Is there a limit within the resolution when calculating with SDR 3.6.0?
I ran the model successfully with 2.5m resolution (the catchment is about 30km² big).
I am now running the model with a **resolution of **50cm (0.5m)****. First it worked well till the point of "visit_stack on stream distance"
it is kind of stuck in that four hours.
how can i bypass that calculation?
what is it exactly?
Any experience with a resolution that high?
thank you
Hi @bastian87 -
It’s not so much a matter of pixel size as overall raster size and the associated memory and disk requirements to process it. How large are your output files? Are you maxing out the memory on your computer? I’ll leave it to the software team to speak to how the models handle large files, but I do know that they’ve put a lot of effort into making calculations rather memory efficient, although there are still limitations to what’s possible logistically.
~ Stacie
thanks for replay Stacie!
the RAM is not the problem. there is still 50% left during calculations. also there is still plenty of space left on the disc.
i stopped the calculations (0.5 x 0,5m) after 6hours.
the model runs successfully with a raster resolution of 1m x 1m.
Hi @bastian87, When you say that the model is ‘stuck’ for several hours, is it that there’s no motion whatsoever, or is it just that the same message (including the stack size) is being printed for several hours? With a large raster like this, I’d expect routing to take at least 4 times however long it takes at a 1mx1m resolution.
It’s always possible there’s some other issue happening, though, which is one of the reasons we’ve greatly improved the routing implementations in InVEST 3.7.0.
well no. it is actually calculating.
after 6 hours the stack size number was barely changing its value though, but i actually think that it could finish…sometime.
i am fine with the 1mx1m result for now, but i will definitely try it again and also with the 3.7.0.
jdouglass thank you very much
since the update 3.7.0 the model is calculating 4 times faster! 1m x 1m (30km²) takes about 15-20 minutes. 0.5 m 0.5 m is also now possible to calculate in a foreseeable time.
Great! I’m glad to hear that the update is working better and faster 
