Era5 wave spectra scaling

Hello,

I had downloaded the 2d wave spectra of 2016 to compute integrated parameters out of reduced frequency range.

I started by checking that I get comparable values to the published significant wave heights if I use all frequencies.

But I get comparable values if I calculate 2*sqrt(integral(E)) instead of 4 times the square root.

I don't understand why the different factor. Am I missing something?


Nir