GIS

Question	Answer
How would you find data freely?	Many publicly funded agencies mandate free access to GIS data – Web is great repository for GIS data (might only be subject to slower download speeds if not purchased -> e.g. QSpatial) – University departments and legacy projects
Where would you buy data?	– Government agencies (local council, state/territory, federal) – GIS (re)sellers (check online) – Can be expensive; may not be able to “try before you buy” (quality of data &/or resolution often unknown); layers of interest may be formatted for incompatible software; raster vs. vector availability – If it’s what you need, no further work to do (quick); can share costs with similar users (legal issues with usage license)
How would you modify paid or free data sources an/or layers?	Change objects of existing data • Example: You have a layer showing local government authorities (LGAs), but need state boundary. Delete internal lines (merge features) & voilà! – Change attributes of existing data • Example: You have 2007 data layers (businesses and locations) but need 2008 data. Spend a lot of money for mostly same information or spend a bit of time updating 2008 details into your 2007 layer. – Combine existing data (attribute join) • Example: Relate georeferenced map file (e.g. Southport properties) with non-georeferenced data file (e.g. property values or census results), each of which has a common field (e.g. street address, or ‘lotplan’ number)
How Would you create your own data sources?	Points and lines from Global Positioning System (GPS) readings  GPS = Device to collect readings on ground (latitude, longitude, elevation, speed, time, etc.) as measured by orbital satellites • GPS satellites have built-in error (1 m – 200 m, normally ~ 5 to 20 m), which may affect the accuracy of your final analyses – Digitize from hard copy or scanned images • Plenty of old maps and photos available online or in libraries • May require special hardware (scanner, digitizing table, digitizing puck) and software • Time- and labour-intensive (especially final corrections), but some software allows automation of the process • Georeferencing likely to be less accurate (3D real-world coordinates to 2D map coordinates to 2D screen estimate of coordinates transformed back to 3D coordinates)
What are the three principles errors in GIS?	1. Systematic 2.Human 3. Random
What is systematic error?	(Instrumental/Cumulative) – (generic) problems with the processes involved with data collection, measure, or analysis • Can be removed (or sometimes corrected) if caught – Example: GPS under high atmospheric disturbances will vary wildly from their “true” positions – usually indicated by some quality index, can be accommodated (error buffer) or corrected (post-processing differential GPS) – Example: automatic conversion of type of attribute when importing EXCEL file (text instead of numeric);
What is Human error?	n (unnecessary gross errors/blunders/stuff-ups) – problems with inattentiveness or carelessness during data collection (manual entry), review, measure, or analysis • Tight protocols and post-collection/analysis checks can be used to minimize, catch, and correct errors – Example: Mix up of units entered (e.g. different users entered angle measurement first in degrees then radians) – Example: typos, e.g. entering text (string) using capital letters in some cases but not others
What is Random Error?	(Compensating) – Errors present after first two types of error have been corrected (= errors not readily explained) • Can’t be corrected (& often not even detected), only minimized. Assume these errors throughout & that they all balance out. Or, to minimize total measurement error, apply function (like square root) to all of your data. – Example: Snap mode was used at some stage during screen digitizing but this was not realised, next user started w/ default setting (snap mode was turned off) & continued; – Example: “Mistakes” from last week’s workshop.
How would you minimise error?	Examine your (attribute and spatial) data • Examine it again, this time more closely (summary stats) • Share examination duties with a co-worker, supervisor, or friend so as to have two sets of eyes to pick up any errors • Read the metadata and compare what is written with what you see to help pick up any problems • Get to know your data. Everything may look fine, but perhaps only because you are unfamiliar with how it should really look. This takes time and experience • Keep back-up copies of everything. Maybe any errors come from your error-checking abilities! You may inadvertently make changes that weren’t needed
What does C.O.A.S.T stand for?	-Computational = incorrect interpolations, transformations, analyses, precision, etc. -Output/Presentational = shoddy maps -Attribute= wrong values applied to features -Spatial/Positional = things in wrong place -Topological = incorrect spatial relationships
How could you have error in Attribute datafiles?	Incorrect Values – Written incorrectly (CAPS, typ0s) – Field types incorrect – Information completely wrong • Missing Values • Basically, attribute errors
Errors in Graphics (Vector map data)	Undershoots & dangles, sliver & weird polygons – Usually derive from bad digitising or mismatched overlays – Can be fixed with snap-, threshold-, & tolerance-setting • But this can introduce errors, too • 3D into 2D (S-road example) • Basically, spatial and topological errors
Errors in Graphics (Raster)	You have less to worry about with rasters • Basically, spatial errors • BUT! This increases importance of attribute errors..
How can you end up with error in analysis in terms of map projections?	Wrongly applied or transformed projections (often dependent on scale and place) Remember your transatlantic cable exercise from the workshops: – unless you use a projection that is specifically designed for your feature of interest, you’ll end up w/ an error (sometimes only a very small one) = just be careful what statement you make based on your input data (possibly with some small errors) and your analyses – (another example: ignoring elevation)
