30m pixels VS 1km variables

[Newton Pimentel de Ulhôa Barbosa]

Hi people,

I'm working with a endemic species to Brazil, with few points of occurrence. I am using altitude, slope and aspect  in a resolution of 30 m. However the bioclimatic variables continue with definition of 1km, although I have resampled then (19 bioclimatic variables) for 30m. Is this correct?

My doubt is: Working with one occurrence point per 30 m' pixel will I have problems? It's because my bioclimatic data, despite having pixels of 30m, contains data with resolution of 1km (Worldclim).

Thank you very much!

Cheers,

--
Newton Pimentel de Ulhôa Barbosa

[Heather Peacock]

Hi Newton,

 

I had this same issue.  My forest cover layer was at 28.5m resolution and the WorldClim data is at 1km.  However, I resampled the forest cover layer to be the same resolution as the climatic variables.  There may be bias introduced in the climate variables with the resampling because 30m does not fit in 1km evenly so in the resampling there will be estimates for the values of the pixels that lie on the edge of the original resolution.  Since you have many more climatic variables it may be advisable to resample your slope and aspect data to match that of the WorldClim data.  You could even do a surface analysis of the altitude file that comes with the WorldClim data.  You will also have to be careful to make sure the occurrence points lie in a sufficient number of larger pixels now.  Depending on the size of your study area the distortion associated with resampling the slope and aspect data could be minimal.  You just need to make sure to make all environmental variables the exact same resolution and extent so that it works in Maxent. 

 

Hope this helps!

 

Heather

 

 

 

 

 

2010/11/25 Newton Pimentel de Ulhôa Barbosa <newto...@gmail.com>

--
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.

[Marwa Waseem]

Hello Heather,

When yu resampled the WorldClim data, did it give you blocky pattern. As when I resampled the WorldClim data from the 1km to 30 m, it gave me a blocky pattern. when I used these layers in Maxent I found that the this pattern was influencing the results that I got. Please let me know what you got or is there any way to smooth this blocky pattern.
 

Marwa Waseem

---

From: Heather Peacock <heather...@gmail.com>
To: max...@googlegroups.com
Sent: Thu, November 25, 2010 3:52:08 PM
Subject: Re: 30m pixels VS 1km variables

[Newton Pimentel de Ulhôa Barbosa]

Hi Heather,

Thank you for your answer. Do you think it would be better do not use these resampled data? I did this because I found that using altitude, slope and aspect in a resolution of 30m I would have better results. However, results can even be affected by errors in the edges of resampling, as you said.

As my species is endemic, with few points, I thought that a better resolution, though only the altitude, slope and aspect; would improve the model. I will try run the model with my resampled data to see.

I do not know what to do about it. Do I have to use the same data from the world 1km clim?

Thanks again!

Cheers,

--
Newton Pimentel de Ulhôa Barbosa

2010/11/25 Heather Peacock <heather...@gmail.com>

[Antonio Trabucco]

Dear Newton,

first of all, SRTM and the following WorldClim has a resolution indeed of 30 seconds (900 and few meters at equator). As the latitude decreases/increases the linear resolution increases, for instance at 60 degrees latitude the WorldClim resolution is about 450 meters. The WorldClim dataset has been built over the SRTM dataset, which in its distribution format has a resolution of 3 seconds (about 90 meters at equator). The WorldClim authors have aggregated 10 x 10 SRTM 3 second pixels into the 30 seconds bioclimate variables. Therefore if you would work rather with the SRTM at 90 meters instead than 30 meters, you may have perfect geographic and data content consistency between DEM and WorldClim. Careful about the fact that WorldClim has been built over CSI SRTM v3 (available from WorldClim web page), rather than the most recent and correct CSI SRTM v4.

 

I think that downscaling the WorldClim dataset into higher resolution more apt for local studies is a very useful option and a solid and standardized methodology should become available. I still have not tested yet a specific methodology, but I have looked little bit into the Geographical Weighted Regression tool, which is available as not-free extension in Arcview 3.x and now I believe on the latest ArcGIS 10. 

