questions from new Maxent user

[Daft Fad]

Hi. I’m interested in possibly using Maxent
for modeling Terrapene carolina and Terrapene ornata species
geographic distributions,
both in a local area and also rangewide.

I’ve installed Maxent and have worked through the “A Brief Tutorial on
Maxent” by Steven Phillips. Now I’m ready to give it a try using as
presence localaties the GPS latitude and longitude readings I have
from hundreds of sightings – I think I’ve got these samples set-up
correctly like this:
Filename: turtle.csv
species,ddlong,dd lat
Terrapene_Caroline,37.382523333,-85.17051667
Terrapene_Caroline,37.382590000,-85.16763333
Terrapene_Caroline,37.832573333,-85.16741667
Terrapene_Caroline,37.832573333,-85.17198333
Terrapene_Caroline,37.832521667,-85.17490000
Terrapene_Caroline,37.182763333,-85.17236667
Terrapene_Caroline,37.832776667,-85.17163333

Then, I tried to set-up a dummy environmental variable as shown below
(not sure if I have the XLLCORNER & YLLCORNER correct for the sample
data).

I have some questions about Maxent::
1) is there any documentation on how these environmental variables
should be set-up?

2) Any doc. on what the various values in the ascii raster grid mean?
I read that they are ascii raster grids in ESRI’s .asc format.

3) Surely these ascii raster grids aren’t created by hand – what is
used to create them?

4) Is there a website where these environmental variables are kept in
the ESRI asc format and that allow you to download them? If so, can
you specify the area that you need the data for?
(so that you don’t have to load down the entire world data)

5) If I wanted to create a ascii raster grid for an environmental
variable by hand, is that allowed?

6) Are there any environmental variables that must be present when you
run Maxent?

7) Is there any interface to Google Earth

8) Where do the map images (e.g. of South America) come from? Can you
provide your own map images? If so, do they have to be in any
particular format?

9) What unit is used for the cell size that appears after the
XLLCORNER & YLLCORNER? Can this be changed? If you change the cell
size, is there anything else on it that needs to be changed?

10) I’ve seen references to ArcGIS, ArcView, ArcEditor, ArcInfo. Can
these products create the . ESRI ascii grids?

11) Are there student versions of any of the Arc… packages and if so,
does anyone known the price?

12) Are there any other grid formats than the ESRI ascii grids that
can be used as environmental variables for MAXENT?

13) tried small test using different values in sample,csv file than
is used in tutorial

Warning: Sample at 37.15033333, -89.66615 in turtle.csv is outside the
bounding box of environmental data, skipping

ncols 1212
nrows 1592
xllcorner -94.8
yllcorner -56.05
cellsize 0.05
NODATA_value -9999

--SO TRIED FOLLOWING CHANGE – no longer get outside bounding box,
But now get missing some env. data
--change to ---
ncols 1212
nrows 1592
xllcorner 37.00
yllcorner -90.00
cellsize 0.05
NODATA_value -9999

Warning: Sample at 37.15033333, -89.66615 in testmonday.csv is missing
some environmental data (e.g. WOLDvap6190_ann)
Warning: Skipping sample at 37.15033333, -89.26615 which has no
environmental data
Time since start: 59.406
699800 points with values for all grids
Adding samples to background in feature space


Thanks for the help

[Lucy Robinson]

Hi Daft

I have attempted to answer some of your Qs below:

1) you may want to check out Steven (Phillips) and Richard (Pearson) species distribution modeling course website: http://biodiversityinformatics.amnh.org/index.php?section_id=82&content_id=140
if you click on the "individual projects" link (at the bottom of the webpage) then you will find a pdf doc with some info on setting up env data. I think there is info on worldclim database in there.

3) Personally I wouldn't create a ESRI ascii file manually. That sounds like hard work


4) The ESRI ascii format is an Arc GIS grid format. So the raster grids that you use in ArcMap can be saved as and ascii output file. I think this is where the format originated? But there are other programs like Matlab and probably many more that will write data to this format. If you use Matlab and want that code let me know.


