Pollination normalisation species input data

Science, Research & Data
1

As I am using raster data (grid points, tif. files) as input data for species abundance and not the guild table with one column of species abundance, I need to do a pre-processing for the normalisation function:

Here the python part I am referring to (standard InVEST python L. 1339 ff).

# species_abundance[species] (string->float)
total_relative_abundance = guild_df[_RELATIVE_SPECIES_ABUNDANCE_FIELD].sum()
result[‘species_abundance’] = {}
for species in result[‘species_list’]:
result[‘species_abundance’][species] = (
guild_df[_RELATIVE_SPECIES_ABUNDANCE_FIELD][species] /
total_relative_abundance)

Here my normalization function (pre-processing before adapted python code):

  • Using a spatially explicit adapted InVEST 2024 this is not feasible, as for every single grid point the total relative abundance of all species should be calculated.
  • I used the total relative abundance from the guild table for all grid points (summary of all probability of occurrence values, e.g. 25%, + 30%, + 40%… in my case 347 as calculated by the standard InVEST python model.
  • Some grid points may be over- oder points under-estimated. However, I will not discuss outputs on a grid point basis, but on national or regional basis. So in overall /average over nation / region, for me the output is acceptable.
  • Output values < 1 and make sense.

What do you think about my pragmatic solution?
Kind regards
Sibylle

2

Hi @sibylles,

I just replied to your other post before seeing this new one. Thanks for trying to isolate your problem and question!

As a general forum etiquette note, it’s discouraged and not usually necessary to “@” a list of people for a post. We typically take a rotation approach to forum support, so you might not always get help from the same people. Often times, the person on rotation might “@” someone they think could uniquely help. If you’ve been communicating with someone on a thread I think it’s totally fine to “@” that person, but just know that if it’s been some time, you might get help from someone else or not hear from that person immediately. We are usually quick with responses, but for non trivial questions it can take some time to find the right person to help answer that particular issue.

I’m not particularly familiar with this model, but will reach out to some of our colleagues to see if they have time to take a look.

Thanks!

Doug

1 Like
3

Hi @dcdenu4

Many thanks for your reply, advice and suggestion.

You are welcome to contact me if natural capital decides for a spatially explicit InVEST pollination version. Our adapted version is fully working.

Yes it would be very helpful to receive a feedback on the pre-processing of the species input data, the normalization function.

Kind regards
Sibylle

4

Hi @sibylles,

Thanks! We certainly want to support work the larger community is doing around ecosystem services and have some exciting features coming to InVEST this year we think will help with that. Excited to share more about this later this spring!

Cheers,

Doug

1 Like
5

Hi @dcdenu4

Ah many thanks. Excited about the news for InVEST pollination.
So there will be no feedback for my specific question how to pre-process the normalization of raster data with many grid points?

Kind regards
Sibylle

6

Hey @sibylles,

I’ve reached out to a few folks about your question and will provide any updates I get. Would love to see others from our broader community chime in too if they have experience around this!

Best,

Doug

1 Like
7

Dear @dcdenu4

Many thanks. In the meantime I discussed the pre-processing data normalization issue with 2 statisticians and 1 species distribution modeller.

Feedback: the pragmatic approach is scientifically sound, specifically as I will not analyse at the grid-level and I have many raster grids per canton, region etc.
They also mentioned that we may bring in error when normalising data at the grid level. We cannot provide this accuracy.

Kind regards
Sibylle