Wave partitioning in WW3 - is there an energy leakage?

[oleaarnes]

I’m doing a WW3 run in a large ocean basin where multiple swell systems occur simultaneously. I’ve used the default partition configuration PTM1 which returns n (i.e. NOSW) swell components in order of decreasing wave height. When comparing the magnitude (i.e. square root of the sum of squares) of the Hs from all partitions (including wind sea partition) with the total Hs derived directly from the 2D wave spectra I notice there is some energy leakage, see figure below. This is principally observed over low to moderate values of total Hs, which is where the multi-modal sea states occur. The leakage decreases when increasing the number of swell partitions to output (e.g. NOSW = 5), however for my application it is not practical to output this many partitions.
Question: Is anyone aware of how I can configure WW3 to output a total of 3 partitions: 1 dedicated for windsea, 1 for the primary swell and 1 representing the residual swell. The residual swell would then be the combined sum of the swell systems with wave height below the primary swell.

[mickaelaccensi]

Dear @oleaarnes ,

I've never looked too deep the energy leakage in partitioning but your graphic shows it well.
About partitioning, there is actually no method to match exactly your request, the only one that could fit is PTM4 which splits it into 2 partitions, wind-sea and swells but you won't have the first swell and the others. Please have a look at the user manual in the section "spectral partitioning"

I hope it helps,

Best,

Mickael