Re: Spatial model for inconvenient pot layout?

Hello alex
My reply is in the email below

Brian Cullis
Research Leader, Biometrics &
Principal Research Scientist
NSW Department of Primary Industries
Wagga Wagga Agricultural Institute

Professor,
Faculty of Agriculture, Food & Natural Resources
The University of Sydney

Phone: 61 2 6938 1855
Fax: 61 2 6938 1809
Mobile: 0439 448 591

Alexander Zwart <Alec.Zwart_at_CSIRO.AU>
Sent by: ASReml users discussion group <ASREML-L_at_AGRIC.NSW.GOV.AU>
09/08/2007 04:44 PM
Please respond to
ASReml users discussion group <ASREML-L_at_AGRIC.NSW.GOV.AU>

To
ASREML-L_at_AGRIC.NSW.GOV.AU
cc

Subject
Spatial model for inconvenient pot layout?

Dear list members

I have data from a series of Cabinet trials and am wondering if
AR-style spatial modelling is going to be appropriate/feasible for my
situation.

Within each trial there are four tanks, with 77 plants in pots
placed in each tank in a 9 x 17 "checkerboard" pattern:

O O O O O O O O O
 O O O O O O O O
O O O O O O O O O
 O O O O O O O O
O O O O O O O O O
 O O O O O O O O
O O O O O O O O O
 O O O O O O O O
O O O O O O O O O

Does anyone know whether an AR*AR spatial model for each Tank in
ASREML could reasonably cope with such a pathological pot layout,
assuming the dataset is appropriately augmented with missing values
for the unused row-column combos? Would some other correlation
structure be more appropriate?
****this is not a difficult problem for asreml or asreml-r. You can fit a
distance based model using the matern class of models. something along the
lines of

rcov=~at(tank):mat(x,y,...
see the asreml-r manual p44 for details. you compute the x,y values from
the tank and away you go.
This is taken from our paper in JABES this year....
Haskard et al (2007) for details
You would also want to constrain the variance parameters across tanks as
well.
I would start by fitting random=~tank and then seeing if there was
evidence of spatial within tanks as it is an effort to do this and then
find out that there in no heterogeneity within tanks

cheers
brian

Thanks,

Alec Zwart
Received on Mon Aug 09 2007 - 18:03:01 EST