Category: GIS Analysis

The tutorial continues…

If you haven’t already read the first three parts of this guide check out:

• Spatial SQL for the Geographer – Introduction
• Spatial SQL for the Geographer – Part 1 – Welcome to Spatial SQL
• Spatial SQL for the Geographer – Part 2 – Basic Spatial SQL Scripts

The first three parts of the guide give some background information on SQL Server, covers some basic spatial and database topics, and provides links to download some software and data. Once the user has completed the first three part of this guide they will be able to understand this tutorial, which will provide some additional basic spatial SQL scripts.

Tutorial

In the last tutorial several scripts were introduced and examples were given.  In the last tutorial I covered the process to validate shapes, calculating centroids, and generate buffers.   In this tutorial several additional scripts will be covered, including generating the bounding box/extent of a feature, intersection and union operations, and difference operators.

Calculate Extent

Calculating the extent of a feature is something I use from time to time.  To find these values, such as the northern bounding latitude or eastern bounding longitude, can be calculated through Spatial SQL.  Spatial SQL has a built in method that will calculate the bounding box of a record, STEnvelope().  In its simplest context the user can generate the bounding box for any feature in a spatial table:

use Spatial_Database
go

select
geom.STEnvelope() as boundingbox
from dbo.States_Provinces
where NAME_1 = 'Massachusetts'

This example will return a field called “boundingbox” for the record where NAME_1 equals ‘Massachusetts’ that contains the minimum area rectangle that encompasses the shape.    The data in the “boundingbox” field is relatively useless the user adds some additional code to the query.  Building on the example provided by Microsoft the user can add the ToString() method to the query:

use Spatial_Database
go

select
geom.STEnvelope() as BoundingBox,
geom.STEnvelope().ToString() as BoundingBoxString
from dbo.States_Provinces
where NAME_1 = 'Massachusetts'

Here the user will see a new column, “BoundingBoxString” that returns a polygon with the five points that represent it.  You may ask yourself, why are there five points when you only need four to create the bounding box?  Well, for SQL Server to create a polygon the first point needs to be represented twice, so that the polygon closes itself properly.

Intersect

The intersect table operator is one of the most popular tools in the GIS work belt. Intersect is an overlay operation, where the area of the input feature is used as the basis to select features from the second feature, the intersect feature.  The following illustration will explain what an intersect is better than I can with words…

In spatial SQL the STIntersects() tool is straight forward and easy to use.   However, it does not return a shape as other overlay methods do.  When run, STIntersects() will return a value of one if the two records spatially intersect and a zero if they do not.  Also, this method only tests the relationship between a single record in one table against a single record in another table.  The following provides an example of intersecting a single line object against a polygon object:

use Spatial_Database
go

DECLARE @poly_geom geometry
select @poly_geom = geom.STAsText()
From dbo.States_Provinces
where dbo.States_Provinces.name_1 like '%Massachusetts%'

DECLARE @line_geom geometry
select @line_geom = geom.STAsText()
from dbo.Roads
where ID = 2974

SELECT @poly_geom.STIntersects(@line_geom) as IntersectValue, @poly_geom as input1, @line_geom as input2

The Massachusetts record form the State and Provinces table is set as the polygon of interest and a single line object from the roads table is also selected.  Now, I know that the line object is in Massachusetts so the value returned in IntersectValue field should be a value of one.

Contains

Another spatial overlay method available in spatial SQL is STContains().  This is a yes/no operation.  If a geometry instance is completely within another geometry a value of one is returned. If the geometry instance is not wholly inside the other, a value of zero is returned.  The following provides a simple example of STContains():

use Spatial_Database
go

DECLARE @poly_geom geometry
select @poly_geom = geom.STAsText()
From dbo.States_Provinces
where dbo.States_Provinces.name_1 = 'Massachusetts'

DECLARE @pointgeom geometry
select @pointgeom = geom.STAsText()
from Spatial_Database.dbo.Populated_Places
where NAMEASCII = 'Boston'

SELECT @poly_geom.STContains(@pointgeom) as ContainValue, @pointgeom as input1, @poly_geom as input2

In this query I am simply testing if the point that represents the city of Boston is within the polygon that represents the state of Massachusetts.  After all the variables are declared the select statement tests the point against the polygon.  The results, in the ContainValue field returns a one, meaning that the selected point is within the selected polygon.

