Methods of Data Collection

1
Methods of Data Collection

• Downloading
• Federal or State sources
• Import
• Data purchases from private sources
• Digitizing
• Hard copy maps or photo sources
• Scanning
• Hard copy maps or photo sources
• Hand coding
• Attribute collection and updating, e.g., census

2
Data Integration

• All data must be geographically integrated
• Topology
• Feature matching/registration
• Projection
• Data must be error checked and cleaned
• Digitizing georeferencing errors
• Attribute errors

3
Accuracy, Precision, Resolution

• High resolution or precision does not
  automatically equal high accuracy
• A watch can be precise to the 1/100th of a
  second, and still show the incorrect time.
• Vector points and lines allow for essentially
  infinite zoom and (false) precision.
• Raster cells of 1cm (e.g., DOQQs) are available
  but may not be georeferenced to the same level of
  accuracy
• Raster cells of 10km square can provide accurate
  climate modeling

4
Relation between positional error and scale
US National Map Accuracy Standard (for paper maps)
Large Scale
Small Scale
The National Map Accuracy Standard (NMAS) for
paper maps defines the requirements for meeting
horizontal accuracy as 90 of all measurable
points must be within 1/30th of an inch for maps
at a scale of 120,000 or larger, and 1/50th of
an inch for maps at scales smaller than 120,000.
5
A increase in accuracy, e.g., from 90-99 results
in significantly higher costs
6
Types of Error

• Positional
• Registration the process by which map layers are
  made to overlay each other exactly
• Many forms of integration require data to be
  transformed between different spatial structures
• Attribute
• Logical consistency
• Completeness

7
Data Preparation and Management

• Spatial cleaning
• Georeferencing errors
• Digitizing errors (spikes, dangles)
• Topological errors
• Attribute cleaning
• Attribute errors
• Missing values

Some checks can be done automatically, but
visualising the data (maps, graphs, correlograms)
and statistical analysis are essential to
identify many errors
8
Spatial Errors A Simple Case
(a) Sites with suspect coordinates such as those
along 0 longitude in France or those falling
outside the designated country.
(b) Sites falling completely outside the study
area.
9
Detroit Area - travel times to select hospitals
10
Processing and Data Errors A Complex Case
(L) Zip code looping example. (R) Zip code 49633
is a large zip code and is part of two distinct
Limited Access Areas.

• Introduce an error in population totals when
  modeled using any area based methodology
• Produce an over counting, depending on the
  number of splits,

11
Data Quality in GIS

• Coverage
• Gaps, biases
• Spatial accuracy
• Errors in georeferencing, digitizing errors,
  limits of spatial resolution, uncertainties in
  projection system
• Measurement accuracy
• Measurement errors, reporting errors, detection
  limits
• Timeliness
• Delays in reporting/data capture

12

• Building Communities to Promote Physical
  Activity Exploring the Lansing Capital Region
• Collect GPS for front door locations and
  sidewalks
• Mail surveys with door to door follow-up

13
Topology

• Topology describes the spatial relationships
  between spatial elements
• Proximity lies near to
• Adjacency lies next to
• Connectivity is joined to
• Intersection - crosses
• Containment lies within (or encloses)
• Sequence starts at.ends at

Knowledge of topology is essential in spatial
modeling and especially for dynamic modeling
14
Using Topology
Study sites and network structures and
accessibilities for each site, well connected
network (left) and fragmented network (right).
15
Topology and Spatial Relationships

• Correspondence
• Co-location (super-position)
• Containment
• Overlap
• Proximity
• Adjacency
• Nearness (distance)

Spatial relationships depend on the way in which
the feature has been conceptualized, the
scale/resolution and the spatial accuracy of the
data
16
Spatial relationships
17
Sample Spatial Relationships

• Proximity
• Distance measures between target and source
• Buffering
• Creates zone of fixed radius around target
  feature, which is then used as a search area
  within which to seek source features
• Topological analysis
• Can be used to measure adjacency or shared
  perimeter of area features, connectivity, etc
• Clustering
• Measures density of features within specific area
• Intersection
• Intersection of source and target features (e.g.
  point in-polygon, line-in-polygon, area overlay)

18
Battle Creek
Distance Calculation Process
4852.2 m.

• Start ArcCatalog and create new Address Locators
• Geocoding
• Check for the correct address
• Rematch manually
• Measure the distance

19
Spatial Relationships in Epidemiology

• Equity
• Assignment of socio-economic status
• Computation of denominator populations for
  calculating disease rates
• Exposure assessment
• Nearest monitoring site
• Monitoring site(s) within census area/city
• Relationship to emission source
• Disease mapping
• Cluster detection
• Spatial pattern and trend
• Modeling disease spread
• Contiguity or connectivity

20
(No Transcript)
21
Buffering Issues Many zip codes fall within
several hospital clusters 30 minute sub service
areas.
Inside 30 Minute Service Areas
22
Closest path
Path Distance Map
Furthest path
23
Lansing area
Sparrow Hospital St. Lawrence
Sparrow Hospital
Ingham Regional
Eaton Rapids Medical Center
24
Emergency Travel Areas close to major highways
tend to have low travel costs
Kent
Ottawa
Allegan
25
The impact of no roads
9 km.
21 km.
Benzie county
26
Results
In support of the state Certificate of Need
process, we identified two areas that met all
criteria for placement of a new health care
facility.
111,781
57,791
27

• MDCH Questions
• Identify areas in Michigan where patients travel
  further than 30 minutes away from home for
  treatment
• Identify hospitals receiving visits from areas
  determined to have a high percentage of patients
  traveling further than 30 minutes away from home
  for treatment.

Geocoded hospitals and street data
Results of 30 minute travel time analysis on
Cluster 1H
Final 30 minute service area based on zip code
overlap