Hi I have a csv file of 160 presence points. Of these points, 5 are missing environmental data in one layer (mean annual precipitation). All layers have same extent and resolution (pixels = 300 m X 300 m) so I know that this is not the problem. The problem is these 5 presence point fall into a cell where there are no data for precipitation.
My question is: will Maxent discard those 5 points in the modelling process?
Thank you
Marie-Hélène
Dear Marie-Hélène,
You could use the least rainy quarter as a climatic variable to assess a possible explanatory condition for its occurrence. Also, the total amount of rain below 60 mm generating a grid indicative of water reduction in the soil.
Keep these five points of occurrence in those 10% of your points to validate your model.
Are you working with average annual rainfall? Also consider other climatic variables such as thermal amplitude. What climate variables have you tested using?
Best regards,
Lucieta
Thank you. but what I would like to know first is: will Maxent use data from these five points to train the model?
No, if there is missing environmental data for any of the layers it will discard those points (and probably throw up a warning message about it depending on your MaxEnt settings).
Dear Marie-Hélène,
The answer of Heather Williams needs a supplement. By default, it is true. But there is a parameter of the command-line MaxEnt tool called 'allowpartialdata'. If you change it from the default false value to true, MaxEnt will use those samples that have at least one known environmental data. So if only one predictor (i.e. the precipitation) is missing, and allowpartialdata is set to true, the sample will not discarded.
Here you can find the list of the available parameters:
https://github.com/mrmaxent/Maxent/blob/master/density/parameters.csv
If you use the graphical user interface, search for the "allow partial data" option. If you use package 'dismo' in R, run function maxent(x, p, path, args = "allowpartialdata=true").
HTH,
Ákos
- Badges
- Science topic
- Similar topics
- Filing