at that time shows an area of divergence south of Long Island, suggesting that trajectories passing through that region would have much greater uncertainty. The effects of timeaveraging will be examined again in subsequent sections.
