Exercise #6




  • Problem #1 - The trajectory clustering tutorial assumed the release would be from Dayton, Ohio. The other tracer release point was Sudbury, Ontario (46.62N 80.78W). Would Sudbury have been a better selection for September?

  • Hint - Retrieve traj_freq_control.txt, change the starting location to Sudbury, and then run the daily trajectories and determine the optimum number of clusters.

  • Solution - clusplot.html. Does the three-cluster solution look any better?

  • Problem #2 - Is the 600 m release height realistic?

  • Hint - Rerun the calculation using a 10 m AGL release height.

  • Solution - The result is not that different.

  • Problem #3 - Repeat the 10 m AGL release but this time for October.

  • Hint - The meteorological data file (RP198310.gbl) should already be in the CAPTEX directory. If not, download it through the ARL Data FTP / Archive menu.

  • Solution - The result dramatically illustrates one of the benefits of clustering, showing four orthogonal trajectory clusters, naturally selected according to the flow patterns for that month. Cluster #4 presents the ideal case for sampling in the CAPTEX network, but only with a probability of 9%.