Assign a date to a cluster

[jwich]

Hi there

I ran the HYSPLIT trajectories at 1 location in Pretoria, South Africa from mid Apr 2017 to mid Apr 2018, so 1 year. I then did the cluster analyses for the entire year. 5 clusters were optimal. The frequency of days in each cluster is:

Cluster 1: 22%
Cluster 2: 31%
Cluster 3: 38%
Cluster 4: 4%
Cluster 5: 5%

3 heights were used every 6 hrs in 24 hrs, so 12 trajectories per day

There is no way to get the dates allocated to cluster numbers in HYSPLIT.... so I then opened CLUSLIST5 output file in Notepad, saving as txt file and then imported into SAS.

For each day I then calculated the mode of the cluster of the 12 trajectories on each day. However in SAS is there is a tie, e.g. mode is 6 for cluster 1 and mode is 6 for cluster 3, then SAS will select the lowest value, so then cluster 1 and not cluster 3 will be selected.... so the frequency of days in each cluster is is slightly different in SAS than the HYSPLIT output:

Cluster 1: 26%
Cluster 2: 32%
Cluster 3: 35%
Cluster 4: 4%
Cluster 5: 4%

How can I solve this issue?

Many thanks

[barbara.stunder]

What do you mean "the dates allocated to cluster numbers in HYSPLIT"? In the CLUSLIST5 file it shows which trajectories (dates) are in each cluster by cluster number. I do not use SAS.