How can you make an error of anaysis in terms of false precision?	Over- or under-generalization (grid cell size & other examples of false precision) Frequent mistakes: – resampling a grid to generate a smaller cell size; – using GPS readings for analyses of spatial relationships at very large scales (eg. < 1:200, building footprints on 1 ha block); – Using weather data interpolated from local weather stations at local scales (e.g. the Gold Coast City local government area)
How can you make an error of analysis in terms of incorrect methodology	Incorrect methodology used during analyses: – Search distances for spatial analyses set too small or too large (e.g. for analysing point patterns -> includes too many points or not enough); – No removal of (spatial) bias due to spatially clustered observations…. • Poor interpolative or model rules – point to raster interpolations: using an algorithm that does not match the assumed or known behaviour of the attribute variable (z-value) used. • Basically, ‘algorithm-selection’ errors
How can you make an error within maps?	TOSSLAD (more about this in Lecture 7 [map design]) – or SADLOST or LASTSOD or ASSDOLT... • No metadata • Presenting erroneous (map extent, scale, etc.) or confusing information (e.g., reversed colour ramps for elevation or temperature data) • Basically, output errors (duh!)
What are some common forms of vector data?	-Shapefile = proprietary (old ArcView 3.x) format for storing vector and tabular data; not topological; editable  Geodatabase = proprietary (ArcGIS 10.x) format for storing vector, raster and tabular data; can be topological; editable
What are some common types of Vector Layers?	Cadastre (DCDB) = vector; specialised layer of land parcel/real estate information (ownership, dimensions, cost, etc.), the basis of administrative boundaries, urban planning instruments, land use layers (often used in conjunction with digital orthophotos) -Road (transport) networks = vector; specialised layer of line network data, the basis for all (car) navigation systems) -ABS census data = vector; several levels of spatial detail (e.g. suburb to State & Territories), linked to several census attribute tables (or information categories)
Common forms of Raster Data?	Grid = raster layer as we have seen before (ESRI grid, GeoTIFF, ASCII grid, CSV..
Common types of Raster Layers?	DEM (DTM) => special raster = digital elevation model; specialized grid containing elevation values (mostly as height of ground above mean -Scanned Aerial Photograph = raster; uncorrected photo taken from plane (B/W, colour, IR; low – high altitude) -Digital Orthophoto = raster; geometrically corrected, aerial photo (no problems with lens distortion, plane angle, etc.) -Satellite Image = raster; picture taken from orbital satellite
Common Types of derived layers	- DSM = as DEM raster, digital surface (NOT terrain!!) model, discrete points interpolated to fill in intermediate spaces (any surface, e.g. tree canopy height) - TIN = triangular irregular network (meshes); vector (nodes and lines) representation of a surface - geodatabase (gdb) = vector and raster, a special database management system (DBMS) for ArcGIS to enable storage of a wide range of spatial and other data, incl. multiple feature (vector) data, raster data, tables
Data within your data base will come in what two forms?	- Continuous = Values change throughout space; points in any particular area likely to contain different values; used most often with raster data (if vector, most likely with point layers) • Examples: elevation; pollution levels; rainfall - Discrete = Values change inconsistently and distinctly through space; points in any particular area more likely to have same value; used most often with vector data (if raster, used to speed spatial analyses) • Examples: habitat type; political names; speed limit
What three ways can your data be saved?	1. Numeric 2. Text 3. Date
What are the three formats are tables stored?	1. Spreadsheet or Flat File = All of your records (objects you can see on map) in one basket – similar to using Microsoft Excel to store data 2. Hierarchical (Parent-Child, One-to-Many) = Several tables that are linked by specific field “pointers” • Can think of as branching somewhat like a tree with roots • Assumes “lower-level” data present in only one “higher-level” table • Good for simple analyses, not so much for more difficult analyses 3. Relational (Many-to-One or Many-to-Many) = Records in several tables linked through commonality in data not specific pointers • Most flexible system & particularly suited for database analyses • Most popular format for GIS
What is the numeric way for your data to be saved?	1. Numeric = Numbers only (no text)  Short Integer = Whole numbers (positive or negative), typically used for coding. Used for lists such as land-use codes, vegetation types, and Booleans (i.e., true/false). Range: +/- 32,768  Long Integer = Whole numbers (positive or negative), typically used to store quantity values such as population figures. Range: +/- 2,140,000,000  Float = Single-precision floating-point numbers that can support numbers with an accuracy to 6 places past the decimal. Floats are used to store simple decimal numbers such as percentages. Range: +/- 3.41,308  Double = Double-precision floating point numbers that can support numbers with an accuracy to 15 places past the decimal. Doubles are used to store decimal numbers with a high level of detail such as latitude and longitude. Range: +/- 3.410,308
What is the text way for you to save data?	2. Text (String) = Stores any character string (names, abbreviations, alphanumeric codes, and numeric codes that begin with 0 such as postal codes). Range: 1 – 255 characters
What is the way you can save data by date?	3. Date = Stored in Coordinated Universal Time (UTC) format (= time and date at Prime Meridian = 0° +/- 7.5° longitude) and are translated into the current day and time in the local time zone. Range: 1 January 100 – 31 December 9999. • Can take on a variety of formats (and sometimes converted to Text or Numeric), so make sure to error-check if involved in analyses

Next up

Description

Resource summary

Similar

	Created by Duran Sealee over 7 years ago