Can BioClim variables be used for a SDM on a small area? Any alternatives?

[mgde...@gmail.com]

Hi folks,

I am going to run Maxent for Principe Island, West Africa. It's only 14,200 ha so I am wondering if there's any high res bioclimatic data available now, anywhere. I've used Chelsa and WorldClim on the past but I believe the highest res was 1km?
Cheers

Marcelo

[Andrés González]

I got the same problem back then when i was trying to use maxent to model past population movements. The study area was 232 km² and i needed the variables in a smaller resolution.

This is what you can do

1: Try to find weather stations in your country, usually that information is public because is from the government, then make a interpolation of the data to have the variables in the scale that you need.

2. The another option (the one that i used) is to use variables from the raster, i mean, use raster information to create the other variables as slope, distance to water an so on. You need to be careful because the results of this method  have the risk of being correlated, so you need to watch for it

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/c1945b82-a4b7-496f-bdda-8700d3807011n%40googlegroups.com.

[Neftalí Sillero]

Hi,

Th best solution is to use remote sensing data. Depending on the products you want to use, you have several spatial resolutions: 10 m, 30 m, 250 m, 500 m, and 1 km.

Best regards,

Neftalí

---

From: max...@googlegroups.com <max...@googlegroups.com> on behalf of Andrés González <mahom...@gmail.com>
Sent: 03 October 2024 15:57
To: max...@googlegroups.com <max...@googlegroups.com>
Subject: Re: [maxent] Can BioClim variables be used for a SDM on a small area? Any alternatives?

 

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CAGp7horM3356%3DVmXr0C1jnfc5esQq%2B1%3DgyqqGuxqztnk3Gfrmw%40mail.gmail.com.

[Marcelo Lima]

Hi Neftali, what products do you mean? For the bio climatic data, I need ground data as well. For Principe Island, their might be data from meteorological stations on the North and Central part of the island but I doubt there’d anything from the South though. I will check. 

I found this paper very interesting. 

High spatial resolution bioclimatic variables to support ecological modelling in a Mediterranean biodiversity hotspot Erika Bazzato a , Leonardo Rosati b , Simona Canu c , Michele Fiori d , Emmanuele Farris e , Michela Marignania,

Dr Marcelo Gonçalves de Lima
(Views are mine)

Research Fellow - Center for Large Landscape
Cambridge Conservation Forum - Connectivity Conservation Work Group Chair
IUCN - WCPA member/Connectivity Conservation Specialist Group - Brazil and UK Lead
IUCN - CEM member
Biologist, PhD in Ecology

Sent from Gmail Mobile

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/AS8PR10MB7398504AF040A5FC9DDA278AAE712%40AS8PR10MB7398.EURPRD10.PROD.OUTLOOK.COM.

[Neftalí Sillero]

Hi,

Remote Sensing provides many products. The most used in biodiversity studies are:

• Land cover/land use
• Land surface temperature
• Vegetation indices
• Primary productivity
• Digital elevation models

These products are global and with temporal resolutions from 16 to 1 day.

---

From: max...@googlegroups.com <max...@googlegroups.com> on behalf of Marcelo Lima <mgde...@gmail.com>
Sent: 03 October 2024 21:08

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CALZ4CaxNNCNJ6p0O9X1W08qU3L_Ab8ovBgQg_tPaGC-ChzZb8w%40mail.gmail.com.

[Marcelo Lima]

Hi, thanks. I do know that, for some 30 years now :) 

. I thought you were talking about a specific high res bio climatic product that was now available. 

Cheers 

Dr Marcelo Gonçalves de Lima
(Views are mine)

Research Fellow - Center for Large Landscape
Cambridge Conservation Forum - Connectivity Conservation Work Group Chair
IUCN - WCPA member/Connectivity Conservation Specialist Group - Brazil and UK Lead
IUCN - CEM member
Biologist, PhD in Ecology

Sent from Gmail Mobile

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/AS8PR10MB739861815CDF8890474F255FAE712%40AS8PR10MB7398.EURPRD10.PROD.OUTLOOK.COM.

