lecture 2 - maps as outcomes of processes

chpt 3 + 4

What are 3 suspicions of display?

1. maps do not compress data - they do not simplify data

2. maps give mixed messages - a collection of signs that can lead to diff interpretations = can fail to communicate the proper message

3. maps are hard to draw accurately - historically = by hand = too hard, modern = too easy to create incorrectly

What is neogeography?

The activities that have map creation as the main objective. They use data collected from private individuals (crowd sourcing or companies like Google)

What is the purpose of scientific visualization

To explore data and info graphically. Focusing in the development of ideas (not presentation of results). Leads to geovisualization.

develop ideas graphically → present them non graphically

What is geovisualization? Is it useful on its own?

The fusion of visualization and cartography. Naur, just use spatial analysis, stats and other methods.

What are the 4 types of geovisualization?

1. cartography: using visual queues to represent information (colour, spacing, location) 

2. linking and brushing: linking aspatial data to spatial data, highlight components of spatial data 

3. projection: projecting information differently 

4. animation: difficult to describe and publish, harder to create an effective animation vs a map 

What are the 3 approaches to geovisualization?

1. pure geovisualization: interactive exploration of data (zoom, pan, linking and brushing) 

2. spatial analytical: modifies data to be mapped mathematically (convert population to density) 

3. transformation: reprojecting the data into a space (cartogram) 

What do dot density maps represent?

• variables - dot spacing, size and # events represented

• locations - arbitrary, if no info = dots are evenly distributed across an area

• does not provide spatial structure/shape of the field data

• requires interpolation to drive a surface

What can a point pattern calculate?

The overall density (an estimate of intensity of events).

# points in a pattern in a study area by the area of that region.

What is KDE? What does it do?

kernel density estimation.

• when pattern can have a density at any location in the study region, not just at location where there is an event

• transforms point objects to a field of density estimates

KDE can:

• result identifies hotspots

• can contour around areas of high/low point density

• can link to other field data

How is KDE calculated? How does the size of the bandwidth change process?

It counts the # events in a region/kernel (circle)…

Creates a circle around a point with a radius that lies within the overall point pattern.

Larger = smooths data, estimated densities will be similar in study region

Smaller = focus surface pattern on individual events

What are isolines?

All locations in a field that are of equal value and connected. It shows the absolute value of a field. The standard spacing provides information about the gradient.

!! spot heights of isolines are estimated with interpolation. this is not original data

How turn non spatial data into spatial data

ADD COORDS/LOCATION INFO DATA WITH GPS

What are the 2 processes?

1. deterministic process

   • same input = same output, only 1 realization

   • ex. coordinates

2. stochastic process

   • same process = may have diff results

   • chance + random variation play a role

   • introduce a random or stochastic element to make it unpredictable

     • random = describe method where symbols are located, not the pattern that results

What are the 2 conditions of IRP?

1. equal probability: any event has an equal probability of being located anywhere on the map

2. independence: positioning of one event has no influence on positioning of any other event

What is IRP?

independent random process.

• aim to derive the expected frequency distribution of the numbers of events for IRP

• # events in equal sized non overlapping areas (quadrats)

• use math to predict the frequency distribution of quadrats → use prob as a standard to compare

[examples of tables]

What is stationarity? How does it relate to 1st/2nd order effects?

Rules that govern a process and control the placement of entities, although probabilistic, do not change, or drift over space.

If chances of event change over space due to 1st or 2nd order effects = no longer stationary.

IRP = STATIONARY = DOES NOT CHANGE OVER SPACE/TIME

isotropic vs anisotropic

• anisotropic: when a process varies with spatial direction 

• isotropic: where directional effects do not occur 

What are 2 weaknesses of spatial statistical analysis in relation to 1st/2nd order effects?

1. both first and second order effects can lead to uniformity or clustering in the distribution of point objects

2. can detect a departure from IRP/CSR (complete spatial randomness), but can’t tell if it’s due to first or second order effects

What are the ways to analyze different stochastic processes? What are some issues with each?

• points

• lines: lines have length, direction and connection (networks)

  • issues: path lengths can take any value

    • they are constrained by perimeter of the area, path lenghts depend on the shape of the area

    • less attention has been paid to stochastic line generation

  • can be simulated

• areas: most complex case to map and analyze

  • can randomly allocate values and create an experimental distribution to compare against

• fields: continuous (not interger values)

  • GEOSTATISTICS