With respect to five layers of GIS,

we discussed the first layer,

spatial reference framework in the last lecture.

In this lecture, we'll discuss spatial data model.

How to represent spatial reality in an abstract manner.

The content of the lecture are the following,

comparison of object view and field view,

and related vector model and raster model will be discussed.

And vector model will be investigated in more detail.

I will present other vector models;

network model and TIN model,

and you will learn spatial data models for spatial big data as well in the end.

As mentioned in the first lecture of the week object,

view assumes that space is composed of discrete features such as building,

parcel, road, point of interest and many others.

And they are generally represented in the vector model.

On the other hand, field view assume that space consists of continuous phenomena.

Such as terrain, like rainfall.

And it can be represented in a matrix form of homogeneous grid cells,

which is called raster model.

One exception should be noted that TIN

triangulated irregular network can model a field view of the space.

Which is vector format now the raster model.

The figure on the slide illustrates the examples of object view

and field view of a given spatial reality of land use and land cover.

Here object view is implemented in vector model and field view is in raster model.

Now, let's take a closer look at vector model and raster model.

Vector model has three basic features point, line, polygon.

Vector model can provide three types of data spatial,

attribute, and relationship among spatial features.

It can accommodate more than one attribute for example,

the parcel polygon in green color can have multiple attributes such as owner name,

address, tax assessment, and so on.

Scale and the corresponding level of generalization

determine the quality and the shape of spatial data in vector format.

This issue will be covered in mapping and

Geo-visualization in more detail later in this week.

Data structure is rather complicated than raster model.

Those are the characteristics of vector model.

Raster model has the basic unit, called pixel

and raster model is simply viewed as

2 dimensional array or pixels, just like an image format.

A pixel can have only one attribute value.

So if you need multiple attributes, then additional rater layers are required.

Pixel size defines the spatial resolution, that is, level of detail.

The size of raster data is generally bigger than

vector data and data structure is rather simple,

compatible with simple image formats.

The examples are land cover data in the above, and the digital elevation model in the below.

As mentioned the vector model can support spatial data,

attribute data, and relationship with spatial features.

Spatial data and attribute data can be handled either in a separate manner,

or an integrated manner, everything in DBMS.

After object relational DBMS was developed.

The relationship among spatial components in vector model is called topology.

Which can be maintained for spatial data integrity and efficient spatial data processing.

The concept of topology was already present in the first week.

Topology is defined as a set of rules

that describe spatial relationship among point line and polygon.

You also learned the two main benefits of topology,

data integrity checking and efficient data of operations.

Topological data structure can confirm if spatial data are well composed for example,

polygon is complete with the a crossed loop of lines.

Polygons are non-overlapped.

And space is a fully filled up without any gaps or holes.

Those kind of things can be conformed with topological data structures.

The given set of topological rules on the slide now,

define the relationship among 0D object-node,

1D object-chain,

and 2D object-polygon.

They describe relationship between node and chain,

polygon and chain, polygon and node.

Imagine if we have a data structure supporting

the rules that is being described on the slide.

Then we can speed up basic spatial operation such as adjacency,

connectivity, containment, operations of spatial features.

The figure is a data model,

or point node, line, chain, and Polygon.

That support the topological rules in the previous slide.

Network data model is a vector model with point and line,

which can accommodate all the properties of network.

Network is defined as a group of interconnected things or peoples.

We can easily find samples of network for example,

like as a physical network, subway network, load network,

water supply and sewage network, river network,

internet connection network and on the other hand

as a virtual networks, SNS friend networks.

Network can be represented with graph data structure.

Graph data structure is composed of a set of vertices and edges that connect vertices.

Vertex is a terminal point or an intersection point of edges.

Edge is a link between two vertices.

It can be either one directional or bi-directional.

There are many variation for implementation of graph data structure.

Here I'm presenting one simple method node table and connectivity matrix.

In the connectivity matrix edge can be formed

as matrix element in yellow color now as you can see.

At the upper left corner of physical network let's say maybe subway network,

can be simplified to net mode on the next.

Then it can be implemented with Graph data structure with node table and

corresponding connectivity matrix that you are looking at.

TIN.

TIN is also an important spatial data model in GIS.

TIN stands for triangulated irregular network,

a vector model which can implement of field view of the space.

With contiguous and non-overlapping triangles.

TIN is mainly used for representing terrain and the related analysis.

At the same time, the basic algorithm to build TIN is Delaunay Triangulation,

which is often used in many spatial analysis such as,

proximity analysis for market analysis.

Now let's briefly take a look at spatial data model for spatial big data.

Basically they are mostly equivalent to

conventional spatial data model; either vector or raster data model.

The list of example,

the list explains different spatial data types and corresponding examples.

In the figure smart transportation card transaction and

floating population can be handled with point event data model on the first row,

text trajectory with polyline data model and satellite image in raster data model.

In this lecture you have studied on spatial data models and related topics,

which is second layer of GIS.

At the same time that is basic data models of spatial data science.

All right. This is the end of this lecture. See you in the next lecture.

Spatial Data Models

Spatial Data Science and Applications

Rencontrer les enseignants