 

 

Antonio Trabucco

Forest Ecology and Management

Division Forest, Nature and Landscape

K.U.Leuven

 

From: max...@googlegroups.com [mailto:max...@googlegroups.com] On Behalf Of Newton Pimentel de Ulhôa Barbosa
Sent: Friday, November 26, 2010 2:21 AM
To: max...@googlegroups.com
Subject: Re: 30m pixels VS 1km variables

 

Hi Heather,

[Newton Pimentel de Ulhôa Barbosa]

Hi Antonio,

thank you for your answer! i will try to downscale the 30 arc-sec bioclim data to 3 arc-sec and use this with 3 arc-sec altitude, slope and aspect. Do you think that this "block" aspect of the data will be bad for my model? Because the elevation and co. will have a ~90m resolution, but every block of ten 3 arc-seconds bioclimatic pixels will have a resolution of 30 arc-sec. My idea is just resample these layers in Arc-Gis, without any special metodology. What do you think about it?

I can't find the SRTM v3 to download. Do you have the link? I found only the version 4. 

Another point is if I will need one occurrence point per 3 arc-sec pixel or one per each block of 10 pixels. Do you understand me?

Thank you again!

Cheers,

--
Newton Pimentel de Ulhôa Barbosa

2010/11/26 Antonio Trabucco <respec....@gmail.com>

[Bruno Arantes]

Hi Newton.

You can find SRTM with 30m resolution at INPE: http://www.dsr.inpe.br/topodata/

Cheers.

--
Bruno Arantes de Andrade Bueno
http://twitter.com/barantes
http://barantesbirdwatching.blogspot.com

[Newton Pimentel de Ulhôa Barbosa]

Hey Bruno, 

Muito obrigado meu amigo. Bom ver brasileiros por aqui. E ainda por cima Mineiro como eu!

Abracao,

--
Newton Pimentel de Ulhôa Barbosa

2010/11/26 Bruno Arantes <bara...@gmail.com>

[Newton Pimentel de Ulhôa Barbosa]

Hey people, 

Following Bruno's suggestion, I get the data from INPE website (INPE is the Brazilian spacial agency). The ~30m SRTM data was created by interpolation from the ~90m SRTM v3. Hey Antonio, what do you think about the use of this data with my ~30m resampled bioclim? 

Thank you again,

Cheers,

--
Newton Pimentel de Ulhôa Barbosa

[MGRM]

There is also the ASTER-derived 30m global DEM,
available at: http://www.gdem.aster.ersdac.or.jp/index.jsp
,which is nice!

Cheers.


On Nov 26, 8:03 pm, Newton Pimentel de Ulhôa Barbosa

<newtonul...@gmail.com> wrote:
> Hey people,
>
> Following Bruno's suggestion, I get the data from INPE website (INPE is the
> Brazilian spacial agency). The ~30m SRTM data was created by interpolation
> from the ~90m SRTM v3. Hey Antonio, what do you think about the use of this
> data with my ~30m resampled bioclim?
>
> Thank you again,
>
> Cheers,
> --
> Newton Pimentel de Ulhôa Barbosa
>
> Em 26 de novembro de 2010 12:31, Newton Pimentel de Ulhôa Barbosa <

> newtonul...@gmail.com> escreveu:

>
>
>
>
>
>
>
> > Hey Bruno,
>
> > Muito obrigado meu amigo. Bom ver brasileiros por aqui. E ainda por cima
> > Mineiro como eu!
>
> > Abracao,
>
> > --
> > Newton Pimentel de Ulhôa Barbosa
>

> > 2010/11/26 Bruno Arantes <baran...@gmail.com>

>
> > Hi Newton.
>
> >> You can find SRTM with 30m resolution at INPE:
> >>http://www.dsr.inpe.br/topodata/
>
> >> <http://www.dsr.inpe.br/topodata/>Cheers.
>
> >> --
> >> Bruno Arantes de Andrade Bueno
> >>http://twitter.com/barantes
> >>http://barantesbirdwatching.blogspot.com
>
> >>  --
> >> 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<maxent%2Bunsu...@googlegroups.com>
> >> .

[Newton Pimentel de Ulhôa Barbosa]

Hello MGRM,

I was using this data. It´s really good!

Cheers,
--
Newton Pimentel de Ulhôa Barbosa

2010/11/27 MGRM <m_mill...@hotmail.co.uk>

To unsubscribe from this group, send email to maxent+un...@googlegroups.com.

[Newton Pimentel de Ulhôa Barbosa]

Hello Antonio and all,

How are you?

You had said that there are no standardized methodologies for downsacaling and suggested seeking the extension "Geographical Weighted Regression tool" for ArcGis. However, I did it using only the "resample" option in the "date managment tools" of ArcMap. Is this correct?

Thank you very much!

Cheers

--
Newton Pimentel de Ulhôa Barbosa

2010/11/26 Antonio Trabucco <respec....@gmail.com>