[R-sig-Geo] how to chose the right cell size to grid an irregular point dastaset.

[Massimo Di Stefano]

Hi All,

i have a dataset based on a oceanographic ship
it is based on a set of point along transect (the points are 2 meters spaced along the track direction and each track are 400 meter spaced)

for each point i have 6 images (the instrument i'm working on is used to collect sea flor images, 6 image/sec ) 
so is mostly like .. i have a "continuum infromation" along the track with a resolution of 2 meters (each image overlap the previouse one with a 50% factor)

the images arte manually analized and for them i've "how many times a single spiecies is recognized in each images"  (each image has also associated other metadata like depth and substrate type)


for each point i have this information :

x (longitude)
y (latiutude)
z (depth)
specie_density (specie_Count / field_of_view)
substrate_type (id identifier for the substrate type)


for now i want try to generate a regular grid from my dataset interpolating the  "specie_density" value

my first question is ... 

how to chose the right cell size for my grid ? beacouse the resolution alog the transect is like 2 meters .. while each transect is distant from the next one about 400 meters ... 

if i want try first simple interpolation methods like IDW, Spline built-in-side grass

then i have to make a more accurate analisys to interpolate my data (specie_density),
i was tinking to include the depth and substrate as "correlated information" to be used in a cokriging interpolation, 
make sense for you ? 
(for bathymatery data i've also a continuos DTM 10 meters cell size that can be used in addition of each point depth, can be helpfull ?)

for now i'm working inside a grass / python environment (easy to try the grass interpolation methods) while to try to krige my data,
using rpy i just generated a spatialPointDataframe in R (including :  x, y, specie_densithy, depth, substrate) 
i'm now tring to learn more on the sp, rgdal amd gstat package in R  to undstand how to use my spatialpointdataframe as input for a cokriging analisys,
any hints on this next step is will give me an huge help!