We often find ourselves in a similar situation, and we usually leave the GPS on, so that it pans the map to the general vicinity of where we want to collect the data. When we go to collect an actual feature, we simply drag the location from where the GPS placed it to the correct location, based on our own high-resolution aerial photo we're using a basemap. If you traveling a distance between locations -- at least the scale of the map -- we find it helpful and saves a bit of time to have the GPS get us close to the right place and do the panning and zooming work for us.
Also, even though we are operating online, we still find the performance is better when you have the high-resolution aerial photo stored locally on the device as your basemap. As our imagery is usually quite large, we leverage the side-loading approach to quickly and easily get it on the devices, rather than downloading it over wifi to each one.