Do we need diffusion?

[jaseeverett]

We need to know what effect diffusion is having on the model. If we can delete diffusion we will automatically shorten the model run-time by ~25 % as it takes ~1/3 of the bottleneck section.

https://github.com/MathMarEcol/ZooMSS/blob/a628e9d46e945e26747b72a9c8369e28b5cf3d7e/fZooMSS_Run.R#L86-L91

[BryceStansfield]

Thanks! I'll get on this tomorrow.

[jaseeverett]

I had a look at this. I ran a full simulation (1628 cells) with diffusion turned off.

We get some very interesting results. I am uploading some figures here which compare a "Control" run with the "No_Diffusion" run (diff = 0).

This plot shows the change in Fish Biomass when turning off Diffusion.

This plot shows the differences by group. The Blue is the geom_smooth, and the red is the 1:1 line.

@BryceStansfield @ric325 @ryanheneghan Perhaps we can discuss this further at Bryce's meeting tomorrow morning. We need to dig into the size-spectra to look at how they are changing along the spectra.

For those who want to take a look, the output (and code for these figures) is on the RDM.

[ric325]

@jaseeverett . Well done! As you say, very interesting. Many questions come to mind.

1. How long did it take to run compared with the standard runs? Was it faster?
2. Why is the delta fish map (almost opposite) to the one in the manuscript, yet you fish biomass plots are very similar?
3. Why are some groups affected little (flagellates, omnivorous copepods, all fish) and others massively (most of the carnivores?)
   And @jaseeverett , is the code from the paper in a workflow to produce those figures (particularly Fig. 3) again relatively easily?

[jaseeverett]

After our discussion about whether or not to update the coming papers with runs which remove Diffusion, I reran the Group Removal Experiments to see what happens. More as an additional test, rather than advocating changing manuscripts. Below you can see Figure 4 from our H.I.P., and the corresponding figure with diffusion removed (Including the control run - so NoOmniCop without diffusion is compared to a control run with No Diffusion either). As we discussed on the phone on Friday, Carnivores have the greatest impact. I will look at increasing the feeding kernel and rerunning.

With Diffusion:

Without Diffusion:

[ric325]

Nice work Jase! Pretty similar, which is good. The interesting thing is that without diffusion there are more fish in the world than with it. Can you say how much more? Cheers Ant

[jaseeverett]

No Diffusion biomass: 3889 g m-3
With Diffusion biomass: 3620 g m-3
(The model is in grams? I forget)
So its a 7 % increase in biomass when we remove diffusion.

Don't forget that the figures above show the change in the biomass with differing zooplankton resolution. So figure D) is saying there is more biomass when we use a full zooplankton resolution, compared to the control run with no diffusion. The 'control' run for each of the D)s is different so we can't necessarily compare them directly. But it is saying that the model is more sensitive to changing zooplankton resolution under the No Diffusion scenario.

[ric325]

Thx Jase. Yes I thought of this when I sent the email. But it does show quite a bit more fish biomass w/o diffusion. And interestingly the spatial pattern is more in oligotrophic areas I thought… From: Jason Everett <notifications@github.com> Reply to: MathMarEcol/ZooMSS <reply@reply.github.com> Date: Tuesday, 12 May 2020 at 8:50 am To: MathMarEcol/ZooMSS <ZooMSS@noreply.github.com> Cc: Anthony Richardson <Anthony.Richardson@csiro.au>, Mention <mention@noreply.github.com> Subject: Re: [MathMarEcol/ZooMSS] Do we need diffusion? (#21) No Diffusion biomass: 3889 g m-3 With Diffusion biomass: 3620 g m-3 (The model is in grams? I forget) So its a 7 % increase in biomass when we remove diffusion. Don't forget that the figures above show the change in the biomass with differing zooplankton resolution. So figure D) is saying there is more biomass when we use a full zooplankton resolution, compared to the control run with no diffusion. The 'control' run for each of the D)s is different so we can't necessarily compare them directly. But it is saying that the model is more sensitive to changing zooplankton resolution under the No Diffusion scenario. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#21 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABLPRX4WIT5VMCCKQRR2Y5LRRB6KBANCNFSM4MJB43GQ>.

[ryanheneghan]

Looking at the figures, I think it's because carnivores are not as important without diffusion. This could be because they have the smallest feeding kernels, which diffusion can help to offset.

