In this post we look at how to use ArcGlobe for 3D visualisations of large areas.

One benefit of using GIS is the creation of 3D visualisations using geographic data.  Often we can use ArcScene to do this, but it doesn’t take long to butt up against its limitations (speed, quality of display, fiddle-iness…), so in this post I’ll advocate for the use of ArcGlobe as one alternative (ArcGIS Pro is another good alternative but on a bit steeper of a learning curve – you’ll need crampons, ropes and safety gear.)  ArcGlobe has the same look and feel as Map and Scene but has some extra added benefits of display and flexibility.  When opened, you’ll see something not unlike Google Earth:

A few things to highlight here before we go on.  The Table of Contents has three sets of layers turned on by default: Floating, Draped and Elevation layers.  Elevation layers give shape to the virtual globe – there are two elevation models built in, one at a 30 m resolution and the other at 90 up to 1 km, depending on what part of the world you’re looking at.  Note that you can add your own elevation layers if you’ve got a higher resolution DEM.  Draped layers are like a bedsheet thrown over a piece of furniture – they take on the shape of the underlying elevation surface.  By default, an Imagery layer is loaded as a disappointingly out of date satellite image.  Floating layers are layers on the globe but not attached to the surface (in the help files they give examples of airplanes and clouds…).

Many of the tools should look familiar though some are new:   These are the navigation control buttons.  The 3D navigate tool at the left is used to move your viewpoint around.  With this tool active, you can left-click-hold and move the globe around.  To zoom in, right-click-hold and drag the mouse down (towards you) – opposite to zoom out.  Or you can use the mouse wheel to move in and out. Hold both buttons down and drag to pan the view

There are two navigation modes: global and surface.  At first the global mode is set and you can pan around much like you would in ArcMap with a view looking straight down at the ground.  In surface mode you can tilt the view and get oblique views.  Toggle between the modes with the mode button: .  In global mode I might navigate here:

Switching to surface mode, holding down the left mouse button and zooming/panning, I might end up with something like this:

Now I’ve got an oblique view with a bit of perspective, so very Google Earth like.  Holding down the left mouse button lets you change the angles.

But wait, the fun’s just getting started.  We can add any spatial data layers we like and drape them over the elevation surface.

The Add Data button gets us access to our layers.  I’ll add the Landcover Database layer (a vector layer) for Canterbury as an example – it lives at J:\Data\Landcover_database_4\lris-lcdb-v40-land-cover-database-version-40-SHP\CanterburyLCDB4.shp.  I get this screen first:

This allows me to set the scale at which this layer is visible.  I tend to leave it at “Show layer at all distances” but this windows allows you to set minimum and maximum scales at which to display layers.  Click Next:

We can use this to set features to real-world size if we want.  I generally accept this and click Finish.  The layer is added to the map but for some strange reason is usually place under the Imagery layer and is therefor not visible – dragging it above that layer, symbolising by landcover category and adding some transparency gives us:

(To be fair, I’ve thrown a lot of data at Globe this took a wee while to render.)  Once on the map it was pretty easy to navigate around.  So that’s how it works for vector data – raster data are a bit different.  I’ll turn the LCDB layer off and add a raster layer of a viewshed originating at the Lyttelto Timeball.  The first window I get when adding the data is this one:

Earlier I mentioned that you can add your own elevation layers.  Globe recongises this as a raster layer and anticipates that I might be adding my own DEM.  Here’s where I can make the choice.  In this case, I just want to drape the raster over the elevation surface to I stick with using this as an image source.  Again, it inserts it under my Imagery layer (argh) so next I drag it on top, play around with the symbology (set the non-visible cells to No Color, reset the visible cell colour, add a bit of transparency and reset the view):

This is a nice way to demonstrate how viewsheds are driven by topography.  Once it’s rendered, it easy to rotate the view and move around – try that in ArcScene.  You’ll only get frustrated.  But let’s not stop there.  I’ve got a vector point layer of the Timeball location so I’ll add that and make it a 3D symbol  I get yet a different Add Data window:

Ticking the first box sentences to point to always be a flat, two-dimensional image.  Who wants that?  Ticking “Display features as 3D vectors” is much more interesting, though it does look a bit horrendous when first added – anyone reminded of Google?:

We can play around with this to get a better result by going to the point layer’s symbology tab.  When you click on the Symbol image you get the Symbol Selector:

No shortage of choices here but let’s get a bit more sophisticated.  Click “Style References” and tick “3D Industrial”:

Back in the Symbol Selector, scroll down and you’ll see a range of industrial type building.  There won’t be an image for the Lyttelton Timeball so we’ll just pick something similar for demonstration purpose:

And click OK:

We can tweak the symbol settings to get it scaled to its actual size and placement but this is probably pretty good for now.  Because we’re zoomed in so far, the raster cells look at bit hazy, but that could be fixed with a bit more tweaking.

So we’ve covered ArcGlobe as an alternative to ArcScene for 3D visualisations.  It’s a little bit harder to use but I would argue gives better results especially for large areas.  Have a 3D go next time you’re needing something like this.