Aspect; best way to quantify

[Julie]

Dear Maxent users

I am struggling to decide on the best way to quantify aspect (of
seabed) as an input variable. The problem with the standard measure of
aspect, degrees from North (0 to 360) is that the model will not
recognise that 359 is very close to 1 degree in aspect. -180 to 180
would have the same problem. I thought of simply using 180 to 0 to
180, (so 0 as north, 180 as south) but the problem with this is that
it wouldn’t differentiate, for example 90 (east) from 90 (west) which
are very different aspects. It's strenght is that it WILL recognise
that 355 (5 in this case) is similar to 5, i.e. both are almost
directly north, and would essentially work as a measure of how
different from North the aspect is. I also thought of trying 180 to 0
to 180, with an additional categorical variable stating whether it is
East or West, but given that there is no option in Maxent to specify
interactions, I'm not sure if/how that would work. Perhaps I shall try
this anyway and see if, by interpretation of the response curves, I
can get an idea of whether aspect itself is important or simply how
differnt from North (or South if I do it the other way) it is.

So far I have ran the model with N, E, S and W categories instead of
aspect as degrees. This is not really showing much of an affect of
aspect, but given the biology of my system I am expecting aspect to
have some influence.

Can anyone shed some light on this, or suggest a quantification which
would meet my requirements?


Thanks

Julie

[Alvar Carranza]

July, try reclassiffying the layer accordingly; for example creating a categorical variable using, say, 5º bins

best

alvar

--
Msc. Alvar Carranza
UNDECIMAR
Facultad de Ciencias, Universidad de la República
Iguá 4225
CP 11440
Montevideo, Uruguay
http://alvar.carranza.googlepages.com

[Milton Cezar Ribeiro]

Hi Julie,

 

how about you convert your aspect in two variables: northness and easterness?

 

cheers

 

miltinho

toronto=brazil

2009/9/17 Julie <julie....@imperial.ac.uk>

[Ruth]

Hello Julie,

You can transform aspect in two components: Eastern and Northern, to
do that you can calculate the cos and sin of the angle (Eastern and
Northern components respectively) in your aspect grid. So you will
have values that go from -1 to 1.

You can do that in ArcMap: Toolbox-spacial analyst tools-Math-
Trigonometric.

To see a paper where they have done that you can check: Macleod C.D.
et al., 2008. A comparison of aproaches for modelling tha ocurrece of
marine mammals. Hydrobiologia 612: 21-32

Good luck,

Ruth

[epiphyte]

here is a previous thread discussing aspect transformation...
http://groups.google.com/group/Maxent/browse_thread/thread/e55034013289acf2

[Julie]

Thanks for the suggestion. I used the northness and eastness as used
in MacLeod et al 2008. It works perfectly, and is exactly the kind of
measure I had been looking for.

Cheers

Julie

On Sep 21, 6:41 am, epiphyte <ecogar...@gmail.com> wrote:
> here is a previous thread discussing aspect transformation...http://groups.google.com/group/Maxent/browse_thread/thread/e550340132...

[andrew]

Using sin and cosine of the radians can produce the north-southness
and east-westness but how do you handle flat values which are depicted
as -1's in a GIS raster? One could interpolate the edges of flat
areas to a sloped center bottom cell based on the dem but not sure if
there is a routine available. Perhaps this question motivates the
search for an aml or other rotune that will perform the calculations.
I've treated N, E, S, W, and flat as a categorical variable and
produced meaningful results.

A. Yost