[Bede-Fazekas Ákos]

Dear Marcelo,

I agree with Andrés - you can always downscale/interpolate the coarse resolution climate data to the desired fine resolution. Either from weather station data or from a gridded data, like WorldClim or Chelsa. Elevation is usually available at fine resolution - so you can use it as an auxiliary variable (along with latitude and longitude) in a multivariate interpolation method, such as regression kriging.
I suggest downscaling the raw monthly climate data and then calculate the bioclimatic variables at fine resolution instead of downscaling the bioclimatic variables. This is especially important if there is a spatial heterogeneity in the specific climate period (https://doi.org/10.1111/gcb.17496; e.g. wettest quarter, coldest month) underlying the BCVs.

Have a nice weekend,
Ákos
___________
Ákos Bede-Fazekas
Centre for Ecological Research, Hungary

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CALZ4CawK9eqp214y0sVquUBdPR4xLucMsWa7b30D4KPBEYFPPw%40mail.gmail.com.

[Marcelo Lima]

Thanks Akos. I am weary though of downscaling the data from, say, Chelsa. It’s like using 200 m NDVi which will provide little/no variance amongst pixels if downgraded. 

I will look into the article. By auxiliary data do you mean an extra layer of data, or adding in some way to existing data?

Cheers 

Marcelo

Dr Marcelo Gonçalves de Lima
(Views are mine)

Research Fellow - Center for Large Landscape
Cambridge Conservation Forum - Connectivity Conservation Work Group Chair
IUCN - WCPA member/Connectivity Conservation Specialist Group - Brazil and UK Lead
IUCN - CEM member
Biologist, PhD in Ecology

Sent from Gmail Mobile

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/b35056f3-dfe2-4eca-af75-109301a50ee8%40gmail.com.

[Amir Sohail Choudhury]

Hello

Prepare your own variables. Otherwise, if you want to use Climatic data then take local meteorological data. 

Please find an attachment for your reference. 

Thank you 

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CALZ4CaykbfgPjb1mW1cXMb0LY_mpakGyXGm6GBj1bYmzHdWSjg%40mail.gmail.com.

--

Dr. Amir Sohail Choudhury

                              B.Sc. (Hons), M.Sc., Ph.D.

[Marcelo Lima]

Thank you Amir

Dr Marcelo Gonçalves de Lima
(Views are mine)

Research Fellow - Center for Large Landscape
Cambridge Conservation Forum - Connectivity Conservation Work Group Chair
IUCN - WCPA member/Connectivity Conservation Specialist Group - Brazil and UK Lead
IUCN - CEM member
Biologist, PhD in Ecology

Sent from Gmail Mobile

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CALnNaER%2B4uDCD2dvZwEw_e0fcfXAkT7vMX5mzHuadtxtNq8UbA%40mail.gmail.com.

[Bede-Fazekas Ákos]

Dear Marcelo,
sure, I agree: a simple interpolation like IDW or ordinary kriging will not produce variance. This is why I suggest using elevation as an auxiliary dataset (an extra layer), which has a similar role in the downscaling as a predictor in a species distribution model. To put it very simply, you will train a modell with a formula of "temperature ~ elevation + latitude + longitude" at coarse resolution, and you'll predict the temperature at fine resolution based on the fine resolution elevation dataset. Of course, regression kriging is a bit more complicated, since it will use not only the abovementioned elevation-temperature model but also a variogram model of the spatial structure of the residuals.

Have a nice weekend,
Ákos
___________
Ákos Bede-Fazekas
Centre for Ecological Research, Hungary

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CALZ4CawDYftccYuS%3Diz2D2t20%3DCfK-q%3Do%3DGOHENk-jO0k%3DzB2w%40mail.gmail.com.

[Marcelo Lima]

Hi Bede, thanks.

