ACLED data occassionally records events at sea. This is relatively rare, but for instance, Cameroon has had one instance in the historical record (which is what led me to detect the issue) and Israel has had 7. There are two issues in the codebase that need addressing.
We currently generate errors if the admin boundaries do not cover the geometries they are plotted with. For instance, we check that the boundaries cover the event points from ACLED and IDMC. We can increase the buffer to cover nearby waters to address this issue. We have extended the buffer in this way in #229.
A larger issue that would need addressing is how do you visualize these points. We currently only visualize a single country. Do we add other countries + boundaries + water bodies to ensure it's clear? But that would mean we also might want to plot points that fall in neighboring countries too, because if we mapped them but didn't plot, might look like there were "no events" when that isn't true. Do we want to not map any sea points? Just drop them for simplicity? Something to discuss.
ACLED data occassionally records events at sea. This is relatively rare, but for instance, Cameroon has had one instance in the historical record (which is what led me to detect the issue) and Israel has had 7. There are two issues in the codebase that need addressing.