Great Roger.

My comments were regarding creating a fast efficient software implemenation to "ballpark" the probabilities rather than to achieve great accuracy.

Of course, reducing the cell sizes (say to 0.5 degs x 0.5 degrees) will enhance accuracy in the probabilities at the expense of more cpu resources / run time.

Can't wait to see what those probabilities look like ...especially over many more cases. I guess this begs a historical data analysis approach doesn't it ?

I wonder what Don has to say about any of this ?

.02