Overview of Spatial Data (GIS 5013) PDF

This lecture looks at the different kinds of spatial data we use in GIS. It is instructive to think of the representation of geographic data and GIS as a three-step process, we start with a spatial entity or event in the real world, which we would like to represent digitally in the computer. Next, b...

This lecture looks at the different kinds of spatial data we use in GIS. It is instructive to think of the representation of geographic data and GIS as a three-step process, we start with a spatial entity or event in the real world, which we would like to represent digitally in the computer. Next, based on the characteristics we wish to abstract and capture for this entity, we must then decide on an appropriate data model and an associated data structure. This data structure is translated into practical format to be processed by the computer. Because the choice of data model carries pros and cons. In terms of how it simplifies real-world entities, we must choose it carefully. This lecture will review three different spatial data models and their routine used in computer mapping. Specifically, the three spatial data models we will introduce include the vector, roster and image data models. You will see that each respective model has important and unique capabilities with respect to managing spatial data. One way to think about space is how we visualize the real-world on a map. In other words, how we represent space. Carter graphically in terms of cartographic representation or how it seeks to represent the real world on a map. The vector data model regards the world as being composed of countable distinct objects, often called features. This object view of the world consider space to be empty except for where objects occur. For example, I can count the number of squids or the number of seahorses occurring in the space shown here. However, the Vector Model counts all other area is empty. We can choose to map the locations of any of these objects are features as well. Here is an additional visualization of the object view of the world. In GIS, this approach to mapping forms the basis of the vector data model. Under the vector data model, we digitally represent features as discrete objects. In the example shown here, we have a set of parcels. Under the vector data model. Each land parcel is considered as an individual discrete object. Another thing to note is that discrete objects, hubs, fixed spatial locations associated with them. While we will explore this concept in more detail next week. At this time, note all discrete objects mapped will have fixed spatial locations associated with them. For example, we can assign our example parcel set a single XY coordinate in space. Also note that each object can be assigned numerous attributes describing that object. For example, in this case, the land parcel is assigned a commercial attribute. We will work more with attributes later in the course. Furthermore, objects can be related to one another in various ways. In other words, they have Topology. Here, topology refers to how discrete objects relate to one another in space. For example, the two partial objects shown are considered adjacent to one another. The vector data model is used to represent geographic objects observed in the real world, such as the streams, trees, and lakes you see here in our example. And they do so using a set of abstract object types. The basic types of objects comprising vector data includes points, lines, and polygons. We will explore these different vector object types in our lab, but let's briefly define them here. Our first vector object type are points. For point's location is the only spatial information recorded. A point is a singular location in space and has no length or area. For example, here we represent the location of three trees as three individual points. Our second vector object type are lines. Lines represent linear features as two points that are joined together by a segment. Also, lines are one-dimensional objects allowing us to represent the distance between those two points. This sense, length is the one dimension captured by lines aside from the locations of its constituent points. For example, here we have a river represented as a set of points connected by a line comprising a line vector object. Our third vector object type are polygons. Polygons represent a real objects or features with the definable area and are constructed from points and lines. For example, a set of points connected by lines which enclose an area in space, forms the polygon representation of the lake shown here. One characteristic of the vector data model overall is that it is well-suited for representing clearly defined objects in space. Some examples you will commonly see represented as vector data include census tracks or state boundaries. Because it is clear where the boundary of one state ends and one state begins. One useful aspect to the vector data model is that it can effectively capture spatial relationships between objects. For example, the river shown here is understood to flow into the lake because these objects share a common point. For this and other reasons. Many of the maps you see communicating a spatial relationship tend to also use vector data as an alternative to the vector data model and corresponding object view of the world. We will review the raster data model. Next. Under the raster data model, we can think of the world is having continuous variation across space or a field view of the world. In other words, the raster data model conceptualizes the world as having a finite number of variables or attributes that we can measure. And each variable has a value at every point in space. For example. Every point in space is associated with a specific land cover type, a temperature, and elevation. Consequently, we can then proceed with mapping these continuous measurements. In GIS, we often represent continuous features using the raster data model. In other words, our need to measure and store continuous spatial phenomena forms the basis of the roster data model. The roster data model digitally represents the world as a continuous field. This is done by dividing the world or space into square grid cells, also known as a lattice or