[RADS] radsstats bug

[Remko Scharroo]

Hi Eric,

I fixed the bug. When doing box-wise averaging, the box indexes could  
just exceed the allocated indices by 1. This would only happen in  
those rare cases that a point was on the boundary (exactly lat=0 in  
your case).

Cheers,
Remko

On 19 May 2008, at 09:15, Eric Leuliette wrote:

> Remko,
>
> I've come across a minor bug in radsstat when I run it on both lark  
> and raven. When I run:
> radsstat -c sat=j1 lim=-1,1 lat=-90,0 sel=0
> I get a segmentation fault after cycle 8.
>
> If I change the latitude range to:
> radsstat -c sat=j1 lim=-1,1 lat=-67,0 sel=0
> on lark, I get a segmentation fault after cycle 23. On raven, I get  
> an out of range value for the time during cycle 24:
> 24 020831   275698************  0.0244 0.1015
>
> The problem appears to be the the upper bound of the latitude range.  
> As a workaround, I can use:
> radsstat -c sat=j1 lim=-1,1 lat=-90,-0.000001 sel=0
>
> While, radsstat -c sat=j1 lim=-1,1 lat=-90,0.000001 sel=0 also fails  
> after cycle 8...
>
> Cheers,
> Eric
>