GIS and Precision Agriculture
This post is about how GIS fits in with precision agriculture, something that’s likely to be an important topic here over the next few years.
Someone recently asked me if GIS could be used for precision ag. The definitive answer is: Yes! Another answer is: How could you not? Precision ag strikes me as an inherently spatial thing to do. If I’ve got it right, the basic idea is to deliver just the right amount of water or nutrients to plants, which means you have to first know how things like soil moisture and nutrient levels vary across a paddock. Precision ag is a good illustration of how GIS functions as an environment to bring together data from various sources for analysis. If you’ve got sensors in a paddock making measurements, and aerial photos, and layers of paddock boundaries, then they can all come together in one place on a map. If you know how nutrients vary across the paddock, you can set up your equipment to deliver just the right amount at just the right place. Here’s an example.
Peter Almond took one of his soils courses out and measured soil moisture on some Lincoln dairy farm paddocks, just down Ellesmere Junction Road, using an EM10 sensor. Here’s an image of the paddocks – the aerial photo sits on top of the topo map so you can see where the paddock is:
The EM10 was connected to a GPS so every reading has a position and a water content. That allows us to map those locations as shown below (zoomed in so you can see the individual points better).
At this scale we can see that there’s a pretty dense collection of points with an average spacing of around 5 m between measurements and, say, 10 – 15 m between rows of readings. Those are pretty high resolution data (and several hours worth of walking up and down the paddock) and give us a sense of how soil moisture varies. We could symbolise the points based on their moisture content – let’s see what that looks like:
Areas of high moisture are reddish, low areas are darker green. This is useful, but we’ve still got to make an educated guess about what happens between the points. A next step up from this is to create a surface from the points. Using some mathematical approaches we can make a more refined (but still educated) guess and then map those variations:
On this surface, the blue areas have the highest levels of moisture, greens and yellows are midrange and the driest areas are orange to brown. Creating this grid is a area of spatial analysis called interpolation and ranges from a very simple approach (such as above) to some heavy duty geostatistics that we could easily teach a whole course on. The method used to generate this surface is called Inverse Distance Weighting and arguably gives us a reasonable estimate of how the soil moisture varies over the paddock.
So what good is this? Well this surface tells us where the areas of high and low water content are at a reasonably high resolution (4m x 4m grid cells but could be finer) and could serve as an input to a precision irrigator. The irrigator would have to have some sense of where it is (via GPS) and the ability to vary its output based on that location. If our sensors were measuring nutrient levels, the surface could be used to programme fertiliser application equipment.
Keep in mind that the density of readings has a direct impact on the resolution of the surface – the more the better (so long as they’re evenly spread out across the paddock). In the great scheme of things, it’s unlikely a farmer would be able to regularly get data at this resolution via this instrument; remote sensing is a viable alternative to sensors in the ground and could approach the same resolution.
I can’t claim to be an expert in precision ag, but it’s pretty clear to me that GIS provides the environment that can bring together the data needed to drive a precision ag system. It can’t do that without data from sensors, a way to manage all those data and then a connection from the surface to the irrigation drive mechanism, but as a spatial modelling tool, GIS sits roughly in the centre of those components.
And it’s not just agriculture. Increasingly, this same idea is being applied to vineyards, so called precision viticulture (just ask Val Saxton about that).
As a side note, I’ve always been intrigued by those lines of higher water contents radiating out from the centre of the paddock. I did a bit of digging and found some old aerial photos, georeferenced them and laid them underneath the surface. Turns our there used to be old fence lines right at those spots, so what you see on the surface grid is apparently an artifact of previous practice.