Request from user

I am working on regional 6y maps for DIN in the Baltic.

Last time I had fixed correlation lengths and I want to improve the result by for example limit the correlation length near the coast with the function lengraddepth. I don’t understand how to choose the parameters L and hmin?

For example: with L = 1000 there is a sharp border at the coast but with L = 10_000 there seems to also be limitations in the open sea:

I also tried fitvertcorrlen but it generated lenz varying between 2 and over 100 m and the result was very patchy. When I tried fithorzcorrlen – the code crashed.

I have tried a lenz dependent on the depth resolution but I get really strange gradients near the surface with high values at 0 m and a decrease at 5 m. I remember I got this result also last time and that is why I ended up with a fixed lenz=25m. In the Baltic in winter, the surface is well mixed down to the halocline and it would be nice to have this result also from the diva model.

An example: this is DIN-profiles from a position in the open sea for the winter. Lenz is twice the depth resolution and in the surface lenz is thus 10 m. There are very sharp gradients in the surface that are not present in observations, do you have an idea on what to tune to get a more realistic result? If I increase lenz in the surface it gets better but still there are these gradients. I also tried to do a run with lenz=0 but that was not good. (lenx=leny=60_000*RL)

Observations from this area (winter 2000-2006):