Site mapping at Arne

The aim of the practical exercise is to extract some key environmental parameters for 100 points chosen at random which fall within 50m of a path within the SSSI management units of the Arne reserve. These parameters could be covariates in a statistical analysis of the data that you will collect on the field trip. The data could therfore form part of a broader analysis based on field work.

In order to extract the data we will need to do the following.

• Derive terrain properties from the digital surface model and digital terrain model . Extract only the paths that lie within the SSSI units
• Buffer around the paths to produce a polygon representing the area in which the points will be placed
• Select random points and extract the values of the terrain models for all the points

Terrain models can be produced in most GIS programs including Arc Map, QGIS or R. The examples here were all produced in R. Advice will be provided in the practicals regarding how to run the same using a point and click GIS interface.

The Arne SSSI polygons are extracted from the Natural England SSSI_Site_Units layers. Each polygon is a management unit of the reserve. Notice that a central part of the area is not managed for nature conservation and so it is not included as an SSSI unit.

The digital surface model is a smoothed, interpolated raster derived from the firs returns from the Lidar point cloud. As such it should represent the surface of vegetation and the tops of hard buildings

The digital terrain model (DTM) is derived from the last returns, again smoothed and interpolated. As such, in theory, it should represent the land surface underneath vegetation and beneath buildings. This may be challenging to achieve perfectyl in practice, particularly in urban areas with larger buildings, as the interpolation of the underlying terrain may be difficult. Specialist software is used to produce both products, and Lidar point cloud processing can be a complex matter.

It is important to be aware of the size of the raster when processing. These are the row and column stats.

[1] 1931

[1] 2054

This is large, but not excessive. Processing will be quick in any GIS.

Contour lines can be produced for either the DTM or the DSM or both. It makes more sense to produce contours for the DTM

Slope is derived through using a “local” algorithm acting on each pixel and its neighbours. This looks at each pixel in turn together with its neighbours and uses basic trigonometry to calculate the slope. Slopes can be measured in degrees or radians.

Calculation of aspect uses a similar approach to the calculation of slope. Again the measurement can be in degrees or radians.

In theory, at least, we should be able to get a reasonable measure of the height of the vegetation at our 2m Lidar resolution simply by subtracting the DTM from the DSM. This may not work in all cases, but the Arne Lidar layers have been professionally produced and do lead to good measures.

We can now find all the paths that fall within the reserve areas from the OSM vector data layer by looking for lines which intersect the SSS boundaries.

Note that the paths extend beyond the boundaries of the reserve. This is a common issue that is resolved by carrying out a formal vector intersection operation to clip the lines. We will now take only the paths that fall inside the polygons.

We now need to produce a buffer around the paths. This is a simple operation in GIS. It is often necessary to produce a set of buffers of differing widths for some forms of analysis in order to carry out sensitivity analysis.

There are a range of typical operations that we conduct on the geometries of vector layers in a GIS that can be illustraed and used here. If we just want the outline of the whole area being managed for conservation we could run an union on the individual SSSI polygons.

[1] 557.6057

As the converse of the buffer around the paths would be the area in the reserve that does not fall into the buffer we can try an operator. The area can be obtained by spatial differencing, i.e. finding the area within the polygon representing the whole reserve that differs from the buffer geometry.

Most GIS programs have a function for choosing random points within any vector or raster layer.

A spatial overlay can be used to extract the values of the raster pixels that match the random points.

        hts     slope     aspect   elev
1 0.1728888  8.459603  91.372504 20.782
2 8.0082174  8.755093 290.302959 10.757
3 0.1282005  3.703272  46.564955 25.110
4 0.1862729  1.223497   7.228865  5.350
5 3.6481428 14.194590 155.752897 23.483
6 0.1408667  3.378708 182.062759 17.157

We can also overlay the polygons that we produced on top of the raster layers. This will produce a large number of values that can be summarised statistically. We might want to see whether there appears to be a difference in the distribution of vegetation heights in the buffer as opposed to the non-buffer area in order to check how representative the buffer area is of the reserve as a whole

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
 0.1000  0.1687  0.4546  1.6677  2.3208 19.2756    1794 

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
 0.1000  0.1659  0.3306  1.5060  1.6377 21.8954    1635 

The practical has demonstrated some common GIS operations. All the analyses can be run using either Arc Map, QGIS or R. In the practicals these operations will be demonstrated using a point and click GUI interface. You will also see how to perform the same operations using reusable scripts that can speed up processing if the same analysis needs to be re run.