1 Raster Analysis

how is the length of linear features determined?

areal measure

examples of linear features in the environment

roads, rivers

3 issues to consider with measuring length

1. feature of interest is much smaller than raster resolution

2. feature of interest is much wider than raster resolution

3. dependence on chosen neighbourhood

what happens when the feature of interest is much smaller than the raster resolution?

the feature might not be represented at all

what happens when the feature of interest is much wider than the raster resolution?

the total number of raster elements will be equal to area instead of length

a thinning algorithm needs to be applied down to a single raster element width

why is raster data analysis considered to have linear complexity?

each cell is visited once so computing time is directly proportionate to the size of the raster

why aren’t raster data sets linear by nature?

each raster element covers an area or areal extent with a length and width

results of a raster analysis are always ____

discrete and there are no fractions

how do you find the real world size from an area feature of a raster data set?

multiply by scale factor

what is classification?

identification of features from field data

a raster analysis operation over fields

is classification a linear process?

yes because each cell is visited once in a linear fashion

2 statistical measures of area features

averages and variations

what is a field?

a cell
not points, lines or area features

what two processes involve fields?

1. sampling

2. resampling

what is sampling?

reduction of continuous signals into a discrete signal

rasterisation is ____

sampling

what is resampling?

deduction of another discrete signal from an already discretised signal

samples a raster to create another raster for a different resolution or orientation

what process does resampling require?

interpolation

what is interpolation?

when the value at a desired location is computed from the values of neighbouring cells

2 methods of interpolation

1. nearest neighbour interpolation

2. bilinear interpolation

where do you start the count for interpolation?

lower left corner

how is nearest neighbour interpolation carried out?

1. find nearest point by finding the grid point with the smallest difference in both columns and rows

2. choose its value as the best estimate

how is bilinear interpolation performed?

1. consider all 4 neighbourhood grid points

2. compute a weighted average

3. multiply the opposite weighted average by the value at the point

resampling is an ____

interpolation task

which is better? nearest neighbour or bilinear?

bilinear because it computes averages rather than extrapolates