Union

The union operator is an overlay analysis, like the intersect, however the results of a union are the combination of the two features, with overlapping areas obtaining the attributes of both.  Again, let’s turn to Esri for a nice graphic:

Like the intersect operation, the STUnion() operation analyzes the spatial relationship between two different records.  The basic union query will return the area represented in the union.  In the following example the query will select two countries from the world countries table to union.  The results of the query return the unioned geomerty of the two selected polygons.

use Spatial_Database
go

DECLARE @poly_geom geometry
select @poly_geom = geom.STAsText()
from dbo.world_countries
where dbo.world_countries.ADMIN = 'Germany'

DECLARE @poly_geom2 geometry
select @poly_geom2 = geom.STAsText()
from dbo.world_countries
where dbo.world_countries.ADMIN = 'Switzerland'

SELECT @poly_geom2.STUnion(@poly_geom) as UnionGeo

The initial input are two distinct polygon records, one representing Germany and the other, Switzerland. If the user were to query these two records from the table they would see the following spatial result:

However, during the union the geometry of these two records are morphed into a single feature, as seen below:

Using the example script the user could then add additional fields to the select query or perform any number of additional operations on the unioned record.

Difference

The final spatial operator that will be covered will be STDifference().  The difference overlay operation uses two layers with the area of one being removed from the overlapping area of the other.

In spatial SQL the difference method, STDifference(), returns the geometry of the difference.  In the following example the user will find the shape that is left behind between the difference of the United States and Massachusetts:

use Spatial_Database
go

DECLARE @poly_geom geometry
select @poly_geom = geom.STAsText()
From dbo.States_Provinces
where dbo.States_Provinces.name_1 = 'Massachusetts'

DECLARE @USA geometry
select @USA = geom.STAsText()
from dbo.world_countries
where dbo.world_countries.ID = 236

SELECT @USA.STDifference(@poly_geom) as Part1andPart2

The results, as seen below, displays the area that Massachusetts represents in the world_countries table.

Conclusion

Well, that is it for this tutorial.  If I have some time over the next month I’d like to explore spatial indexes within spatial SQL and perhaps cover some more in-depth spatial queries.  All these queries will work within 2008 R2 and the user can, if desired, use tables from ArcSDE within the queries.

The tutorial and the examples are very simple and I do not profess to be an expert in spatial SQL, so, if there is a way to represent any of these queries more efficiently let me know.

The tutorial continues…

If you haven’t already read the first two parts of this guide please read:

• Spatial SQL for the Geographer – Introduction
• Spatial SQL for the Geographer – Part 1 – Welcome to Spatial SQL

The first two parts of the guide give some background information on SQL Server, covers some basic concepts, provides some links to download some software and data.  Once the user has down completed the first two part of this guide they will be able to understand this section of the guide, which will provide some basic spatial SQL scripts.

Tutorial

Here the user will learn about a few basic Spatial SQL scripts.  Nothing too complicated, but enough to get the user started.  Hopefully the user will start to think about ways to expand the scripts into their own ideas.

This tutorial will cover the MakeValid, calculating centroids, and buffering a polygon.

Checking the Validity of Shapes

Before the user starts to calculate buffers, generate centroids, union or intersect tables or perform any other Spatial SQL methods the user should first run the MakeValid command. MakeValid will check the shape to see if the data set breaks any of SQL’s rules for storing the geomerty of the shape. As mentioned in the first part of this guide the geometry datatype is very flexible, but the user should run MakeValid first before doing any analysis because there still might be self intersecting polygons, bad ring order or any other number of topoligical errors within the dataset.

Many GIS programs won’t worry about these problems on the surface, but SQL will and users should perform these queries on all spatial datasets, especially polygons data, as it is usually full of errors.

