[Biomod-commits] Help with data formatting

Robin Engler robin.engler at gmail.com
Fri Jun 3 08:41:39 CEST 2011

Hi Shannon,

> I have a DEM at 3 differing resolutions containing 2.6 million, 23 million,
> and 220 million grid cells.   When preparing data for import to R for
> modelling species distributions in BIOMOD, do I need to include the
> lat/longs and  values of each of my environmental variables for every grid
> cell of the DEM??, and then specify presence/absence for my species, or just
> the values in the grid cells where I have species presence???

If you plan to do projections of your model onto your entire study
area, then the answer to your question is yes, you need to import all
values for every pixel of your study area (well, lat/long would be
sort of optional, and only needed if you want to plot your map
directly in R).
If you do not plan to carry out projections for your entire study area
(e.g. you might just be interested in studying the ecology of your
species using SDMs, but not in making projections actual for your
entire study area), then you do not need to import all the data. The
data associated with presence/absence locations would be enough.

> Also, what is the best format for my data??  At the moment, I am importing
> my data to .txt format as .xlsx can only cope with 1.04 million rows.

You could work with .txt files, but I think that it is likely that you
run into problems when trying to load 220 million pixels into R (well,
this will depend on your computer's hardware, but I think loading 220
million pixels requires a lot of RAM). Since you have a lot of data to
project on, you might want to look at importing your data as "raster"
objects (see the R raster package) and project directly on these
raster objects (Biomod has functions for that, see the manual).

hope this helps,

Robin Engler
Spatial Ecology Lab
University of Lausanne
CH - 1015 Lausanne

More information about the Biomod-commits mailing list