9) The units of the cell size would be dependent on the projection that you are using. You can change the cell size but this is likely to change all of your grid values, xll, yll, no. of cols and rows. As you are either taking data that was recorded at a particular resolution or interpolating to a higher or lower res. So you are likely to get different values for a particular variable at a particular location depending on the scale (or cell size).


12) I think ESRI ascii is the only grid file format that maxent will accept but have you looked at the SWD (species with data) format which is just a .csv file that can be easily manipulated in excel?

13) do you actually have data for those locations that you have changed your extent to? As it is no good just changing your extent in the grid file manually if you don't have the data for that location in the file...
Plus are all extents for each grid file equal? I am assuming so as it probably wouldn't have run if not.



Hope this helps.


Cheers
Lucy

-----Original Message-----
From: Max...@googlegroups.com on behalf of Daft Fad
Sent: Tue 1/6/2009 1:33 PM
To: Maxent
Subject: questions from new Maxent user



Hi. I'm interested in possibly using Maxent
for modeling Terrapene carolina and Terrapene ornata species
geographic distributions,
both in a local area and also rangewide.

I've installed Maxent and have worked through the "A Brief Tutorial on
Maxent" by Steven Phillips. Now I'm ready to give it a try using as
presence localaties the GPS latitude and longitude readings I have

from hundreds of sightings - I think I've got these samples set-up

correctly like this:
Filename: turtle.csv
species,ddlong,dd lat
Terrapene_Caroline,37.382523333,-85.17051667
Terrapene_Caroline,37.382590000,-85.16763333
Terrapene_Caroline,37.832573333,-85.16741667
Terrapene_Caroline,37.832573333,-85.17198333
Terrapene_Caroline,37.832521667,-85.17490000
Terrapene_Caroline,37.182763333,-85.17236667
Terrapene_Caroline,37.832776667,-85.17163333

Then, I tried to set-up a dummy environmental variable as shown below
(not sure if I have the XLLCORNER & YLLCORNER correct for the sample
data).

I have some questions about Maxent::
1) is there any documentation on how these environmental variables
should be set-up?

2) Any doc. on what the various values in the ascii raster grid mean?
I read that they are ascii raster grids in ESRI's .asc format.

3) Surely these ascii raster grids aren't created by hand - what is

used to create them?

4) Is there a website where these environmental variables are kept in
the ESRI asc format and that allow you to download them? If so, can
you specify the area that you need the data for?
(so that you don't have to load down the entire world data)

5) If I wanted to create a ascii raster grid for an environmental
variable by hand, is that allowed?

6) Are there any environmental variables that must be present when you
run Maxent?

7) Is there any interface to Google Earth

8) Where do the map images (e.g. of South America) come from? Can you
provide your own map images? If so, do they have to be in any
particular format?

9) What unit is used for the cell size that appears after the
XLLCORNER & YLLCORNER? Can this be changed? If you change the cell
size, is there anything else on it that needs to be changed?

10) I've seen references to ArcGIS, ArcView, ArcEditor, ArcInfo. Can
these products create the . ESRI ascii grids?

11) Are there student versions of any of the Arc. packages and if so,

does anyone known the price?

12) Are there any other grid formats than the ESRI ascii grids that
can be used as environmental variables for MAXENT?

13) tried small test using different values in sample,csv file than
is used in tutorial

Warning: Sample at 37.15033333, -89.66615 in turtle.csv is outside the
bounding box of environmental data, skipping

ncols 1212
nrows 1592
xllcorner -94.8
yllcorner -56.05
cellsize 0.05
NODATA_value -9999

--SO TRIED FOLLOWING CHANGE - no longer get outside bounding box,

[Julian Ramirez]

Hi Daft,

As I see you'll also need some kind of GIS-oriented tutorial (and not only a Maxent one). You can take a look into the ESRI's homepage, they have some interesting courses.

