Projecting onto new geographic area

[gro...@yahoo.com.au]

Hi all,

I have a project in which I need to take the predicted distributions
from one area and project them onto another geographic area. I am
unclear how to achieve this, does the projections file have to contain
the environmental data for the new geographic area and the file names
have to be the same as the file names used for the original
prediction?

Many thanks
Ivor

[Melissa Lynn Reed-Eckert]

That's it. Just make sure the environmental data layers you project to have the same exact names as the corresponding files used for training the model. And be sure to perform clamping and select write clamp grid. The maxent tutorial explains all of this.
Melissa...

>--
>You received this message because you are subscribed to the Google Groups "Maxent" group.
>To post to this group, send email to max...@googlegroups.com.
>To unsubscribe from this group, send email to maxent+un...@googlegroups.com.
>For more options, visit this group at http://groups.google.com/group/maxent?hl=en.
>

[Martin Damus]

Yes, exactly. The projections folder has to contain the same environmental variables (by name and unit / scaling) as the training environmental variable folder. So if you've got "JanTemp.asc" in the training folder and the units are degrees celsius, then you need "JanTemp.asc" in the projections folder with data in units of degrees celsius, etc.

Cheers,
Martin Damus

----- Original Message ----
From: "gro...@yahoo.com.au" <gro...@yahoo.com.au>
To: Maxent <max...@googlegroups.com>
Sent: Wed, June 23, 2010 2:52:14 AM
Subject: Projecting onto new geographic area

Hi all,

Many thanks
Ivor

--

[gro...@yahoo.com.au]

Thanks to you both.

Melissa when you say the MAXENT Tutorial, do you mean the brief
tutorial written by Steven Phillips? I couldn't find anything about
clamping in that.

Cheers

Ivor

> For more options, visit this group athttp://groups.google.com/group/maxent?hl=en.- Hide quoted text -
>
> - Show quoted text -

[fiscus2]

You might also change the regularization multiplier - more rigid if
the training area has a much wider span of climatic conditions (e.g.
world) than your projection area (be careful you could over fit the
model easily) and somewhat loser if it is more restricted. The problem
is of course that you do not have a criterion for doing this right
unless you have a sample from your study area that allows you to
perform the test statistics. But I did it for California (California
vs. world training vs Spain for brachypodium distachyon) and the
prediction was too widespread for world data and to restricted for
Spain. Changing the regularization generated better AUCs. So be
careful and do not over-interpret projections from other areas
especially if abundance or absence seem unlikely. Another problem
might be having less control over the data quality.

On Jun 23, 3:26 pm, "grow...@yahoo.com.au" <grow...@yahoo.com.au>
wrote:

> > For more options, visit this group athttp://groups.google.com/group/maxent?hl=en.-Hide quoted text -