tessellation. Characteristics about the real world are represented by assigning attribute values to each and every cell in the raster. In the example shown here, each cell is assigned an attribute value representing a particular landcover type at that sell. These landcover values correspond to land cover types found at each cell location in the real world. Each cell in our roster is associated with a particular spatial location on Earth. Each raster is also associated with a particular spatial resolution. A given roster cell spatial location is recorded as an XY location in coordinate in space. Additionally, a cell spatial resolution denotes the size of the solid space. We will explore spatial location and resolution in more detail in a later lecture. It should also be noted that rosters are very uniform and shape compared to vector data sets, which can have varying shapes. Essentially, every roster you will encounter will be rectangular in shape. In other words, it must have a fixed number of rows and columns. Another property of the roster data model is that only one attribute will be recorded per roster. Because of this, raster, attribute tables will differ from the attribute tables attached a vector objects. Vector objects can have multiple fields in a single attribute table where rasters will only have one. For example, here we see four attributes we want to represent about the real-world vegetation, roads, elevation, and buildings. If we store these four attributes for a given study area as rosters, we will require four separate roster data sets to store these variables. It is also important to note that the raster data model, unlike the vector data model, is good for representing continuous features. Common use cases for the roster data model include recording elevation and temperature, which continuously vary across space. Another common roster, you will see our land use and land cover data sets. Because every location on Earth is associated with some land use or land cover type. Finally, we will look at the image data model. Remotely sensed images differ from vector and raster data. And as such, these are usually covered separately and in far greater detail in courses dedicated to remote sensing. While the image data model share some similarities to the raster data model, it is important to note what those differences are. Additionally, many roster data sets are derived from images. Images do have some, I'll be at few uses in computer mapping, which is why we will briefly cover them here. Images used in computer mapping are often the result of a collection process known as remote sensing. At each pixel in an image, values representing the intensities of electromagnetic radiation at specific wavelengths are recorded. Each element in the sensor array corresponds to a pixel in the image. A pixel, meaning picture element. Recall how roster store a single attributes for each raster dataset. While images look similar to rosters. Unlike rosters, images do not document a particular physical attribute of an object or area because of these differences. Instead of referring to them as cells, they are referred to as pixels. Similar to raster cells, pixels of course, have associated spatial properties. In other words, each pixel has an X, Y coordinate that corresponds to a ground location. Pixels also have a spatial resolution. For example, the spatial resolution of this particular pixel is 25 by 25 units of ground distance. Similar to how grid cells are arranged in a raster, pixels in an image follow a rectangular lattice pattern. Now despite these similarities and their spatial configurations, it's important to know that the images are fundamentally different from rasters. Again, pixels comprising an image data set record the intense. Now despite the similarities in their spatial configuration, it's important to note that images are fundamentally different than rasters. Again, pixels comprising an image DataSet record the intensity of spectral reflectance at a particular location and time. An additional note with respect to the nature of spectral reflectance data, is that reflectance as a phenomena of various continuously values recorded for spectral reflectance are different from fixed measures of phenomena carried by raster cells or vector object attributes. For example, the intensity of reflectance is recorded as a digital number, or DNS for short, along an arbitrary scale defined by some number of bits. Digital number values and images used in this course will range from 0 to 255 for each band. These are also called brightness values. For example, the images we will examine in lab have a radiometric resolution of eight bits, corresponding to 255 possible hues or shades. In essence, the higher the dN number or brightness value, the higher the electromagnetic radiation intensity is recorded at that pixel for a given band. How these image bands and digital numbers are combined to produce analysis products is beyond the scope of this course. However, you can learn more about this by enrolling in remote sensing courses. What I would like you to note here is that while there are relatively few uses for images and computer mapping, they are useful for digitizing new data and also for providing context to other features we wish to map. We will explore how satellite imagery and aerial photographs are used in computer mapping in later labs. To summarize data models and GIS describe how geographic data will be represented and inform what type of data structure will be used. How a particular way data is organized and stored on a computer depends on the data model selected. In other words, we must select a data model by how we want to best represent the location, shape, and attributes that geographic object. We introduce three spatial data models and explored how they differ and representing real-world geographic features. This course we'll focus primarily on the vector data model and the raster data model. So it's important to note the difference in how we represent the real world using these two data models. Since our goal for this course will be to cover the fundamental operations used in GIS analysis. We won't cover images much beyond this lecture as they are covered thoroughly and the fundamentals of remote sensing course.