Spatial data of dragonfly observations from the Fijian island of Taveuni posed some interesting mapping problems .  This post covers how we resolved them using a new projection.

You may recall an earlier post where we talked about collecting dragonfly sightings across the Pacific.  My colleague, Milen, is quite the dragonfly aficionado and has been traveling the Pacific building up a database of observations.

Earlier we set up a webmap that he could use on his phone to collect locations and photos without a 4G or wireless internet connection and have them later synced to our GIS server here at Lincoln.   So far this has worked really well and he has used it on other islands, including the lovely Fijian Island of Taveuni.  Here’s a bit of a map to give you an idea of where it is:

There could be worse places to have to collect data and many times I have offered to carry his bags, but so far, no luck…

Anyway, after he had done his field work and the data were all synced, I got to work on putting together a map for Milen’s report.  So I dutifully opened ArcMap and added the data – but something looked immediately wrong:

All I can see are two points!  What’s going on here?  Check out the scale of the map – 1:156,000,000+!  At that scale I must be way way zoomed out so I’ll zoom in to the one of the left side to see what’s going on:

Definitely have some points there – a healthy 58 in total as shown in the table, but after a quick count only 49 of them here.  Let’s zoom to the other side:

The remaining nine are over here.  What’s the story?  To help figure this out, I’ll add a basemap:

Well, this starts to explain things – my points are at opposite sides of my basemap!  Let me zoom in again on the left hand side:

Poor Taveuni!  It is an island rent in two by well meaning cartographers!  Separated at birth at opposite ends of the world!  Here’s the other side:

What we’re really dealing with here is a coordinate system problem.  If you’ve been paying attention to the screenshots you might have seen the coordinates in the lower right hand corner:

These are clearly longitude and latitude coordinates in decimal degrees.  This comes from the fact that Milen’s phone used its on board GPS receiver to collect the points and the coordinate system GPS speaks is latitude and longitude.  Lat/Long is based on a roughly spheroidal planet and, when coupled with an elevation, is a 3D system.  We will always face the fundamental problem of how to translate 3D spheroidal coordinates to flat 2-dimensional maps – that’s where projections come in.  We’ve covered projections ad nauseum previously but in essence, projections allow us to take 3D coordinates and translate them to a flat 2D map.  If you think about that wall map that you had on your wall as a kid (maybe it’s still there?), it had a top and bottom and two sides (duh!).  Most of those maps were probably centred horizontally on Greenwich as the prime meridian.  This means that some places will end up on the edges, like Taveuni, and be split in two.  That’s the case here.

So how do I deal with this?  What I need is a different projection that centred on the Pacific.  The basemap projection is known as WGS 84 (World Geodetic System 1984).  After a bit of searching I managed to find one – WGS 84 PDC Mercator (The PDC stands for Pacific Disaster Center, who developed this projection).  This is a projection and accompanying coordinate system that is Pacific centred so should reunite the opposite ends of Taveuni.

Not to get too complicated, but there are two ways I can move forward: I could either just change the coordinate system of the map (from Layers > Properties > Coordinate System), or I could be a bit more thorough and “project” the data (transform its coordinate system) to the new coordinate system.  This means creating a new layer with the projection set to PDC Mercator.  Here’s how:

From ArcToolbox > Data Management Tools > Projections and Transformations I chose the Project tool.

The input file is my original data layer.  The tool automatically picks up the coordinate system from the data themselves.  I’ve specified a new output layer called “TaveuniObservationsPDCMercator.shp”.  The new coordinate system can be found at Projected Coordinate Systems > World > WGS 84 PDC Mercator.  In some cases, specific geographic transformations are needed (such as going from NZMG to NZTM) and if one is needed, you’ll see some options under that box.  One is not needed here so we can click OK and create a new layer.  I’ll add that to a new, blank map and:

Now my points are more in the centre of the map and look to be in the right place.  Zooming in, I can imagine many happy Taveunians:

Now that’s more like it.  Families reunited.  Huge fuel savings in getting from one end of the island to the other.  I can now carry on and finish my map, which ended up looking like this and went into his report on the observations:

I’ve used a hillshade layer as a backdrop for two reasons – it’s a simpler layer that also allows our dear readers to get a sense of topography, and frankly, I didn’t have anything that worked better.

Just to recap, the dragonfly data from Taveuni unfortunately ended up on opposite ends of the world.  A Pacific centred projection helped matters by reuniting the torn asunder island.  And it all ended well with a decent map for Milen’s report.

(To be honest, I think I’d do a few things differently with the map if I had the chance, but there you go. IAh well, I must let it go, like a dainty dragonfly.)

In another post we’ll go over how radar-derived elevation data from a space shuttle mission helped finish off this map.

C