…

On 12 May 2020, at 08:52, Anthony Richardson ***@***.***> wrote: Thx Jase. Yes I thought of this when I sent the email. But it does show quite a bit more fish biomass w/o diffusion. And interestingly the spatial pattern is more in oligotrophic areas I thought… From: Jason Everett ***@***.***> Reply to: MathMarEcol/ZooMSS ***@***.***> Date: Tuesday, 12 May 2020 at 8:50 am To: MathMarEcol/ZooMSS ***@***.***> Cc: Anthony Richardson ***@***.***>, Mention ***@***.***> Subject: Re: [MathMarEcol/ZooMSS] Do we need diffusion? (#21) No Diffusion biomass: 3889 g m-3 With Diffusion biomass: 3620 g m-3 (The model is in grams? I forget) So its a 7 % increase in biomass when we remove diffusion. Don't forget that the figures above show the change in the biomass with differing zooplankton resolution. So figure D) is saying there is more biomass when we use a full zooplankton resolution, compared to the control run with no diffusion. The 'control' run for each of the D)s is different so we can't necessarily compare them directly. But it is saying that the model is more sensitive to changing zooplankton resolution under the No Diffusion scenario. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#21 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABLPRX4WIT5VMCCKQRR2Y5LRRB6KBANCNFSM4MJB43GQ>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#21 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMUXJVEUTLXEIFTYCMF5KGTRRB6SHANCNFSM4MJB43GQ>.

[jaseeverett]

Yes. Running the wider feeding kernels is my next stop. Should I double all groups, or just the carnivores?

[ryanheneghan]

All groups.

…

On 12 May 2020, at 09:00, Jason Everett ***@***.***> wrote: Yes. Running the wider feeding kernels is my next stop. Should I double all groups, or just the carnivores? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#21 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMUXJVF6IQN5WNJTXYGVONDRRB7RDANCNFSM4MJB43GQ>.

[ric325]

Very interesting Ryan! From: Ryan heneghan <notifications@github.com> Reply to: MathMarEcol/ZooMSS <reply@reply.github.com> Date: Tuesday, 12 May 2020 at 8:56 am To: MathMarEcol/ZooMSS <ZooMSS@noreply.github.com> Cc: Anthony Richardson <Anthony.Richardson@csiro.au>, Mention <mention@noreply.github.com> Subject: Re: [MathMarEcol/ZooMSS] Do we need diffusion? (#21) Looking at the figures, I think it's because carnivores are not as important without diffusion. This could be because they have the smallest feeding kernels, which diffusion can help to offset.

On 12 May 2020, at 08:52, Anthony Richardson ***@***.***> wrote: Thx Jase. Yes I thought of this when I sent the email. But it does show quite a bit more fish biomass w/o diffusion. And interestingly the spatial pattern is more in oligotrophic areas I thought… From: Jason Everett ***@***.***> Reply to: MathMarEcol/ZooMSS ***@***.***> Date: Tuesday, 12 May 2020 at 8:50 am To: MathMarEcol/ZooMSS ***@***.***> Cc: Anthony Richardson ***@***.***>, Mention ***@***.***> Subject: Re: [MathMarEcol/ZooMSS] Do we need diffusion? (#21) No Diffusion biomass: 3889 g m-3 With Diffusion biomass: 3620 g m-3 (The model is in grams? I forget) So its a 7 % increase in biomass when we remove diffusion. Don't forget that the figures above show the change in the biomass with differing zooplankton resolution. So figure D) is saying there is more biomass when we use a full zooplankton resolution, compared to the control run with no diffusion. The 'control' run for each of the D)s is different so we can't necessarily compare them directly. But it is saying that the model is more sensitive to changing zooplankton resolution under the No Diffusion scenario. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#21 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABLPRX4WIT5VMCCKQRR2Y5LRRB6KBANCNFSM4MJB43GQ>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#21 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMUXJVEUTLXEIFTYCMF5KGTRRB6SHANCNFSM4MJB43GQ>.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#21 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABLPRX5GRDXZL4ORJPZPVDTRRB67JANCNFSM4MJB43GQ>.

[ric325]

Wasn’t the idea to set the feeding kernel to the same width for all groups? From: Ryan heneghan <notifications@github.com> Reply to: MathMarEcol/ZooMSS <reply@reply.github.com> Date: Tuesday, 12 May 2020 at 9:02 am To: MathMarEcol/ZooMSS <ZooMSS@noreply.github.com> Cc: Anthony Richardson <Anthony.Richardson@csiro.au>, Mention <mention@noreply.github.com> Subject: Re: [MathMarEcol/ZooMSS] Do we need diffusion? (#21) All groups.

On 12 May 2020, at 09:00, Jason Everett ***@***.***> wrote: Yes. Running the wider feeding kernels is my next stop. Should I double all groups, or just the carnivores? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#21 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMUXJVF6IQN5WNJTXYGVONDRRB7RDANCNFSM4MJB43GQ>.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#21 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABLPRXZGSA2E6SVP7JMDKVDRRB7VNANCNFSM4MJB43GQ>.