Beginner mapping issue

[Jessica Brey]

Hi everyone! 

This group was shared when I inquired on APGA forums about ArcGIS mapping for public view of our plant collections. I have just submitted a mentee request, I saw that post further down the inbox. I am coming at ArcGIS with no knowledge. 

I was wondering if anyone would want to offer some advice on my current map predicament. 

We have a drone map of our site, collected and assembled by a consultant. 

Those files were passed on to a GIS consultant who tiled them to be used in Hortis and uploaded them to ArcGIS online so I could test the accuracy with some of my gps coordinates recorded with a Juniper Geode. 

This map seems to have a consistent shift off. I am comparing the accuracy to a google map. Maybe I shouldn't do that? 

I was wondering if anybody knew what the heck is going on. 

The GIS consultant and Hortis have reprojected this map so many different ways. The one currently on view is a reproject the original drone map to Web Mercator (EPSG: 3857). We were all hoping that this would fix it. It is not ideal. 

Is the drone map just bad? 

Are the projections wrong?

Is it the visualization? 

Thanks for anyone brave enough to join my rag tag team of folks trying to figure this out. 

Map viewer: https://arcg.is/r04Ge1

Google map comparison: https://www.google.com/maps/d/u/0/viewer?mid=1ipsLdXAS4q_NPZ5l4S4-bFAiw7vOyDs&ll=40.89880910112281%2C-73.91174990297225&z=21

Best,

Jess

[Veronica Nixon]

Hi Jessica,

I don’t think your imagery link is publicly shared so I can’t see the imagery, but my two cents is that aerial imagery is always annoyingly shifted between years and between sources (like drone v Google Maps V NAIP).

 

As long as you don’t zoom in too closely it’s usually not a problem, but botanical gardens due tend to need to zoom in farther.

NAIP imagery doesn’t claim to be more accurate than 5’ and I’m guessing Google Maps is similar, so you’re probably better off comparing the location of sharp features in your drone imagery to your (sub-cm?) GPS locations. 

To correct the shift you could georeference your drone imagery product by snapping visible sharp corners to your GPS points. Or try to track down an original version before projection in case that process went awry?

-Veronica

--
You are receiving this message because you are a member of the Plant Mapping Meetup Google Group
---
You received this message because you are subscribed to the Google Groups "Plant Mapping Meetup" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plant-mapping-me...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/plant-mapping-meetup/fc472525-16f9-497a-afea-c15478ace4f5n%40googlegroups.com.

[Veronica Nixon]

NAIP only guarantees 6-meter accuracy rather! I was getting that mixed up with my county’s imagery product which does guarantee 5 feet. Neither are good enough for garden scale application so you’d need to georeference using local gps data.

Veronica

[Jessica Brey]

Adding screenshots, since Map Viewer is not allowing the link to be shared publicly. 

ArcGIS Map Viewer: 

[Kayla Flamm]

Unfortunately, the images shared aren't showing up for me. 

But the description of your experience can certainly be because Google imagery and other sources of not-super-detailed imagery are not so concerned with being exact. For example, NAIP imagery in Missouri is only guaranteed to +/- 2 ft accuracy. So if you compared your drone image to something of that accuracy, there is a very good chance it will be a little off.

To test your imagery using your Juniper Geode unit, you can take a few points at locations that are very discrete and highly visible on your drone imagery. Like a corner of concrete, for example. Then see if the point lines up where you expect it to be when displayed over the drone imagery. However, there is a caveat to this approach...

The results of testing your own imagery's accuracy against your Juniper Geode will depend largely on your Geode's accuracy. If the Geode is only getting down to 1 ft (or 0.3 m) accuracy through SBAS corrections, for example, then any point you can will be off by an average of +/- 1 ft.... But if you can get down to centimeter-level accuracy using RTK corrections, then that will give you much better results.

If you find that the points are still off even after ensuring your accuracy is down to 1-2 cm, that could easily be because the points themselves need to be re-projected, as they may have been collected in a different coordinate system compared to the imagery. Gotta compare apples to apples, or same projection to same projection, after all!

If you haven't figured this by now, projections are annoying, and fine-scale mapping has a lot of challenges. Makes you want to give it up and just go pull some weeds sometimes!