I am using the crop pollination model in InVEST 3.14.1 and seem to run into an issue when the model gets to the pygeoprocessing.geoprocessing INFO convolution worker approximately 0.0% complete on floral_resources_apis mellifera.tif. Everything runs well until this point where it does not progress any further even when left for more than half an hour.
I have seen a previous post that had a similar issue but in the urban cooling model although in my case there is no progress when this stage is reached, slow or otherwise. I would have expected it to take some time but not for it to halt processing any further once this stage is reached.
I have tried reducing the number of pollinators but whilst the overall run time reduces this problem still occurs.
It may be because the resolution for the LULC is 10mx10m and so it is taking a long time to process but not processing past 0.0% even when left for a long period of time is the issue I’m facing.
Hello @sumaya , welcome to the forums, and thanks for providing your logfile!
Based on the logs, I see that the convolution is printing out the same information, but it is still logging updates, which indicates that the model is still running.
Is it possible that you’re running the InVEST pollination model at 10m resolution across the whole country of Brazil? If so, I’m sorry, but this model is just going to take a long time to compute given these parameters. As you can see from the logs, a key step in this model is the convolution, which happens once per species, and relies on the species alpha to determine the search radius. When you have a particularly fine-scale LULC combined with a large alpha, the convolution has to read in a ton of data in order to compute. The main way you could speed things up a bit is by making your LULC coarser-resolution. If that isn’t an option, I’d suggest running this model on a fast computer that you can just let run for a long time, such as on an institutional computer cluster.
Please let us know if you have any further questions!
James