This first example will update the State_Provinces table, checking the geometry of each record in the table.  This query will update the spatial data type column in a table so that the shape is valid, meaning that the shape meets SQL’s requirements for the datatype

In this example the State_Provinces table is being checked. The user will need to
run the MakeValid command against the column that stores the spatial data type, in this case, the geom column.

use Spatial_Database
go 
update dbo.States_Provinces
set geom = geom.MakeValid()

This query make not catch all errors in the data set.  The following scripts will check the data for validity as well.  The third script will check to see if any of the records still contain any invalid geometries.  These scripts were modified for the geometry datatype from Beginningspatial.com, a great resource the user should check out.

--Fix invalid shapes
update dbo.States_Provinces
set geom = geometry::STGeomFromWKB(geom.MakeValid().STAsBinary()
,geom.STSrid)

--Correct the order of the points in the polygon ring
update dbo.States_Provinces
set geom = geometry::STGeomFromWKB
(geom.STUnion(geom.STStartPoint()).STAsBinary(),geom.STSrid)

--Check to see if any records still have any invalid geometries. 
--The results should return an empty table.
select  ID, geom.STIsValid() as ValidCheck,
geom, geom.ToString() as Spatial_Text_String
from dbo.States_Provinces
where geom.STIsValid() != 1

When these queries are run the geometry in the table be updated.  From there the user will be able to repeat this process for each of the tables loaded in the first tutorial.  If the user does not run the MakeValid query some of the examples in this tutorial may not work correctly.

Calculating Centroids

The next set of queries will take a polygon layer, calculate the centroids, and convert the centroids into latitude and longitude.  Using the world_countries table in the Spatial_Database database the user can run this query and return the name of the feature and the X and Y of the polygon’s centroid.  This query uses a subquery where the centroid is first calculated.  Once this is done the outer query converts the centroid into latitude (Centroid.STX) and longitude (Centroid.STY).

select U.Region_Name, Centroid.STY as Longitude,
Centroid.STX as Latidude
 from
(	 select geom.STCentroid() as Centroid, NAME as Region_Name
	 from dbo.world_countries
)U

The following will expand on the previous query.  In the next example the user will calculate the centroid for a specific polygon and then buffer the centroid.  In the subquery a where statement is set to select a specific record that meets a certain requirement. In this example the user will select records from the name_1 column that equal ‘Massachusetts’.  Also, when the STCentroid() method is called no parameters are needed, unlike other spatial methods which can require several methods.

In the outer query the centroid is buffered by 0.05 decimal degrees, set in the STBuffer method. Why are decimal degrees used here?  Well, it is because the data were loaded in as WGS84.   There will be a future tutorial on how to tackle this problem.

The user can run this query and return a buffer around the centroid for the state of Massachusetts.

select U.Centroid, Centroid.STBuffer(0.05) as buffer, U.StateBound
from
(	select geom.STCentroid() as Centroid, geom as StateBound
	from dbo.states_Provinces
	where name_1 =  'Massachusetts'
) U

Calculating a Buffer

The last query that will be demonstrated will introduce the buffer, even though I just gave an example. The buffer operation is a common operation within GIS tools. The following script will give a basic example of how to generate a buffer in Spatial SQL.

The user will notice that the shape column, geom, is set and the buffer method, STBuffer, is called. STBuffer accepts a parameter, the size of the buffer.  In this example the buffer size will be 0.05 decimal degrees.  Also, in this example a where statement is set to query a specific record from the table. When this is done only that record would be buffered. Again, the States_Provinces table is used.

select geom.STBuffer(0.05) as buffer
from dbo.States_Provinces
where name_1 =  'Massachusetts'

What’s Next

That is it for this section of the tutorial. As I said earlier, pretty basic stuff. In the next tutorial I will get into some other topics including unions, intersects, spatial indexes, joins and other methods and operations will be covered. And, it looks like I may end up creating several more tutorials than I originally planned!

Unitl next time…