Vector or Raster, Spatial Data is Important to your Farm
All the map information used in farming can be classified as either vector or raster spatial data. Knowing the difference can make you better at what you do.
Jun 6, 2016
Data churns tirelessly through most modern “connected” farm implements. While distances are covered, rates are applied, and bushels are harvested, all that data’s gotta go somewhere; and where data is concerned, it usually ends up packaged as either vector or raster spatial data. If you’ve never given much thought to either format, this GIS nerd is gonna change that for you today.
So why should I care?
At the heart of the matter here, we’re talking about your data. You pay the bills, you manage the land. As wheels roll or propellers spin, equipment is collecting data that can enable you to better manage your operation. Not knowing how data can fit together once you’re out of the field and back at your desk could mean you’re missing out on valuable insight; and that’s just like leaving money on the table.
Raster data – The original “big data”.
Raster data refers to data stored in a grid; and in the digital world what fills in the grid are pixels. Pixels are a fundamental unit of display on just about every modern screen you put your eyeballs to; but pixels also happen to translate directly to image size of your digital photos as well. Most cameras capture three separate bands of spectral wavelength (i.e. “color”) each diced up into 256 different shades. Each band is represented in the pixel by 8 bits or a “byte” of information. Twenty-four bit or “true” color is simply the product of three 8 bit bands. The Sony shooter we fly with yields images of 5456×3632, or just shy of 20 million [mega] pixels. Bootstrapping it at 3 bytes per pixel, that’s a 60 megabyte photo if you didn’t have compression (that’s what jpeg does). High-resolution imagery is dense with information.
Vector data – Connect the dots.
Vector data is stored in point, line, or polygon format. To the point (<- ugh, didn’t mean to throw a pun at you that soon), when we talk about “vectors” think back to that algebra or calculus class taken eons ago when you were made to do all that graphing. “X-axis”, “Y-axis”, “coordinate pairs”; vector data is data that’s described in this context, and the sheet of graph paper just happens to be “Mother Earth” (Wanna know more? This post explains that coordinate system relationship).
If we’re mapping in two-dimensional space (let’s ignore elevation for simplicity’s sake), an x-y coordinate pair is a point. If we need to represent a line, we simply need to group a bunch of points together and let the system know we mean to connect them. Polygons are the same way; except the system with which we’re collecting data needs to know that we mean for this line to end back on itself and that area needs to be calculated for the encircled space. Each independent record is considered to be a feature; and the number of features depends a lot on how the data is stored. For example, with four coordinate pairs, you might be storing four points; but you could also be storing two lines or even a single polygon. Context is important, and with the right software, other vector types can be derived from the original.
Formats such as shapefiles take this concept a step further by indexing an associated database or table along which contains attribute data for each feature record. Think of each feature as a little card file in which that attribute data associated with the feature is stored (if you’re too young to remember what a card file is, you probably don’t need this analogy either). In the case of something like a yield monitor, bushels are sampled at a set interval as the combine moves across the field. The combine knows where it’s at, and the computer records the x-y coordinate. Yield is recorded at the same time the point is acquired, and that number is recorded and associated to the coordinate. Collect enough of those points and assign color values to the yield numbers behind them, and voila! You have your yield map.
Advantages of vector data.
Vector data is great because it’s very compressible. Four lines of coordinate pairs could define an area encompassing thousands of acres. Small file sizes means faster load times, less wait, and easier transportability between compatible platforms. A comparable raster image could be over hundreds of times larger. If you’ve ever tried to attach a large photo to an email over a slow Internet connection, you’ve experienced the inherent difficulty with using raster imagery.
When is raster the way to go?
Raster data has its place; and part of being a good “data handler” is knowing when to use it. In the case of simply “looking at a picture”, raster is the only way to go. In other cases however, raster is a means to an end. Using techniques referred to as spatial analysis, a person can cluster and classify a raster, averaging the results into a grid that can be “vectorized”. In a typical workflow, this output may be a shapefile that stores an application rate derived from values gleaned from the original raster. A sprayer doesn’t need (nor could it use) all that extra information contained in a typical drone-flown orthomosaic. So, averaging the data and “distilling” it into a more easily digestible vector format makes sense.
Be a better data “harvester”.
This is all a lot to take in to be sure; but if you get nothing else out of it, then know that understanding whether vector or raster spatial data is right for your job makes you a better collector of said data; and that can only improve your results down the road. Got more questions about data formats? Don’t know a vertex from a dangling node? Post your questions and comments in the discussion!