Differences in projected output between 3.3.3.and 3.4.1 MaxEnt versions

[Jacek Stefaniak]

Hi guys,

I have a rather weird bug (or problem): I try to project my model to another period of time (LGM in this case) and I get some abnormal results.. but using previous (mean 3.3.3) version of MaxEnt, i got rather good maps (this weird line on the north).

All settings, variables, samples (and output types) are the same... and still I got two different results in models prediction, depending on version. Any advices or explanations why results are so different?

PS: I attached my results and map of differences in projected layers.

[Dimitris Poursanidis]

Hi

The default new version provide results after cloglog transformation while the previous in logistic.

You can change the new at logistic and compare results.

D.

>->->->->->->->->->->->->->->->-

Dimitris Poursanidis, Ph.D

http://rslab.gr/

http://www.terrasolutions.eu

Google Scholar

ResearchGate

ORDIC

Remote Sensing in Ecology and Conservation - Editorial Board

The latest papers !!! 

1) https://www.frontiersin.org/articles/10.3389/fmars.2017.00347/abstract

2) http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0187143

--
You received this message because you are subscribed to the Google Groups "Maxent" group.
To unsubscribe from this group and stop receiving emails from it, send an email to maxent+unsubscribe@googlegroups.com.
To post to this group, send email to max...@googlegroups.com.
Visit this group at https://groups.google.com/group/maxent.
For more options, visit https://groups.google.com/d/optout.

[Jacek Stefaniak]

PS: Differences between current variables output exist, but they are neglible, no more than 0.08 in extreme cases. Problem occurs only with projected variables.

[Jacek Stefaniak]

Hi,

I tried that, and nothing changes - additionally, "current" models (without projected layers) are almost identical between versions, so isnt this case.

[Adam Smith]

The newer version also does not use threshold features by default. And when it does the procedure that finds the "breaks" is slightly different.

Maybe that's why...?

Adam

[Jacek Stefaniak]

Hi,

I found source of problem: for some reasons, new MaxEnt do not interpret correctly -9999 value as NO DATA but as a variable value - and probably compute that as a good range for my species. Older version do not that, and just exclude points with NO DATA (mean -9999).

Problem is (now) how to tell the new MaxEnt algorithm to treat -9999 as NO DATA - I have it set in "Settings", but obviously it didn't work. Any ideas?

[Weverton Carlos]

Are you sure that in all rasters the NO DATA value is set to be -9999?

I use Maxent 3.4.1 with default NO DATA VALUE and cells with -9999 value are ignored by Maxent.

I have ever gotten weird results too, but because in some rasters the NO DATA value was different from -9999.

[Husam El ALqamy]

I think the best thing to do is to open the ASCII file in a text editor and replace "no Data" with"-9999". I think this will get things going.

 

		Hossameldin ELALKAMY, MPhill., PhD. Conservation Mapper  Conservation Data Center|  Victoria Ministry of Environment & Climate Change P. 250.614.7521 C. 778.896.3229|525 Superior St.   Victoria, BC., V8W 9M1
		Webpage | GIS Requests | Profile Antelope Specialests Group, IUCN

--

You received this message because you are subscribed to the Google Groups "Maxent" group.

To unsubscribe from this group and stop receiving emails from it, send an email to maxent+un...@googlegroups.com.

[Jacek Stefaniak]

I've got this... but things are rather weird ;)

it seems that problem lay in way how the MaxEnt support BIL file format - the same files opened in 3.3.3 with probelr -9999 recognition (as NO DATA) but in 3.4.1 version exactly this same BIL file with exactly the same settings are handed with error - and 3.4.2 treat -9999 value as variable score - and make the model according to this (despite fact that in settings I have -9999 set as NO DATA).

This problem do not occurs with ASC files apparently - and after converting all BIL to ASC, 3.4.2 suddenly start "see" -9999 as NO DATA. So real source of problem is that 3.4.1 MaxEnt cannot handle correctly NO DATA values in BIL files, but somehow 3.3.3 do it pretty good. Odd, but this is it ;)

Thanks for all support and advices!

[Adam Smith]

I haven't tried this, but you might be able to do a quick fix using the "NAvalue()" function in the raster package to define the value that represents NA.
- Adam

[huiq...@163.com]

Hi,

    I downloaded .tif from the worldclim.Before importing it to the maxent, I clipped the .tiff and transtered it to .asc with ARCGIS.I got a good map when I didn't import any hotlayer. but when I tried to project my model to another period of time, I got nothing( a map in blue) .Can you give me some advices ?

     

            Ivy
在 2018年2月28日星期三 UTC+8下午10:43:36，Jacek Stefaniak写道：

[Husam El ALqamy]

One or more of your rasters in the past time have a spatial extent bigger than the others and this is causing the weird extent in the prediction map. You need to clip an snap your past rasters with the extent of obe of your cyrrent rasters in ArcGIS.

 

 

Hossameldin ELALKAMY, MPhill., PhD. Conservation Mapper  Conservation Data Center|  Victoria Ministry of Environment & Climate Change P. 250.614.7521 C. 778.896.3229|525 Superior St.   Victoria, BC., V8W 9M1

Webpage | GIS Requests | Profile Antelope Specialests Group, IUCN

--

[Ivy]

 thank you!

 I clip the past and the current data based on the extent of my interest regional at the same time, but it doesn't change anything ou the results. Can you give me more datials about how to clip an snap my past rasters with thr extent of obe my current rasters?

 --Ivy

在 2018年4月3日星期二 UTC+8下午10:19:22，alqamy写道：