search

USF-IMARS / python-tech-workgroup

IMaRS python technical working group
2 stars 0 forks source link

GEE next steps: seagrass distribution modeling #21

Open 7yl4r opened 2 years ago

7yl4r commented 2 years ago

Now that data is loaded in from GEE the best next step seems to be to pursue species distribution modeling using the GEE data.

The first steps towards this that come to mind:

  1. [ ] explore basic usage of scikit-learn distribution modeling
  2. [ ] load in species occurrence data to work with
    • this might include a lot of filtering/modifying/converting @luislizcano 's seagrass coverage data
  3. [ ] identify possible input data in the GEE catalog
    • ideas: depth, sst, chl-a, other water quality?

@hart2 : do you have any advice on applying species distribution modeling for seagrasses? How is the seabird distribution modeling going?

hart2 commented 2 years ago

@7yl4r You have to have exact location of seagrass (lon, lat) populations, but I'd go with individual plants. You need a large number of these occurrence records (depends on the various species you look at that's why I can't give you an exact number) as well as environmental/ecological/biological data for the model. I prefer to use Maxent, but depending on how simplistic of a model you wish to create you may want to use BIOCLIM. I have several papers which discuss Maxent and even the effect of sampling bias and presence only occurrence records on SDMs in my box.