[gradsusr] Query regarding scorr function in grads

[Andrew Friedman]

Hi Suman,

You can’t use all of the grid points as your degrees of freedom. The issue is that your points are likely spatially correlated, meaning that points near each other are similar to each other. This will likely reduce your degrees of freedom considerably.

Here’s a paper that discusses the issue:
The Effective Number of Spatial Degrees of Freedom of a Time-Varying Field. Christopher S. Bretherton et al., Journal of Climate 1999.
DOI: http://dx.doi.org/10.1175/1520-0442(1999)012<1990:TENOSD>2.0.CO;2

Best, Andrew

> On Feb 8, 2017, at 7:01 AM, Suman Maity <suman.buie at gmail.com> wrote:
> 
> Hi all
> This is Suman Maity from Indian Institute of Technology Kharagpur, India. I have a very specific question. If I calculate spatial correlation over a domain, it ill provide a single value considering values over all the grid points of the studied domain. Now how to check the confidence level of that correlation? In general we use t-distribution for finding that. Now since in these cases, the no. of grid points is huge (~100000 or more), therefore degrees of freedom (dof) is huge and there is no tabulated value for those huge dof. I want to know how to declare the confidence level of those kind of correlation where dof is huge?
> Any kind of help will be highly solicited.
> 
> Thanks and Regards
> 'S Maity'
> *******************************************************
> Suman Maity
> Research Scholar
> Center for Ocean, Rivers, Atmosphere and Land Sciences(CORAL)
> Indian Institute Of Technology Kharagpur.
> ******************************************************* 
>  
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr