MST species models differ across links

[timpwhite]

When I click on some species links found and referenced in an older document, for example, long-finned pilot whale in the Gulf of America: Globicephala melas distribution (mammal: Long-Finned Pilot Whale; mdl_seq: 1434) from AquaMaps SDM | BOEM Marine Sensitivity the species model occurs when it should have been masked out.

However, when I open a new instance of the MST viewer, the long-finned pilot whale model is not in the Gulf of America: Phoca vitulina distribution (mammal: Harbor Seal; mdl_seq: 51909) from Merged Model | BOEM Marine Sensitivity

The 'mdl_seq' appears to be referencing different models.

[bbest]

Hi @timpwhite,

Note that in the first link provided that model is for ONLY the AquaMaps SDM (i.e., before applying a mask):

• Globicephala melas distribution (mammal: Long-Finned Pilot Whale; mdl_seq: 1434) from AquaMaps SDM | BOEM Marine Sensitivity
  

In the original version 1, that was the only spatial data applied, but then we used the IUCN range map to limit its extent in the final the Merged Model, which now says "No data to display — species may lack this layer or fall outside program areas":

• https://app.marinesensitivity.org/mapsp/?mdl_seq=51873
  

You can expand the sidebar and see all the inputs to the model, and individually inspect each input layer, including the masks. For the IUCN Range model, if you select mask by Ecoregions, you can see the overlap with the US Atlantic coast (outside the current Program Areas):

• https://app.marinesensitivity.org/mapsp/?mdl_seq=23515
  

We can discuss ways to layout and better explain the user interface, but fundamentally these have been 'corrected' for being out of range.

[bbest]

Harbor seal (Phoca vitulina) does look like a problem of not having ingested the IUCN range map (I suspect it was a weird geometry issue, which was the case for a few small handful.).

• https://app.marinesensitivity.org/mapsp/?mdl_seq=51909
  

• https://www.iucnredlist.org/species/17013/45229114
  

[timpwhite]

When I follow an older link for the long-finned pilot whale (https://shiny.marinesensitivity.org/mapsp/?mdl_seq=1434), the MST viewer only shows the species in the Gulf of America — a region where this species is not known to occur — while showing no distribution in the Atlantic, where it regularly occurs.

This suggests the viewer isn't rendering the full AquaMaps model. If this were the complete AquaMaps output pre-masked to BOEM Program Areas, we would expect distribution to appear in the Atlantic, not the Gulf. It seems like something is either truncating the model or this particular model sequence was never valid to begin with.

Please see the range map:
https://www.fisheries.noaa.gov/species/long-finned-pilot-whale

[bbest]

Hi @timpwhite,

We made a very intentional decision to mask the app to the Program Areas by default, hence the default view for the the long-finned pilot whale AquaMaps SDM (not the merged model; https://shiny.marinesensitivity.org/mapsp/?mdl_seq=1434) only shows the Gulf of America, since there are no Atlantic Program Areas:

• Mask: Program Areas
  

By selecting the other mask, you can see the Atlantic distribution of the original AquaMaps SDM (again, not the merged model):

• Mask: Ecoregions
  

The Merged Model has no data because the IUCN range does not overlap with the BOEM Program Areas:

• Model: Merged Model
  

These finer points are not intuitive in the display, so we need to think deeply about how to best present the data, but remind ourselves the content (at least in this case) is fully "correct."

[bbest]

Hi @timpwhite,

I've update the mapsp app presentation of the models (without altering their content, nor the unique model sequence identifier mdl_seq) to show a pill bar that is orange when displaying an $\textcolor{orange}{\textbf{input model}}$:

• https://shiny.marinesensitivity.org/mapsp/?mdl_seq=1434
  

Then clicking on the "show Merged Model" link or the "Merged Model" pill shows green for $\textcolor{green}{\textbf{merged model}}$:

• (which technically does not have a mdl_seq since a merged model does not exist)
  

I look forward to hearing whether this does the trick in explaining app distribution layer for you. We can discuss more when we meet next.