1. Environmental variables are rasters (and can be found in any format ESRI ASCII, ESRI grid, bil, grd, and many others) which contain values describing a single variable that can be measured in the atmosphere (i.e. temperature, precipitation, altitude, relative humidity, radiation). These variables must be obtained from local or global databases and if necessary the should be interpolated from ground data.

2. An ASCII grid contains a header (which is formed by the lines you're trying to manually modify) and the values (separated by spaces) corresponding to each one of the pixels of the raster itself. Each value in the large list you see above the header is a pixel.

3, 4 and 5. You can download environmental data from different portals (www.worldclim.org, http://srtm.csi.cgiar.org/SELECTION/inputCoord.asp, http://ionia1.esrin.esa.int/index.asp, www.nasa.gov) or from local sources you may know. Most of these portals allow downloading different formats and selecting only the region of your interest. You may download them in any format and convert them to asciis using any GIS software (ArcGIS, Arc/Info, GRASS, R, DIVA-GIS). Assuming what you mean with "by hand" is that you'll use a text editor such as notepad (or any other more advanced) and will start "writting down" all the data; you can, but should not create grids by hand just because: what values would you put on it?.

You can, however, use ground data (meteorological stations data) to interpolate surfaces in a single and small area. But this is another case (look for interpolation methods).

6. No, you can use any variable you want. You should select them according to the species you're working with. For example, if you're modeling certain plant's distribution then you may use for example altitude, total annual rainfall, dry period rainfall, wet period rainfall, sunny hours, temperature ranges, and any other variable you think is important and you have the gridded data.

7. Regarding to the Google Earth stuff there's a software named "Super Overlay". It's a bit annoying and weak but You can use it freely for 30 days after which you'll have to buy it, crack it, install in another computer, or do anything to get it working again. Super Overlay uses GEO-TIFF to create the KML. I developed a Java-based code (but it has no interface so you have to run it in the console or command prompt) which works fine and uses ascii grids directly to create the KML.

8. Dont know what do you mean

9. It depends, as Lucy says, on the coordinate system. If you have your data in geographic coordinate system then your units will be Degrees-Minutes-Seconds, but if you have projected data for example in UTM (Universal Transverse Mercator) then your units will be meters, kilometers, etc.

If you want to change the cellsize then you will have to do either a resample (if youre making your cellsize value greater -this means making the pixel bigger-) or an interpolation (if you're making your cellsize value lower -this means making the pixel smaller-). This must be done with GIS software and not manually in the header as it will change both the number cells in the grid and its values.

10. ArcGIS is from ESRI, so ESRI grids are to be created with ESRI's software. However, the ESRI ascii format is very easy to handle and you can use other soft. to create them (DIVA-GIS, R [with respective packages], GRASS). You can also write some small code under Java, Matlab, or other languages to handle ascii grids.

11. Dont know, but you can try the "evaluation version". It's free and if you ask for it in ESRI's webpage they will send you a free copy. It will work fine for more/less two months. After this time you can buy it, crack it, or do whatever you want to make it work again. You can try free software: R, DIVA-GIS, GRASS, Quantum GIS, Map Window GIS; some of them will allow only visualization but others will allow importing/exporting and other stuff.

12. I think you can use .bil, .grd or .asc.

13. This is occurring because you changed the coordinates, so the points went outside of the extent of the grid. you may be aware that each grid has a certain geographical coverage so if you change the points' coodinates then you need to have a bigger grid or move the points just slightly.

Then, to solve this, you "manually" changed the extent (with a text editor i guess). You may not create/edit rasters' headers manually. In this case you will have to download a grid with greater spatial coverage or merge this grid with another one (describing the same variable) but spatially complementary.


I strongly suggest you to take some GIS tutorials so that you can understand what rasters are and which characteristics do they have. If not, you will be in trouble for interpreting maxent results and handling them to create maps.

Hope this helps,


Julián Ramírez-Villegas
International Centre for Tropical Agriculture, CIAT
Cali, Colombia