Apparently, there is only one weather station on the Island of Principe, at least one government-run one. The topography changes a lot, and it is pretty different south of the station, with a more mature primary forest. Can the data from that station be somehow extrapolated to the rest of the Island? It's only 14,200 ha. Maybe an extrapolation would be better than the usual course resolution. 
I am unsure how to train the model in the way you suggested; any references would be helpful.

Cheers

Marcelo

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/3be96729-1188-483b-8369-c8fa1eeab53a%40gmail.com.

--

Dr Marcelo Gonçalves de Lima

(views are mine)

Research Fellow - Center for Large Landscape Conservation

Cambridge Conservation Forum - Connectivity Conservation Work Group Chair

IUCN - WCPA member/Connectivity Conservation Specialist Group - UK and Brazil Lead

IUCN - CEM member

Biologist, PhD in Ecology

https://www.linkedin.com/in/marcelo-gon%C3%A7alves-de-lima-35b3ba20/

[Amir Sohail Choudhury]

Hello,
One option is to collect data from portable weather stations installed at schools, colleges, universities, or industries. I've noticed that many institutions and organizations set up weather stations for various purposes. You can gather data from these sources and use IDW interpolation in ArcGIS or similar software to create your own raster.

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CALZ4CaxdOMv9x82VVG-Zary0fwSwGMznnHmag4Lofhem%3DT7rJw%40mail.gmail.com.

[Marcelo Lima]

Indeed.

The issue is that the south part, where the Principe Natural Park is located is very isolated, with no communities. Research has been rather uncoordinated. I plan to establish some ecosystem services monitoring devices which will include weather stations but that’s in the near future.

Cheers 

Dr Marcelo Gonçalves de Lima
(Views are mine)

Research Fellow - Center for Large Landscape
Cambridge Conservation Forum - Connectivity Conservation Work Group Chair
IUCN - WCPA member/Connectivity Conservation Specialist Group - Brazil and UK Lead
IUCN - CEM member
Biologist, PhD in Ecology

Sent from Gmail Mobile

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CALnNaERsas6%2BrS2_iK-ZcTynKXNvKP-V41uQCyuN-0nVhWDARQ%40mail.gmail.com.

[Bede-Fazekas Ákos]

Dear Marcelo,
I think that most of the statistical downscaling methods like IDW or kriging need several training points, so one weather station is not enough. But you can use the 1-km resolution WorldClim dataset as the coarse resolution training points.
Regarding the regression kriging, please refer to e.g. the "Regression Kriging" section of this tutorial: https://rpubs.com/leydetd/geostatistics2
Have a nice week,

Ákos
___________
Ákos Bede-Fazekas
Centre for Ecological Research, Hungary

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CALZ4CazrU1cxke-JE5MBLO26eySkMDidTJ5ny4_8F1gOxqcyDQ%40mail.gmail.com.

[Marcelo Lima]

Thank you!

Dr Marcelo Gonçalves de Lima
(Views are mine)

Research Fellow - Center for Large Landscape
Cambridge Conservation Forum - Connectivity Conservation Work Group Chair
IUCN - WCPA member/Connectivity Conservation Specialist Group - Brazil and UK Lead
IUCN - CEM member
Biologist, PhD in Ecology

Sent from Gmail Mobile

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/4b671679-ce07-4c3d-a304-f2f84b6fac2c%40gmail.com.

[Pablo García Díaz]

Hi there, 

Have you tried packages microclima and NicheMapR in R? 

Cheers 

Pablo 

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CALZ4CayNKW52eDRFWceCA51cZnQu2H7D8LnUqVHt%3DcqCipExww%40mail.gmail.com.

[Marcelo Lima]

Hi Pablo, no. I haven't used R much to be honest. It's difficult for old dogs to learn new tricks! I will have a look into it though.

Cheers

Marcelo

To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/CAGJ2pxwB4xQAh13Nxiyqn_xVuoOYNVc1%2BZDjZwDp6VkgAtdYTA%40mail.gmail.com.