Just posting to comment that you don’t ‘have’ to download the full data set - I put off doing this for ages as the comments about having to d/load 5Gb of data put me off. There is a way to remotely query for what you need but you can manually limit the range of data with some very basic trial and error experimentation - this should give you a much more manageable data set. Then you have to think how to search for the type of data you’re looking for. There are still a couple of gotchas after that! I found this hard and took me well out of my comfort zone but was very happy when I finally cracked it - thanks @Sm4rtK1dz
I’d love to see how other people solved this as I bet there are a few different ways.