Based on previous discussions on this forum, most people’s preference is to attempt to capture the location of the organism observed, instead of the location of the observer. For one such discussion, read down from here
There are some gray areas, but in any case, I would say try to use the minimum accuracy distance that you are confident encompasses the location of the organism.
For example, If you believe your frog was about a kilometer away in a certain direction, you could map the observation there, and include an accuracy distance of 1000 m (or whatever seems right to you).